Вы находитесь на странице: 1из 457


Parallax Re-visions
Cultureand Society

StephenG. Nichols,GeraldPrince,and

O I 9 9 2 , I 9 9 7 ,2 0 0 6T h e J o h n sH o p k i n s U n i v e r s i t yP r e s s
A l l r i g h t sr e s e r v e dP
. u b l i s h e d2 0 0 6
P r i n t e di n t h e U n i t e dS t a t e so f A m e r i c a
9 8 7 6 5 4 3 2 1

T h e J o h n sH o p k i n s U n i v e r s i t yP r e s s
2 7 15 N o r t h C h a r l e sS t r e e t
B a l t i m o r e ,M a r y l a n d2 12 18 - 4 3 6 3

L i b r a r yo f C o n g r e s sC a t a l o g i n g - i n - P u b l i c a t iD

Landow,Ceorge P.
H y p e r t e x t3 . 0 : c r i t i c a lt h e o r ya n d n e w m e d i a i n a n e r a o f g l o b a l i z a t i o n/ C e o r g eP .
Landow.- [3rd ed.]
p.cm.- (Parallax)
R e v .e d . o f : H y p e r t e x t2 . 0 . 1 9 9 7 .
I n c l u d e sb i b l i o g r a p h i c ar le f e r e n c e a
snd index.
I S B N 0 - 8 0 1 8 - 8 2 5 6 -(7h a r d c o v e r :a l k . p a p e r )- I S B N 0 - 8 0 . 1 8 - 8 2 5 7(-p5b k . : a l k .p a p e r )
l . C r i t i c i s m .2 . L i t e r a t u r ea n d t e c h n o l o g y3. . H y p e r t e x st y s t e m s . l . T i t l e : H y p e r t e x t
t h r e e p o i n t z e r o .l l . L a n d o w G t . 0 l l l . T i t l e .l V . P a r a l l a x( B a l t i m o r e M
, e o r g eP . H y p e r t e x 2 , d.)
P N 8 r . L 2 82 0 0 5
80] 2005007788

A catalog record for this book is availablefrom the British Library.

ForRuth-qlways 1,0

Preface:Why Hypertext 3.01 xi

Acknowledgments xv

HypertexLAn Introduction
Hypertextual Derrida, PoststructuralistNelsonl t I The Definition of
Hypertext and Its History as a Concept 2 I Yery AcIwe Readers 6 /
VannevarBush and the Memex 9 / Forms of Linking, Their Uses
and Limitations 13 I Linking in Open Hypermedia Systems:Vannevar
BushWalkstheWeb 22 I Hypertextwithoutlinksl 27 / ThePlace
of Hypertextinthe Historyof InformationTechnoloCy29 | Interactive
or Ergodic) 4t I BaudrlTlard,Binarity, and the Digital 43 / BooksAre
Technology,Too 46 / Analoguesto the Gutenberg Revolution 49

Hypertextand CriticalTheory
Tex|ual Openness 53 / Hypertext and Intertextuality 55 / Hyperte>ct
and Multivocality 56 / Hypertext and Decentering 56 I Hypertext as
Rhizome 58 / The Nonlinear Model of the Network in Current Critical
Theory 62 | Causeor Convergence,Influence or Confluencel 65

the Text
From Text to Hlpertext 69 I The In MemoriamWeb 71 / New Forms
of Discursive Prose-Academic Writing and Weblogs 77 I Problems
with Terminology: What Is the Object We Read,and What Is a Text in
Hypertextl 82 / Visual Elements in Print Text 85 / Animated Text 89 /
Stretchtext93 / TheDispersedText 98 / HypertextualTranslationof

CONTENTS Scribal Culture 99 I AThird Convergence:Hypertext and Theories of

ScholarlyEditing L02 I Hypertext, ScholarlyAnnotation, andthe
ElectronicScholarlyEdition 103 / Hypertextandthe ProblemofText
Structure L07 | kgumentation, Organization, and Rhetoric 109 I
Beginnings in the Open Text 110 / Endings in the Open Text Il2 I
Boundariesof the Open Tert 113 I The Statusof the Text, Statusin the
Text 118 / H)?ertext and Decentrality:The PhilosophicalGrounding 123

the Author
Erosion of the Self I25 I How the Print Author Differs from the
HypertextAuthor 131 / VirrualPresence135 / CollaborativeWriting,
CollaborativeAuthorship 136 / ExamplesofCollaborationin Hypertext 142

The ProblematicConceptof Disorientation t44 / The Conceptof
Disorientationinthe Humanities 146 I The Loveof Possibilities148 /
The Rhetoric and Stylisticsof Writing for E-Space;or, How Should We
Write H)?ertex'tl 151 / Hlpertext as CollageWriting 188 / Is This
HypertertAny Goodl Or, How Do We EvaluateQuality in Hypermedial 198

guring Narrative
Approachesto Hypertext Fiction-Some Opening Remarks 215 /
Hypertext and the Aristotelian Conception of Plot 218 / Quasi-
Hypertextuality in Print Texts 219 / Answering Aristotle: Hlpertext
and the Nonlinear Plot 221 / Print Anticipations of Multilinear
Narrativesin E-Space223 | NananveBeginningsand Endings 226 I
Michael Joyce'safiemoon 229 / Stitching TogetherNarrative, Sexuality,
Self: Shelleyfackson'sPatchworkGirl 234 I Quibbling A Feminist
Rhizome Narrative 242 I Storyworldsand Other Forms of Hypertext
Narratives 245 I Computer Games,Hlpertext, and Narrative 250 /
Digitizing the Movies: Interactiveversus Multiplied Cinema 254 I
Is HypertextFiction Possiblel 264

Reconfiguri ng LiteraryEducation
Threats and Promises 272 | Reconfiguring the Insrructor 275 |
Reconfiguring the Student 278 I Learning the Culture of a
Discipline 280 / Nontraditional Students:Distant Learnersand Readers
outside EducationalInstitutions 28t I The Effectsof Hypermedia in
Teachingand Learning 284 / Reconfiguring Assignments and Methods

CONTENTS of Evaluation 286 I AHypertextExercise 287 I ReconceivingCanonand

Curriculum 292 I Creatingthe New DiscursiveWriting 302 I Frcm
Intermedia to the Web-Losses and Gains 309 / Answered Prayers,or the
Academic Politics of Resistance 312 I WhaI ChanceHas Hlpertext in
Educationl 313 / Getting the Paradigm Right 314

The Politicsof Hypertexl Who Controlsthe Textl

Can Hypertext Empower Anyonel Does Hypertext Have a Political
Logicl 321 / The Marginalization of Technologyand the Mystification
ofliterature 330 / The Politicsof ParticularTechnologies335 /
Technologyas Prosthesis 336 I The Political Vision of Hypertext;
or, the Messagein the Medium 343 | Hpertext and Postcolonial
Literature,Criticism, and Theory 345 I Infotech, Empires, and
Decolonization 347 I Hypertext as Paradigmfor Postcoloniality 351 /
Forms of PostcolonialAmnesia 354 / Hypertext as Paradigmin
PostcolonialTheory 356 I The Politics of Access:Who Can Make Links,
Who DecidesWhat Is Linked) 358 I Slashdot:TheReaderas Writer and
Editor in a Multiuser WebIog 362 / Pornography,Gambling, and Law on
the Internet-Vulnerability and Invulnerability in E-Space364 / Access
to the Text and the Author's Right lCopyrighr) 367 / Is the Hypertextual
World of the Internet Anarchy or Big Brother'sRealml 376

Notes 377
Bibliography 399
lndex 425

Why Hypertext 3.0? When I wroreHypertert2.0 inl997,rheneed was

obvious: developments in hardware and software since the appearanceof
the first version led me to remove most referencesto Intermedia, replacing
them with discussionsof the World Wide Weband other hlpermedia systems
(Storyspace,Microcosm,CD-ROMproprietaryenvironments).In addition,I
addeda chapter on writing for e-space,included examplesfrom new hyper-
tefi fiction, and so on. Since the appearanceof Hypertext2.0, severaldevel-
opments have occurred that again led to the need for a new version. These
changesinclude (1) the enormous growth of the Web and its use in literary
business, and political applications; (2) the development of Weblogs, or
blogs, as a widely availableform of read-write hypertext-the first widely
availableWeb mode that begins to approachthe vision of the first hypertext
theorists; (3) the rapid growth of interest in animated text, using Flash,now
that enough Web users havebroadbandaccessto make such large files prac-
ticable in Internet applications;(4) the increasing importance of our under-
standing of postcolonialityand globalization:and (5) some first stepstoward
a theory of digital cinema (Hypertext2.0briefly discussedthis topic with em-
phasis on examplesrather than on their theoreticalimplications.) To the ear-
lier discussionsofthe convergenceofhypertext, critical theory and editorial
theory, I also propose to consider two additional possible points of conver-
gence,postcolonialculture and interactivecinema.
Perhapsmost important, given the optimistic and even celebratorytone
taken by most writers on hypermedia, has been the notorious dotcom bust,
which Vincent Moscohas so effectivelydescribedin terms of its relations
to cyberspaceas a cultural myth. In The Digital Sublime:Myth, Power,and

PREFACE (2004),he explains "the extraordinaryboom-and-bustcycle" (6) of

the 1990sby placing it in the context of mythic notions of cyberspaceprom-
ulgated by those like Nicholas Negroponteand other proponents of "techno-
mania" (21).Accordingto Mosco,cyberspace functionedas one of thosecul-
tural myths that provide "stories that animate individuals and societiesby
providing paths to transcendencethat lift peopleout ofthe banality ofevery-
day life. They offer another reality, a reality once characterizedbythe prom-
ise of the sublime" (3).

Convinced bythedemiseof theColdWarandthemagicof a newtechnology,people

acceptedtheviewthathistoryasweonceknewit wasendingandthat,alongwiththe
endof politicsasweknewit,therewouldbeanendto thelawspromulgated bythat
mostdismalof sciences,
Constraints onceimposedby scarcitiesof re-
wouldend,or at leastloosensignificantly,
anda neweco-
nomicsof cyberspace
economics") wouldmakeit easier for societies
growand,especially, to growrich. . . Whatmadethedotcombooma mythwasnot
thatit wasfalsebutthatit wasalive,sustained
a newworldbytranscending

These "m1ths," Mosco argues,point "to an intense longing for a prom-

ised community, a public democtacy"(15).Like all myths, they make "socially
and intellectually tolerablewhat would otherwise be experiencedas incoher-
ence" (29) and otherwise shield people from political and economic realities
(31)becausethey "mask the continuities that make the power we observeto-
day,for examplein the global market and in globe-spanningcompanieslike
Microsoft and IBMI' Cyberspacemyths, which purport to lead us to a golden
future in which geographyand history end, create"amnesia about old poli-
tics and older myths" (83).Mosco'ssolution involvesvaluably reminding us
that during the past two centuries almost every major developmentof tech-
nology-electricity, telegraph,telephone, radio, television, cable television,
and so on-brought with it similar mythic claims.
One of the few weaknessesin his convincing, if limited, analysislies in
the fact that it so emphasizesmyth as a socialconstruction born from a com-
munity's need that it never inquires if any of these myths about cyberspace
proved to have roots in fact. Thus, although he severaltimes assuresthe
readerthat "myths are not true or false,but living or dead" (3),in practicehe
alwaysacts as if all statementsabout cyberspaceare false. Unlike William f .
Mitchell in Me++: The CyborgSelfandthe NetworkedCity (2003),Mosconever
inquires if someof the claims about location-independentwork, business

PREFACE applicationsof the Internet, or hypermedia in education proved correct.

After all, a great many computer-relatedenterprises-educational, artistic,
and commercial-continue to thrive.
In fact, since I wrote the first version of Hypeflert the situation of com-
puting in humanities, arts, and culture has changed dramatically.When I
first tried to explain the nature and possibilities of hypermedia, most of my
readershad little contactwith computing, but that had changedby Hypertext
2.0.The situation has now changed dramaticalTyonce more, and a book like
this one now finds itself situatedvery differently within our culture, particu-
larly the humanities, than was the caseonly a short time ago. For example,
when I first explained the characteristicsof a document within a hypertext
environment, contrasting it to a page of print, I had to describeand explain
three things: (L)how one used a computer-even how one used a mouse and
drop-downmenus; (2)the basiceffectsofdigital information technology;and
(3) the characteristicqualities and experienceofhypertext itself. Such is no
longer necessary,and such is no longer adequate.It is not simply a matter that
many of you have become skillfirl users of e-mail, discussion lists, Google,
and the World Wide Web. Equally important, you have experiencednumer-
ous digital applications,genres,and media that do not take the specificform
of hypertext. Some of these, such as Weblogs,show a important relation to
hypermedia,but others,like computer games,haveonly a few points of con-
and cultural
vergencewith it. Still othersof increasingeconomic,educat.ional,
importance, such as animated text, text presented in PDF (portable docu-
ment format) format, and streaming sound and video, go in very different
directions, often producing effectsthat fundamentall y differ fromhypermedia.
Let me emphasizehere that I do not proposeto evaluatenonhypertextual
developmentsof digital information technology according to the degreeto
which they resemblehypertext and hypermedia. I am also not interested in
presenting hypermedia as an overarchingumbrella conceptunder which to
gather all other digital forms. I shall, however,comparethese other kinds of
digitality to hlpermedia on the assumption that doing so will help us better
understand characteristiceffectsand applicationsof all these new media.
The situation-in particular, the academicstanding and fashionability-
of poststructuralism has also changed markedly since the first version of
Hypefiert, though in a way perhapsoppositeto that of hlpermedia. Whereas
hypertext and other forms of digital media have experienced enormous
growth, poststructuralism and other forms of critical theory have lost their
centrality for almost everyone,it seems,but theorists of new media. One
might claim to seea parallel betweenthe dotcom bust and the generalloss of

PREFACE academicstanding by critical theory,but websites,blogs, discussionlists,

and new media arts flourish despite the bankruptcy of many ill-conceived
computer-relatedbusinesses,some of which never managedto produceany-
thing more than vaporware.
I don t believethis changein situation lessensthe valueof one of the main
approachesof this book, its use of hypertext and late-twentieth-centurycriti-
cal theory to illuminate each other. As I stated repeatedlyin the earlier ver-
sions of this book, the writings of Roland Barthes,facquesDerrida, and other
critical theoristsneither causedthe developmentof hypermedianor coincided
exactlywith it. Nonetheless,their approachto texhrality remains very helpful
in understanding our experienceof hypermedia. And vice versa.I havehad
many students in my hypertext and literary theory classwho have told me
that they found the writings of Barthes,Derrida, Michel Foucault,and Gilles
Deleuzeand F6lix Guattari easierto understand after the experienceofread-
ing and writing hypertexts.Others haveagreedthat these theorists, particu-
larly Derrida and Barthes,provide useful ways to think about hypertext.
Perhapsthe single most important developmentin the world of hyper-
media has been the steadydevelopmentof read-write systems-of the kind
of systems,in other words, that the pioneering theorists VannevarBush and
Theodor H. Nelson envisioned.Blogs,wikis, and the Portal Maximizer by Ac-
tive Navigationall representattempts to bring to the Web the featuresfound
in hypertext softwareof the 1980sthat made readersinto authors.

Becausemy first acquaintancewith the idea of hypertext goesback to 1986or

1987,when members of Brown University'slong-vanishedInstitute for Re-
searchin Information and Scholarship(IRIS) recruited me to ioin the Inter-
media project,I owe specialthanks to its founding director,William G. Shipp,
to its later co-directors,Norman K. Meyrowitz and Marry f . Michel, andto my
friend and colleaguePaul Kahn, who was proiect coordinator during the cre-
ationof TheDickensWebandlater Intermedia projects andwho servedasthe
institute's final director. Nicole Yankelovich,IRIS project coordinator during
the initial developmentand application stagesof Intermedia, alwaysproved
enormously resourcefirl,helpful, and good humored even in periods of cri-
sis, as did fulie Launhardt, assistantproject coordinator.In the final yearsof
the project, the late fames H. Coombs,who createdmany of the key parts of
the secondstageof Intermedia, provided invaluableassistance.
fay Bolter enticed me into using Storyspace,and I am most grateful to
him, Michael foyce,and Mark Bernstein of EastgateSystemsfor their con-
tinuing assistance.
I owe an especialdebt to my enthusiastic and talented graduateand un-
dergraduateresearchassistantsbetween t987 and 1992,particularly Randall
Bass, David C. Cody, ShoshanaM. Landow Jan Lanestedt,Ho Lin, David
Stevenson,Kathryn Stockton,GaryWeissman,GeneYu, andMarcZbyszyn-
ski. My studentsat Brown University,the University ScholarsProgram at the
National University of Singapore(NUS), and the Faculty of Computer Sci-
enceat NUS haveprovided a continual sourceof inspiration and delight.
The development of Intermedia was funded in part by grants and con-
tracts from International BusinessMachines, Apple Computer, and the

Annenberg/Corporation for Public BroadcastingProject,and I am grateful to
them for their support. A Mellon Foundation grant and one from Dr. Frank
Rothman, the provost of Brown University,enabledme to transfer the Inter-
media materials createdfor English and creativewriting coursesinto Story-
space.The generosityof Daniel Russell,then of Apple Computers, made it
much easierfor me to carry out my researchin the 1990safter the closing
of IRIS, when my university found itself able to offer little assistanceor
Since2000NUS has funded the web serversin New york and Southeasr
Asia on which residethe most recent descendantsof materialsoriginally cre-
atedin Intermedia and storyspace-The victorian, postcolonialLiteratureand.
Culture, and Cyberspace, Hypertext,and Critical Theorysites-and in 2O0L-2
NUS funded postdoctoralfellows and senior researchfellows, who created
materials for the sites, including Philip V. Allingham, Marjorie Bloy,Leong
Yew,TamaraS.Wagner,and Johnvan Whye. I alsohaveto thank the hundreds
of international contributors, particularly Philip v. Allingham, contributing
editor of Ihe Victoian Web,who have shared so many thousands of docu-
ments with readersof thesesites.I would like to thank peyton Skipwith of the
Fine Art Society,London, and PeterNahum for generouslygranting permis-
sion to include the imagesand text from their catalogues,thus permitting me
to createthe victoian web'ssectionson painting and the decorativearts. I am
especiallygrateful to the authors oftwo dozen out-of-print scholarly books
and contributors of many other victorian texts who have generouslyshared
their work with victoian web, thus making possiblethe victoiqn web Books
sectionthat exploreswhat is happening to the forms of humanistic scholar-
ship in a digital age.Thanks, too, to the readersof my websiteswho were re-
sponsiblefor their receiving 17 million hits/page views in March2002 (95%o
of them for the Victorian Web).Noysius Tay Wee Kok, head of information
technology at the University Scholarsprogram, and his crew of technicians
have set up and maintained the serversin both the United Statesand Singa-
pore with the assistanceof fosephAulisi of Macktez.com.
I alsoowe a debt of gratitudeto many colleaguesand studentswho shared
their work with me: Mark Amerika, f . David Bolter, Alberto Cecchi, Robert
Coover,Daniela Danielle, Cicero da Silva,fay Dillemuth, Carolyn Guyer,Ter-
ence Harpold, Paul Kahn, Robert Kendall, David Kolb, Deena Larson, Gary
Marchionini, Stuart Moulthrop, and Marc Nanard kindly provided me with
draft, prepublication, or prereleaseversions of their work; and Cambridge
UniversityPress,Dynamic Diagrams,EastgateSystems,MetaDesignWest,

PWS Publishing, Oxford University Press,Routledge,and Voyagerhavepro-
vided published versionsof their electronicpublications.
I would alsolike to thank for their advice,assistance,and encouragement
Irina Aristarkhova, David Balcom, Bruno Bassi,Gui Bonsiepe,GeorgeBorn-
stein, Katell Briatte,LeslieCarr, LauraBorrdsCastanyer,Hugh Davis,Marilyn
Deegan,Emanueladel Monaco, facquesDerrida, Umberto Eco,Markku Es-
kelinen, Susan Farrell, Niels Ole Finnemann, Patrizia Ghislandi, Antoni f .
Gomez-Bosquet,Diane Greco, Robert Grudin, Anna Gunder, Wendy Hall,
E. W: B. Hess-Lithich,ElaineYeeLin Ho, RaineKoskimaa,Iean-LouisLebrave,
fos6 Lebrero, Michael Ledgerwood,Gunnar Liestol, Peter Lunenfeld, Cathy
Marshall, Graham McCulloch, Bernard Mcguirk, Tom Meyer,f . Hillis Miller,
Andrew Morrrison, Elli Mylonas, Palrizzia Nerozzi, Geoffrey Nunberg,
Sutayut Osornprasop,Allesandro Pamini, Paolo Petta, Allen Renear,Mas-
simo Riva, Peter Robinson, Lothar Roisteck,Luisella Romeo, fames Rosen-
heim, Daniel Russell,Marco Santoro,Valentina Sestini,Ture Schwebs,Shih
Choon Fong, RosemaryMichelle Simpson, Christine Tamblyn, feffTaylor,
Robert Trappl, Paul Tucker, Frank Turner, Gregory Ulmer, Andy van Dam,
Karin Wenz, Rob Wittig, and the members of CHUG.
Among the many studentsand others who havesharedtheir hypermedia
projects with me since the late 1980sI have to thank Mark Amerika, Diego
Bonilla, Don Bosco, Sarah Eron, Ian Flitman, Nicholas Friesner, Amanda
Ian M. Lyons,Abigail New-
Griscom,leremy Hight, Taro Ikai, ShelleyJackson,
man, Nitin Sawhney,David Balcom,JeffPack,Ian Smith, Owen Strain, Noah
Wardrip-Fruin, DavidYun, and Leni Zumas,
When I presentedthe idea for the first version of this work to the fohns
Hopkins University Press, Eric Halpern, then editor in chie{, was open-
minded enough to haveenthusiasm for a project that editors at o*rer presses
thought too strangeor too unintelligible to consider.I greatly appreciatethe
encouragementI receivedfrom him and the support for the secondversion
by Douglas Armato and Willis Regier,then director of the Press.Michael
Lonegro, my editor for 3.0, has added to my experienceof fohns Hopkins
University Press assistancewith his valuable encouragement and sugges-
tions. Jim fohnston, design and production manager when the first version
was produced,and Glen Burris, the book'sdesigner,deservethanks for tack-
ling something new in a new way.Thanks, too, to Maria denBoer,who copy-
edited this version, for contributing much to whatevergrace,clarity, and ac-
curacythis book may possess.
Finally, I would like to thank my children, Shoshanaand Noah,who have

A C K N O W L E D G M E N T Slistenedfor yearsto my effusionsabout links, webs,lexias,web views,and

local tracking maps. Noah's technical erpertise about information architec-
ture, blogging, and countlessarcanedetails of hardware and softwaremade
many of my projects possible,and he keepsintroducing me to new areasof
digital culture. My most important debt, of course, is to my wife, Ruth, to
whom this book is dedicated.It was she who coined the titles Hypefiert 2.0
and 3.0 and who taught me everything I know about Internet shopping. In
the courseof encouragingmy explorationsof hypermedia,shehas becomea
true member of the digerati-someone who has worn offthe characterson
severalkeyboardswhile editing arnagazineon the other side of the world via
the Internet and who sendsme a stream of e-mail even when we are in the
same room. Of all the debts I haveincurred while writing this book, I enjoy
most acknowledgingthe one to her.
An Introduction

When designersof computer software examine the pagesof

HypertextualDerrida, GIasor Of Grammatology, they encounter a digitalized, hyper-
textual Derrida; and when literary theorists examine Literary
Machines,they encounter a deconstructionist or poststruc-
turalist Nelson. Theseshocksof recognition can occur becauseover the past
severaldecadesliterary theory and computer hypertext, apparently uncon-
nectedareasof inquiry haveincreasinglyconverged.Statementsby theorists
concernedwith literature, like those by theorists concernedwith computing,
show a remarkableconvergence.Working often, but not always,in ignorance
of eachother,writers in theseareasoffer evidencethat providesus with a way
into the contemporary epistemein the midst of maior changes.A paradigm
shift, I suggest,has begun to take place in the writings of facques Derrida
and Theodor Nelson, Roland Barthes and Andries van Dam. I expect that
one name in eachpair will be unknown to most of my readers.Thosework-
ing in computing will know well the ideas of Nelson and van Dam; those
working in literary and cultural theory will know equally well the ideas of
Derrida and Barthes.l
All four, like many others who write on hlpertext and literary theory ar-
gue that we must abandon conceptual systemsfounded on ideas of center,
margin, hierarchy,and linearity and replacethem by ones of multilinearity,
nodes,links, and networks. Almost all parties to this paradigm shift, which
marks a revolution in human thought, seeelectronic writing as a direct re-
sponseto the strengths and weaknessesof the printed book, one of the ma-
jor landmarks in the history of human thought. This responsehas profound
implications for literature, education,and politics.

3.0 The many parallelsbetweencomputer hypertext and critical theory have
many points of interest,the most important of which, perhaps,lies in the fact
that critical theory promises to theorize hypertext and hypertext promises to
embody and thereby test aspectsof theory,particular\ those concerning tex-
tuality, narrative, and the roles or functions of reader and writer. Using hy-
pertext, digital textuality, and the Internet, students of critical theory now
havea laboratorywith which to test its ideas.'zMost important, perhaps,an ex-
perienceof readinghypertextor readingwith hypertextgreatlyclarifiesmany
of the most significant ideasof critical theory.As f . David Bolter points out in
the courseof explaining that hypertextualityembodiespoststructuralistcon-
ceptions of the open text, "what is unnatural in print becomesnatural in the
electronic medium and will soon no longer need sayingat all, becauseit can
be shown" (Witing Space,143).

In S/2, Roland Barthesdescribesan ideal textuality that pre-

The Definitionof Hypertextand cisely matchesthat which has come to be called computerhy-
perturt-textcomposed of blocks of words (or images)linked
Its Historyas a Concept
electronicallyby multiple paths, chains, or trails in an open-
ended, perpetually unfinished texruality described by the terms link, node,
network,web,andpath."In this idealtext,"saysBarthes,"thenetworkslrdseaux]
are many and interact,without any one of them being ableto surpassthe rest;
this text is a galaxyof signifiers, not a structure of signifieds; it has no begin-
ning; it is reversible;we gain accessto it by severalentrances,none ofwhich
can be authoritativelydeclaredto be the main one; the codesit mobilizes
extend asfar as the eyecan reach,they are indeterminable . . . ; the systems
of meaning can take overthis absolutelyplural text, but their number is never
closed,basedas it is on the infinity of language"(5-6 [Englishtranslation];
Like Barthes,Michel Foucaultconceivesof text in terms of network and
links. In TheArchaeologyof Knowledge, he points out that the "frontiers of a
book are never clear-cut,"because"it is caught up in a system ofreferences
to other books, other texts, other sentences:it is a node within a network . . .
[a]network ofreferences"(23).
Like almost all structuralistsand poststructuralists,Barthesand Foucault
describetext, the world of letters, and the power and statusrelations they in-
volve in terms shared by the field of computer hypertext. Hypefiert, a term
coined by Theodor H. Nelson in the 1960s,refers alsoto a form of electronic
text, a radically new information technology,and a mode of publication.3"By
'hypertext,"' Nelson explains, "I mean non-sequential writing-text that

AN lNrRoDUcrloN branchesand allows choicesto the reader.best read at an interactivescreen.

As popularly conceived,this is a series of text chunks connectedby links
which offer the readerdifferent pathways"(LiteraryMachines,}l2\. Hypertert,
as the term is used in this work, denotestext composedof blocks of text-
what Barthesterms a lexia-andthe electroniclinks that join them.a Hyper-
media simply extendsthe notion of the text in hypertext by including visual
information, sound, animation, and other forms of data.sSince hypertext,
which links one passageof verbal discourseto images,maps, diagrams, and
sound as easily as to another verbal passage,expands the notion oftext be-
yond the solely verbal, I do not distinguish between hypertext and hyper-
media. Hyperturt denotesan information medium that links verbal and non-
verbal information. In this network, I shall use the terms hypermed.iaand
hypertextinterchangeably. Electronic links connect lexias "external" to a
work-say, commentary on it by another author or parallel or contrasting
texts-as well as within it and thereby createtext that is experiencedas non-
linear, or, more properly, as multilinear or multisequential. Although con-
ventional readinghabits applywithin eachlexia, onceone leavesthe shadowy
bounds of any text unit, new rules and new experienceapply.
The standardscholarlyarticle in the humanities or physical sciencesper-
fectly embodies the underlying notions of hypertext as multisequentially
readtext. For example,in reading an article on, say,|ames ]oyce'sUlysses,
readsthrough what is conventionallyknown as the main text, encounters a
number or symbol that indicatesthe presenceof a footnote or endnote, and
leavesthe main text to readthat note, which can contain a citation of passages
in Ulysses
that supposedlysupport the argument in question or information
about the scholarly author's indebtednessto other authors, disagreement
with them, and so on. The note can also summon up information about
sources, influences, and parallels in other Titerarytexts. In each case, the
readercan follow the link to another text indicatedby the note and thus move
entirely outside the scholarly article itself. Having completed reading the
note or having decidedthat it does not warrant a careflrl reading at the mo-
ment, one returns to the main text and continues reading until one encoun-
ters another note, at which point one again leavesthe main text.
This kind of reading constitutesthe basic experienceand starting point
of hypertext. Supposenow that one could simply touch the page where the
symbol ofa note, reference,or annotation appeared,and thus instantly bring
into view the material contained in a note or eventhe entire other text-here
al7of Uysses-to which that note refers. Scholarlyarticles situatethemselves
within a field of relations,most of which the print medium keepsout of sight

T. 0 and relatively difficult to follow becausein print technology the referenced
(or linked) materials lie spatially distant from the referencesto them. Elec-
tronic hypertext, in contrast,makes individual referenceseasyto follow and
the entire field ofinterconnections obvious and easyto navigate.Changing
the easewith which one can orient oneselfwithin such a contextand pursue
individual referencesradically changesboth the experienceof reading and
ultimately the nature of that which is read. For example,if one possesseda
hypertextsystemin which our putative foycearticlewaslinked to all the other
materials it cited, it would exist as part of a much larger systemin which the
totality might count more than the individual document; the article would
now be woven more tightly into its contextthan would a printed counterpart.
As this scenariosuggests,hypertextblurs the boundariesbetweenreader
and writer and therefore instantiates another quality of Barthes'sideal text.
From the vantage point of the current changesin information technology,
Barthes'sdistinction betweenreaderlyand writerly texts appearsto be essen-
tially a distinction between text basedon print technologyand electronic
hypertext,for hlpertext fulfills

thegoalof literary
work(ofliteratureaswork)[which] isto makethereaderno longer
a consumer, buta producer of thetext.Our literature bythepitiless
is characterized
institution between
maintains of thetextand
its user,between between This
itsauthorandits reader.
readeris thereby he is,in short,
intoa kindof idleness-heis intransitive;
serious: of functioning
instead himself,
insteadof gaining to themagicof the
to thepleasure
signifier, of writing,he is leftwithno morethanthepoorfreedom
thertoacceptor rejectthetext:
reading is nothingmorethan a referendum.
its negative,
thewriterlytext,then,is its countervalue, reactivevalue:whatcanbe
read,butnotwritten:thereaderly. Wecallanyreaderlytext (S/2,4)
a classictext.

Comparethe way the designersof Intermedia, one of the most advanced

hypertext systemsthus far developed,describethe activereader that hyper-
text requiresand creates:

Bothanauthorttoolanda reader'smedium,a hypertext

thorsor groups together,
to linkinformation
ofauthors pathsthrough
create a corpus
ofrelatedmaterial, existingtexts, to either
text. . . Readers
dataor thebodyofthereferenced
bibliographic canbrowsethrough
cross-referenced, textsin anorderly
annotated butnonsequential (17)6

To get an idea of how hypertextproducesBarthes'swriterly text, let us ex-

amine how the print version and the hlpertext version of this book would

AN lNrRoDUcrloN differ. In the first place,insteadof encountering it in a paper copy,you would

read it on a computer screen (or alreadyhave if you've read the ]ohns Hop-
kins translation of the first version into hlpertext). ln L997, computer
screens,which had neither the portability nor the tactility of printed books,
made the act of reading somewhatmore difficult than did the print version.
For those peoplelike myself who do a large portion of their reading reclining
on a bed or couch, screenson desktop machines are markedly less conven-
ient. For the past four years,however,I haveworked with a seriesof laptops
whose displaysdo not flicker and whose portability permits enjoyableread-
ing in multiple locations. Of course,myApple G4laptop still doesn't endow
the documents read on it with the pleasurabletactility of the printed book,
but since my wife and I use wireless accessto the Internet, we can both read
Internet materials an;'where in the house or sitting outside in a recliner on
the porch. Although I used to agreewith peoplewho told me that one could
never read large amounts of text online, I now find that with these new dis-
plays I prefer to read the scholarlyliterature on my laptop; taking notes and
copying passagesis certainly more convenient.Nonetheless,back in the late
1980s,reading on Intermedia, the hypertextsystemwith which I first worked,
offered certain important compensationsfor its inconveniences.T
Reading an Intermedia, Storyspace,or World Wide Web version of this
book, for example,you could changethe size and even style of font to make
reading easier.Althou gh you could not make such changespermanently in
the text as seenby others, you could make them wheneveryou wished. More
important, sinceon Intermedia you would readthis hypertextbook on a large
two-pagegraphics monitor, you would have the opportunity to place several
texts next to one another.Thus, upon reachingthe first note in the main text,
which follows the passagequoted from S/Z,youwould activatethe hypertext
equivalent of a referencemark (glyph, button, link marker), and this action
would bring the endnote into view.A hypertextversion of a note differs from
that in a printed book in severalways. First, it links directly to the reference
symbol and doesnot reside in some sequentiallynumbered list at the rear of
the main text. Second,once opened and either superimposed on the main
text or placedalongsideit, it appearsas an independent,ifconnected, docu-
ment in its own right and not as some sort of subsidiary,supporting, possi-
bly parasitic text.
Although I have since converted endnotes containing bibliographic
infotmation to in-text citations, the first version of Hypertert had a note
containing the following information: "Roland Barthes, S/2, Irans. Richard
Miller (New York: Hill and Wang, L974),5-6: A hypertextlexia equivalentto

3.0 this note could include this sameinformation, or, more likely,takethe form of
the quoted passage,a longer sectionor chapter,or the entire text of Barthes's
work. Furthermore, in the various hypertext versions of this book, that pas-
sagein turn links to other statements by Barthes of similar import, com-
ments by studentsof Barthes,and passagesby Derrida and Foucaultthat also
concern this notion of the networked text. As a reader, you must decide
whether to return to my argument, pursue some of the connectionsI suggest
bylinks, or, using other capacitiesof the system,searchfor connectionsI
have not suggested.Readingon the World Wide Web producesthis kind of
reading experience.The multiplicity of hypertext,which appearsin multiple
links to individual blocks oftext, calls for an activereader.
A full hypertext system,unlike a book and unlike some of the first ap-
proximationsof hypertextavailable-HyperCardt', Guide", and the current
World Wide Web (exceptfor blogs)-offers the reader and writer the same
environment. Therefore,by opening the text-processingprogram, or editor,
as it is known, you can take notes, or you can write against my interpreta-
tions, againstmy text. Although you cannot changemy text, you can write a
responseand then link it to my document. You thus have read the readerly
text in severalways not possiblewith a book you have chosenyour reading
path, and since you, tike all readers,will chooseindividualized paths, the
hypertefi version of this book would probablytake a very different form, per-
haps suggesting the values of alternate routes and probably devoting less
room in the main text to quoted passages.You might havealsohavebegun to
take notes or produce responsesto the text asyou read,some of which might
take the form of texts that either support or contradict interpretations pro-
posedin my texts.

When one considersthe history of both ancientliterature and

VeryActive Readers hardly
recentpopular culture,the figure ofthe reader-as-writer
appears at all strange, particularly since classical and neo-
classicalcultural theory urged neophyte authors to learn their craft by read-
ing the mastersand then consciouslytrying to wdte like them. Anyone who's
taken an undergraduatesurvey coursewill know that Vergil seliconsciously
read and rewrote Homer, and that Dante read and rewrote both Homer and
Vergil, and Milton continued the practice. Such very active readers appear
throughout the past two centuries. To an important extent,Jane Eyretepre-
just as North and Southand
sentsa very activereading of Prideand Prejud.ice.
Aurora Leigh represent similar readings and rewritings of the two earlier
texts. In fact, all four works could have been entitled "Pride and Prejudice,"

AN INTRODUCTION and all four presentwomen of a supposedlylower socialand economic class

disciplining their men; in Victorian versionsof this plot the man not only has
to apologizefor his shortcomings but he also has to experiencemajor pun-
ishment-bankruptcy, severeinjury blindness, or a combination of them.
Literary scholarsare quite accustomedto chains of activereadings that
produce such rewritings. We call it a tradition. We also, following Harold
Bloom, call it the anxietyof influence,the later author challenging the earlier
one. Readersof , say,Aurora Leighrecognizethat ElizabethBarrett Browning's
novel-poem simultaneously assertsthe existenceof a female literary tradi-
tion while also challenging its creators,the poet'spredecessors,for pride of
placewithin that tradition.
Such aggressivelyactive reading has proved particularly popular with
postcolonial and postimperial authors. Thus Jean Rhys'sWide SargassoSea
offers a very different, Caribbeanreadingof JaneEyre,teTlingthestory almost
entirely from Bertha'spoint of view. We encounter the empire again writing
back in Peter Carey'sJack Mags, a novel told from the vantagepoint of the
Magwitch character;in this version, which includes a Dickens-like novelist,
the illegally returned convict doesnot die with Pip at his side: realizing what
a dreadful person the Pip-characterhas turned out to be, Maggs returns to
wealth, fatherhood,and fame in Australia. Takethat, Dickens!
Given the history of high culture, one is not surprised to encounterthese
activereadings and rewritings, but such approachesalso appearin so-called
genre fiction, such as detectivestories and sciencefiction. In JapserFforde's
The EyreAfair, for example,we learn how Brontd'snovel receivedits happy
ending. Al1 the examplesof such very activereading thus far belong to the
upper reachesofthe culture industry: major commercial firms publish them,
they win prestigiousprizes, and they quickly earn canonicity by being taught
in universities.There are, however,large numbers of very activereaderswho
receivelittle notice from the publishing and academicestablishments.The
wide availability of low-cost information technologies-first mimeographs
and photo-offsetprinting and later desktoppublishing and finally the World
Wide Web-permitted the creation of self,published rewritings of popular
entertainment, such as Star Trek,that first appearedin books,television,and
cinema.Active readingsof the popular sciencefiction series,ConstancePen-
ley explains,haveexistedsincethe mid-1970s.

Mostof thewritersandreaders
startedoffin "regular"StarTrekfandom,
in it, evenwhiletheypursue
in whatis called
"K/S"or "slash"fandom.Theslashbetween K(irk)and5(pock)serves asa codeto

3.0 t h o s ep u r c h a s i n
bgym a i la m a t e u r f a n z i n( oers" z i n e s "t )h a t t h es t o r i e sP,o e m sa,n d
artwork published thereconcern a same-sex relationship between thetwomen.Such
a designation standsin contrast to "ST,"forexample, withno slash, whichstands for
actionadventure stories
based or "adultST,"which
on IheStarTrekfictional
refersto storiescontaining onesonly,saybetween
but heterosexual
Captain KirkandLieutenant Uhura,or SpockandNurseChapel.Othermediamale
coupfeshavebeen"slashed" andHutch(S/H). . . or Miami
in thezines,likeStarsky
Vica's andTubbs.(i 37)

According to Penley,women produce most of thesesamizdattexts,and these

readers-as-writerstake "pride in having createdboth a unique, hybridized
genre that ingeniously blends romance, pornography,and utopian science
fiction and a comfortable social spacein which women can manipulate the
products of mass-producedculture to stagea popular debatearound the is-
suesof technology,fantasy,and everydaylife" (137).As one might expect,the
development of the V/orld Wide Web has stimulated this active reading even
more, and one can find all kinds of works by readerswho want towtite their
versionsof rnaterialscommercially published. The presenceand productions
of very active readers answer the critics of digital information technology
who claim it cannot demonstrateany examplesof cultural democratization.
whether one actuallylikes this or other kinds of cultural democratizationis
another matter.
Very activereaders(or readers-as-writers)havetendedto go unnoticed for
severalreasons.First, although some of thesefanzines mayhave circulations
aslarge as first novelspublished by prestigious publishers,they representan
underground culture of which mass media and educationalinstitutions re-
main unaware. Another reason why the continuations and rewritings they
produce receivelittle attention derivesfrom some of the obviousqualities
of print culture: like Carey'svery activereading of Dickens'snovel, theseun-
derground texts, eventhose that appearon the Internet, take the form of dis-
crete works separatedin time and spacefrom the texts they rewrite. The In-
ternet works, however,appear in a very different context than do the print
ones.Anyone who stumblesupon any of thesewritings is likely to find them
linked to a personal or group site containing biographies of the site owner,
explanations of the imaginative world, and lists of links to similar stories.
The link, in other words, makes immediately visible the virtual community
createdby these activereaders.
How does such activereading-as-writingrelate to the hlpertert reader?
First ofall, this kind ofprint-based activereaderencountersa supposedlydis-

AN lNrRoDUcrloN crete,finished texUthe reader'sresponse-writing a new text-demonstrates

that this kind ofreader both acceptsthat fact and also doesnot want to accept
its limitations. This activereading characterizesreadersof blogs:they take an
existing text and add to it, but becausethey write in a networked computer
environment the commented-onblog, employing TrackBack,can link to the
activereaderk text, incorporating it into the ongoing discussion.8
Like blogs, by-now atypicalhypertext systemsthat permit readersto add
their own links and materials (Intermedia, Storyspacein the authoring envi-
ronment) or even websitesthat solicit reader contributions represent ways
that readerscan assumethe role of authors.AII of theseforms of activeread-
ing differ from the experienceof the hypertext reader in read-onlysystems,
whose writing takes the form not of adding new texts but of establishing an
order ofreading in an already-writtenset oftexts. Readersoflarge bodies of
informational hypermediacreatethe document they readfrom the informed
choicesthey make. It might appeatthat such is rarely true of readersof fic-
tional hypertextswho may not know where particular links lead. Nonethe-
less,the best hJperfictions, I submit, permit the readerto deduceenough ba-
sic information, sometimes,asin Michael Joyce'safiemoon,byretracingtheir
steps,to make informed (thus creative)decisionswhen they arrive at links.
Still, no matter how much power readershaveto choosetheir ways through
a hypertext,they never obtain the same degreeof power-or haveto expend
as much effort-as those who write their texts in responseto another's.

Writers on hypertext trace the conceptto a pioneering article

VannevarBush and the Memex by Vannevar Bush in a t945 issue of Atlantic Monthly that
called for mechanically linked information-retrieval ma-
chines to help scholarsand decision makers facedwith what was alreadybe-
coming an explosionof information. Struckby the "growing mountain of re-
search"that confronted workers in everyfield, Bush realizedthat the number
of publications had already"extendedfar beyondour presentability to make
real use of the record. The summation of human experienceis being ex-
pandedat a prodigious rate,and the means we use for threading through the
consequentmaze to the momentarily important item is the sameaswasused
ships" (17-18).As he emphasized,..theremay be
in the daysof square-rigged
millions of fine thoughts, and the account of the experienceon which they
are based,all encasedwithin stonewalls of acceptablearchitechrralform;but

According to Bush,the main problem lies with what he termed "the mat-

HYPERTEXT ter of selection"-information retrieval-and the primary reasonthat those
who need information cannot find it lies in turn with inadequatemeans of
storing, arranging, and tagging information:

Ourineptitudein gettingattherecordis largely

caused of systems
Whendataof anysortareplaced
of indexing. theyarefiledalphabetically
in storage,
or numerically, it downfromsubclass
isfound(whenit is)bytracing
lt canbein onlyoneplace,unlessduplicates
to subclass. areused;onehasto have
rulesasto whichpathwilllocateit,andtherulesarecumbersome.Havingfoundone
item,moreover,onehasto emerge fromthesystem on a newpath'(31)

As Ted Nelson, one of Bush's most prominent disciples,points out,

,,thereis nothing wrong with categorization.It is, however,by its nature tran-

sient: category systems have a halflife, and categorizationsbegin to look

'Pong Balls, Ping'
fairly stupid after a few years . . . The army designationof
has a certain universal characterto ir" (LiteraryMachines,2 1491.Accotdingto
Bush and Nelson, then, one of the greateststrengths of hlpertext lies in its
capacityof permitting users to find, create,and follow multiple conceptual
structures in the same body of information. Essentially,they describe the
technologicalmeans of achieving Derrida'sconceptof decentering.
In contrast to the rigidity and difficulty of accessproduced by present
means of managing information basedon print and other physical records,
one needs an information medium that better accommodatesthe way the
mind works. After describing present methods of storing and classifying
knowledge,Bush complains, "The human mind doesnot work that way" ('As
We May Think," 31)but by association.With one fact or idea "in its grasp,"
the mind "snaps instantly to the next that is suggestedby the associationof
thoughts, in accordancewith some intricate web of trails carded by the cells
of the brain" (32).
To liberate us from the confinementsof inadequatesystemsof classi-
fication and to permit us to follow natural proclivitiesfor "selectionby as-
sociation, rather than by indexing," Bush therefore proposes a device,the
"memex," that would mechanize a more efficient, more human, mode of
'A memex," he explains, "is a devicein
manipulating fact and imagination.
which an individual stores his books, records, and communications, and
which is mechanized so that it may be consulted with exceedingspeedand
flexibility. It is an enlargedintimate supplement to his memory" (32).Wrir
ing in the daysbefore digital computing (the first idea for a memex came to
him in the mid-1930s),Bush conceivedof his deviceas a desk with translu-
cent screens,levers,and motors for rapid searchingof microform records'

AN lNrRoDUcrloN In addition to thus searchingand retrieving information, the memex also

permits the readerto "add marginal notes and comments, taking advantage
ofone possibletype ofdry photography,and it could evenbe arrangedso that
[an individual] can do this by a stylus scheme,such as is now employedin the
telautographseenin railroad waiting rooms, just as though he had the phys-
ical pagebeforehim" (33).Two things demand attention aboutthis crucial as-
pect of Bush'sconceptionof the memex. First, he believesthat while reading,
one needsto appendone'sown individual, transitory thoughts and reactions
to texts. 'With this emphasis Bush in other words reconceivesreadingas an
active processthat involveswriting. Second,his remark that this active,in-
trusive reader can annotate a text "just as though he had the physical page
before him" recognizesthe need for a conception of a virhral, rather than a
physical,text. One of the things that is so intriguing about Bush'sproposalis
the way he thus allowsthe shortcomingsof one form of text to suggesta new
technology,and that leads,in turn, to an entirely new conceptionoftext.
The "essentialfeature of the memex," however,lies not only in its capac-
ities for retrieval and annotation but also in those involving "associativein-
dexing"-what present hypertext systems term a link-'the basic idea of
which is a provision whereby any item may be causedat will to selectimme-
diatelyand automaticallyanother" (34).Bush then providesa scenarioof how
readerswould create"endlesstrails" ofsuch links:

W h e nt h eu s e ri s b u i l d i nagt r a i l h, en a m e ist ,i n s e r ttsh en a m ei n h i sc o d eb o o ka, n d

taPsit outon hiskeyboard. Before himarethetwoitemsto bejoined,projected onto
At thebottomof eachtherearea numberof blankcode
anda pointer
is setto indicate
oneoftheseon eachitem.Theusertaosa
singlekey,andthe itemsarepermanentlyjoined.In eachcodespaceappears the
codeword.Out ofview,butalsoin thecodespace,is inserted
a setofdotsfor oho-
andon eachitemthesedotsbytheirpositions designate
the index
numberof theotheritem.Thereafter,
at anytime,whenoneof theseitemsis in view,
theothercanbe instantly
merelybytappinga buttonbelowthecorrespon-

Bush'sremarkably prescient description of how the memex user creates

and then follows links joins his major recognition that trails of such links
themselvesconstitute a new form of textuality and a new form of writing. As
he explains, "when numerous items havebeen thus joined together to form
a trail . . . it is exactlyasthough the physicalitems had been gatheredtogether
from widely separatedsourcesand bound together to form a new book."
In fact, "it is more than this," Bush adds, "for any item can be ioined into

HYPERTEXT numerous trails" (34),and thereby any block of text, image, or other infor-
mation can participate in numerous books.
Thesenew memex booksthemselves,it becomesclear,are the new book,
or one additional version of the new book, and, like books, these trail setsor
webs can be shared.Bush proposes,again quite accurately,that "wholly new
forms of encyclopediaswill appear,ready-madewith a mesh of associative
trails running through them, ready to be dropped into the memex and there
amplified" (35).Equallyimportant, individual reader-writerscan sharedocu-
ment setsand apply them to new problems'
Bush, an engineer interested in technical innovation, provides the ex-
ample of a memex user

studying whytheshortTurkish bowwasapparently superior to theEnglish longbow

s .e h a sd o z e n os f p o s s i b lpye r t i n e nbto o k sa n d
i n t h es k i r m i s h eosf t h eC r u s a d eH
articles in hismemex. Firstherunsthroughanencyclopedia, findsan interesting but
sketchyarticle, leaves it pro.jected. Next,in a history, hefindsanother pertinent item,
andtiesthetwotogether. Thushegoes,building a trail of many items. Occasionally
l rj o i n i n gi t b ya
h ei n s e r tasc o m m e notf h i so w n ,e i t h elri n k i n igt i n t ot h em a i nt r a i o
item.Whenit becomes
sidetrailto a particular evident of
availablematerials hada greatdealto do withtheboWhebranches offona sidetrail
whichtakeshimthroughtextbooks andtablesofphysical
on elasticity constants.He
insertsa pageof longhand of hisown.Thushe buildsa trailof hisinterest
to him.(34-35)
throughthemazeof materials

And, Bush adds,his researcher'smemex trails, unlike those in his mind,

"do not fade," so when he and a friend severalyearslater discuss "the queer
ways in which a people resist innovations, even of vital interest" (35),he can
reproduce his trails createdto investigateone subject or problem and apply
them to another.
Bush'sidea of the memex, to which he occasionallyturned his thoughts
for three decades,directly influenced Nelson, Douglas Englebart, Andries
van Dam, and other pioneers in computer hlpertext, including the group at
the Brown University's Institute for Researchin Information and Scholar-
'As We May Think" and "Memex
ship (IRIS) who createdIntermedia. In
Revisited"Bush proposedthe notion of blocks of text joined by links, and
he alsointroducedthe terms tinks,linkages, trails,andwebto describehis new
conception of textuality. Bush's description of the memex contains several
other seminal, evenradical,conceptionsof textr-rality.It demands,first of all,
a radical reconfiguration of the practiceof reading and writing, in which both
activitiesdraw closertogether than is possiblewith book technology.Second,

AN lNrRoDUcrtoN despitethe fact that he conceivedof the memex before the advent of digital
computing, Bush perceivesthat something like virtual textuality is essential
for the changeshe advocates.Third, his reconfiguration oftext introduces
three entirely new elements-associative indexing (or links), trails of such
links, and setsor webs composedof such trails. Thesenew elementsin turn
produce the conception of a flexible, customizabletext, one that is open-
and perhapsvulnerable-to the demands of eachreader.They also produce
a concept of multiple textuality, since within the memex world texts refers
to individual reading units that constiflite a traditional "work," those entire
works, sets of documents createdby trails, and perhaps those trails them-
selveswithout accompanyingdocuments.
Perhapsmost interesting to one considering the relation of Bush'sideas
to contemporarycritical and cultural theory is that this engineerbeganby re-
jecting some of the fundamental assumptionsof the information technology
that had increasingly dominated-and some would say largely created-
thought since Gutenberg.Moreover,Bush wished to replacethe es-
sentially linear fixed methods that had produced the triumphs of capitalism
and industrialism with what are essentiallypoeticmachines-machines that
work accordingto analogyand association,machines that captureand create
the anarchic brilliance of human imagination. Bush, we perceive,assumed
that scienceand poetry work in essentiallythe sameway.

Beforeshowing some of the waysthis new information tech-

Formsof Linking,Their Uses nology sharescrucial ideasand emphaseswith contemporary
critical theory I shall examine in more detail the link, the ele-
and Limitations
ment that hypertext adds to writing and reading.eThe very
simplest, most basicform of linking is unidirectional lexia to lexia (Figure 1).
Although this type of link has the advantageof requiring little planning, it
disorientswhen usedwith long documents,sincereadersdo not know where
a link leads in the entered document. It is best used. therefore. for brief lex-
ias or in systemsthat use card metaphors.
Next in complexity comesbidirectional linking of two entire lexiasto one
another-identical to the first form exceptthat it includes the ability to re-
trace one's steps (or jump). Its advantagelies in the fact that by permitting
readersto retracetheir steps,it createsa simple but effectivemeans of orien-
tation. This mode seemsparticularly helpful when a reader arrives at a lexia
that has only one or two links out, or when readersencounter something, say,
a glossarydefinition or image, that they do not want to consult at that point
in their readine.
[exia to LexiaUnidirectional

simpte,requireslittle ptanning.

disorientswhenusedwith long documents,
sincereadersdo not knowwhere[ink leads;bestusedfor
brief [exiasor in systemsthat usecardmetaphor.

Lexiato LexiaBidirectionat

by permittingreadersto retracetheir steps

createssimplebut effectivemeansof orientation.
hetpfulwhenarrivingat lexiasthat haveonty
oneor two departurelinks.

String (word or phrase)to Lexia

Advantages:(1) attowssimplemeansof orientingreaders;

(2) permitslongerlexias;(3) encourages
of annotation

disorientswhenusedwith long documents,
sincereadersdo not knowwherelink leads;bestusedfor
brief [exiasor in systems
that usecardmetaphor.

Figurel. ThreeFormsof Linking

Linking a string-that is, word or phrase-to an entire lexia, the third

form of linking, has three advantages.First, it permits simple means of ori-
enting readersby allowing a basic rhetoric of departure (Figure 1). When
readersseea link attachedto a phrase,such as 'Arminianism" or "Derrida,"
they havea pretty good idea that such a link will take them to information re-
lated in some obvious way to those names. Second,becausestring-to-lexia
linking thus providesa simple means of helping readersnavigatethrough in-
formation space,it permits longer lexias.Furthermore, since one can choose
to leavethe lexia at different points, one can comfortably readthrough longer

AN INTRODUCTION texts.Third, this linking mode also encouragesdifferent kinds of annotation

and linking, since the ability to attachlinks to different phrases,portions
of images, and the like allows the author to indicate different kinds of link
destinations.One can, for example,use icons or phrasesto indicate that the
readercan go to, say,another text lexia, one containing an illustration, bibli
ographicalinformation, definitions, opposing arguments, and so forth.
The difficulties with string-to-lexia links, the form most characteristic
of links in World Wide Web documents, arise in problems encountered at
the destination lexia. Readerscan find themselvesdisoriented when enrer-
ing long documents, and therefore string-to-lexialinking works best with
brief arrival lexia.The fourth form oflinking occurswhen one makesthe link
joining a string to an entire lexia bidirectional. (Most linking in HTML
[HyperText Markup Language]documents takes this form in effect-"in
effect,"becausethe return function provided by most browsers createsthe
effect of a bidirectional link.)
The fifth form, unidirectional string-to-stringlinking, has the obviousad-
vantageof permitting the clearestand easiestway to end links and thereby
createa rhetoric ofarrival. By bringing readersto a clearly defined point in a
text, one enables them to perceive immediately the reason for a link and
hence to grasp the relation between two lexias or portions of them. Readers
know, in other words, why they havearrived at a particular point. The anchor
feature in HTML, which is createdby the <a na me> tag, thus permits authors
to link to a specific section of long document. The possible disadvantageof
such a mode to authors-which is also a major advantagefrom the readerk
point of view-lies in the fact that it requiresmore planning, or at least,more
definite reasonsfor eachlink. Making such links bidirectional, our sixth cat-
egory,makes navigatinghyperspaceeveneasier.
Full hyperterfuality in a reading environment depends,I argue, on the
multisequentiality and the readerchoicescreatednot only by attaching mul-
tiple links to a single lexia but by attaching them to a single anchor or site
within a single lexia.A fully hypertextualsystem(or document) thereforeem-
ploys a seventhform, one-to-manylinking-linking that permits readersto
obtain different information from the same textual site (Figure 2). One-to-
many linking supports hypertextuality in severalways. First, it encourages
branching and consequentreaderchoice.Second,attachingmultiple links to
a single text allows hypertext authors to createefficient overyiewsand direc-
tories that serveas efficient crossroaddocuments,or odentation points, that
help the rcadet navigatehyperspace.Multiple overviewsor sets of overviews
have the additional advantageof easily permitting different authors to pro-
String to String

permitsclearestwayto end links.


requiresmoreptanningthan do tinksto futl



(1) encourages
Advontages: branching and consequent
readerchoice;(2) permitsefficientauthor-generated
overviewanddirectorydocuments; (3) whencombined
with systemsthat providelink menusand other preview
functions,helpsgreatlyin orientingreaders.

can producesenseof an atomizedtext.


Figure2.Two Formsof Linking

vide multiple ways through the same information space.Third, when com-
bined with software,such as Microcosm, Storyspace,or Intermedia, that pro-
vides link menus and other so-calledpreview functions, one-to-manylinking
greatly helps in orienting readers.The major disadvantageof this kind of
link, which plays a major role in most hypertextfiction, lies in its tendencyto
produce a senseof atomized text.
The eighth kind of link-many-to-one linking-proves particularlyhandy
for creating glossaryfunctions or for creatingdocumentsthat make multiple
referencesto a single text, table, image, or other data (Figure 3). DynaText,

(1) handyfor gtossary

Advantages: functiorsor for texts
that makemutfiptereferences
to a singtetext, tabte,
image,or otherdata; (2) encourages
(3) attowssimplemeansof
producingdocumentsfor readerswith differinglevetsof

systemsthat createmany-to-onelinking
can producea distractingnumberof

Figure3. Many-to-OneLinking

Microcosm, and the World Wide Web exemplify hypertext environments in

which one can have many links lead to a single document, an arrangement
that has major advantagesin educationaland informational applications.
In particular, many-to-onelinking encouragesefficient reuse of important
information. For example,having once createdan introductory essayon, say,
CharlesII, Lamarckianism,or Corn Law agitation,the original (andlater) au-
thors simply use linking to provide accessto it as the occasionarises. Fur-
thermore, by providing an easy,efficient means of offering readersglossaries
and other basic information, many-to-onelinking also permits webs to be
used easilyby readerswith differing levelsof expertise.
The major disadvantageof such linking involvesnot the links themselves
but the means various systemsuse to indicate their presence.Systemsthat
createmany-to-onelinking, particularly thosethat createit automatically,can
producea distracting number of link markers.The World Wide Webusescol-
ored underlining to indicate hot text (link anchors),and in the DynaTextver-
sion ofthe first version ofthis book, Paul Kahn chosered text to signify the
presenceof links. In both casesthe readerencountersdistracting markup in-
truding into the text. Experiencewith these systemsquickly convincesone of

T. 0 the need for a means of easilyturning on and off such link indicators, such
asone can do in EastgateSystem'sStoryspace.The disadvantageswith many-
to-onelinks derive not from this form oflinking itselfbut from other aspects
of individual hypertext environments, and any such disadvantagesbecome
amplified by the inexperienceof readers:in the first yearsof the Web,for ex-
ample, authors and designers generally agreed that users, many of whom
had little experiencewith computing, required colored underlining to find
links; otherwise, it was correctly reasoned,readerswould not know what to
do. In the very earliestdaysof the Web,in fact, one often encounteredlinked
underlined text immediately beneath a linked icon becauseweb designers
knew that many neophyteusers would not realize they could follow links by
clicking on the icon. As peoplebeganto use the Web everyday,however,they
recognizedthat when they moved a cursor acrossthe surface of a web
browser, it changedfrom an arrow to a hand when placed over a link. Expe-
rienced usersthus no longer required the once ubiquitous blue underlining,
and many sites now do not use it.
As we shall observeshortly,some systems,such as Microcosm, include a
particularly interesting and valuable extension of many-to-onelinking that
permits readersto obtain a menu containing two or more glossaryor similar
documents.While creating a hypertextversion of my book on Holman Hunt
and Pre-Raphaelitepainting for the World Wide Web, an environment that
does not permit either link menus or one-to-manylinks, I had to choose
whether to link (connect)muitiple mentions of a particular painting, say,the
arlist's Finding of the Saviourin the Temple,to an introductory discussion of
the picture or to an illustration of it. In contrast,while creating a hlpertext
version of the samebook in Microcosm, I easilyarrangedlinks so that when
readersfollow them from any mention of the painting, they receivea menu
containing titles of the introductory text and two or more illustrations,
thereby providing readerswith convenientaccessto the kind of information
they need when they need it (seeFigure 6).
Typed links, our ninth category take the form of limiting an electronic
link to a specific kind of relationship, such as "exemplifies," "influences,"
"conlrary argttment,""derivesfrom" (or "child of "), and so on (Figure4). Soft-
ware that includes such link categorizationrange from proposed research
systemsthat, in attempting to help organize argument, permit only certain
kinds of connections, to those like Marc and focelyne Nanard's MacWeb,
which allows au*rors to createtheir own categories.In fact, any system,such
as Intermedia, Storyspace,or Microcosm, that permits one to attachlabelsto
individual links allows one to createtyped links, since labelspermit authors

Advantages:(1) if cteartyLabeted,
actsas a form of link
previewand aidsreadercomfor| (2) can produce
differentkindsof link behavior,

Disadvantage:can clutterreadingareaor confuseby

producingtoo manydifferentactionswhenonefotlows


to indicateeverythingfrom document type (essay,illustration, statistics,time-

line) to a parlicular path or trail of links that overlaya number of lexias.In fact,
asthe experienceof the World Wide Web reveals,one can use icons or text to
createwhat are essentiallytyped icons evenwhen the systemmakes no pro-
vision for them. Thus, one can make clear (asI havein the Viaorian Web)that
a link leadsto bibliographicalinformation, an illustration, or an opposing
argument by simply linking to a word, such as sourceor illustration,within
The advantageof typed links includes the fact that, when clearlylabeled,
they offer a generalizedkind of previewing that aids readercomfort and helps
navigatinginformation space.Suchlabeling can take the form of icons in the
current lexia (DynaText,VoyagerExpandedBook, World Wide Web),similar
indications in a secondwindow (Intermedia'sWebView and similar dynamic
hypergraphs,such as that created experimentally for Microcosm), and dy-
namic link menus (Intermedia, Storyspace).In systemsthat include pop-up
windows overlayingthe current lexia (DynaText,the proprietary one created
by Cognitive Systemsfor the Microsoft Art Gallery and ones createdby Java
for the World Wide Web),typed links can alsoproduce different kinds of link
behavior.A potential disadvantagefor readersofthe typed link might be con-
fusion producedwhen they encountertoo many different actionsor kinds
of information; in fact, I havenever encounteredhypertextswith theseprob-
lems, but I'm sure some might exist. A greaterdanger for authors would

3.0 exist in systemsthat prescribe the kind of links possible. My initial skepti-
cism about typing links arose in doubts about the effectivenessof creating
rules of thought in advanceand a particular experiencewith Intermedia. The
very first version of Intermedia used by faculty developersand students dif-
ferentiatedbetweenannotationand commentarylinks, but sinceone person's
annotation turned out to be another's commentary no one lobbied for re-
taining this feature,and IRIS omitted it from later versions.
An equally basic form of linking involves the degree to which readers
either activateor even create links. In contemporary hypertext iargon, the
opposition is usually phrased as a question of whether links are author or
wreader determined, or-putting the matter differently-whether they are
hard or soft. Most writing about hypertextfrom Bush and Nelson to the pres-
ent assumesthat someone,author or readerfunctioning asauthor, createsan
electronic link, a so-calledhard link. Recently,workers in the field, particu-
larly the University of Southampton'sMicrocosm developmentgroup, have
posedthe question, "Can one havehypertext 'without links')"-that is, with-
out the by-now traditional assumption that links have to take the form of
always-existingelectronicconnectionsbetweenanchors.This approachtakes
the position that the reader'sactions can createon-demand links. In the late
1980swhen the first conferenceson hlpertext convened,such a conception
of hlpertext might havebeen difficult, if not impossible,to advocate,because
in those days researchersargued that information retrieval did not consti-
tute hypertext, and the two representedvery different, perhapsopposed,ap-
proachesto information. Part of the reason for such views lay in the under-
standableattempts of people working in a new field in computer scienceto
distinguish their work-and thereby justify its very existence-from an
establishedone. Although some authors, such as the philosopher Michael
Heim, perceivedthe obvious connectionbetweenthe activereaderwho uses
searchtools to probe an electronictext and the activereaderofhypertext, the
need of the field to constitute itself as a discretespecialtyprompted many to
juxtaposehypertext and information retrieval in the sharpestterms. When
the late fames H. Coombs createdboth Interlex and full-text retrieval in
Intermedia, many of these oppositions immediately appearedfoolish, since
anyonewho clicked on a word and used Intermedia'selectronicversion of the
Ameican Heitage Dictionary-whether they were aware of it or not-in-
evitably used a secondkind of linking. After all, activating a word and fol-
lowing a simple sequenceof keys or using a menu brought one to another
text (Figure 5). Of course,Web users now havenear-immediateaccessto the
I ntermedia Font R Print ltrlll<+ flttir*r*{'*
pro-phase(pro"faz')r, Thefirststagein celldivision
Cut mitosis,
duringwhichchromosomes fsrmfromthechromatin
[0pu ofthenucleus,- prorpha"sic{-fa"zik}adTl
{}*sIrt f f H l l :a n " a - p h a s e
ll*ste ll+itfl 6-s-phase (an"e-faz') n, The stageof mitosisin which
{.l+:*r the daughterchromosomesmovetowardthe polesofthe
l)* lt Mitosis Stfloes
l>ri*t I t
Find... Metaphase
f i*rl !{** t
lineupat equator.
Chromosomal is maxtmal.
f,heck Spelling E
Eachchromatid whichi$attached
hasa kinetochore
l *s{r}t Ilar{r*'<+ to a poleby kinetochore
I *sel't fl{t r+} E
S e l e c tH l l Chromatids to opposite
separate lfi@,
onserv leadthepoleward
kinetochores m0vement.
iltftj {il's+*t SfIi themselves
Poles moveapad.

Figure5. HypertextLinks and InformationRetrieval:The Interlex Featurein Intermedia

fourth edition of TheAmericanHeritageDictionary (www.bartleby.com)or to

dictionaries in dozensof languagesranging from Abenaki and Armenian to
Walloon and Yemaon www.yourdictionary.com.
Microcosm, a systemon which work began in the early daysof Interme-
dia, has built this idea of reader-activatedlinks into its environment in two
ways. First, using the "Compute Links" function, readersactivatewhat are
essentiallyinformation-retrieval software tools to produce menus of links
that take exactlythe same form as menus of links createdby authors. Soft
links-links created on demand-appear to the reader identical to hard
ones createdby authors. Second,readerscan activateimplicit or generalized

3.0 links. When readers click on hard links, they activatea connection estab-
lished by a hypertextauthor,who, in some systems,could be a previoushy-
pertext reader.When readersactivate"Compute Links," they use what are es-
sentially information-retrieval devicesto createa dynamic relation between
one text and another. In contrast to both these previous approaches,Micro-
cosm'sgeneralizedlink function producesa different form of electroniccon-
nections that we call term sofilinking,linking activatedonly on demand. Es-
sentially,Microcosm'sgeneralizedlinks createa link that only appearswhen
a readerasksfor it (Figure 6). No link marker, no code,indicatesits existence,
and nothing deforms the text in a lexia to announce its presence.In fact, only
a reader'sinterest-a reader'senergy,activeinterest, or aggressiverelation to
the text-brings such a link fully into being. Readerswill recognizethat this
approach,this kind of linking, permits the many-to-more-than-onelinking
that permitted me to havereadersobtain an introductory discussionand two
plates of a painting by a Victorian artist by clicking on the title of one of his
paintings. Microcosm'sgeneralizedlinking facility, in fact, permitted me to
recreatein a matter of hours links that had taken weeksto createmanually in
another system.
The final forms of linking-action links, warm links (or reader-activated
data-exchangelinks), and hot linking (automaticdata-exchangelinks)-rep-
resent, in contrast,kinds that carry the hard, author-createdlink in other
directions. These author-createdlinks do more, in other words, than allow
readersto traverseinformation spaceor bring the document to them. They
either initiate an action or they permit one to do so.
In later chapterswhen we examine examplesof hypermedia containing
animation and video, we shall observeyet other permutations of the link.
Nonetheless,thesepreliminary remarks permit us to graspsome of the com-
plex issuesinvolved with adding the link to writing, with reconfiguring tex-
tuality with an element that simultaneouslyblurs borders,bridges gaps,and
yet draws attention to them.

For more than a quarter century, many computer scientists

Linkingin Open Hypermedia haveproposeda conceptionof linking that differs fundamen-
tally from the one used by HTML, Storyspace,and earlier sys-
tems, such as Guide and HyperCard. This different way of
Walksthe Web conceivingthe link, not surprisingly, is also associatedwith a
different theory of how hypermedia systems should work.
HTML and Storyspacehave accustomedmost of us to the idea that links
exist as integral parts of documents in which they appear.To anyonewho has
apter Two, Typolagical $ymbolism in Hunt's Major

r Findingof the Saviourin theTemple

&eF,ixdrt*goJr&aS*vi6,urix l&e ?erapJaHunt *pprsed rnnarandcut*r

es v.rhrleat the sametrme lhE Finding ol the g*irk
W. H. Hunt The Findinqol the $aviour
lennplethc buildersare H. Hunt,ThBFinding
: wrthinrtswallsthe
H. l'{rrrt, Ths finding
led the conrp*sibon ints
rdants,ofir:rrs ofthe Te
lged tn a ser$i-clrcle,
:*d &at thepainterrhose
r*yedhis deepestfe*lings
r. arrdwe shouldalscsb
re of the c*nsen'ative

The old
.{nd g+d 8rli1
Lest one

$+"&amllolnran Hu$t ?&s ,+lxdixg o/l$* fuvtagr in the

Terxple.1854-60 Oil on calas, 33 3/4 x 55 U2in
Eirninof,am flihr ll{rrcplmq rn,.{ Art li*1lenr

Figure6. GenericLinkingin Microcosm.This screenshot showsthe resultsoffollowing a genericlink eitherfrom the word oFind-
ing'or from the phrase"The Findingof the Saviourin the Temple"in the Microcosmversionof my book William Holman Hunt

andTypologicalSymbolism,Thisactionproducesa menu (at rightl with threechoices:a sectionof my originalbookcontainingthe

principaldiscussionof this painting and two discussionsof it. Sincechoosing"Follow Link" (or double clicking)on any word or

phrasethat servesas an anchorproducesthesethree choices,this screenshot representsmany-to-manylinking. Furthermore,al-

though readersexperiencethe resultsofgenericlinking (herethe menu with threedestinationlexias)just as ifthe author had man-
ually linked each anchorto the discussionand two illustrations,in fact the links only come into existencewhen readerscall for

them. One can thereforeconsiderthis screento exemplif soft many-to-manylinking.Although Microcosmpermitsauthorsto cre-
ate the usual manual form ofone-to-one and one-to-manylinks, the generic link function takes a great deal ofthe work out ofcre.

ati ng i nformational hypertextwebs.


3.0 ever createda link in HTML that point seemsobvious, and, in fact, placing
links within each lexia has major benefits, including simplicity, easeof
creating them, and permanence-they dont move or get lost. This concep-
tion of the link, however,representsa fundamental deparhrrefrom the kind
of medium proposedby Vannevar Bush. The user of the memex, we recall,
createdtrails of associativelinks on top of alreadyexisting texts, savedthose
trails, and sharedthem with others. Different readerscould createvery differ-
ent collectionsof links for the sametexts. Links, in other words, exist outside
the individual lexia in this kind of hypermedia.
Many hlpertext researchers,inspired by Bush,havedesignedand imple-
mented such open hypermedia systemsand infrastructures, a defining char-
acteristicof which is the link databaseor linkbase (seeRizk and Sutcliffe for
a list of such systems).Intermedia, one such system,drew upon its separation
of links and data to permit users to generatemultiple webs from the same
body oftexts and images;depending on an individual userk accessrights, he
or she could view the webs createdby others. In educationalterms, using a
linkbasehad the effectof permitting studentsto usethe main courseweb plus
links addedby students or to screen out links createdby them. It also per-
mitted instructors, aswe shall seein chapter4, to use links to incorporatema-
terials createdby those in other disciplines within their webs without affect-
ing either the original author's text or web. In practice,readersexperienced
an Intermedia web, such as Context32,much as they do its HTML descen-
dant, TheViaorian Web.Infact, eachhypermediacollectionof documentsex-
isted only as a virfual web calledinto being by the linkbase and linkserver.
The linkbase and its associatedserver,which combine to createlink ser-
vices,lie at the heart of open hypermedia systemslike Hyperbase,Multicard,
Sun's Link Service,Microcosm, and its various later incarnations. David C.
De Roure, Nigel G. Walker, and LeslieA. Carr offer the following definition
of thesekey terms:

At itssimplest,
a hypermedia
a source
anchorin a multimedia
a link
a linkbase)
for linkscontaining
identifyspecific or objectsin particular
locations multimedia
documents; alterna-
tivelytheymighthavebroader applicability,
Thelinkbasequerymightalsoberefined bytheuser's
basedon theirprofile,
duringor afterdocument
an in-
for linkcreation

AN INTRODUCTION The Multimedia ResearchGroup at the University of Southhamptonun-

der the leadershipof Wendy Hall and Hugh Davis standsout as the team
of computer scientiststhat has the longest continuous experiencewith open
hypermedia.Their articles dominate the literature in the field, and they have
produced a number of commercial systems.Microcosm, at which we looked
earlier, appearedin commercial form as Multicosm (1994),and as the World
Wide Web becameincreasingly prominent, the Southampton team applied
the heart of Microcosm-its link sewices-to the Internet, creating Distrib-
uted Link Services(1995),Multicosm (1998),and Portal Maximizer (2001).
Multicosm, the company formed to provide commercial versions of the
groupk link-services-basedapplications,has recently becomeActive Naviga-
tion, but the open hypermedia approachremains the same.
As Hays Goodmanpoints out about Active Navigation,"the coretechnol-
ogy behind the company'sproducts is the ability to insert activehyperlinks
on-the-fly in almost any textual format document." The already-observed
forms of linking possiblewith Microcosm show what an immensely power-
ful system it is, but that power came at considerablecost-or, rather, at two
different kinds of costs.Like all open systems,Microcosm and all its descen-
dants require a separateserverfor the linkbase,and the team also had to cre-
ate the softwareto make it work. A different kind of cost appearsin the way
Microcosm has createdanchors.At first, Microcosm recordedlinks solely
in terms of the anchork position-essentially counting offnumbers of char-
actersor units of spatial measurement to record where in the document a
phrase (or image) begins and ends. This method proved to have enornous
advantages.Initially, the Southamptonteam had the goalof creatingthe kind
of hypermedia system that Vannevar Bush would love, since it could create
links not just in other people'sdocuments but also in other software: one
could, for example,link a document in MS Word to another in Word Perfect
to another in a PDF file. This version of Microcosm worked, and much
researchwent into devising ways of linking among different kinds of appli
cations; one of the most interesting of these projects involved placing links
inside a very large CadCam document used by architects, and part of the
difficulty included creating tiny, yet accurate,summaries of the visual data.
Eventuallythe team discoveredthat some featuresthey wished to add to the
systemcould function only if all text had the sameformat, and so they turned
to a more closed system. In the Microcosm version of my book on Pre-
Raphaelitepainting, all the text documents were createdin Word and saved
as in the RTF (rich text format) file format, and although the system,like cur-

3.0 rent HTML, permits linking to images,in practicethe needto attachcaptions
to them resulted in placing imageswithin text documents.
This wonderfully powerful system,which was convenientfor both author
and reader,permitted linking all kinds of data,but it had one Achilles heel:
the computer files to which the system addedlinks could not be modified in
any way. Unlike Intermedia'slinkbase, Microcosm'srequired freezing a doc-
ument once it had links; adding or deleting words would move the link to an
irrelevant phrase.
To solvethis problem, one had to add a secondmethod of identifying link
anchorsin the linkbase, one that required "matching content rather than po-
sition" (De Roure,Walker,and Carr, 67).This method hasthe greatadvantage
of enabling powerfirl generic linking, but it is also much less suited to non-
alphanumeric media. This seemsto be the form of linkbase storagethat al-
lowed Microcosm-Multicosmto becomea Webapplication.Goodmanexplains
how one version marketed by Active Navigationworks:

is implementedessentiallyasa Webproxyserver.
questsa Webpage,the browserwillbedirectedto theWebcosmproxy.Webcosm
fetch thepagefrom the originallocationandannotatethe pagewith ertra linksbefore
passingthe modifedWebpagebacktotheuser's browser.
Whenthewebmaster hasac-
theuserwillseeportions ofthetexttransformed intohyperlinks,
fromwhatis knownasa linkbase.
Thislinkbase at a min-
imuma sourcewordor phrase,
a destination
of thelink.Thelinkbase
is generatedautomaticallybycrawling theWeb
siteat predeterminedintervals,
with the resultsfullytunableso that by movinga
slideronecandecidehowbroador narrowparticular
everywordin a document
but by selectively
linkbasescanbe used,sothatdifferent
on theirprofileor interests.
depending (Emphasis

By storing links apart from text, images,and other media forms, open hy-
permedia systemscan place links in someoneelse'sWeb document without
ever affecting that document. VannevarBush walks the Web. Depending on
the desiresof those who own the server,these addedlinks can be viewed by
anyonewho visits their website,or they can be screenedfrom outsiders.The
capacityof open hypermedia applicationslike Portal Maximizer to add links
to documents coming from another site has important implications for our
conceptionsofauthorship, intellectual property,and political rights, particu-
larly the right offree speech.
As we have observedin the discussion of the hypermedia
Hypertextwithout Linksl pioneers Bush and Nelson, they believed that one of the
greateststrengths of hypertext lies in its capacityto permit
users to discover or produce multiple conceptual structures in the same
body of information. A respected group of computer scientists, however,
reject Nelsonian style link-and-lexia hypertext representedby Intermedia,
Microcosm, and the World Wide Wbb. In'As We Should Have Thought"-
the title an obviousplay on Bush'sseminal essay-Peter |. Niirnberg, fohn |.
Leggett,and Erich R. Schneider assert, for example, that "linking is more
than harmful-it is downright deadly" (96). For anyone whose chief ex-
perience of hypermedia has involved reading computer help files and
materials on the World Wide Web, these statementsseem to come from
authors dwelling in some ThroughtheLookingGlassalternateuniverse-par-
ticularly when they explain that the "two main problems . . . with hyper-
media researchtoday" derive from "our current notion of linking. Firstly,
linking implies a certain kind of structural paradigm, one in which the user
(or occasionallya program) links information together for purposesof navi-
gation . . . Secondly,linking implies the primacy of data, not structure" (96).
They certainly describewhat is commonly understood to be hypertext, and
their reasonfor rejecting it becomesclear when they explain their emphasis
on structure:

thathypermedia isjusta special
caseof a general
ofcomputing which is
structure moreimportant thandata. should
Structure bethe
ubiquitous, blockavailable at alltimesandfromwhich
to allsystems
allother are
data) derived.Herewe ofthe
primacy "structural
of structure (95)

Clearly,Niimberg, Leggett,and Schneiderhave an entirely different set of

concernsthan do Bush and Nelson.
In fact, they representa different approachto information technology-
spatialhypertext-which they point out "has alwayspushedthe limits of our
notions of hypertext . . . Structure in spatial hypertext systemsis dynamic
and implicit. It is defined by the placement of data objectsin a space.This
structtlre is not traversedexplicitly for the pu{pose of navigating the infor-
mation. Instead, it is traversed (by the system) for the purpose of finding
higher-level compositions of atomic data objects and lower-level composi-
tions" (97).They are concernedprimarily with information systemsthat an-
alyzestructure computationally,and, despitetheir opening salvo,it turns out
that they do not in fact reject link-and-lexiaor navigationalhypermedia at all

T. 0 but simply wish to place their fundamentally different approachto comput-
ing "on a par with navigationsystems"(97).
Ntrnberg, Leggett,and Schneiderpoint to VIKI, a systemdevelopedby
CatherineC. Marshalland Frank M. Shipman III, as an exampleof spatial
hypertext.As Marshall and Shipman explain,VIKI, which functions as a con-
ceptual organizer,"providesusers with visual and spatial affordancesfor or-
ganizingand interpreting information" ("Information Triage,"125)."spatial
hypertext," they explain elsewhere, "has its origins in browser-basedap-
proachesin which the emerging hypertext network is portrayed graphically,
in an overview . . . In browser-basedhypertext, boxes generally symbolize
nodes;lines representthe links among them. In a completelyspatialview
of hypertext, the lines-links-may be removed from the picture, and the
nodes may move about freely against their spatial backdrop" ("Spatial Hy-
pertext,"online version). Systemslike VIKI rely on our "spatial intelligence,"
using graphicinterfacesto organizecomplex ideas.Boxlikeicons,which may
contain text, representconcepts,and users can arrangetheseboxesand nest
one inside another to explore or expresstheir relationship. The Storyspace
view, which functions in this manner, exemplifies one feature of spatial hy-
pertext, but to this VIKI adds "stmcture finding algorithms that analyzethe
spatial layout and the visually salient properties of the information objects,"
so that authors do not haveto construct explicit structuresthemselves.Many
discussionsof spatialhypertextso emphasizeconceptualstructuresthatthey
make it seem purely an organizational tool, but Marshall and Shipman be-
Iieve that both readersand writers can benefit from it "For readers,the sys-
tem providesan opporhrnity to readin context,with awarenessofthe related,
nearby nodes." For writers, it supports exploring various conceptual struc-
tures. For both, the graphicdisplay of manipulatableinformation hierarchies
"helps keep complexity tractable" ("Spatial Hypertext,"online version).
Niirnberg, Leggett,and Schneider'sdefinition of the conceptmakes me
suspectthat what they term spatialhyperte.rt
has little to do with hypertert and
hypermedia,though it certainly representsan important areaof research
in computer science.In contrast, Marshall and Shipman'sdescription of a
graphic overviewin which one can hide the lines representing links (which
one can do in Storyspace)suggeststhat such spatial display of information
doesplay a role in specifichypertext systems,though I don't know if by itself
that feature constitutesa form of hypermedia.Graphic sitemaps,suchas The
Victoian Web'sopening screen,the Storyspaceview, and EastgateSystemss
Tinderbox all exemplify graphic presentation of conceptual structures, but
they dont haveVIKI's ability to analyzeand represent such structrres com-

AN rNrRoDUcrloN putationally. Since the computer scienceliterature uses the term spatialhy-
peftert,I shall, too: in later chaptersit refers to those aspectsof hypertext en-
vironments, such as Storyspace,that use the graphic arrangement of lexias
to convey structural information.

The appearance of any new information technology like

The Placeof Hypertext hypertext provides conditions for major societal change,
though any change, such as the democratizing effects of
in the Historyof
writing, which took millennia, can take a very long time to
Information Technology occur. Such changesof information regimes alwaysproduce
both loss and gain. In fact, let's propose a fundamental law
of media change:no free lunch; or, there is no gain without some loss.Thus,
if writing offers us the ability to contemplate information and respond to
it at our leisure, thereby permitting personal reflection and considered
thought, it also lacks the immediacy of the spoken voice and the clues that
we receivewhile observing the person to whom we are speaking. Similarly,
if we gain large audiences,new forms of text preservation, and standardi-
zation of the vernacular from print, we also lose what Benjamin termed
the "aura" provided by the unique ob1'ect.When people find that any par-
ticular gain from a new information technology makes up for the corollary
loss, they claim it representsprogress;when they feel loss more than gain,
they experiencethe new information technology as cultural decline. Print-
ing, an information technology that has so shaped our culture that most
see it as an unqualified benefit, had its bad effects,too. Late medieval and
early Renaissanceconnoisseurs,who mourned the loss of the scribal hand
and pages that integrated words and images, considered printing a crude
technology that destroyed aestheticquality and blamed it for removing an
important source of beauty from the world. For this reason they paid
scribes to copy printed books and create manuscripts.lo Far more impor-
tant, the printed book, as Elizabeth Eisenstein has shown, led directly to
centuries of religious warfare. Historians of the printed book point out the
way it has shapedour culture, influencing our notions of self,,intellectual
properrf language, education, and scholarship, and they present it in a
largely favorablelight but admit it had other effectsas well. So when we con-
sider the potential of hypermedia to changethe way we do things, we must
ask what the gains are and how they balancethe lossesthat any new infor-
mation regime causes.
Evaluatingthe relative effects and values of various media in relation to
one another alwaysturns out, however,to be more than a simple matter of

3.0 loss and gain, becauseas f . David Bolterand RichardGrushin convincinglyar-
gue, everynew medium

a p p r o p r i a t tehse t e c h n i q u efso,r m s ,a n ds o c i asl i g n i f i c a n oc feo t h e rm e d i aa n d

to rivalor refashion
themin thenameof thereal.A mediumin ourculture
in isolation, it mustenterintorelationships
because of respect
withothermedia.Theremaybeor mayhavebeencultures
in whicha single
paintingor song)exists
withlittleor no reference
doesnot seem possible
for us today,whenwe cannot
the representational
evenrecognize powerof a mediumexceptwith reference

For an instantly convincing example of the ways in which newer and

older information technologiesinfluence each other, we need go no farther
than Bolterand Grushin'sobservationthat CNN and other televisionnetworks
haveincreasinglyresembledwebpages,and at the same time the CNN "web
site borrows its senseof immediacy from the televisedCNN broadcasts"(9).
Someexamplesof remediation produce results that can strike us as very odd
indeed. Take the caseof the arrival of printing within a scribal culture. "Ty-
pography was no more an addition to the scribal art than the motorcar was
an addition to the horse. Printing had its 'horse-lesscarriage'phaseofbeing
misconceivedand misapplied during its first decades,when it was not un-
common for a purchaser of a printed book to take it to a scribe to have it
copied and illustrated" (Mcluhan , IJnderstanding
Media, 189).
Beforeexamining the relation of hypertext to previous media, I propose
to look briefly at the advantagesand disadvantagesofvarious forms of infor-
mation technology,a term that today is often mistakenly understoodto refer
solely to computing. Digital information technology certainly begins with
the electronic digital computer, but information technology itself has been
around for millennia. It begins with spokenlanguage,which makespossible
communal or community memory that in turn permits cultural development.
Unlike biological or Darwinian development,such cultural (or Lamarkian)
changepermits groups of peopleto accumulateknowledgeand practiceand
then passthem on to later generations;writing servesasindividual prosthetic
memory which in turn createsa prosthetic group or community memory.
Speechas an information technologyhas certain qualities,which can be
experiencedas advantageousor disadvantageous,depending on the specific
situations in which it occurs. "Language,like currency,"Mcluhan reminds
Media,"actsas a store of perceptionand as a transmitter
us in Understanding
of the perceptions and experienceof one person or one generation to an-

AN INTRODUCTION other. As both a translator and storehouseofexperience, languageis, in ad-

dition, a reducer and a distorter ofexperience. The very great advantageof
acceleratingthe learning process,and of making possible the transmission
of knowledge and insight acrosstime and space,easily overridesthe disad-
vantagesoflinguistic codificationsofexperience" (151-52).Theseare not the
only advantagesand disadvantagesofspoken language,for as Derrida (fol-
lowing Plato)urges, it is fundamentally a technologyof presence.
listener haveto be present in the same place and time, though, as Christian
Metz points out, both do not haveto be in sight of eachother; one can,for ex-
ample, hear words spokenby someoneon the other side of a door or in a dark-
ened room. What advantagesand what disadvantages,then, doessuch an in-
formation technologybasedon presencehavel This question turns out to be
an especiallycrucial one, becausemany info-pundits automatically assume
that presencehas more importance than all other qualities and effectsof any
particular information technology.I've often obsewedthat many writers on
media, particularly its educationalapplications,often reactto improvements
in computing, such asincreasedspeedof Internet connections,asif the most
important result of any such changelies in the possibility of affordabletele-
presence.They imply that speakingin sight of the listener alwayshas more
value than writing or other forms of communication. In other words, con-
sidering the possibility of sending and receivinglarge quantities of informa-
tion over electronic networks, the first reaction of many educatorsand busi-
nesspersonsis that we can replacewritten text by talking heads.
This typical reaction exemplifies Bolter and Grushin's point that the
advocatesof all new information technologies always claim theirs possess
immediacy,and this claim derives from "the desire to get past the limits of
representationand to achievethe real" (53).Such assumptions,which ignore
the very different strengths of speechand writing, demonstrate that many
people believethat being in the presenceof someonetrumps all the advan-
tages, including reflection, abstraction, organization, and concision, that
writing enables.Moreover, presenceis itself not always a desirable,much
lessthe most important, quality when communicating with another person.
We can all think of situations in which we feel more comfortable talking on
a telephonethan speakingto someonefaceto face:when we do not look our
best-say, becausewe've just awakened-or when we wish to fend offsome-
one trying to solicit money for a charity or sell us something. Absence,in
other words, also has great value in certain communicative situations, a
crucial factor to take into account when considering the gains and losses
involvedin writing.

3.0 Writing, probablythe most important technologyhuman beings everde-
veloped,exchangespresenceand simultaneity for asynchronouscommuni-
cation-for the opportunity to respond at one'sown convenience.Becauseit
does not base the act of communication on presence,writing does not re-
quire the person communicating to be in either the same place or the same
time as the person receiving the communication. The person communicat-
ing information placesit in a form that permits someone else to receiveit
later.Writing, printing, cinema,and video are all forms of asynchronouscom-
munication, which, as Mcluhan points out in TheGutenbergGalaxy,permits
reflection, abstraction,and forms of thought impossible in an oral culture.
Writing's combinationof absenceand asynchronisityobviouslypermits a new
kind of education,as well as itself becoming a goal of education,sinceteach-
ing reading and writing becomesa primary function of early instruction in
erasin which these skills are important.
For millennia, writing, which eventually leads to silent reading, none-
thelessremained a technologythat oddly combined orality and literacy.The
explanationfor this situation lies in economic and material factors.The
high cost and scarcity of writing surfacesprompted scribes to omit spaces
between words and adopt a bewildering array of abbreviations,all so they
could cram as many charactersaspossibleon a scroll or page.Thesematerial
conditions produced a kind of text that proved so difficult to read that it
chiefly servedas a mnemonic device,and readersoften read aloud. Eventu-
ally, around the year 1000,cheaperwriting materials led to the development
of interword spacing, which in turn encouragedsilent reading-a practice
that tended to exchangeexpressiveperformance and a communal experi
encefor privary,increasingreading speed,and the senseof personalor inner
place.Interword spacing,like the codex(whatwe generallycall a book),even-
tually changedreading from a craft skill to an ordinary one required ofevery
Sincethe invention of writing and printing, information technologyhas
concentratedon the problem of creating and then disseminating static, un-
changing records oflanguage. As countless authors since the inception of
writing have proclaimed, such fixed records conquer time and space,how-
evertemporarily, for they permit one person to sharedata with other people
in other times and places.As Elizabeth Eisensteinargues,printing addsthe
absolutelycrucial element of multiple copiesof the sametext; this multiplic-
ity, which preservesa text by dispersing individual copiesof it, permits read-
ers separatedin time and spaceto refer to the same information (116).As
Eisenstein,Marshall Mcluhan, William M. Ivins, J. David Bolter, and other

AN tNrRoDUcroN studentsofthehistoryoftheculturaleffectsofprinttechnologyhaveshown,
Gutenberg'sinvention produced what we today understand as scholarship
and criticism in the humanities. No longer primarily occupiedby the task of
preservinginformation in the form of fragile manuscripts that degradedwith
frequent use, scholars,working with books, developednew conceptionsof
scholarship,originality, and authorial properfy.
Hand-set printing with movable type permits large numbers of readers
widely separatedin time and spaceto encounter essentiallythe same text-
and hence createsa new kind of virrual community of readers and many
other things basicto modern culture. The existenceof multiple copiesof the
samete)ctpermits readershundreds of miles and hundreds of yearsapart to
refer to specificpassagesby page number. Printing, which thus exemplifies
asynchronous,silent communication, provides the conditions for the devel-
opment of a humanistic and scientific culture dependenton the ability to cite
and discuss specific details of individual texts. And of course it drastically
changesthe nature of education,which moves from dictating primary texts
to the student to teaching the student modes of critical analysis."Even in the
early eighteenth century," Mcluhan reminds us, "a 'textbook'was still de-
fined as a 'ClassickAuthor written very wide by Students,to give room for an
Interpretation dictated by the Master, &c., to be inserted in the Interlines'
(O.E.D.).Beforeprinting, much of the time in schooland collegeclassrooms
was spent in making such texts" (UnderstandingMedia, t89).
High-speed printing, which appearedin the nineteenth century truly
acted as a democratizing force, producing many of our conceptionsof self
intellectual property,and education.In addition to creatinga virhral commu-
nity ofreaders,the relativelyinexpensivetextscreatedby high-speedprinting
radically changedthe notions of an earlier manuscript culture about how to
preservetexts:with printing, one preservestextsby creating and distributing
multiple copies of them rather than, as with manuscripts, which eventually
degradeafter many readings,protecting the text by permitting fewer people
to have accessto it. As we all know the book also functions as a kind of self-
teaching machine that turns out to be far more accessibleand hence more
quickly democratizing than manuscript texts can everbe.
Although the fixed multiple text produced by print technology has had
enormous effects on modern conceptions of literature, education, and re-
search,it still, asBushand Nelsonemphasize,confrontsthe knowledgeworker
with the fundamental problem of an information retrieval system basedon
physical instantiations of text-namely, that preserving information in a
fixed, unchangeablelinear format makes information retrieval difficult.

3.0 We may statethis problem in two ways. First, no one arrangementof
information proves convenient for all who need that information. Second,
although both linear and hierarchical arrangementsprovide information in
some sort of order, that order doesnot alwaysmatch the needsof individual
users of that information. Over the centuries scribes, scholars,publishers,
and other makers ofbooks have invented a range ofdevices to increasethe
speedof what today are called information processingand retrieval. Manu-
script culture graduallysawthe invention ofindividual pages,chapters,para-
graphing, and spacesbetweenwords. The technologyof the book found
enhancementby pagination, indices, and bibliographies. Such deviceshave
made scholarshippossible,if not alwayseasyor convenientto carry out.
The next great changein information technology-and that which most
concernsus-came with the developmentof digital information technology.
For the first time, writing, which had alwaysbeen a matter of physicalmarks
on a physicalsurface,insteadtakesthe form ofelectronic codes,and this shift
from ink to electronic code-what fean Baudrillard calls the shift from the
"tactile" to the "digital" (Simulations,115)-produces an information tech-
nology that combines fixity and flexibility, order and accessibility-but at a
Using Diane Balestri'sterminology,we can saythat all previous media
took the form of hard text (citedin Miles, " softvideography");computing pro-
duces soft text, and this fundamental change, like all developmentsin in-
fotech, comes with gains and losses.For example,although electronic writ-
ing has the multiplicity of print, it does not have the fixity-and hence the
reliability and stability-of either written or printed texts.
As Bolter and Grushin point out, over the past half century digital com-
puting has undergone what they call a "processof 'remediatization " during
which societyunderstood it as having fundamentally different purposes:

r The "programmable digital computer was invented in the 1940sas a calcu-

lating engine(ENIAC, EDSAC,and so on)" (66) for military and scientific

r During the next decade"large corporationsand bureaucracies"(66)used it,

instead, for accounting.

r About the same time, a few pioneers saw the computer as "a new writing
technology" (66).

r Turing and those involvedwith AI (artificial intelligence)sawthe computer

primarily asa "symbolmanipulator"that could "remediateearlier technologies
of arbitrary symbol manipulation, such as handwriting and printing" (66).

AN lNrRoDUcloN r "In the 1970s,the firstword.processors

appeared,and in the 1980sthe desk-
top computer. The computer could then becomea medium becauseit could
enter into the social and economic fabric of businessculture and remediate
the typewriter almost out of existence"(66).

r More recently,the computer has been seen as an imagecapturer,presenter,

and manipulator:"lf eventen yearsago we thought of computers exclusively
as numerical enginesand word processors,we now think of them also as de-
vices for generating images, reworking photographs,holding videoconfer-
ences,and providing animation and specialeffectsfor film and television"(23).

This fundamental shift from tactile to digital, physical to code,and hard to

soft media produces text with distinctive qualities. First of all, since elec-
tronic text processingis a matter of manipulating computer codes,all texts
that the reader-writer encounters on the screen are virhral texts. Using an
analogyto optics, computer scientistsspeakof "virtual machines" createdby
an operattng system that provides individual users with the experienceof
working on their own individual machines when they in fact share a system
with as many as severalhundred others.13According to the Oxford.English
Dictionary,"virhral" is that which "is so in essenceor ffia, although not for-
mally or actually;admitting of being calledby the name so far as the effect or
result is concerned,"and this definition apparentlyderivesfrom the use of
the term in optics, where it refers to "the apparentfocus or image resulting
from the effect of reflection or refraction upon rays of lightJ' In computing,
the virrual refers to something thatis "not physicallyexistingassuchbut made
bysoftwareto appearto do sofrom the point ofview ofthe program or the user"
(emphasisadded).As Marie-LaureRyan points out, the powerfi.rlconceptof
virtr-ralizafion"leadsfrom the here and now, the singular, the usableonce-for-
all, and the solidly embodiedto the timeless,abstract,general,multiple, ver-
satile,repeatable,ubiquitous, and morphologically fluid" 1177.'o
Similarly, all texts the reader and the writer encounter on a computer
screenexist as a version createdspecificallyfor them while an electronicpri-
mary version residesin the computer'smemory. One therefore works on an
electronic copy until such time as both versions convergewhen the writer
commands the computer to "save"one's version of the text by placing it in
memory.Atthis pointthe texton screenand in the computer'smemorybriefly
coincide,but the reader alwaysencountersa virtual image of the stored text
and not the original version itself; in fact, in descriptions ofelectronic word
processing,such terms and such distinctions do not make much sense.
As Bolter explains, the most "unusual feature" of electronic writing is

T. 0 that it is "not directly accessibleto either the writer or to the reader.The bits
of the text are simply not on a human scale.Electronictechnologyremovesor
abstractsthe writer and reader from the text. If you hold a magnetic tape or
optical disk up to the light, you will not see text at all . . . In the electronic
medium severallayers of sophisticatedtechnology must intervene between
the writer or reader and the codedtext. There are so many levelsof deferral
that the readeror writer is hard put to identify the text at all: is it on the screen,
in the transistor memory or on the disk?" (Writing Space,42-43).Further-
more, whereasa printed book has weight and mass, its digital form appears
immaterial."Ifyouwantto getpickyaboutthephysics,"Mitchellelegantlyex-
plains, "we can say that the corpus of classicalliterature is now embodied
electromagnetically,and, yes,electronsdo havemass.But that is irrelevant at
the level of everydayexperience.My briefcasequickly getsweighed down if I
load volumes of the Loeb ClassicalLibrary into it, but my laptop doesnot get
any heavierif I downloadthe TIG onto its hard drive" (Me++,23Ln.7).
The "'virrual' and the 'material,"' Ned Rossiterreminds us, "are always
intimately and complexly intertwined" (177),and so emphasizing the virfu-
ality of electronicturt and imogeinno way implies that the actualreading ex-
perienceinvolveseither a disembodiedreaderor a nonmaterial presentation
of text itself. As N. Katherine Hayles emphasizes,we haveto find new ways
"to think about embodiment in an ageof virfuality" (193).1s
We must, for ex-
ample, come to the absolutelynecessaryrecognition that the physical,mate-
rial conditions of computer deviceswe use affect our experienceof virhral
text. As I havepointed out elsewhere,the size of monitors, the changefrom
bitmap to grayscaleto color displays,the portability of computers, and our
physical distancefrom them make dramatic differencesin kinds of texts we
can read and write ("What's a Critic to Dol" and "Connected Images,"82).16
Computer text may be virtual, but we who read it are still physical,to read it
we rely on physicaldevices,and it has effectson the physicalworld. "Bits just
don't sit out there in cyberspace,"Mitchell reminds us, and therefore "it
makesmore senseto recognizethat invisible, intangible, electromagnetically
encodedinformation establishesnew types of relationships amongphysical
eventsoccurring in physicalplaces" (Me++, 4).
The code-basedexistence of electronic text that makes it virrual also
makes it infinitelyvariable. If one changesthe code,one changesthe text. As
Hayleshas pointed out, "When a text presentsitself asa constantlyrefreshed
image rather than as a durable inscription, transformations can occur that
would be unthinkable if matter and energy,rather than informational pat-
terns, formed the primary basisfor the systemicexchanges"(30).Further-

AN lNTRoDUcrtoN more, since digital information technologystoresboth alphanumerictext

(words) and images as codes,it seesno essentialdifference between them.
With images,aswith words, if one manipulatesthe code,one manipulatesthe
text that this codepreservesand produces.Furthermore, as anyonewho has
everresizeda web browser window or enlargeda font in a Microsoft Wbrd or
PDF file knows, this text-as-codeis alwaysadaptable.Becauseusers only ex-
periencea virh.ralimage of the text, they can manipulate the version they see
without affecting the source.Many forms of computer text, in other words,
grant the reader more power than doesany example of writing or print, though
occasionallyat the costof a loss of powerfirl graphicdesign. E-textdocuments
also have permeable limits: borders and edges,like spaces,are matters of
physicality,materiality,embodiment, but digital text-text woven of codes-
does not have and cannot have such unity, such closure. The digital text,
which existsindependent of the placein which we experienceit, e-mergesas
dispersedtext. When we discusshypertext later, we shall seethat hypertex-
tual linking relatesin important ways to this property of electronictext.
The codedbasisof digital text permits it to be processedin various ways,
producing documents,for example,that are both searchableand analyzable.
Thus users can searchelectronictextsfor letters and other characters,words,
or various groups of them. Userscan alsotake advantageof such code-based
textuality to checkthe spelling, grammar, and styleof digital text. Processable
text also permits text as simulation since changing the code makes the text
move to show things impossible to present with a static image or text. As we
shall seewhen we examine examplesof animation in chapter 3, such capac-
ities permit one to argue by demonstrating things often too difficult to show
easilywith linguistic argument.
Digital text can be infinitely duplicated at almost no cost or expenditure
of energy.Duplicate the code,duplicate the text-a fact true for images (in-
cluding imagesof text, as above)or alphabetictext. As Mitchell explainswith
characteristicclarity, "Digital texts, images, and other artifacts begin to be-
havedifferently from their heavier,materially embeddedpredecessors.They
becomenontrivial assets-they areneither depletednot dividedwhen shared,
they can be reproducedindefinitely without cost or loss of quality, and they
can be given awaywithout loss to the giver" (Me++,83). One can just dupli
catethe codeand therebyrepeat-reproduce-the text, thereby affecting the
cost (and value) of the text and the potential size of one'saudience.
Becausethe codesthat constitute electronic text can move at enormous
speedover networks, either locally within organizations or on the Intemet,
they createthe conditions for new forms of scholarlyand other communica-

T. 0 tion. Beforenetworkedcomputing, scholarlycommunication relied chiefly on
moving physicalmarks on a surfacefrom one placeto another with whatever
costin time and money such movement required. Networkedelectroniccom-
munication so drasticallyreducesthe time scaleof moving textual informa-
tion that it producesnew forms of textuality. fust as transforming print text
to electronic coding radically changedthe temporal scaleinvolved in manip-
ulating texts, so too has it changedthe temporal scaleof sharing them. Net-
worked electronic communication has both dramatically speededup schol-
arly communication and createdquickly accessibleversions of older forms
of it, such as online, peer-reviewedscholarlyjournals, and new forms of it,
such as discussion lists, chat groups, blogs, and IRC (Internet RelayChat)
(Landow,"Electronic Conferences,"350). In networked environments users
also experienceelectronic text as location independent, since wherever the
computer storing the text may residein physicalreality,usersexperienceit as
beinghere,on their machines.When one movesthe text-as-code, it movesfast
enough that it doesn't matter where it "is" becauseit can be everywhere ...
and nowhere.tTFinally, electronic Iertis net-work-able,alwayscapableof be-
ing joined in electronic networks. Thus, hypertext and the World Wide Web.
Like many featuresof digital textuality,the sheer speedof obtaining in-
formation has its good and bad sides.Its advantagesinclude increasingly so-
phisticated World Wide Web searchtools, such as Google,that can provide
neededinformation nearly instantaneously.For example,as part of the pro-
cessof writing Hypertert3.O I wantedto look up sometechnicalterms (RSS,
Atom feed)relatedto blogs.Typing one of theseterms into Google,I pressed
the "return" key and receiveda list of relevantweb documents in less than a
second-O.22 second,to be exact;the information I found most useful
occurred in the first and third listed items. The convenienceof such infor-
mation retrieval has increasinglyled studentsand faculty to use such search
tools insteadofphysical libraries. Indeed, "'one ofthe rarestthings to find is
a member of the faculty in the library stacks,"'-so Katie Hafner's article in
the Nep YorkTimesquotesan instructor at a major researchuniversity.
True, Hafner slightly sensationalizesthe use of Googlein researchby not
clarifying the difference between Internet searchesand online resources,
such as large collections of scholarly journals that originally appearedin
print. Facultyand studentsdevotea good deal of their researchtime to locat-
ing and reading these scholarly journals, so online versions of them are
enormously convenient:one can locateindividual articlesin a few minutes at
most, multiple users can read them at the same time, and one can obtain
them when the library is closed;some journals are actuallymore to pleasant

AN lNTRoDUcrloN to read online, since one can increasethe size of print in the online copy.
Nonetheless,not all researchinvolves back issues of specialistperiodicals,
and depending on Internet searchtools at the present time might causeone
to miss a good deal:

of information
to theso-called
whereinformation is contained
in isolated
Searchenginesseekso-called staticWebpages,
do nothavesearch
oftheirown.Information onthedeepWeb,ontheotherhand,comesto the
surfaceonlyas the resultof a database
queryfromwithina particular
Coogle,for instance,
to research
UptonSinclair's 1934campaign
of Cal-
andyouwillmissan entirecollection
of pamphlets
of California
at LosAngeles's
of digitized

With an estimated 500 billion webpageshidden from searchengines, com-

panies like Google and Yahoo have entered into agreementswith major li-
braries to index their collections. Still, as many observershave pointed out,
researcherswho only Googlefor their information-yes, it's actuallybecome
a much-used verb-miss not only a good deal of valuable material but the
pleasures of working with printed books and materials, including the de-
lightful serendipity of stumbling onto something particularly interesting
while looking for something else. If history provides any lessons,then the
marked convenienceof Internet resourceswill increasingly dominate both
scholarlyresearchand far more common everydaysearchesfor information:
the appearanceof the printed book did not make individual manuscripts any
more difficult to use than they had been before Gutenberg,but eventuallythe
rapidly growing number of texts in print, their standardizedvernacular,and
their increasedlegibility made them so convenient that only scholarswith
very specificinterests consulted manuscripts, or still do.
Frankly, I think the consequencesfor literary education, criticism, and
scholarship are vastly exaggeratedfor two simple reasons.First, compara-
tively little-indeed, almost no-literary researchrequiring this kind of in-
accessibleinformation takesplacein collegesand universities.Much of what
is now termed literary researchsimply takes the form of reading secondary
materials, and the rest involves working with materials contemporary with
the texts one is studying-materials almost alwayscatalogued.As far as
undergraduate education is concerned, I believe electronic resources like
ISTOR provide a far greaterrange of information than do Twentieth-Century
Viewsand other prepackagedcollectionsof secondarymaterials.

3.0 Far more important, libraries frequently do not have that kind of infor-
mation missed by Internet searchtool in handwritten, Qped, or printed form
becausemany of thesematerials are out of fashion and hencefall beneaththe
radar. As an old-fashionedhide-bound scholar whose first books depended
on manuscripts and extremelyrare printed material, I quickly discoveredthat
my own university library and others had neither the information I needed
nor the information aboutthe information. For example,looking for the pub-
lished transcripts of sermons fohn Ruskin commented on in his diary while
in the midst of an agonizing religious crisis, I discoveredthat major New York
libraries had no record of whatwas once an extremelypopular and profitable
genre (at least three weekly British periodicals dispatchedstenographersto
take down the sermons of popular preachers;I stumbled onto the fact of their
existencewhen Ruskin quoted from one in a famous letter to The Times;the
greatVictorian scholarGeoffreyTillotson, then my Fulbright advisor,told me:
"I think you're on to something important. Follow it up."). I finally found un-
cataloguedcopiesstoredin a carton in the basementof a theologicalseminary.
Another example:the catalogueof the Beinekerare book library at Yale-"ac-
cessed"by snailmail and the good offices of a librarian-listed the manu-
script of one of Ruskin'sown childhood notes on sermons (his mother made
him do it), but I unexpectedlydiscoveredthe valuablefirst draft ofthese notes
in a display casein the tiny museum in Coniston, where Ruskin lived for
many years.Evenif one knows where materialsarelocatedthrough the schol-
arly grapevine,they may not be maintained in easily searchableform. After
traveling to the Isle of Wight to work with the vastcollection of Ruskin letters
and diaries at the Bembridge School, I discoveredthey were uncatalogued.
Evenlocating the catalogueentry for an item (the information about the in-
formation) doesnt mean you will find it. Thus, when I thought I had located
in the then-British Museum Library a crucial anonymous exhibition pam-
phlet in fact written by the artist V7.Holman Hunt himsel{, I submitted my
call-slip,waited forty-fiveminutes, and discoveredthat it had been "destroyed
by enemy action" during the Blitz; I unexpectedlybumbled onto a copyat the
bottom of a trunk when, as I was leavinghis adoptedgranddaughter'shome,
sheasked,"Would you like to look through some things in the garagel"That's
enough of what van Dam called van "barefoot-in-the-snowstories" (889) in
his keynoteaddressat the world's first hlpertext conference.Two points: first,
it's obviously better to be lucky than good, and, second, digitizing all the li-
brary cataloguesand deepWeb material in the world doesnot help if the in-
formation you need is not there in the first place-and for much of the most
interesting kind of researchthat cataloguinginformation doesnot exist.

AN lNrRoDUcrloN Far more important a problem with digital searches,as EugeneProvenzo

warned two decadesago,is not what necessaryinformation we can t find but
what personalinformation governmentsand corporationscan near-instantly
discoverabout us. One examplewill suffice.Google,which has so shaped
the world of education and scholarship, is currently offering free e-mail
accountswith enormous storage(1 gigabyte),and the company urges users
neverto discardanything: "You never know when you might need a message
again, but with traditional webmail services,you delete it and it's gone for-
ever.With Gmail, you can easily archiveyour messagesinstead,so they'll still
be accessiblewhen you need them." According to a website whose URL is
gmail-is-too-creepy.com, "Google admits that even deletedmessageswill
remain on their system,and may also be accessibleinternally at Google,for
an indefinite period of time." The danger, according to Public Information
Research,which createdthe site, is that the company pools its information,
keeps it indefinitely, and can share it with anyone they wish. 'Al1 that's re-
quired is for Googleto'have a good faith beliefthat access,preservationor
disclosureof such information is reasonablynecessaryto protect the rights,
property or safetyof Google,its users or the public." Theseprivacy advocates
claim that the company'sstatementsabout terms of use and privacy "mean
that all Gmail account holders have consentedto allow Googleto show any
and all email in their Gmail accountsto any official from any government
whatsoever,evenwhen the requestis informal or extralegal,at Google'ssole
discretion."Moreover,nothing in Gmail's statedpolicy darifies if it will save
and index incoming mail from thosewho havenot agreedto use their system.
When one uses Googleas a searchtool, its software,like that of many other
sites, placesa so-calledcookie with a unique ID number on your computer
that doesnotexpire until2038. Bythat means itkeeps track ofany searchyou
haveevermade.Accordingto variousprivacyadvocatesand consumergroups,
connecting e-mail to this powerful tool createsthe inevitability of enormous
abusesby corporateand government interests, many of whom are not sub-
ject to U.S. law-this last a particularly relevant point since two-thirds of
Googleusers live outside the United States,many in countries without pri-
vacylaws. No free lunch.

Readersmay havenoticedthat in the precedingdiscussionsof

Interactiveor Ergodicl electronic media I have not employed the words interqctive
and interqctivity.As many commentators during the past
decadeand a halfhave observed,thesewords havebeen used so often and so
badly that they have little exact meaning anymore. fust as chlorophyll was

3.0 used to sell toothpastein the 1950sand aloewas used to sell hand lotion and
other cosmeticproductsin the 1970sand 1980s,interac-tivehas beenusedto
sell anything to do with computing, and the word certainly playeda support-
ing role in all the hype that led to the dotcom bust. The first time I heard the
two terms criticized, I believe,was in 1988,when a speakerat a conference,
who was satirizing falseclaims that computersalwaysgiveuserschoices,pro-
jected a slide of a supposeddialogue box. To the question, "Do you want me
to eraseall your datal" the computer offered two choices:"Yes"and "OK"'18
Espen Aarseth, who has particular scorn for inleracliveand interactivity,
quite rightly points out that "to declarea systemis interactiveis to endorseit
with a magic power" (aS).He proposesto replaceitby ergotic,"using a term
appropriatedfrom physicsthat derivesfrom the Greekwords ergonandhodos,
meaning'work' and'path.' In ergodic literature, nontrivial effort is required
to allow the reader to traversethe text. Ifergodic literature is to make sense
as a concept,there must alsobe nonergodicliterature, where the effort to tra-
versethe text is trivial, with no extranoematicresponsibilities placed on the
readerexcept(for example)eyemovement and the periodic or arbitrary turn-
ing of pages" \1-2\. Ergodic,which has the particular value of being new and
thus far not used in false advertising,has receivedwide acceptance,particu-
larly by thosewho study computer gamesas cultural forms. Still, Marie-Laure
Ryans Nanative as Virtual Reality(2001),one of the most important recent
books on digital culture, retains "interactive,"and a glancethrough the pro-
ceedings of 2003 Melbourne Digital Arts Conference reveals that people
working with fi1m and video also prefer the term.le
Ergodic,whenusedasa technicalterm, has its problems, too, sinceit's not
clearthat the reader's"eyemovement" and turning pages,which result from
intellectual effort, are in fact trivial-a point Aarseth himself seemsto accept
when he emphasizesBarthes'spoint that readerscan skip about a page (78).
Ergodicnonethelessappearsa useful coinage,and so is the word interactive
when used,as in Ted Nelson'swritings, to indicatethat the computer user has
power to intervene in processeswhile they take place, as opposedto the
power to act in a way that simply producesan effect,such as flipping a switch
to turn on a light. The wide misuse of an important term is hardly uncom-
mon. After alI, deconstructionhasbeen used in academicwriting and news-
papers to mean everything from "ordinary interpretation" to "demolition"
while the term classicalhasrneant everything from a "historical period," to an
"aestheticstyle,"to an "eternal principle found throughout human culture."
Before writing these paragraphs I checked the earlier version of Hypefiert
and found only four usesof interactiveotherthan in quoted material;this one

AN rNrRoDUcrloN usessix. Eventhough I do not employ it very much, I think interaaive,like

ergodic,has its uses.

fean Baudrillard, who presentshimself as a follower of Wal-

Baudrillard,Binarity,and ter Benjamin and Marshall Mcluhan, is someonewho seems
both fascinated and appalled by what he sees as the all-
the Digital
pervading effects of digital encoding, though his examples
suggestthat he is often confusedabout which media actuallyemploy it.roThe
strengthsandweaknessesof Baudrillard'sapproachappearinhis remarks on
the digitization of knowledge and information. Baudrillard correctly per-
ceivesthat movement from the tactile to the digital is the primary fact about
the new information technology,but then he misconceives-or rather only
partially perceives-the implications of his point. According to him, digital-
ity involvesbinary opposition: "Digitality is with us. It is that which haunts
all the messages,all the signs of our societies.The most concreteform you
seeit in is that of the test, of the question/answer,of the stimulus/response"
(Simulations,115).Baudrillardmost clearlypositsthis equivalence, which he
mistakenly takes to be axiomatic,in his statementthat "the true generating
formula, that which englobesall the others, and which is somehow the sta-
bilized form of the code,is that of binarity, of digitality" (145).From this he
concludesthat the primary fact about digitality is its connection to "cyber-
netic control . . . the new operational configuration," since "digitalization is
its metaphysicalprinciple (the God of Leibnitz),and DNA its prophet" (103).11
Tiue, at the most basiclevel of machine codeand at the far higher one of
program languages,digitalization, which constitutesa fundamental of elec-
tronic computing, does involve binarity. But from this fact one cannot so
naivelyexffapolate,as Baudrillard does,a completethought-world or episteme.
Baudrillard,of course,may well haveit partially right: he might haveperceived
one key connectionbetweenthe stimulus/responsemodel and digitality. The
fact of hypertext,however,demonstratesquite clearlythat digitality doesnot
necessarilylock one into either a linear world or one of binary oppositions.
Unlike Derrida, who emphasizesthe role of the book, writing, and writ-
ing technology,Baudrillard never considersverbal text, whose absenceglar-
ingly runs through his argument and reconstitutesit in ways that he obvi-
ously did not erpect. Part of Baudrillard's theoretical difficulty, I suggest,
derivesfrom the fact that he bpasses digitized verbaltext and moveswith too
easygracedirectly from the fact of digital encoding of information in two di-
rections: (1) to his stimulus/response,either/or model, and (2) to other non-
alphanumeric (or nonwriting) media, such as photography,radio, and televi-

3.0 sion. Interestingly enough, when Baudrillard correctly emphasizesthe role
of digitality in the postmodern world, he generally deriveshis examplesof
digitalization from media that, parficularly at the time he wrote, for the most
part dependedon analoguerather than digital technology-and the different
qualities and implications of each are great.Whereasanaloguerecording of
sound and visual information requires serial,linear processing,digital tech-
nology removesthe need for sequenceby permitting one to go directly to a
particular bit of information. Thus, if one wishes to find a particular passage
in a Bach sonata on a tape cassette,one must scan through the cassettese-
quentially, though modern tape decks permit one to speed the processby
skipping from spaceto spacebetween sectionsof music. In contrast,if one
wishes to locatea passagein digitally recordedmusic, one can instantly travel
to that passage,note it for future reference,and manipulate it in ways im-
possiblewith analoguetechnologies-for example,one can instantly replay
passageswithout having to scroll back through them.
In concentratingon nonalphanumeric media, and in apparentlyconfus-
ing analogue and digital technology,Baudrillard misses the opporrunity to
encounter the fact that digitalization also has the potential to prevent,block,
and bypass linearity and binarity, which it replaceswith multiplicity, true
reader activity and activation,and branching through networks. Baudrillard
has describedone major thread or constituent of contemporaryreality that is
potentially at war with the multilinear, hypertextualone.
In addition to hypertext, severalaspectsof humanities computing derive
from virruality of text. First of all, the easeof manipulating individual al-
phanumeric symbols produces simpler word processing.Simple word pro-
cessingin turn makes vastly easierold-fashioned,traditional scholady edit-
ing-the creationof reliable,supposedlyauthoritativetextsfrom manuscripts
or published books-at a time when the very notion of such single, unitary,
univocal texts may be changing or disappearing.
Second,this same easeof cutting, copytng,and otherwise manipulating
texts permits different forms of scholarlycomposition, ones in which the re-
searcher'snotes and original data exist in erperientially doser proximity to
the scholarlytext than everbefore. According to Michael Heim, as electronic
textuality frees writing from the constraints of paper-print technology,"vast
amounts of information, including further texts, will be accessibleimmedi-
ately below the electronic surface of a piece of writing . . . By connecting a
small computerto a phone,a professionalwill be ableto read'books'whose
footnotes can be expandedinto further'books'which in turn open out onto

AN lNrRoDUcrloN a vast seaof data basessystemizingaTlof human cognition" (ElectricLan-

guage,l0-ll). The manipulability of the scholarly text, which derives from
the ability of computers to searchdatabaseswith enormous speed,also per-
mits full-text searches,printed and dynamic concordances,and other kinds
of processingthat allow scholarsin the humanities to ask new kinds of ques-
tions. Moreover,as one writes, "the text in progressbecomesinterconnected
and linked with the entire world of inform anon" (Electic Language,161).
Third, the electronicvirfual text, whoseappearanceand form readerscan
customize as they see fit, also has the potential to add an entirely new ele-
ment-the electronic or virhral link that reconfigures text as we who have
grown up with books have experiencedit. Electronic linking createshyper-
text, a form of textuality composedof blocks and links that permits multilin-
ear reading paths. As Heim has argued, electronic word processing in-
evitablyproduceslinkages,and theselinkagesmove text,readers,andwriters
into a new writing space:

rilff[ ffi:::::17
I meannotsomeloosephysicalconnectionlike
a commonphysical spacein thelibrary.
andfor interwoven
andit hascometo have
accuracy ofmeaning inthecaseofwordprocessing.
is interactive,

The presence of multiple reading paths, which shift the balance between
reader and writer, thereby creating Barthes'swriterly text, also createsa text
that exists far less independently of commentary, anaTogues, and traditions
than doesprinted text. This kind of democratizationnot only reducesthe hi-
erarchical separation between the so-calledmain text and the annotation,
which now exist as independent texts, reading units, or lexias, but it also
blurs the boundaries of individual texts. In so doing, electronic linking re-
configures our experienceof both author and authorial property, and this
reconceptionofthese ideaspromisesto affectour conceptionsofboth the au-
thors (and authority) of terts we study and of ourselvesas authors.
Equallyimportant, all these changestake place in an electronic environ-
ment, the Nelsonian docuverse,in which publication changesmeaning. Hy-
pertext,far more than any other aspectof computing, promises to make pub-
lication a matter of gaining accessto electronicnetworks. For the time being

3.0 scholarswill continue to rely on books,and one can guessthat continuing rm-
provements in desktop publishing and laser printing will produce a late
effiorescenceof the text asa physicalobject.Nonetheless,thesephysicaltexts
will be produced (or rather reproduced)from electronictexts, and as readers
increasinglybecomeaccustomedto the convenienceof electronicallylinked
texts, books, which now define the scholar'stools and end-products, will
graduallylose their primary role in humanistic scholarship.

We find ourselves,for the first time in centuries, able to see

BooksAre Technology,Too the book as unnatural, as a near-miraculoustechnologicalin-
novation and not as something intrinsically and inevitably
human. We have, to use Derridean terms, decenteredthe book. We find
ourselvesin the position, in other words, of perceiving the book as tech-
nology.I think it no mere coincidencethat it is at preciselythis period in
human history we have acquired crucial intellectual distance from the book
as object and as cultural product. First came distant writing (the tele-
graph), next came distant hearing (the telephone), which was followed by
the cinema and then the distant seeing of television. It is only with the
addedpossibilities createdby these new information media and comput-
ing that Harold Innis, Marshall Mcluhan, fack Goody,Elizabeth Eisenstein,
Alvin Kernan, Roger Chartier, and the European scholars of Lesengeshichte
could arise.
Influential as these scholarshave been, not all scholarswillingly recog-
nize the power of information technologieson culture. As Geert Lovink, the
Dutch advocateofthe sociopoliticalpossibilities ofthe Internet, haswryly ob-
served,"By and large, [the] humanities have been preoccupiedwith the im-
pact oftechnology from a quasi-outsider'sperspective,as ifsociety and tech-
nology can still be separated"(Dark Fiber,13).This resistanceappearsin two
characteristicreactionsto the proposition that information technology con-
stitutes a crucial cultural force. First, one encounters a tendency among
many humanists contemplating the possibility that information technology
influences culture to assume that before now, before computing, our intel-
lectual culture existedin some pastoral nontechnologicalrealm. Technology,
in the lexicon of many humanists, generallymeans "only that technologyof
which I am frightenedl' In fact, I have frequently heard humanists use the
word technologyto mean "some intrusive, alien force like computing," as if
pencils, paper,t)?ewriters, and printing presseswere in some way natural.
Digital technologymay be new but technology,particularly information tech-
nology,has permeatedall known culture sincethe beginnings of human his-

AN lNrRoDUcrroN tory. If we hope to discern the fate of readingand writing in digital environ-
ments, we must not ffeat all previous information technologiesof language,
rhetoric, writing, and printing as nontechnological.
As fohn Henry Cardinal Newman'sldca of a Universityremindsus, writ-
ers on education and culture havelong tended to perceiveonly the negative
effects of technology.To us who live in an age in which educatorsand pun-
dits continually elevatereading booksas an educationalideal and continually
attacktelevision as a medium that victimizes a passiveaudience,it comesas
a shock to encounter Newman claiming that cheap,easilyavailablereading
materials similarly victimized the public. According to him,

the printingpressis to do withmind;it is
to actmechanically,
is to bepassively,
bythe meremultiplication
of volumes. it be

Part of Newman's rationale for thus denouncing cheap, abundant reading
materialslies in the belief that they supposedlyadvancethe dangerousfallacy
that "learning is to be without exertion, without attention, without toil; with-
out grounding, without advance,without finishing"; but, like any conserva-
tive elitist in our own day,he fears the people unsupervised,and he cannot
believethat reading without proper guidance-guidance, that is, from those
who know from those in institutions like Oxford-can produce any sort of
valid education,and, one expects,had Newman encounteredthe self-taught
mill-workers and artisansofVictorian Englandwhomadediscoveriesin chem-
istry, astronomy,and geologyafter reading newly availablebooks, he would
not havebeen led to changehis mind.
Like Socrates,who feared the effects of writing, which he took to be an
anon)rynous,impersonal denaturing of living speech,Newman also fears an
"impersonal" information technology that people can use without supervi-
sion. And also like Socrates,he desires institutions of higher learning-
which for the ancient took the form of face-to-faceconversationin the form
of dialectic-to be sensitive to the needs of specific individuals. Newman
therefore argues that "a University is, according to the usual designation,
an Alma Mater,knowing her children one by one, not a foundry, or a mint, or
a treadmill."
Newman'scriticism of the flood of printed matter produced by the new
technologysuperficiallyechoesThomas Carlyle,whose "Signs of the Times"

3.0 (1829)had lambastedhis age for being a mechanical one whose "true Deity
is Mechanism."In fact, claims this first of Victorian sages,

aloneis managed
by machinery,
butthe internal
also.Heretoo nothingfollowsitsspontaneous nothingis lefttobe
accomplished . . . lnstruction,
byold,naturalmethods thatmysterious
of Wisdomwithlgnorance,
is no longeran indefinable process,
tentative requiring
studyofindividual anda perpetual
aptitudes, ofmeansandmethods,
variation to at-
tainthesameend;buta secure,
to beconducted
in thegross,bypropermechanism,
ascomesto hand.(101)

Severalthings demand remark in this passage,the first and most obvious of

which is that it parallels and might have provided one of the major inspira-
tions for Newman'sconceptionsof education.The secondrecognition,which
certainly shocksus more than doesthe first, is that Carlyle attacksthose like
Newman who proposeeducationalsystemsand design institutions.
In sentencesthat I have omitted from the quoted passage,Carlyle ex-
plained that everything,with his contemporaries,"has its cunningly devised
implements, its preestablishedapparatus;it is not done by hand but by
machinery. Thus we have machines for education: Lancasterianmachines;
Hamiltonian machines;monitors, maps,and emblems."Or, as Carlylemight
say today,we have peer tutoring, core curricula, distribution requirements,
work-study programs, and junior yearsabroad.
What is not at issue here is the practicality of Carlyle'scriticisms of the
mechanization of education and other human activities-after all, it would
seem that he would attack any organizational changeon the same grounds.
No, what is crucial here is that Carlyle,who apparentlydeniesall possibilities
for reforming existing institutions, recognizessomething crucial aboutthem
that Newman, the often admirable theorist of education,doesnot. Car$e, in
other words, recognizesthat all institutions and forms of social organization
are properly to be considered technologies. Carlyle, who pointed out else-
where that gunpowder and the printing press destroyedfeudalism, recog-
nized that writing, printing, pedagogicalsystems, and universities are all
technologiesof cultural memory. Newman, like most academicsof the past
few hundred years,considersthem, more naively,as natural and inevitable,
and consequentlynoticesthe effectsof only those institutions new to him or
that he doesnot like.
The great value ofsuch a recognition to our project here lies in the fact
that it reminds us that electronifying universities does not take the form of
technologizing them or adding technologyto them in some way alien to their

AN tNrRoDucnoN essentialspirit. Digital information technology,in other words, is only the

latest to shapean institution that, as Carlyle reminds us, is both itself a form
of technologXa mechanism, and has alsolong been influenced by thosetech-
nolosies on which it relies.
,i ,..o.rd form of resistanceto rccognizingthe role of information tech-
nology in culture appearsin implicit claims that technology,particularly in-
formation technology, can neverhave cultural effects. Almost always pre-
sentedby speakersand writers as evidenceoftheir own sophistication and
sensitivity, this strategy of denial has an unintended effect: denying that
Gutenberg'sinvention or television can exist in a causal connection to any
other aspectof culture immediately transforms technology-whatever the
author means by that term-into a kind of intellectual monster, something
so taboo that civilized people cannot discuss it in public. In other words, it
takestechnology,which is both an agent and effect ofour continuing chang-
ing culture(s),and denies its existenceas an element of human culture. One
result appearsin the strategiesofhistorical or predictive studies that relate
cultural phenomenato all sorts of economic,cultural, and ideologicalfactors
but averttheir eyesfrom any technologicalcausation,asifit, and only it, were
in some way reductive.The effect, of course,finally is to deny that this par-
ticular form ofcultural product can haveany effect.
We haveto remind ourselvesthat if, how, and wheneverwe move beyond
the book, that movement will not embody a movement from something nat-
ural or human to something artificial-from nature to technology-since
writing, and printing, and books are about as technological as one can get.
Books, after all, are teaching and communicahng machines.Therefore,if we
find ourselvesin a period of fundamental technologicaland cultural change
analogousto the Gutenberg revolution, one of the first things we should do
is remind ourselvesthat printed books are technology,too.

What can we predict about the future by understanding the

Analoguesto the "logic" ofa particular technology or set oftechnologiesl Ac-
cordingto Kernan,"the 'logic' of a technology,an idea,or an in-
stitution is its tendency consistentlyto shapewhatever it af-
fects in a limited number of definite forms or directions" (a9).The work of
Kernan and others like Chartier and Eisensteinwho have studied the com-
plex transitions from manuscript to print culture suggestthree clear lessons
or rules for anyoneanticipating similar transitions.
First of all, such transitions take a long time, certainly much longer than
early studies of the shift from manuscript to print culture led one to expect.

3.0 Studentsoftechnology and reading practicepoint to severalhundred yearsof
gradual change and accommodation,during which different reading prac-
tices, modes of publication, and conceptionsof literature obtained. Accord-
ing to Kernan,not until about 1700did print technology"transform the more
advancedcountries of Europe from oral into print societies,reordering the
entire socialworld, and restructuring rather than merely modifying letters"
(9). How long, then, will it take computing, specificallycomputer hypertext,
to effect similar changesl How long, one wonders, will the change to elec-
tronic languagetake until it becomesculturally pervasivelAnd what byvays,
transient cultural accommodations,and the like will intervene and thereby
createa more confusing, if culturally more interesting, picturel
The secondchief rule is that studying the relations of technologyto liter-
ature and other aspectsof humanistic culture does not produce any me-
chanical reading of culture, such as that feared by Jamesonand others. As
Kernan makes clear,understanding the logic of a particular technologycan-
not permit simple prediction becauseunder varying conditions the same
technology can produce varying, even contradictory,effects.J. David Bolter
and other historians of writing have pointed out, for example,that initially
writing, which servedpriestly and monarchical interests in recording laws
and records,appearedpurely elitist, evenhieratic;later,asthe practicediffused
down the social and economic scale,it appeareddemocratizing, even anar-
chic. To a large extent, printed books had similarly diverseeffects,though it
took far lesstime for the democratizingfactorsto triumph overthe hieratic-
a matter of centuries,perhapsdecades,instead of millennia!
Similarly, as Marie-Elizabeth Ducreux and Roger Chartier have shown,
both printed matter and manuscript books functioned as instruments of
"religious acculturation controlled by authority, but under certain circum-
stances[they] also supportedresistanceto a faith rejected,and proved an ul-
timate and secretrecourseagainst forced conversion."Booksof hours, mar-
riage charters,and so-calledevangelicalbooks all embodieda "basictension
betweenpublic, ceremonial,and ecclesiasticaluse of the book or other print
obj ect, and personal,private,and internali zed r eadingl'22
Kernan himself points out that "knowledge of the leading principles of
print logic, such as fixity, multiplicity, and systematization,makes it possible
to predict the tendenciesbut not rhe exactwaysin which they were to mani
fest themselvesin the history of writing and in the world of letters. The ide-
alization of the literary text and the attribution to it of a stylistic essenceare
both developmentsof latent print possibilities, but there was, I believe,no
precisenecessitybeforehandthat letters would be valorizedin theseparticu-

AN tNrRoDUcrloN lar ways" (181).Kernan alsopoints to the "tension,if not downright contra-
diction, betweentwo of the primary energiesof print logic, multiplicity and
fixity-what we might call 'the remainderhouse' and the '7ibrary'effects"(55),
eachof which comesinto play,or becomesdominant, only under certain eco-
nomic, political, and technolosicalconditions.
The third lesson or *1. Jn. can derive from the work of Kernan and
other historians of the relations among reading practice,information tech-
nology,and culture is that transformations havepolitical contextsand politi
cal implications. Considerationsof hypertext, critical theory, and literature
haveto take into accountwhat famesoncallsthe basic "recognition that there
is nothing that is not socialand historical-indeed, that everythingis 'in the
last analysis'political" (Political l.Jnconscious,
If the technology of printing radically changedthe world in the manner
that Kernan convincingly explains,what, then, will be the effectsof the par-
allel shift from print to computer hypertextl Although the changesassociated
with the transition from pdnt to electronictechnologymay not parallel those
associatedwith that from manuscript to print, paylng attention to descrip-
tions of the most recent shift in the technologyof alphanumeric text provides
areasfor investigation.
One of the most important changesinvolvedfulfilling the democrattzing
potential of the new information technology.During the shift from manu-
script to print culture "an older systemof polite or courtly letters-primariTy
oral,aristocratic,authoritarian,court-centered-was sweptaway. . . and grad-
ually replaced by a new print-based, market-centered,democratic literary
system" whose fundamental values "were, while not stictly determined by
print ways, still indirectly in accordancewith the actualities of print" (Ker-
nan, PintingTechnologies,4).If hipertextuality and associatedelecffonic in-
formation technologieshave similarly pervasiveeffects,what will they bel
Nelson, Miller, and almost all authors on hypertext who touch upon the po-
litical implications of hypertext assumethat the technologyis essentiallyde-
mocratizing and that it therefore supports some sort of decentralized,liber-
ated existence.As our earlier brief glance at Internet search technology
shows, networked electronic media have at least two contradictory logics-
empowerment of individual readersand their vastly increasedvulnerability
to surveillanceand consequentloss ofprivacy and security.
Kernan offers numerous specific instancesof ways that technoloW "ac-
tually affectsindividual and sociallife." For example,"by changing their work
and their writing, fprint] forced the writer, the scholar,and the teacher-the
standardliterary roles-to redefine themselves,and if it did not entirely cre-

3.0 ate,it noticeablyincreasedthe importance and number of critics, editors,bib-
liographers,and literary historians." Print technologysimilarly redefinedthe
audiencefor literature by transforming it from

a s m a lgl r o u po f m a n u s c r i rpet a d e rosr l i s t e n e r. .s. t o a g r o u po f r e a d e r.s. . w h o

boughtbooksto readin theprivacy of theirhomes.Printalsomadeliterature objec-
tivelyrealfor the firsttime,andthereforesubjectively
conceivableas a universal
of printedbookscontaining
fact,in greatlibraries oftheworld's
largecollections writ-
of lettersto otherpartsofthe social
the relationship
ing . . . Printalsorearranged
worldby,for example, thewriterfromtheneedfor patronage
freeing andtheconse-
quentsubservience and reducing
to wealth,by challenging authority's
controlofwritingbymeansofstatecensorship,andbypushingthrougha copyright
lawthatmadetheauthortheownerof hisownwriting.(4-5)

Electroniclinking shifts the boundaries betweenone text and another as

well as between the author and the reader and between the teacherand the
student. It alsohas radical effectson our experienceofauthor, text, and work,
redefining each.Its effectsare so basic,so radical,that it revealsthat many of
our most cherished, most commonplace, ideas and attitudes toward litera-
ture and literary production turn out to be the result of that particular form
of information technologyand technologyof cultural memory that has pro-
vided the setting for them. This technology-that of the printed book and its
closerelations,which include the typed or printed page-engenders certain
notions of authorial property,authorial uniqueness,and a physicallyisolated
text that hypertext makes untenable. The evidence of hypertext, in other
words, historicizes many of our most commonplace assumptions, thereby
forcing them to descendfrom the ethereality of abstraction and appear as
corollary to a particular technology rooted in specific times and places. In
making availablethese points, hlpertext has much in common with some
major points of contemporary literary and semiologicaltheory, particularly
with Derrida'semphasison decenteringand with Barthes'sconceptionof the
readerlyversus the writerly text. In fact, hypertext createsan almost embar-
rassingly literal embodiment of both concepts,one that in turn raisesques-
tions about them and their interesting combination of prescienceand his-
torical relations {or embeddedness).
Hypertextand CriticalTheory

Like Barthes,Foucault,and Mikhail Bakhtin, facquesDerrida

TextualOpenness continually uses the terms link (liasons),web (toile), network
(riseau), and interwoven(s'ytissent),which cry out for hyper-
texfuality; but in contrastto Barthes,who emphasizesthe writerly text and its
nonlineariry Derrida emphasizestextual openness,intertexhrality, and the
irrelevanceof distinctions betweeninside and outsidea particular tert. These
emphasesappearwith particular clarity when he claims that "like any text,the
text of 'Plato'couldnt not be involved,or at leastin a virtual, dynamic, lateral
manner, with all the worlds that composedthe systemof the Greeklanguage"
(Dissemination,L29).Derrida in fact here describesextant hypertext systems
in which the activereader in the processofexploring a tert, probing it, can
call into play dictionaries with morphological analyzersthat connectindivid-
ual words to cognates,derivations,and opposites.Here again something that
Derrida and other critical theorists describe as part of a seemingly extrava-
gant claim about languagetLlms out preciselyto describethe new economy
of reading and writing with electronicvirtual, rather than physical,forms.
Derrida properly recognizes(in advance,one might say)that a new,freer,
richer form oftext, one truer to our potential experience,perhapsto our
actual ifunrecognized experience,dependson discretereading units. As he
explains,in what Gregory Ulmer terms "the fundamental generalizationof
his writing" (AppliedGrammatology,58),there also exists "the possibility of
disengagementand citational graft which belongs to the structure of every
mark, spoken and written, and which constitutesevery mark in writing
before and outside of every horizon of semiolinguistic communication . . .
Everysign, linguistic or nonJinguistic, spokenor written . . . canbe cited, put

3.0 between quotation marks." The implication of such citability, separability,
appearsin the fact, crucial to hypertext,that, as Derrida adds,"in so doing it
can break with every given context, engendering an infinity of new contexts
in a manner which is absolutelyillimitable" ("Signature,"185).
Like Barthes,Derrida conceivesoftext as constituted by discretereading
units. Derrida'sconceptionof text relatesto his "methodologyof decomposi-
tion" that might transgressthe limits of philosophy. "The organ of this new
philospheme," as Ulmer points out, "is the mouth, the mouth that bites,
chews,tastes. . . The first step of decomposition is the bite" (AppliedGram-
matology,5T).Derrida, who describestext in terms of something close to
Barthes'slexias,explainsin Glasthat "the object of the presentwork, its style
too, is the 'mourceau,"' which Ulmer translatesas "bit, piece, morsel, frag-
ment; musical composition; snack,mouthful." This mourceau,addsDerrida,
"is alwaysdetached,as its name indicatesand so you do not forget it, with the
teeth," and these teeth, Ulmer explains,refer to "quotation marks, brackets,
parentheses:when language is cited (put between quotation marks), the
effect is that of releasingthe grasp or hold of a controlling context" (58).
Derrida'sgroping for a way to foreground his recognition of the way text
operatesin a print medium-he is, after all, the fierce advocateof writing
as againstorality-shows the position, possiblythe dilemma, of the thinker
working with print who seesits shortcomingsbut for all his brilliance cannot
think his way outside this mentalit|. Derrida, the experienceof hypertext
shows,gropestoward a new kind of text: he describesit, he praisesit, but he
can only presentit in terms of the devices-here thoseof punctuatio
ciatedwith a particular kind of writing. As the Marxists remind us, thought
derivesfrom the forcesand modes of production, though, aswe shall see,few
Marxists or Marxians ever directly confront the most important mode of lit-
erary production-that dependent onthe techneof writing and print.
From this Derridean emphasison discontinuity comesthe conceptionof
hypertext as a vast assemblage,what I have elsewheretermed the metatert
Derrida in fact employs the word asseru-
and what Nelson calls the docuverse.
blagefor cinema,which he perceivesas a rival, an alternative,to print. Ulmer
points out that "the gram or trace provides the 'linguistics'for collage/mon-
tage" (AppliedGrammatology,26T),
andhe quotes Derrida'suse of assemblage
in Speechand Phenomena:"The word 'assemblage'seemsmore apt for sug-
gestingthat the kind of bringing-together proposedhere has the structure of
an interlacing, a weaving,or a web, which would allow the different threads
and different lines of senseor force to separateagain, as well as being ready
to bind others together" (131).To carry Derrida's instinctive theorizing of

T ND hypertextfurther, one may also point to his recognition that such a montage-
C R I T I C A LT H E O R Y like textuality marks or foregroundsthe writing processand therefore rejects
a deceptivetransparency.

H)?ertext, which is a fundamentally intertextual system,has

Hypertextand Intertextuality the capacityto emphasize intertextuality in a way that page-
bound text in books cannot. As we have already observed,
scholarlyarticles and books offer an obvious example of explicithypertextu-
ality in nonelectronicform. Conversely,any work of literature-which for the
sakeof argument and economy I shall here confine in a most arbitrary way
to mean "high" literature of the sort we readand teachin universities-offers
an instance of implicit hypertext in nonelectronic form. Again, take Joyce's
Ulyssesfor an example. If one looks, say,at the Nausicaasection, in which
Bloom watchesGerfy McDowell on the beach,one notes that foyce'stext here
"alludes" or "refers" (the terms we usually employ) to many other texts or
phenomena that one can treat as texts, including the Nausicaasectionof the
Odyssey,the advertisements and articles in the women's magazines that
suffuse and inform Gerty'sthoughts, facts about contemporary Dublin and
the Catholic Church, and material that relatesto other passageswithin the
novel.Again, a hypertextpresentationofthe novel links this sectionnot only
to the kinds of materials mentioned but also to other works in foyce'scareer,
critical commentary and textualvariants.H)?ertext herepermits one to make
explicit, though not necessarilyintrusive, the linked materials that an edu-
catedreaderperceivessurrounding it.
Thais Morgan suggeststhat intertextuality, "as a structural analysis of
textsin relation to the larger systemof signifying practicesor usesof signs in
culture," shifts attention from the triad constituted by author/work/tradition
to another constitutedby text/discourse/cu1rure.In so doing, "intertextuality
replacesthe evolutionary model of literary history with a structural or syn-
chronic model of literature as a sign system.The most salient effect of this
strategic change is to free the literary text from psychological,sociological,
and historical determinisms, opening it up to an apparentlyinfinite play of
relationships" (l-21. Morgan well describesa ma1'orimplication of hypertext
(and hypermedia) intertextuality: such opening up, such freeing one to cre-
ate and perceiveinterconnections, obviously occurs. Nonetheless,although
hypertext intertextuality would seem to devalueany historic or other reduc-
tionism, it in no way preventsthose interestedin reading in terms of author
and tradition from doing so. Scholarshipand criticism in hypertext from In-
termedia and HyperCard to Weblogsdemonstratesthat hypertext does not

3.0 necessarilyturn one'sattention awayfrom such approaches.What is perhaps
most interesting about hypertext, though, is not that it may fulfill certain
claims of structuralist and poststructuralist criticism but that it provides a
rich means of testing them.

In attempting to imagine the experienceof reading and writ-

Hypertextand M ultivocality ing with (or within) this new form of text, one would do well
to pay heed to what Mik'hail Bakhtin has written about the
dialogic, polyphonic, multivocal novel, which he claims "is constructed not
as the whole ofa single consciousness,absorbing other consciousnessesas
objects into itself, but as a whole formed by the interaction of severalcon-
sciousnesses,none of which entirely becomesan object for the other" (18).
Bakhtin's description of the polyphonic literary form presents the Dosto-
evskiannovel as a hypertextualfiction in which the individual voicestake the
form oflexias.
If Derrida illuminates hypertextualityfrom the vantagepoint of the "bite"
or "bit," Bakhtin illuminates it from the vantage point of its own life and
force-its incarnation or instantiation of a voice, a point of view, a Rortyian
conversation.lThus, accordingto Bakhtin, "in the novel itself, nonparticipat-
ing 'third persons'are not representedin any way.There is no placefor them,
compositionallyor in the larger meaning of the work" (18).In terms ofhyper-
textuality this points to an important quality of this information medium:
complete read-write hypertext (exemplified by blogs and Intermedia) does
not permit a tyrannical, univocal voice. Rather,the voice is alwaysthat dis-
tilled from the combined experienceof the momentary focus, the lexia one
presentlyreads,and the continually forming narrativeof one'sreading path.

As readersmove through a web or network of texts,they con-

Hypertextand Decentering tinually shift the center-and hence the focus or organizing
principle-of their investigation and experience.Hypertext,
in other words, provides an infinitely recenterablesystemwhose provisional
point of focus dependson the reader,who becomesa truly activereaderin yet
another sense.One of the fundamental characteristicsof hypertext is that it
is composedof bodies of linked texts that have no pimary axis of organiza-
tion. In other words, the metatextor document set-the entity that describes
what in print technologyis the book, work, or single text-has no center.
Although this absenceof a center can createproblems for the readerand the
writer, it also means that anyone who uses hypertext makes his or her own
interests the de facto organizing principle (or center) for the investigation

T ND at the moment. One experienceshypertext as an infinitely decenterableand
C R I T I C A LT H E O R Y recenterablesystem,in part becausehypertexttransforms any document that
has more than one link into a transient center,a partial sitemapthat one can
employ to orient oneself and to decidewhere to go next.
Western culture imagined quasimagical entrancesto a networked real-
ity long before the developmentof computing technology.Biblical typology,
which playedsuch a major role in English culture during the seventeenthand
nineteenth centuries,conceivedsacredhistory in terms of typesand shadows
of Christ and his dispensation.Thus, Moses,who existed in his own right,
alsoexistedas Christ, who firlfilled and completedthe prophet'smeaning. As
countlessseventeenth-centuryand Victorian seflnons, tracts, and commen-
taries demonstrate,any particular person, event,or phenomenon acted as a
magical window into the complex semiotic of the divine schemefor human
salvation.Like the biblical type, which allows significant eventsand phenom-
ena to participate simultaneously in many realities or levelsof reality,the in-
dividual lexia inevitablyprovidesa way into the network of connections.Given
that EvangelicalProtestantismin America preseryesand ertends thesetradi-
tions of biblical exegesis,one is not surprised to discoverthat some of the
first applicationsofhypertext involvedthe Bible and its exegeticaltradition.,
Not only do lexiaswork much in the manner of types,they also become
BorgesianAlephs, points in spacethat contain all other points, becausefrom
the vantagepoint each provides one can see everything else-if not exactly
simultaneously,then a short way distant, one or two jumps away,particularly
in systemsthat havefull text searching.Unlike forge Luis Borges'sAleph, one
does not have to view it from a single site, neither does one have to sprawl
in a cellar resting one'shead on a canvassack.3The hypertext document
becomesa traveling Aleph.
As Derrida points out in "Strucrr-rre,Sign,and Playin the Discourseof the
Human Sciences,"the processor procedurehe calls decenteringhas played
an essentialrole in intellectual change.He says,for example,that "ethnology
could have been born as a scienceonly at the moment when a de-centering
had come about: at the moment when European culture-and, in conse-
quence,the history of metaphysicsand of its concepts-had been dislocated,
driven from its locus, and forced to stop considering itselfas the culture of
reference" (251).Derrida makes no claim that an intellectual or ideological
centeris in any way bad,for, ashe explainsin responseto a query from Serge
Doubrovsky,"l didn't say that there was no center, that we could get along
without a center.I believethat the center is a function, not a being-a reality,
but a function. And this function is absolutelyindispensable" (27L\.

3.0 A11hypertext systemspermit the individual reader to choosehis or her
own center of investigation and experience.What this principle means in
practiceis that the readeris not lockedinto any kind ofparticular organization
or hierarchy.Experienceswith various hypertextsystemsrevealthat for those
who chooseto organizea sessionon the systemin terms of authors-moving,
say,from Keatsto Tennyson-the systemrepresentsan old-fashioned,tradi-
tional, and in many ways still useful author-centeredapproach.On the other
hand, nothing constrainsthe readerto work in this manner, and readerswho
wish to investigatethe validity ofperiod generalizationscan organize their
sessions in terms of such periods by using the Victorian and Romantic
overviewsas starting points or midpoints while yet others can begin with ide-
ological or critical notions, such as feminism or the Victorian novel. In prac-
tice most readersemploy the materialsin TheVictoian Webas abrt-centered
system,sincethey tend to focus on individual works, with the result that even
if they begin sessionsby looking for information about an individual author,
they tend to spend most time with lexias devotedto specific texts, moving
betweenpoem and poem (Swinburne's"Laus Veneris" and Keats's"La Belle
Dame Sans Merci" or works centering on Ulyssesby foyce,Tennyson, and
Soyinka)and betweenpoem and informational texts ("LausVeneris"and files
on chivalry,medieval revival, courtly love, Wagner,and so on).

Shortly after I began to teach hypertext and critical theory

Hypertextas Rhizome Tom Meyer,a member of my first class,advisedme that Gilles
Deleuzeand F6lix Guatrarl's1,000Plateausdemandeda place
in Hypertert.And he is clearly right. Anyone considering the subject of this
book has to look closely at their discussion of rhizomes, plateaus,and no-
madic thought for severalobviousreasons,only the most obviousof which is
that they present 1,000Pleteausas a print protohypertext.Like |ulio Cortzir's
Hopscotch,their volume comeswith instructions to read it in various reader-
determined orders, so that, as Stuart Moulthrop explains, their "rhizome-
book may itselfbe consideredan incunabular hypertext. . . designedas a
matrix of independent but cross-referentialdiscourseswhich the reader is
invited to enter more or less at random (Deleuzeand Guattari, >or)"and read
in any order. "The reader'simplicit task," Moulthrop explains, "is to build a
network of virtual connections (which more than one reader of my acquain-
tancehas suggestedoperationalizingas a web of hypertextlinks)" ("Rhizome
and Resistance,"
Certainly,many of the qualities Deleuzeand Guattari athibute to the rhi
zome require hypertextto find their first approximationif not their complete

T NO answer or fi.rlfillment. Thus, their explanation of a plateau accuratelyde-
CRITICALTHEORY scribes the way both individual lexias and clusters of them participate in a
web. 'A plateau,"they explain, "is alwaysin the middle, not at the beginning
or the end. A rhizome is made of plateaus.Gregory Batesonuses the word
'plateau' to designate something very
special: a continuous, selivibrating
region of intensities whose developmentavoidsany orientation toward a cul-
mination point or external end" (2t-221, such as orgasm, victory in war, or
other point of culmination. Deleuzeand Guattari,who criticize the "Western
mind" for relating "expressionsand actionsto exterior or transcendentends,
instead of evaluating them on a plane of consistencyon the basis of their
intrinsic value," take the printed book to exemplify such characteristiccli
mactic thought, explaining that "a book composedof chaptershas culmina-
tion and termination points" (22).
Like Derrida and like the inventors of hypertext, they propose a newer
form of the book that might provide a truer, more efficient information tech-
nology,asking: "What takesplacein a book composedinsteadof plateausthat
communicate with one another acrossmicrofissures,as in a brainl We call a
'plateau' any multiplicity
connected to other multiplicities by superficial
underground stemsin such a way as to form or extend a rhizome" (22).Such
a description, I should add, perfectly matches the way clusters or subwebs
organizethemselvesin large networkedhypertextenvironments, such asthe
World Wide Web.In fact, reducing Deleuzeand Guattari'sgrand prescription
to relativelypuny literal embodiment, one could take the sectionsconcerning
Gaskelland Trollope inThe VicTorianWeb,or the individual diary entries in
Phil Gyford'sWeblog version of Samuel Pepys'sDiaies, as embodiments of
plateaus.Indeed, one of the principles of reading and writing hypermedia-
as in exploring a library of printed books-lies in the fact that one can begin
anylvhere and make connections,or, as Deleuze and Guattari put it, "each
plateaucan be readstarting anywhereand canbe relatedto any other plateau."
Such a characteristicorganization (or lack of it) derives from the rhi
zome's fundamental opposition to hierarchy, a structural form whose em'
bodiment Deleuzeand Guattari find in the arborescent:"unlike treesor their
roots, the rhizome connectsany point to any other point, and its traits are not
necessarilylinked to traits of the same nature; it brings into play very differ-
ent regimes of signs, and even nonsign states" (21).As Meyer explains in
Plateaus,a Storyspaceweb that has since been published as part of Witing at
theEdge,wegenerallyrely on "arborescentstruchrres,"such asbinary thought,
genealogies,and hierarchies,to divide the "seemingly endlessstream of
information about the world into more easilvassimilable bits. And. for this

3.0 purpose, these structures serve admirably." Unfortunately, these valuable
"organizationaltools end up becoming the only methods of understanding,"
and limit insteadof enhanceor liberateour thought. "ln contrast,Deleuzeand
Guattariproposethe rhizome asa usefirl model for analysingstructures-the
potato,the strawberryplant, with their thickenings and shifting connections,
with their network-like structureinsteadof a tree-likeone" ("Tree/Rhizome").
This fundamental network structure explainswhy

is reducible to theOnenorthemultiple. . . lt hasneither
neither begin-
a middle(milieu)fromwhichit growsandwhichit over-
ningnorend,but always
spills. . . Whena multiplicity
of thiskindchanges it necessarily
dimension, changes
in natureaswell,undergoes a metamorphosis . . . Therhizomeis an antigenealogy.
It is a short-term
memory, Therhizome
or antimemory. operates expan-
capture, Unlikethe graphic
offshoots. or photography,
unliketracings, to a mapthatmustbeproduced,constructed,
a mapthatisalways
detachable, reversible,
connectable, andhasmultiple
andexitsandits ownlinesofflight. . . In contrastto centered
entrylvays (evenpoly-
modesof communication
centric)systemswith hierarchical and preestablished
paths,therhizomeis an acentered, nonsignifring
nonhierarchical, withouta
Ceneral memoryor centralautomaton,
andwithoutanorganizing defined
a circulation

As we explorehlpertext in the following pages,we shall repeatedlyencounter

the very qualities and characteristicsDeleuzeand Guattari here specify:like
the rhizome, hypertext, which has "has multiple entrywaysand exits," em-
bodies something closer to anarchy than to hierarchy, and it "connectsany
point to any other point," often joining fundamentally different kinds of in-
formation and often violating what we understand to be both discreteprint
texts and discretegenresand modes.
Any reader of hypertext who has experiencedthe way our own activities
within the networked text produce multiple versions and approachesto a
single lexia will seethe parallel to hypertext in Deleuzeand Guattarik point
that "multiplicities are rhizomatic, and erpose arborescentpseudomultiplic-
ities for what they are. There is no unity to serveas a pivot in the object,or to
divide in the subject" (8).Therefore,like hypertextconsideredin its most gen-
eral sense,"arhizome is not amenableto any structural or generativemodel.
It is a strangerto any idea ofgenetic axis or deep structure" (12).As Deleuze
and Guattari explain, a rhizome is "a map and not a tracing. Make a map, not
a tracing. The orchid does not reproducethe tracing of the wasp; it forms a
map with the wasp,in a rhizome. What distinguishes the map from the trac-

TND ing is that it is entirely oriented toward an experimentation in contact with
C R I T I C A LT H E O R Y the rea1"(12).Maps and hypertextsboth, in other words, relatedirectly to per-
formance,to interaction.
Like some statementsby Derrida, some of Deleuze and Guattari'smore
cryptic discussionsof the rhizome often become clearer when considered
from the vantagepoint of hypertext. For example,when they state that the
rhizome is a "a short-term memory or antimemory," something apparently
in complete contrast with any information technology or technology of cul-
tural memory they nonethelesscapturethe provisional,temporary changing
quality in which readersmake individual lexiasthe temporary center of their
movement through an information space.
Perhaps one of the most difficult portions of A ThousandPlateausin-
volvesthe notion of nomadic thought, something, again,much easierto con-
vey and experiencein a fluid electronic environment than from within the
world of print. According to Michael foyce,the first important writer of
hypertextfiction and one of the creatorsof Storyspace,Deleuzeand Guattari
reject "the word and world fully mappedaslogos,"proposing insteadthat "we
write ourselvesin the gap of nomos, the nomadic" (Of TwoMinds,207).They
offer or propose,he explains,"being-for spaceagainstbeing-in space.Weare
in the water,inscribing and inscribed by the flow in our sailing. We write our-
selvesin oscillationbetweenthe smooth spaceofbeing for-time (whathappens
to us as we go aswell as what happensto the spacein which we do so) and the
striated spaceof in-time (what happensoutside the spaceand us)" (207).
Those who find the ruptures and seams as important to hypertext as
the link that bridges such gapsfind that the rhizome has yet another crucial
aspectof hypertextuality.Moulthrop, for example,who "describeshypertexts
as composed of nodes and links, local coherencesand linearities broken
acrossthe gap or synapseof transition," takes this approach:"In describing
the rhizome as a model of discourse,Deleuzeand Guattari invoke the
ciple of asignifying rupture' (9), a fundamental tendency toward unpre-
dictability and discontinuity. Perhapsthen hypefiext and hypermedia repre-
sent the expressionof the rhizome in the socialspaceof writing" ("Rhizome
and Resistance,"
We must take carenot to push the similarity too far and assumethat their
descriptionsof rhizome, plateau,and nomadic thought map one to one onto
hypertext, since many of their descriptions of the rhizome and rhizomatic
thought appearimpossible to fulfill in any information technologythat uses
words, images, or limits of any sort. Thus when Deleuzeand Guattari write
that a rhizome "has neitherbeginning nor end, but alwaysa middle (milieu)

3.0 from which it grows and which it overspills,"they describe something that
has much in common with the kind of quasi-anarchicnetworked hypertext
one encounters on the World Wide Web, but when the following sentence
addsthat the rhizome "is composednot of units but of dimensions, or rather
directions in motion" (21),the parallel seems harder to complete. The rhi
zome is essentiallya counterparadigm,not something realizablein any time
or culture, but it can serveas an ideal for hypertext,and hypertext,at leastNel-
sonian, ideal hypertext,approachesit as much as can any human creation.

Discussionsand designs of hypertext share with contempo-

The NonlinearModelof rary critical theory an emphasison the model or paradigm
of the network. At least four meanings of networkappearin
the Networkin Current
descriptionsof actual hypertext systemsand plans for future
CriticalTheory ones. First, individual printworks when transferredto hyper-
text take the form ofblocks, nodes, or lexias joined by a net-
work of links and paths. Network, in this sense,refers to one kind of elec-
tronically linked electronicequivalentto a printed text. Second,any gathering
of lexias, whether assembledby the original author of the verbal text, or by
someoneelsegathering togethertexts createdby multiple authors, alsotakes
the form of a network; thus document sets, whose shifting borders make
them in some sensesthe hypertexh,ralequivalent of a work, are called in
somepresentsystemsa web.
Third, the term networkalso refers to an electronic system involving ad-
ditional computers as well as cablesor wire connectionsthat permit individ-
ual machines,workstations, and reading-and-writing sitesto share informa-
tion. Thesenetworks can take the form of contemporaryLocalArea Networks
(LANs),such as Ethernet,that join setsof machines within an institution or
a part of one, such as a department or administrativeunit. Networksalsotake
the form of Wide Area Networks (WANs) that join multiple organizationsin
widely separatedgeographicallocations.Earlyversionsof such wide-areana-
tional and international networks include JANET (in the United Kingdom),
ARPANET (in the United States),the proposed National Researchand Edu-
cation Network (NREN),and BITNET, which linked universities,research
centers,and laboratoriesin North America, Europe, Israel, and |apan.aSuch
networks, which until the arrival of the World Wide Web had been used
chiefly for electronicmail and transfer of individual files, havealso supported
international electronic bulletin boards, such as Humanist. More powerfirl
networks that transfer large quantities of information at great speedwere
necessarybefore such networks could fully support hypertext.

T ND The fourth meaning of network in relation to hypertext comes close to
C R I T I C A LT H E O R Y matching the use of the term in critical theory. Network in this fullest sense
refersto the entiretyof all thoseterms for which there is no term and for which
other terms stand until something better comes along, or until one of them
gathers fuller meanings and fuller acceptanceto itself: Iiterature,infoworld,
docuverse,infact, allwritinginthe alphanumeric aswell as Derridean senses.
The future wide areanetworksnecessaryfor large-scale,interinstitutional and
intersite hypertext systemswill instantiate and reify the current information
worlds, including that of literature. To gain accessto information, in other
words, will require accessto some portion of the network. To publish in a
hyperterh.ralworld requires gaining access,howeverlimited, to the network.
The analogy,model, or paradigm of the network so central to hypertext
appearsthroughout structuralist and poststructuralist theoretical writings.
Relatedto the model of the network and its componentsis a rejection of
linearity in form and explanation,often in unexpectedapplications.One
exampleof such antilinear thought will suffice.Although narratologistshave
almost alwaysemphasizedthe essentiallinearity of narrative,critics have
recently begun to find it to be nonlinear. BarbaraHerrnstein Smith, for ex-
ample, arguesthat "by virrue of the very nature of discourse,nonlinearity is
the rule rather than the exceptionin narrativeaccounts"("NarrativeVersions,
Narrative Theories," 223). Since I shall return to the question of linear and
nonlinear narrativein a later chapter,I wish here only to remark that nonlin-
earity has become so important in contemporary critical thought, so fash-
ionable,one might say,that Smith's observation,whether accurateor not, has
becomealmost inevitable.
The generalimportance of non- or antilinear thought appearsin the fre-
quencyand centrality with which Barthesand other critics employ the terms
link, network,web,andpoth. More than almost any other contemporarytheo-
rist, Derrida usesthe terms link, web,network,matix, and interweavir.gasso-
ciatedwith hypefiexruality; and Bakhtin similarly employslinks (Problems,9,
25),linkage(9),interconnectedness(19),and interwoven(72).
Like Barthes,Bakhtin, and Derrida, Foucaultconceivesof text in terms of
the network, and he relies preciselyon this model to describehis project, "the
archaeologicalanalysisof knowledge itselfl' Arguin g in The Order of Things
that his project requires rejectingthe "celebratedcontroversies"that occupied
contemporaties,he claims that "one must reconstitutethe generalsystemof
thought whosenetwork, in its positivity,rendersan interplay of simultaneous
and apparentlycontradictoryopinions possible.It is this network that defines
the conditions that make a controversyor problem possible,and that bearsthe

3.0 historicityofknowledge" (75).Order,for Foucault,is in part "the innerlaw, the
hidden network" (n<);and accordingto him a "network" is the phenomenon
"that is able to link together" (727)a wide range ofoften contradictory tax-
onomies, obsewations,interpretations, categories,and rules of observation.
Heinz Pagels'sdescription of a network inThe Dreamsof Reasonsuggests
why it has such appealto those leery of hierarchical or linear models. Ac-
cording to Pagels,'A network has no 'top' or 'bottom.' Rather it has a plural-
ity of connectionsthat increasethe possible interactions between the com-
ponents ofthe network. There is no central executiveauthority that oversees
the system" (20).Furthermore, as Pagelsalsoexplains,the network functions
in various physical sciencesas a powerful theoretical model capableof de-
scribing-and hence offering researchagendafor-a range of phenomena
at enormously different temporal and spatial scales.The model of the net-
work has capturedthe imaginations of those working on subjectsas appar-
ently diverseas immunology, evolution, and the brain.

The immunesystem,likethe evolutionary

system,is thus a powerfulpattern-
of learningandmemory.
Thisfeatureof the
immunesystemhassuggested to a numberof peoplethata dynamical
the immunesystem,couldalsolearnandhavememory. . . The
evolutionary workson thetimescaleofhundreds
system ofthousands
i m m u n es y s t e mi n a m a t t e o
r f d a y sa, n dt h e b r a i ni n m i l l i s e c o n dHs e. n c ei f w e
howthe immunesystemrecognizes
it will
teachus abouthowneuralnetsrecognize
Afterall,boththe im-
consist of highlyspecialized
of billions cellsthat

Terry Eagletonand other Marxist theorists who draw on poststructuralism

frequently employ the kind of network model or image to which the connec-
tionists subscribe(seeEagleton,LiteraryTheory,14,33,78,I04,165,169, L73,
201).In contrast,more orthodox Marxists,who havea vestedinterest (or sin-
cerebelief) in linear narrativeand metanarrative,tend to usenetworkandweb
chiefly to characterizeerror. Pierre Machereymight therefore at first appear
slightly unusual in following Barthes,Derrida, and Foucaultin situating nov-
els within a network of relations to other texts. According to Machery "The
novel is initially situated in a network of books which replacesthe complex-
ity of real relationsby which a world is effectivelyconstituted."Machery'snext
sentence,however,makes clear that unlike most poststructuralistsand post-
modernists who employ the network as a paradigm of an open-ended,non-

TND confining situation, he perceivesa network as something that confines and
C R I T I C A LT H E O R Y limits: "Locked within the totality of a corpus, within a complex system of
relationships,the novel is, in its very letter, allusion, repetition, and resump-
tion of an objectwhich now begins to resemblean inexhaustibleworld" (2681.
Frederic fameson, who attacks Louis Althus ser in The Political Uncon'
sciousfor creatrngimpressions of "facile totalization" and of "a seamlessweb
of phenomena" (271,himself more explicitly and more frequently makes
these models the site of error. For exampTe,when he criticizes the "anti
speculativebias" of the liberal tradition in Marxism and Fonn, he notes "its
emphasis on the individual fact or item at the expenseof the network of re-
lationships in which that item may be imbedded" as liberalism's means of
keeping people from "drawing otherwise unavoidable conclusions at the
political level" (x).The network model here representsa firll, adequatecon-
textualization,one suppressedby an other-than-Marxistform of thought, but
it is still only necessaryin describing pre-Marxian society.fameson repeats
this paradigm in his chapteron Herbert Marcusewhen he explainsthat "gen-
uine desirerisks being dissolvedand lost in the vast network of pseudosatis-
factions which make up the market system" (100-101).Once again, network
provides a paradigm apparentlynecessaryfor describing the complexitiesof
a fallen society.It does so again when in the Sarlre chapter he discusses
Marx's notion of fetishism, which, according to fameson, presents "com-
'objective' network of relationships which they entertain
modities and the
with eachother" asthe illusory appearancemasking the "reality of sociallife,"
which "lies in the labor processitself" (296).

What relation obtains between electronic computing, hyper-

Causeor Convergence, text in particular, and literary theory ofthe past three or four
decadeslj. Hillis Miller proposesthat "the relation . . . is
or Confluencel
multiple, non-linear, non-causal,non-dialectical,and heavily
overdetermined.It doesnot fit most traditional paradigmsfor defining
tionship"' ("LiteraryTheoryi' ll\. Miller himself providesa fine exampleof the
convergenceofcritical theory and technology.Beforehe discoveredcomputer
hypertext,he wote abouttext and (interpretative)text processingin waysthat
sound very familiar to anyonewho has read or worked with hypertext. Here,
for example, is the way Fiction and Repetitiondescribesthe way he reads a
novel by Hardy in terms of what I would term a Bakhtinian hypertextuolity:
"Each passageis a node, a point ofintersecfion or focus, on which converge
lines leadingfrom many other passagesin the novel and ultimately including

3.0 them all." No passagehas any particular priority overthe others, in the sense
of being more important or as being the "origin or end of the others" (58).
Similarly, in providing "an 'example' of the deconstructive strategy of
interpretation,"in "The Critic as Host" (1979),he describesthe dispersed,
linked text block whose paths one can follow to an ever-widening,enlarging
metatext or universe. He applies deconstructivestrategy"to the cited frag-
ment of a critical essaycontaining within itself a citation from another essay,
like a parasitewithin its host."Continuing the microbiological analogy,Miller
next explains that "the 'example' is a fragment like those miniscule bits of
some substancewhich are put into a tiny test tube and explored by certain
techniquesof analyticalchemistry. [One gets]so far or so much out of a little
piece oflanguage, contert after contextwidening out from thesefew phrases
to include as their necessarymilieux a1lthe family of Indo-European lan-
guages,all the literature and conceptualthought within theselanguages,and
all the permutations of our social structures of household economy, gift-
giving and gift receiving" (223).
Miller does point out that Derrida's "Glas and the personal computer
appearedat more or less the same time. Both work self-consciouslyand
deliberatelyto make obsoletethe traditional codexlinear book and to replace
it with the new multilinear multimedia hypertextthat is rapidly becoming the
characteristicmode of expressionboth in culture and in the study of cultural
forms. The 'triumph of theory' in literary studiesand their transformation by
the digital revolution are aspectsof the same sweeping change" (.,Literary
Theory" 20-21]t.This sweeping change has many components, to be sure,
but one theme appearsboth in writings on hypertert (andthe memex) and in
contemporarycritical theory-the limitations of print culture, the culture of
the book. Bush and Barthes,Nelson and Derrida, like all theorists of these
perhaps unexpectedlyintertwined subl'ects,begin with the desire to enable
us to escapethe confinements of print. This common project requires that
one first recognizethe enormous power of the book, for only after we have
made ourselvesconsciousof the ways it has formed and informed our lives
can we seekto pry ourselvesfree from some of its limitations.
Looked at within this context, Claude Ldvi-strauss'sexplanationsof
preliterate thought in The SavageMind and in his treatises on mythology
appearin part as attempts to decenterthe culture of the book-to show
the confinements of our literate culture by getting outside of it, however
tenuously and however briefly. In emphasizing electronic, noncomputer
media, such as radio, television,and film, Baudrillard, Derrida, fean-Frangois

T ND Lyotard,Mcluhan, and others similarly argueagainstthe future importance
C R I T I C A LT H E O R Y of print-basedinformation technology,often from the vantagepoint of those
who assumeanaloguemedia employing sound and motion as well as visual
information will radically reconfigure our expectationsof human nature and
human culture.
Among major critics and critical theorists, Derrida standsout as the one
who most realizesthe importance of free-form information technologybased
on digital, ratherthan analogue,systems.As he points out, "the development
of practical methodsof information retrieval extendsthe possibilities of the
'message'vastly,to the point where it is no longer the 'written translation of

a language,the transporting of a signified which could remain spokenin its

integrity" (10).Derrida, more than any other major theorist, understandsthat
electronic computing and other changesin media have erodedthe power of
the linear model and the book asrelatedculturally dominant paradigms."The
end of linear writingi' Derrida declares,"is indeed the end of the book," even
if, he continues, "it is within the form of a book that the new writings-liter-
ary or theoretical-allow themselvesto be, for better or worse, encased"(O/
Grammatology,86).Therefore, as Ulmer points out, "grammatalogicalwrit-
ing exemplifies the struggle to break with the investiture of the book" (13)'
'book' is now going through a
According to Derrida, "the form of the
period of generalupheaval,and while that form appearsless natural, and its
history less transparent,than ever. . . the book form alone can no longer
settle. . . the caseof thosewriting processeswhich,inpractically questioning
that form, must also dismantle it." The problem, too, Derrida recognizes,
is that "one cannot tamper" with the form of the book "without disturbing
everything else" (Dissemination,3)in Westem thought. Always a tamperer,
Derrida doesnot find that much of a reasonfor not tampering with the book,
and his questioning begins in the chain of terms that appearas the more-or-
less title at the beginning pages of Dissemination:"Hots Liwes: Outwork,
Hors D'oeuwe, Extratext,Foreplay,Bookend,Facing,and Prefacing."He does
so willingly because,as he announcedin Of Grammatology, "All appearances
to the contrary,this death ofthe book undoubtedly announces (and in a cer-
tain sensealwayshas announced)nothing but a deathof speech(of a so-called
full speech)and a new mutation in the history of writing, in history as writ-
ing. Announces it at a distance of a few centuries. It is on that scalethat we
must reckon it here" (8).
In conversationwith me, Ulmer mentioned that since Derrida's gram
equalslink, grammatologyis the art and scienceof linking-the art and

3.0 science,therefore,of hlpertext.sOne may add that Derrida also describes
disseminationas a descriptionof hypertext:"Alongwith an orderedextension
of the conceptof text, dissemination inscribes a different law governing the
effectsof senseor reference (the interiority of the 'thing,' reality, objectivity,
essentiality,existence,sensible or intelligible presence in general, etc.), a
different relation betweenwriting, in the metaphysicalsenseof the word, and
its 'outside' (historical,political, economical,sexual,etc.),,(Dissemination,42\.
Reconfiguringthe Text

Although in some distant, or not-so-distant,future all indi

FromTextto Hypertext vidual texts will electronicallylink to one another, thus creat-
ing metatextsand metametatextsof a kind only partly imagi
nable at present, less far-reaching forms of hypertextualiry have already
appeared.Translationsinto hlpertexfual form alreadyexist ofpoetry fiction,
and other materials originally conceivedfor book technology.The simplest,
most limited form of such translation preservesthe linear text with its order
and fixity and then appendsvariouskinds of textsto it, including criti cal com-
mentary textual variants, and chronologicallyanterior and later texts.l
Hypertext corpora that employ a single text, originally createdfor print
dissemination, as an unbroken axis offwhich to hang annotation and com-
mentary appearin the by-now common educationaland scholarlypresenta-
tions of canonicalliterary texts (Figure 7). At Brown University my students
and I first used Intermedia and Storyspaceto provide annotatedversions
of storiesby Kipling and Lawrence,and I have since createdmore elaborate
World Wide Webpresentationsof Carlyle's" Hudson'sStatue"and other texts.
TheDickensWeb,acorpusofmaterialsfocusedon GreatExpec-tationspublished
in Intermedia (IRIS, 1990)and Storyspace(Eastgate,l992l,differcfromtlese
projects in not including the primary text, as does Christiane Paul's Unreal
City:A Hypertert Guideto T. S. Eliot's"The WasteLand" (19941.
A second case appearswhen one adaptsfor hypertextual presentation
material originally conceivedfor book technology that divides into discrete
lexias,parricularly if it has multilinear elementsthat call for the kind of mul-
tisequential reading associatedwith hypertext.An early exampleof this form
of hypertext appearsin Brian Thomas'searly HyperCard version of Irnitqtio
of electronicbooks versus
andscholarlybookswith foot- or endnotes Networkstructureof hypertext

1. Wheredoesthe readerenterthe text?

2. Wheredoesthe readerleavethe text?

3. Wherearethe bordersof the text?

FigureT. Axial versus Network Structure in Hypertext

Cisti, and anotheris the electronicedition of the NewOxfordAnnotatedBible

(1995),a hypertext presentation ofthe RevisedStandardVersion that uses
AND Software'sComplex system. Like many commercially availableelec-
tronic texts, the New Oxford AnnotatedBible appearsmore a digitized book
than a true hypertext,though it is nonethelessvaluablefor that. Readerscan
supplement the biblical text with powerful searchtools and various indices,
including ones for Bible and annotation topics, and substantial supplemen-
tary essays,including thoseon approachesto Bible study,literary forms in the
Gospels,and the characteristicsof Hebrewpoetry.The NewOxfordAnnotated
Bible'shypertextuality consistslargely of variant readings (indicated by link

RECONFIGURING icons in the form ofred crosses)and the fact that readerscan add both book-
THE TEXT marks and their own annotations.
A more elaborateform of hypertexluality appearsin the earlier CD Word:
The InteractiveBibleLibrary,which a team basedat DallasTheological Semi-
nary createdusing an enhancedversion of Guide"t. This hypertext Bible cor-
pus, "intended for the student, theologian,pastor,or lay person" rather than
for the historian of religion, includesthe King ]ames,New International, New
American Standard,and RevisedStandardversions of the Bible, as well as
Greektexts for the New Testamentand Septuagint.Thesematerials are sup-
plementedby three Greeklexica,two Bible dictionaries,and three Bible com-
mentaries (DeRose,CD Word,L, 117-26).Using this system,which storesthe
electronic texts on a compact disc, the Bible reader can juxtaposepassages
from different versions and compare variants, examine the original Greek,
and receiverapid assistanceon Greek grammar and vocabulary.
A similar kind of corpus that usesa more sophisticatedhypertext system
is Paul D. Kahn'spioneering ChineseLiteratureIntermedia web, which offers
different versionsof the poetry of Tu Fu (772-7701,ranging from the Chinese
text, Pin-yin transcriptions,and literal translationsto much freer onesby Ken-
neth Rexroth and others. ChineseLiteraturealso includes abundant second-
ary materials that support interpreting Tu Fu's poetry. Like CD Word.,Kahns
Intermedia corpuspermits both beginning and advancedstudentsto approach
a canonicaltext in a foreign languagethrough various versions,and like the
hypertext Bible on compact disc, it also situatesits primary text within a net-
work of links to both varying translations and referencematerials.
Beforeconsidering other kinds of hypertext,we should note the implicit
justifications or rationales for these two successfulprojects. CD Word pres-
ents its intended readerswith a particularly appropriatetechnologicalpres-
entation of the Bible becausethey habitually handle this text in terms of brief
passages-or, as writers on hypertext might put it, as if it had "fine granu-
larity." Becausethe individual poems of Tu Fu are fairly brief, a body of them
invites similar conversionto hypertext.

to the CD Word Bible and the ChineseLiterature

Theln MemoriomWeb Web, which support study chiefly by electronically linking
multiple parallel texts, the In Memoriamweb (Figure 8), an-
other Intermedia corpus createdat Brown University and since published in
Storyspace(Figure 9) after an extensiveexpansionby fon Lanestedtand me
(EastgateSystems,1992),useselectroniclinks to map and hencereify a text's
internal and externalallusions and references-its inter- and intratextuality.2
F6 410:24:g lFl
lSlh h.sru ovow.ddir.cidu)
tuFS 410:25:351991
IN MEMORIAM MonF.b 4 l0:26:01l99l
t.. t
ln Memorian; MohFS4 l0:26:06l99l
E An overyiewoflinked Ft
hdividud sstiore imaoes
- and notifsG P L lSJrr rmrEeru ov
M d F . b4 1 0 : 2 6 : @
hrE lffi Ern v"' r
M o n F .4b1 0 : 2 6 : 1 19 9 1
oflnk.d h4.s Fl
hcr(scc pdfd s.dions)E lEr m.g"4 w
[ldfonto TRrylon's E MonF.b 4 l0:26:23l99l
Ern m.. r
bFeb 410:26:291991
Bh M.mt te
Monfeb 4 l0 26:32l99l
lEr mre.s ov
MonFd4 10126:45 l99l
lilN dEMov
In Memoriam hFeb 410:26r491991
SEtion? (compared to 119) Scction llt Fl
lSlrv rmaqeru
E E E Monreb 4 t0126:5e
Dark house, by which once more I stand, D@8, where myhsad was u*d lobeal
Here in the lontunlovely streel, So quickly, nol as one lhat weep
I @me once horej lhe citysleepEj
D@s, where my hean wa$ used tobeal F* 4 lO:21 20 1991
E I smll lhe meadow in tbe streetl
So quickly, wailintfDrd hand,
I hea!a cbirp ofbirdsj I s
A hand that canbe clasp'd no hcc - Betwixt ihebhck flonts lohBwiihdnwn
Behold me forl cannot sleep, A litht-blue lane of early dawn,
And lhinl of.arlydalE 6nd lh*,
And like a Builtythin8l (reep
Al earliest mominSto the door, And bless lhee, ior lhy lipr an bland,
And briSht the fliendship of thine eye;
He ir not hEjbut faraway And tn mythou8hls with s.ane a sith
The noi* of Iife begns again, E
I take th. Fre$uE ofthine hand.
And Ehasllythm' the drialinthin
On thebald streel bnai(s lhe blank day. E E
(Unr b ,edimr lte<eding 6nd toudhg)
(bnr ta pht.di4 and laudq r.di6r)

Figure8. The Original lntermediaVersionof TheIn Memoriam Web.ln this snapshotof a typicalscreenduring a sessionon Inter-
media, the active document, In Memoriam, section 7 ("ln Mem 7"), appears at the lower left center ofthe screenwith a darkened

striP acrossits top to indicate its status. Using the capacitiesofhypertext to navigatethe poem easily,a readerhas juxtaposed sec-

tions I l9 and 7, which echo and completeeachother.ln Memoriamoverview(lN M EM OV), which appearsat the upper left, is a
graPhicdocumentthat seryesas a sitemap:it organizeslinked materialsunder generalizedheadings,such as "CulturalContext:

Victorianism,"or (lmages and Motifs." The ln Memoriarflimageryoverview(ulM lmageryOV"), a secondvisualindexdocument,

overlies the right border for the entire poem. On the right appearsthe Web View,which the system automatically createsfor each

documentas the documentbecomesactiveeither by beingopened,or, if it is alreadyopen on the deshop, by beingclickedon. In

contrast to the hierarchicallyorganizedoverviewsthe author creates,the Web View shows titled icons representingall documents

connectd electronicallyto the activedocument, here section 7 ofthe poem. Touchingany link markerwith the arrow-shapedcursor
darkensthe iconsrepresentingthe documentslinkedto it; in this case,the readerhasdarkenedthe markerabovethe phrase(com-
paredto I 19" and therebydarkenedicons representingboth the text ofsection 7 and a studentessaycomparingit to sectionI 19.
tlr tdttt*gl&M{)t t
n tl #eesf .rr
.4i. .fr " _jEl*--]
In Memoriam ffi
Sectioni ik*,
, S {a t} *! f'.t6rlsiis8 *
I soi
T( S(tion 6
revq 4 &i']f t r{}a
:A ln sem*rdft :

And \t Thi In Memoriam

Art Settion 7
nut, anf,,

Io follo*
rddd a. euu.
lng A hand that canbe cl Time and Hour
Alfind Tenny*on
: c'e4r1
- rrrl tseholdryr for I can
(hr*n*logy : L i li,i And like a gurltv th
; caH l. At earlhst fiDminfl t{}
Artislir N.*lation* t'a'n n {LL r\ i,ldri*lJ uv
. d,:".r':
\ti{ual arls : . - The Retigion in
: f,ngland 0v*ryiew '-
: e f} * *s{ rhir {an ,* {!*ss, i
Annotation to li Memoriam'
Rkks 871c{rmparesI 3 with L 9 ot virt$r Htlur$"
i Tennvsqn's"If I wer* hvsd", rvhich rc$ "e{rrr Tfurp,and ieach nx, truny yeir$"
prebabh'rvrillento llenru llaliam "Cl :
hour For private sormwk barren
hard-in-hrndwrth thee I le aho drav
compariton with I,r Prinrrs"c6t 1S8:"lt '?ll was good
that Tinx could bring"
handsro latelvrlasut rvith vours." I : "ihe
'Thefkh of tinr"
hrl* draws nearlhe birth of Christ"

Figure 9. The StoryspaceVersion of The In Memoriom Web. Readerscan make their way through this body of interlinked docu-

ments in a number ofways. One can proceed by following links from principal overviews,such as that for the entire web (at \fi1,

religion in England (lower right), or individual motifs-in this casethat for tim e (middle right). One can also explorethe folderlike
structure ofthe Storyspaceview (upper rightl, which can contain a dozen or more layers,or one can follow links from individual

sections ofthe poem. This screen shot indicates how multiwindow hypertext systems, such as Storyspace,Intermedia, and Mul-

ticosm, enableauthorsto fix the locationof windows,therebypermitting one to arrangethe screenin ways that help orient the

reader.Readerscan easily move between parts ofthe poem and commentary on it.

Tennyson'sradically experimental ln Memonam provides an exemplifi-

cation of the truth of Benjamin's remark that "the history of every art form
shows critical epochs in which a certain art form aspires to effects which
could be fully obtained only with a changedtechnical standard,that is, to say,
in a new art form" (Illuminations, 2371.Another manifestation of this prin-
ciple appearsin Victorian word-painting, particularly in the hands of Ruskin
and Tennyson,which anticipatesin abundant detail the techniques of cine-
matography.Whereasword-painting anticipatesa future medium (cinema)

T. 0 by using narrative to structure description, In Memoriam anticipates elec-
tronic hypertextuality precisely by challenging narrative and literary form
basedon it. Convincedthat the thrust of elegiacnarrative,which drives the
reader and the mourner relentlesslyfrom grief to consolation,falsified his
own experiences,the poet constructeda poem of 131fragments to commu-
nicate the ebb and flow of emotion, particularly the way the aftershocksof
grief irrationally intrude long after the mourner has supposedlyrecovered.
Arthur Henry Hallam's death in 1833forced Tennyson to question his
faith in nature, God, and poetry. In Memoiam revealsthat Tennyson, who
found that brief lynics best embodied the transitory emotions that buffeted
him after his loss,rejectedconventionalelegyand narrativebecauseboth pre-
sentedthe readerwith a too unified-and hence too simplified-version of
the experiencesofgriefand acceptance.Creatingan antilinear poetry offrag-
ments, Tennyson leads the reader of In Memoiam from grief and despair
through doubt to hope and faith; but at each step stubborn, contrary emo-
tions intrude, and one encountersdoubt in the midst of faith and pain in the
midst of resolution. Instead of the elegiacplot of "Lycidas,"'Adonais," and
"Thyrsis," In Memoiam offers fragments interlacedby dozensof imagesand
motifs and informed by an equal number of minor and major resolutions,
the most famous of which is section 95's representation of Tennyson'scli
mactic, if wonderfully ambiguous, mystical experienceof contactwith Hal-
lam's spirit. In addition, individual sections,like 7 and 1!9 or 28,78, andL04,
variously resonatewith one another.
The protohypertextuality of In Memoiam aromizesand dispersesTen-
nyson the man. He is to be found nowhere, exceptpossibly in the epilogue,
which appearsafter and outside the poem itself. Tennyson, the real, once-
existing man, with his actual beliefs and fears, cannot be extrapolatedfrom
within the poem's individual sections,for each presentsTennyson only at a
particular moment. Traversingthese individual sections,the reader experi
encesa somewhatidealizedversion of Tennyson'smoments of grief and
recovery.In Memoiam thus fulfills Paul Val6ry's definition of poetry as a
machine that reproducesan emotion. It also fulfills another of Benjamin's
observations,one he makes in the courseof contrastingpainter and camera-
man: "The painter maintains in his work a natural distancefrom reality,the
cameramanpenetratesdeeplyinto its web. There is a tremendous difference
betweenthe pictures they obtain. That of the painter is a total one, that of the
cameraman consists of multiple fragments which are assembledunder a
new law" (Illuminations, 233-34). Although speaking of a different infor-
mation medium, Benjamin here capturessome senseof the way hypertext,

RECONFIGURING when comparedto print, appearsatomized; and in doing so,he also conveys
THE TEXT one of the chief qualities of Tennyson'santilinear, multisequential poem.
The In. Memoiamweb attemptsto capturethe nonlinear organization of
the poem by linking sections,such as 7 and 119,2 and 39,or the Christmas
poems, which echo acrossthe poem to one another. More important, using
the capacitiesof hypertext, the web permits the reader to trace from section
to section severaldozen leitmotifs that thread through the poem. Working
with section 7,for example,readerswho wish to move through the poem fol-
lowing a linear sequencecan do so by using links to previous and succeeding
sections,but they can also look up any word in a linked electronicdictionary
or follow links to variant readings,critical commentary (including a compar-
ison of this sectionand 119),and discussionsof the poem's intertertual rela-
tions. Furthernore, activating indicated links near the words dark, house,
hand, andguilty producesa choiceof severalkinds of materials.Choos-
inghand instantly generatesa menu that lists all the links to that word, and
these include a graphic directory of In Mernoiam's maior images, critical
commentary on the image of the hand, and, most important, a concordance-
like list of each use of the word in the poem and the phrase in which it ap-
pears;choosing any one item in the list producesthe linked document, the
graphicoverviewof imagery a critical comment, or the full text of the section
in which a particular use of hand appears.
Using the capacitiesof Intermedia and Storyspaceto join an indefinite
number of links to any passage(or block) of text, the reader moves through
the poem along many different axes. Although, like the previously men-
tioned hypertext materials, the In Memoriarnwebcontains referencemateri-
als and variant readings,its major difference appearsin its use of link paths
that permit the reader to organize the poem by means of its network of leit-
motifs and echoing sections.In addition, this hypertext presentationof Ten-
nyson'spoem also contains a heavily linked graphic overviewof the poem's
literary relations-its intertextual relations,sources,analogues,confluences,
and influences-that permits one to read the poem along axesprovided by
sets of links relating to the Bible and to works by thirty-eight other writers,
chiefly poets, including Vergil, Horace, Dante, Chaucer, Shakespeare,and
Milton as well asthe Romanticsand Victorians.Although Lanestedt,various
students, and I createdthese links, they represent a form ofobjective links
that could have been createdautomatically by a full-tert search in systems
such as Microcosm. Here, as in other respects,this web representsan adap-
tive form ofhypertext.
In contrast to adapting texts whose printed versions alreadydivide into

3.0 sectionsanalogousto lexias, one may, in the manner of Barthes'sffeatment
of "sarrasine" in S/2, impose one's own divisions on a work. Obvious ex-
amples of possible projects of this sort include hlpertert versions of either
"sarrasine" alone or of it and Barthesk S/2. Stnrt Moulthrop's version of
ForkingPaths:An lwteractionafierJorgeLuisBorges(1987)adaptsBorges's"Fork-
ing Paths"in an electronicversion that activatesmuch of the work's potential
for variation (seeMoulthrop, "Readingfrom the Map"). Other fiction that ob-
viously calls for translation into hlpertext includes Julio Cortiza{s Hopscotch
and Robert Coover's"The Babysitter."
These instances of adaptivehypertext all exemplify forms of transition
betweentexhralityand hypertexruality.In addition,works originally conceived
for hypertextalreadyexistaswell. Thesewebselectronicallylink blocksof text,
that is, lexias, to one another and to various graphic supplements, such as
illustrations, maps, diagrams, and visual directories and overviews,some of
which are foreign to print technology.In the future there will be more meta-
texts formed by linking individual sectionsof individual works, although the
notion of an individual, discretework becomesincreasinglyundermined and
untenablewithin this form of information technology,asit alreadyhaswithin
much contemporary critical theory. Such materials include hypertextual
poetry and fiction, which I shall discusslater in this volume, and the already
abundant World Wide Web equivalentsof scholarlyand critical work in print.
One of the first such works in this new medium-certainly the first on
Intermedia-was Barry J. Fishman's"The Works of Graham Swift: A Hyper-
text Thesis," a 1989 Brown University honors thesis on the contemporary
British novelist. Fishman'sthesis takesthe form of sixty-twolexias,of which
fifty-five are text documents and sevendiagrams or digitized photographs.
The fifty-five text documents he created,which range from one-half to three
single-spacepages in length, include discussions of Swift's six published
bookJength works, the reviewseachreceived,correspondencewith the nov-
elist, and essayson themes, techniques, and intertexfual relations of both
eachindividual book and Swift'sentire oeuvreup to 1989.Although Fishman
created his hypermedia corpus as a relatively self-contained set of docu-
ments, he linked his materials to severaldozen documents alreadypresent
on the system,including materialsby faculty members in at leastthree differ-
ent departments and comments by other students. Since Fishman created
his web, it has grown as many other students addedtheir own lexias,and it
moved first to Storyspaceand, more recently,to the World Wide Web,where
it constitutesan important part of a web containing materials on recent An-
glophone postcolonialand postimperial literature.
I have been describing new kinds ofdiscursive prose, for at
New Formsof Discursive the very least hypertext enablesnew forms of the academic
essay,book review and thesis. More than a decade'swork by
thousandsofscholars using the Internet has shown that these
and Weblogs academicgenrescantake three basicforms. At their simplest,
the author simply placesa text without links into an HTML
template that includes navigation links. As Peter Brusilovsky and Riccardo
Rizzo have pointed out in a prize-winning conferencepaper, a great deal of
current academicwriting for the Webfollows this model, which doesnot take
advantageof the possibilities of hypertext.
In a secondkind of hypertext prose,the author createsa document with
links to documents on the same as well as on other websites.In essencethis
means, as I urge my students,that we must write with an awarenessthat we
are writing in the presenceof other texts. These other texts may support or
contradictour argument, or some of them may serveasvaluableannotations
to it. For example,my review in The Vicloian Webof foseph Bizup's Monu-
factuing Culture:Vindicationsof Early Victoion Industry(2Q03)containsmore
than a dozen links to on-site materials about authors, novels,and historical
events. In contrast, a review of Dale H. Porter's The ThamesEmbqnkment:
Environrnent,Technology, and.Societyin VictorianLondon(1998)contains few
links to existing lexiasbut more than a dozen to brief samplepassagesfrom
Porter'sbook on topicsincluding Oxford in 1850,the invention of urban green
spaces,civil engineering as a profession,and Victorian wagesfor skilled and
unskilled labor. In addition to making links to these brief lexias from the
main text, I also appendeda list of them plus a few others at the end of the re-
view. Links to Porter'smaterials were also addedto the sitemapsfor science,
technology,social history and economics. Both author and publisher were
delighted with this approachto reviewing becausethey believed,correctly,
that it spreadword about Porter'swork in a particularly effectiveway.
The third kind of hlpertext essay,as we have seenfrom Fishman'shon-
ors thesis, takesthe form of a set of networked documents,createdeither to
stand alone,as it largelywas, or to take part in a larger web. Either way,an au-
thor wanting to conceiveof an argument in terms of networked documents
can write a conciseessayto which she or he links a wide rangeof supporting
evidence. Readerscan then choosewhat areas they want to investigatein
greaterdepth, and these auxiliary materials thereby becomeparatexts,easily
accessibleadd-onsto the lexia one is currently reading.
The Weblog,or blog, asit is commonly known, is anothernew kind of dis-
cursive prose in digital form that makes us rethink a genre that originally

3.0 arose when writing took the form of physical marks on physical surfaces.
Blogging, the latest Internet craze,has major importance for anyone inter-
ested in hypertext becauseone form ofit provides the first widely available
means on the Web of allowing the activereader-authorenvisagedby Nelson,
van Dam, and other pioneers.Blogstake the form of an online journai or
diary most commonly written by a single person, and, like paper journals
and diaries, they present the author'swords in dated segments.Unlike their
paperpredecessors,they presententries in reversechronologicalorder.They
can employ two different forms of hypertextuality.First, unlike discussion
lists, all bloggers can link chronologically distant individual entries to each
other, thereby "allowing readersto put events in context and get the whole
story without the diarist having to erplain again" (McNeill, 30).The second
form of hypertexruality occurs only in those blogger systems that permit
readersto comment on entries. Here's how it works: encountering a com-
ment on my son'sblog about the legality of China's revoking the patent for
Viagra, I clicked on the word "Comment" and thereby opened a form into
which I pastedmy remarks plus a few sentencesfrom Vincent Mosco'shis-
torical accountof nineteenth-centuryAmerican information piracy (which I
use in chapter 8, below). Before I could submit it, the form containing my
comment requestedthree bits of information: my e-mail address(required),
my name ("real is appreciated"),and the URL of my website,if any (optional).
Returning to the blog, I discoveredthat the zero next to "comment" had
changedto "1." Clicking on the word "comment" opened a document con-
taining what I had just submitted plus a spacefor other people to add their
responsesto my comment. There are more than a dozen kinds of blogging
software,and many, including b2Evolution, MoveableType,and Serendipity,
havethe Trackbackfeature that also allows bloggersto post links back to the
site of anyonewho commented on them.
Visually,blogstakemany forms, but most haveseveralcolumns,the widest
dedicatedto the dated entries and one or more others containing links to
archivesof older entdes, personal information, associatedblogs, and major
topics of interest to the person who owns the site. Many contain images and
evenvideo, and most contain a personal statementor description of the site,
which may be very brief, such as "i stashthings here so i can find them again.
sometimesother peoplecome visit. or so the tracker implies." Somebloggers
maintain two or more sites, one devotedto their academicor professional
interestsand another to their personaldiary. Many usersprefer reading blogs
of friends and family members to e-mail, becausethey haveno spam, and for
that and other reasonsthey havebecome enormously popular. Many sophis-

RECONFIGURING ticatedbloggersuse specialsoftwareto subscribeto their favoritesites,thereby

THE TEXT ensuring that they know when something new is posted on them. RSSand
Atom feed represent the two main standards for such subscription tools.
WhereasRSS sendsthe subscriberonly the headlineof a new blog, Atom feed
addsa summary and includes its links as well. So-calledfeed readersobtain,
organize,and displaymaterialsfrom large numbers of websitesand Weblogs.
Blogs themselves can take as many forms and have as many principal
organizingideasasother forms of websites,but the majority of bloggers(and
also the most intense ones) are highly skilled computer users whose profes-
sional activities demand technical information. According to www.techno-
rati.com, which claims to have watched 3,145,522Weblogs (entries) and
tracked 456,L40,934links,Slashdot,a famous multiuser techie site, proved
most popular,with 12,904 blogs and27,041,links.Fark, another popular blog
that claims to have receivedover 350 million pageviewsin 2003, offers a
digest of the news. Each entry takes the form of a one-line summary that
links to sourcesaccompaniedby little icons containing comments, such as
"amusing," "cool," "obvious," "scary,"and "stupid." An example of this last
categoryon l:uJyt7,2004: "Martha Stewartcomparesherself to Nelson Man-
dela," which links to its source, CNN. This item provoked ninety-two com-
ments that statedalmost everypossibleopinion about the criminal case.
NGOs and other organizationsdealingwith economicand political issues
haveblogs,aswell asthoseconcentratingon specificdiseases, suchasAIDS,
asthma,tuberculosis,and cancer.Diseaseblogstake two very different forms,
the first createdby organizationsthat work toward the prevention and cure of
a particular illness, the secondwritten by people suffering from the disease;
some of their intensely personal online diaries have acquired large follow-
ings. On a lighter note, many hobbiesor leisure activities,such as model rail-
roading and gardening, have blogs, though I found very few forhunting or
flying; I did discoveran Australian one, though, on fear offlying. Perhapsthe
most exLremepersonal-interestblogs involve sexualfetishes, including one
by a nonsmoker who finds pictures of women smoking in public erotic.
Phil Gyford'stranslation of Samuel Pepys'sfamous seventeenth-century
diary into a Weblog,which exemplifies the way ingenious people find unex-
pectedusesfor computer genres,createsa new form ofparticipatory scholar-
ship.As Gyfordexplainedin an interview availablein the online versionof BBC
News (World Edition),"l thought Pepys'diary could make a greatWeblog.The
published diary takesthe form of nine hefty volumes-a daunting prospect.
Readingit day by day on a website would be far more manageable,with the
real-time aspectmaking it a more involving experience."As I am writing this
3.0 on July 22,2004,I find Pepys'sdiary entry for July 2L,I66L,which has been
up at least since the 19th, since a comment on the place name "Sturtloe" by
Mark Ynys-Monpoints out that it has changedto "Stirtloe,"and a secondsub-
mitted the next day by Vincente points out that "sam on his nag could have
had a nice ride down by the Ouse" and provides a link to a map of the area.
Gyford has tried to ensure that most of the annotations are helpful by advis-
ing contributors: "Before posting an annotation pleaseread the annotation
guidelines. If your comment isnt directly relevant to this page, or is more
conversational,try the discussiongroup." This blog, which contains approxi
mately two hundred words, has thirteen in-text links, one by Gyford himself
(a cross-referenceto another Diary entry\ and others leading to one or more
readers'comments.The entryfor 1uIy21.,1661, dominatesthe screen,though
if one scrollsdown one can find entries back to the 13th.At the top of the win-
dow Gyford has provided links to an introduction to the Diary,background
information, archives,and a summary ("Readthe Story SoFar") for first-time
readers.At the top right a searchwindow appearsand below it a column of
links to seventeencategoriesof backgroundinformation starting with art and
literature, including food and drink, and ending with work and education.As
this brief description makes clear, Gyford has not only made an appropriate
Webtranslation of a classictext but he has also contributed importantly to the
creation ofa new form ofpublic, collaborativeonline scholarship.Two inter-
esting points: (1) Gyford's name does not appearon the main lexias of the
'About this site,"one can find it, and following a
blog, though if one explores
link to his personal site, one can learn about his fascinating careeras an art
student, professionalmodel maker, systemadministrator, and web designer.
(2) This elaboratescholarly proiect, which one expectsthat any Web-sawy
undergraduateand graduate student will use, exists complelely outsidethe
Academy.Pausefor a moment and think about the implications of that.
Many special-interestblogs, like some famous ones by AIDS and cancer
victims, exemplify the Intemet versionof the personaljoumal or diary.Laurie
McNeill's excellentarticle on the blog aspersonaldiary (aboutwhich I learned
from Adrian Miles's blog) points to "an unparalleled explosion of public life
writing by private citizens. By March 2002,more than 800,000blogswere reg-
'Bloggerblogs' were created
istered on the Net; in lvly 2002an averageof 1.5
per minute (blogger.com6 Aug. 2002)" (32).When I checkedtwo yearslater,
some hosts of blogs boastedmillions of users, estimatesof the total ranging
between two and eight million, though one commentator pointed out that
only a quarter of peoplewho begin blogs keep them going.
Googlingthe phrase"how many bloggers,"I receivedthe URLs for several

RECONFIGURINC siteswith someof the information for which I waslooking, but among the top
THE TEXT entries appearedone from the blog of a young woman enumerating her sex-
ual experiences(I hadn t meant that "how many"!). Her entry,which appeared
as a separatelexia, containedlinks to another blog with similar material, and
when I clicked on the link in the original blog labeled "Home," I found a site
whosecontentsreminded me ofthe HBO televisionshow "Sexandthe City"-
more for the comedy,though, than the sex.Although the blogger identifies
herself only as "Blaise K. ," she includes enough personal information, in-
cluding photographsand the asserLionthat she is black and fewish, that her
anon;rnity doesnt seem very well protected. I assume the blogger intends
the site for her friends, but Googlemistakenly brought me there, as it may
well bring her parentsand employers.It is verydifficult to maintain this kind
ofpublic privacy.
McNeill points out that such sites "often reinforce the stereotypeof the
diary as a genre for unbridled narcissism" becausethey assume that other
people care about what bloggers have to say.That narcissism, McNeill ad-
mits, often turns out be justified, for some online diaries receivethousands
of visitors and make their authors famous. They also place the author's re-
marks about private matters in a very public space.In fact, one of the most
interesting effectsof blogging lies in the way it unsettlesour accustomedbor-
ders betweenthe private and public spheres."In their immediacy and acces-
sibility, in their seeminglyunmediated state,Web diaries blur the distinction
between online and offiine lives, 'virnral reality' and 'real life,' 'public' and
'private,' and
most intriguingly for auto/biography studies,betweenthe life
and the text" (McNeill, 25).Thoseblogs that acceptcomments allow McNeill
claims, the "readerof an online diary" to participateactively

in constructing
heor shetakeson in the
Thoughactiveandevenintimate, however,
remains vir-
tual,disembodied.Theconfessor staysbehindthe "grille"of the Internet,allowing
the diarist*and the reader-the illusionof anonymitynecessary for "full" self-
exposure.JanetMurraynotesthat"somepeopleputthingson theirhomepage. . .
thattheyhavenottoldtheirclosest friends.Theenchantmentofthe computer cre-
atesforusa publicspacethatalsofeelsveryprivate inti
and mate" (99). . . Fortheon-
meansthatthediaristhasboth.joinedandcreated com-
actsthatinformthetextsheor shewillproduce.

Many bloggers dont in fact allow comments, or else they screen them,
and some intend their online diaries solelyfor a circle of friends and control
accessto them by using passwords.Nonetheless,once an entry goesonline,

3.0 Internet searchtools can bring it to the attention of Web surfers. The edges
of a blog, like the borders of any document on the Internet, are porous and
provisional at best. Most of the time when we considerthe way digital media
blur the borders of documents,we mean that links and searchtools limit the
power ofauthorship. In blogs we encounter a new prose genre that also un-
settlesour long-standing assumptions about public and private.

Writing about hypertext in a print medium immediately pro-

Problemswith Terminology: ducesterminological problems much like those Barthes,Der-
rida' and others encounteredwhen trying to describea textu-
what ls the object we Read,
' and
ality neither instantiated by the physical object of the printed
What ls a Textin Hypertextl book nor limited to it. Since hypertext radically changesthe
experiencesthat reading,witing, and turt signify, how, with-
out misleading, can one employ these terms, so burdened with the assump-
tions of print technology,when referring to electronic materialsl We still
read accordingto print technology,and we still direct almost all of what we
write toward print modes of publication, but we can alreadyglimpse the first
appearancesof hypertexfualityand begin to ascertainsome aspectsof its pos-
sible futures. Terms so implicated with print technologynecessarilyconfuse
unless handled with ereat care.Two exampleswill suffice.
An instanceof thJ kina of probl"rr, *" f.ce appearswhen we try to decide
what to call the object at which or with which one reads. The object with
which one readsthe production ofprint technologyis, ofcourse, the book, or
smaller print-bearing forms, such as the newspaperor instruction sheet;for
the sakeof simplicity I shall refer to "book" as the most complex instance
ofprinting technology.In our culture the term bookcanrefer to three very dif-
ferent entities-the object itself, the text, or the instantiation of a particular
technology.Calling the machine one uses to read hypertext an "electronic
book," however,would be misleading, since the machine at which one reads
(and writes, and carries out other operations,including sending and receiv-
ing mail) does not itself constitute a book, a text: it does not coincide either
with the virtual text or with a physical embodiment of it.
Additional problems arise when one considersthat hypertext involves a
more activereader,one who not only choosesher reading paths but also has
the opportunity (in true read-write systems)of reading as someonewho cre-
atestext; that is, at any time the person reading can assumean authorial role
and either attachlinks or addtext to the text being read.Therefore,a term like
reader,suchas some computer systemsemploy for their electronicmailboxes
or messagespaces,doesnot seemappropriateeither.3

RECONFIGURINC One earlier solution was to call this reading-and-writing site a worksta-
THE TEXT tion by analogyto the engineer'sworkstation, the term assignedto a relatively
high-poweredmachine, often networked with others, that in the early 1990s
had far more computing power,memory and graphiccapacitiesthan the per-
sonal computer. However, becauseworkstationseems to suggestthat such
objectsexist only in the workplaceand find applicationonly for gainful labor
or employment, this choice of terminology also misleads. Nonetheless,I
shall employ it occasionally,if only becauseit seemscloserto what hypertext
demands than any of the other terms thus far suggested.The problem with
terminology arises,as has now becomeobvious,becausethe roles of reader
and author change so much in hypermedia technology that our current
vocabularydoesnot havemuch appropriateto offer.
Whatever one wishes to call the reading-and-writing site, one should
think of the actual mechanism that one will use to work (and play) in hyper-
text not as a free-standingmachine, like today'spersonal computer. Rather,
the "object one reads"must be seenas the entrance,the magic doorway,into
the docuverse,since it is the individual reader'sand writer's means of partic-
ipating in-of being linked to-the world of linked hypermediadocuments.
A similar terminological problem appearsin what to do with the term
turt, which I have already employed so many times thus far in this study.
More than any other term crucial to this discussion,turthas ceasedto inhabit
a single world. Existing in two very different worlds, it gatherscontradictory
meanings to itself, and one must find some way of avoiding confusion when
using it. Frequently,in trying to explain certain points of difference, I have
found myself forced to blur old and new definitions or have discoveredmy-
self using the old term in an essentiallyanachronisticsense.For example,in
discussingthat hypertext systemspermit one to link a passage"in" the "text"
to other passages"in" the "text" aswell as to those "outside" it, one confronts
preciselysuch anachronism.The kind of text that permits one to write, how-
everincorrecdy,ofinsides and outsidesbelongsto print, whereaswe arehere
consideringa form ofelectronic virhral textuality for which thesealreadysus-
pect terms have become even more problematic and misleading. One solu-
tion has been to vse text as an anachronistic shorthand for the bracketed
material in the following expression: "lf one were to transfer a [complete
printed] text (work),say,Milton's ParadiseLost,intoelectronicform, one could
link passageswithin [what had been] the [original] text (Milton's poem) to
eachother; and one could alsolink passagesto a wide range of materials out-
side the original text to it." The problem is, ofcourse, that as soon as one con-
verts the printed text to an electronic one, it no longer possessesthe same

3.0 kind oftextuality. In the following pagessuch referencesto text haveto be un-
derstood,therefore,to mean "the electronicversion of a printed text."
The question of what to call "text" in the medium of hypermedia leads
directly to the question of what to include under that rubric in the first place.
This question in turn immediately forces us to recognizethat hypertext
reconfiguresthe text in a fundamental way not immediately suggestedby the
fact of linking. Hypertertuality, like all digital texruality,inevitably includes a
far higher percentageof nonverbalinformation than doesprint; the compar-
ative easewith which such material can be appendedencouragesits inclu-
sion. Hypertext,in other words, to some degreeimplements Derrida'scall
for a new form of hieroglyphic writing that can avoid some of the problems
implicit and thereforeinevitablein Westernwriting systemsand their printed
versions.Derrida arguesfor the inclusion of visual elementsin writing as
a means of escapingthe constraints of linearity. Commenting on this thrust
in Derrida's argument, Gregory Ulmer explains that grammatology thereby
"confronts" four millennia during which anything in languagethat "resisted
linearization was suppressed.Briefly stated,this suppressionamounts to the
denial of the pluridimensional characterof symbolicthought originally pres-
'mlthogram' (Leroi-Gourhan'sterm), or nonlinear writing (picto-
ent in the
graphic and rebus writing)" (AppliedGrammatology,8).Derrida, who asksfor
a new pictographicwriting as a way out of logocentrism, has to some extent
had his requestsansweredin hypertext.N. Katherine Haylesarguesthat dig-
ital text alone, evenwithout links, emphasizesthe visual, because"the com-
puter restoresand heightens the senseof word as image-an image drawn
in a medium as fluid and changeableas water" (26).
Becausehypertext systemslink together passagesof verbal text with
images as easily as they link two or more passagesof text, hypertext in-
cludeshypermedia,and I thereforeuse the two terms interchangeably.More-
over, since computing digitizes both alphanumeric symbols and images,
electronictext in theory easily integratesthe two. In practice,popular word-
processingprograms,such as Microsoft Word, haveincreasinglyfeaturedthe
capacityto include graphic materials in text documents,and, as we shall see,
this capacityto insert still and moving images into alphanumeric text is one
of the characterizingfeaturesof HTML. Linking, which permits an author to
send the readerto an image from many different portions of the text, makes
such integration of visual and verbal information eveneasier.
In addition to expandingtlre quantity and diversity of alphabeticand non-
verbal information included in the text, computer text provides visual ele-
ments not found in printed work. Perhapsthe most basic of these is the cur-

RECONFIGURING sor, the blinking arrow line, or other graphic element that represents the
THE TEXT reader-author'spresencein the text. The cursor,which the user moves either
from the keypadby pressingarrow-markedkeysor with deviceslike a mouse,
rollerball, or trackpad,providesa moving intrusive image of the reader'spres-
ence in the text. Holding the mouse over a footnote number in Microsoft
Word producesthe text in the note in a pop-up window. The reader can also
changethe text by using the mouse to position the cursor betweenthe letters
in a word, say,betweentandhinthe.Pressinga button on the mouseinserts
a vertical blinking line at this point; pressing the backspaceor deletekey
removesthe t. Typing will insert charactersat this point. In a book one can
alwaysmove one's finger or pencil acrossthe printed page,but one's intru-
sion alwaysremains physicallyseparatefrom the text. One may make a mark
on the page,but one'sintrusion doesnot affect the text itself.
The cursor, which adds reader presence,activity, and movement, com-
bines in most previous and extant hypertext systemswith another graphic
element, a symbol that indicatesthe presenceof linked material.aThe World
Wide Web offers severalkinds of changing cursors:the cursor changesfrom
an arrow to a hand when positioned over a linked word or image, and com-
monlyused favascriptschangethe appearanceoflinked objectsupon mouse-
over-when, that is, the user positionsthe mouse overthem. Yetother scripts
produce drop-down menus of links. All these graphic devicesremind read-
ers that they are processingand manipulating a new kind of text, in which
graphic elementsplay an important part.

This description of visual elements of all computer tert re.

VisualElementsin PrintText minds one that print also employs more visual information
than people usually take into account: visual information is
not limited, as one might at first think, merely to the obviousinstances,such
as illustrations, maps, diagrams, flow charts, or graphs.sEven printed tert
without explicitly visual supplementarymaterials alreadycontains a good bit
of visual information in addition to alphanumeric code.The visual compo-
nents of writing and print technologyinclude spacingbetweenwords, para-
graphing, changesof type styleand font size,formatting to indicate passages
quoted from other works, assigning specificlocationson the individual page
or at the end of sectionsor of the entire document to indicate reference
materials (foot notes and endnotes).
Despite the considerablepresenceof visual elements in print text, they
tend to go unnoticed when contemporaty writers contemplatethe nature of
text in an electronic age. Like other forms of change,the expansionof writ-

T. 0 ing from a systemofverbal languageto one that centrally involvesnonverbal
information-visual information in the form of syrnbols and representa-
tional elementsaswell asother forms of information, including sound-has
encounteredstiffresistance, often from those from whom one is leastlikely
to expectit, namely, from those who alreadyemploy computers for writing'
Eventhosewho advocatea changefrequently find the experienceofadvocacy
and changeso tiring that they resist the next stage,evenif it appearsimplicit
in changesthey havethemselvesadvocated.
This resistanceappearsparticularly clearlyin the frequently encountered
remark that writers should not concernthemselveswith typesettingor desk-
top publishing but ought to leavethose activities to the printer. Academics
and other writers, we are told, do not design well; and even if they did, the
argument continues, such activities are a waste of their time. Such advice,
which has recently become an injunction, should make us ask why. After
all, when told that one should not avail oneself of some aspector form of
empowerment,particularly as awriter, one should askwhy.Whatif someone
told us: "Here is a pencil. Although it has a rubber apparatusat the opposite
end from that with which you write, you should not use it. Realwriters don't
use it"l At the very least we should wonder why anyone had included such
capacitiesto do something; experimenting with it would show that it erases;
and very likely, given human curiosity and perversity,which may be the same
thing in certain circumstances, we would be tempted to try it out. Thus a
capacitywould evolveinto a guilty pleasure!
Anyone with the slightest interest in design who has even casuallysur-
veyedthe output of commercial and university presseshas noticed a high
percentageofappallingly designedor obviously undesigned books. Despite
the exemplary work of designers like P.f . Conkwright, Richard Eckersley,
and Glen Burris, many pressescontinue to produce nasty-lookingbookswith
narrow margins and gutters, tl?e too small or too coarsefor a particular lay-
out, and little senseofpage design. Financial constraints are usually offered
as the soledeterminant of the situation, though good design doesnot haveto
produce a more costlyfinal product, parricularly in an age of computer type-
setting. In severalcasesI am awareof,,publishers haveassignedbook design
to beginning manuscript editors who havehad no training or experiencein
graphic design. As one who has been forflrnate enough to have benefited
from the efforts of first-rate,talented designersfar more than I havesuffered
from those of poor ones, I make these observationsnot as a complaint but
as a preparation for inquiring why authors are told they should not concern

RECONFIGURING themselveswith the visual appearanceof their texts and why authors readily
THE TEXT acceptsuch instruction.
They do so in part becausethis injunction clearlyinvolvesmatters of sta-
tus and power. In particular, it involves a certain interpretation-that is, a
socialconstruction-of the idea ofwriter and writing. According to this con-
ception,the writer's role and function is just to write. Writing, in turn, is con-
ceived solely as a matter of recording (or creating) ideas by means of lan-
guage.On the surface,such an approachseemsneutral and obviousenough,
and that in itself should warn one that it has been so naturalized asto include
cultural assumptionsthat might be worth one'swhile to examine.
The injunction "just to wdte," which is basedon this purely verbal con-
ception of writing, obviously assumesthe following: first, that only verbal
information has value, at least for the writer as a writer and probably for the
readeras reader;6second,that visual information has less value. Making use
of such devaluedor lesser-valuedforms of information (or doesvisual mate-
rial deservethe description"real information" at alll) in someway reducesthe
statusofthe writer, making him orher lessof a real writer. This matter of sta-
tus againraisesits headwhen one considersanotherreasonfor the injunction
"just to write," one tied more tightly to conceptionsof division of labor, class,
and status. In this view of things, it is thought that authors should not con-
cern themselveswith matters that belong to the printer. Although troubled by
this exclusion, I acceptedthis argument until I learned that until recently
(say,in the 1930s)authors routinely wanderedaround the typesettingshop at
Oxford University Presswhile their bookswere being set and were permitted
to render advice and judgment, something we are now told is none of our
business, beneath us, and so on. The ostensiblereason for instructing au-
thors to refuse the power offered them by their writing implement also in-
cludesthe idea that authors do not havethe expertise,the sheerknow-how to
produce good design. Abundant papers bybeginningundergraduates and
beginning PC users, cluttered with dissonant typefacesand font sizes,used
to be thrust forward to support this argument, one that we receivetoo readily
without additional information. Now peoplepoint to ugly websitesand blogs.
The fact that beginners in any field ofendeavor do a fairly poor quality job
at a new activity hardly arguesforcefully for their abandoningthat activity.If
it did, we would similarly advisebeginning studentsimmediately to abandon
their attempts at creativeand discursivewriting, at dravingand philosophy,
and at mathematicsand chemistry.One reasonwe do not offer such instruc-
tions is becausewe feel the skills involved in those endeavorsare imDt-rr-

3.0 tant-apparently in contrastto visual ones.Another reason,of course,is that
teaching involvesour livelihood and status.The question that arises,then, is
why is visual information less importantl The very fact that people experi-
ment with visual elementsof text on their computers showsthe obviousplea-
sure they receivein manipulating visual effects.This pleasure suggestsin
turn that by forbidding the writer visual resources,we deny an apparently
innocent sourceof pleasure,something that apparentlymust be castasideif
one is to be a true writer and a correct reader.
Much of our prejudice againstthe inclusion of visual information in text
derivesfrom print technology.Looking at the history of writing, one seesthat
it has a long connectionwith visual information, not leastthe origin of many
alphabeticsystemsin hieroglyphics and other originally visual forms of writ-
ing. Medieval manuscripts present some sort of hypertext combination of
font sizes, marginalia, illustrations, and visual embellishment, both in the
form of calligraphyand that of pictorial additions.
This same prejudice against visual elements appearsin recent suppos-
edly authoritative guidelines for creating websites.jakob Nielsen'sDesigning
WebUsability,for example,advisesweb designersto avoid graphic elements,
particularly for opening screens (homepages),becausethey unnecessarily
consume both bandwidth and screenreal estate.I certainly understand the
reasonsfor such advice.Like many other users in the early daysof commer-
cial sites, I've waited many minutes for the opening screenof a national air-
line's site to download even though I had high-speednetwork access,finally
giving up. Earlyweb designersfound themselvessounderstandablyenthralled
by elaborate graphics and animation that they cluttered sites with nonfunc-
tional elements that consumed important resources.As the airline website
shows,this approachproved disastrousfor commercial applicationsat a time
when most potential customers had slow Internet connections. Obviously,
designersmust balanceeaseof accessagainst visual elements that encour-
agepeople to accessthe site in the first place,but avoiding graphic elements
as a basic design principle doesnt make much sensefor one obvious reason:
images and other graphic elements are the single most important factor in
the astonishing growth of the Word Wide Web. The invention of the image
tag (<img src:"picture.jpg">), which instructs the web browser to placea pic-
ture, icon, or other graphic element within text, made the World Wide Web
immensely appealing,turning it into a medium rich with visual pleasures.
The embed tag, which placesQuicktimeVR, sound, and video in an HTML
document, similarly convertsthe Web to a multimedia platform. Therefore,
whether or not we believeit has an identifiable loeic-a Mcluhan-esque mes-

RECONFIGURINC sagein the medium-the Webcertainlyis significantlypictorial.Recommend-

THE TEXT ing that one should not use static or moving images in a medium popular-
ized by their very presencetherefore seemsparticularlybizarre.T
This blindness to the crucial visual components of texruality not only
threatensto hinder our attemptsto learn how to write in electronic spacebut
has also markedly distorted our understanding of earlier forms of writing.
In particular, our habits of assuming that alphanumeric-linguistic-text is
the only text that counts has led to often bizarre distortions in scholarlyedit-
ing. As Jerome J. McGann reminds us in The Tertual Condition, "literary
works typically secure their effects by other than purely linguistic means"
(77),alwaysdeploying various visual devicesto do so. Hence leaving such as-
pectsof the text out of consideration-or omitting them from scholarlyedi-
tions-drastically reconfiguresindividual works. 'All poetry,evenin its most
traditional forms, asksthe readerto decipherthe text in spatialas well as lin-
ear terms. Stanzaicand generic forms, rhyme schemes,metrical orders: all
ofthese deploy spatial functions in scripted texts, as their roots in oral po-
etry's 'visual' arts of memory should remind us" (113).One cannot translate
such nonprint and evenantiprint works like that of Blakeand Dickinson into
print without radically reconfiguring them, without creating essentiallynew
texts, texts a large portion ofwhose resourceshave been excised.Although
"textual and editorial theory has heretofore concerned itself almost exdu-
sively with the linguistic codes,"McGann urges, "the time has come, how-
ever,when we haveto take greatertheoreticalaccountofthe other coding net-
work which operatesat the documentary and bibliographical level of literary
works" (78). Once again, as with the scholarly editing of medieval manu-
scripts and nineteenth-centurybooks,digital word and digital image provide
lensesthrough which we can examine the preconceptions-the blinders-
of what Michael foycecalls "the late ageof print" (Of TwoMinds, Lll).

The essentialvisual componentsof all text find perhapstheir

Animated Text firllest instantiation in the form of animated text-text that
moves,evendances,on the computer screen,sweepingfrom
one side to the other, appearingto move closerto readersor retreatawayfrom
them into a simulated distance.Text animation, which has becomevery pop-
ular in recent digital poetry,derivesfrom the nature of computer text, which
takesthe form of code.Until the developmentof digital texruality,all writing
necessarilytook the form of physicalmarks on physical surfaces.With com-
puters, writing, which had always been physical, now became a matter of
codes-codes that could be changed, manipulated, and moved in entirely

3.0 new ways. "Change the code, changethe text" becamethe rule from which
derivethe advantagesof so-calledwordprocessing(which is actuallythe com-
position, manipulation, and formatting of text in computer environments).
The advantagesof word processingover typewriters becameso immediately
obviousthemselvesin businessand academiathat dedicatedword processors
and then personalcomputers swiftly made tlpewriters obsolete."Changethe
code,changethe text" aiso produces the "styles" option in word-processing
software,such as Microsoft Word, which permits a writer to createand deploy
stylescontaining font, type size,and rules for varioustext entities (paragraph,
inset quotation, bibliography,and so on). By simply highlighting a word, sen-
tence,or paragraph,the user of such softwarecan easilymodify the appear-
ance of text, whether it is intended to remain on screen or issue forth as a
printout or as a t)?eset book.
The sametext-as-codethat permits word processingalsopermits moving
words. In its simplest form, text animation simply involvesmoving the text on
screena line at a time, essentiallydispensing the poem at a rate determined
by the author. Kate Pullinger and Talan Memmot's elegant Branded.(2003;
Figure 10)functions in this way. Pearl Forss'sAuthorship(2000),which com-
bines sound and text animation, exemplifiesthe use of this kind of animated
text to create experimental discursive writing for e-space.First, to the ac-
companiment of a driving drumbeat, the words "What is" appearin white
block letters against a black screen to which are quickly added in the red-
orange words "an authorl" The question mark then danceson screen,after
which the sentencemovesdownward as words of Roland Bartheson au*ror-
ship move on screen;these in turn are replacedby Forss'spronouncements
about authorship; then in greenappearthe words "what matterswho's speak-
ingl"-a question immediately identified as having been asked by Beckett
(whosename, in white, undulates on screen).Next, an image of a rose fills
the entire screen, and on top of it appear many pink letters, which soon
arrangethemselvesto state,'A roseby any other name would smell as sweet,"
an assertionimmediately challengedby the question (in green) "or would it?"
And this screenis rapidly obliteratedby the appearanceof imagesof theorists
on authorship and coversof their books,all of which build to a collage.What
I've describedmakes up the opening section or movement, severalof which
follow, eachpunctuated by the same assemblingcollage.
Such tert animation, often accompaniedby sound, appearsmore fre-
quently in digital literary art than in discursiveor informational projects.For
example,severalof the animated poems onthe DotzeSentits:Poesiacatalona
d'avuiCD-ROM(1996),suchas ]osepPalaui Fabre's"La Noia" and Feliu For-
Figure10. AnimatedText.In KatePullingerand TalanMemmot's elegantanimatedpoem, Bronded,textanimationtakesthe form

of moving the text on screen a line at a time, dispensing the poem at a rate determined by the authors.

mosa's"Ell sort de setaI'aigua," accompanythe sound of the poet'sreading

by movtngwords and phrasesof different sizesand colors acrossthe screen
from top to bottom and from one edge ofthe screento another; words pop
in and out of existence,too, as the text performs itself.8More radical experi-
mentation with animated text appearsin Philadelpho Menezesand Wilton
Azvedo'sInterpoesia:PoesiaHipmed.iaInterativa(1998),in which elements (or
fragments) of both spoken and written words reactto the reader'smanipula-
tion of the computer mouse. Lettersmove,parts ofwords changecolor or dis-
appear,and sounds become layered upon one another as the reader essen-
tially performs the text using the sounds provided.
Moving text on screen,which has only become possible for most users
with the advent of inexpensivecomputing power and broad bandwidth, has
had an effect on digital literary arts almost as dramatic as that of word pro-
cessing on academicinstitutions and the workplace. But are such projects
hypertextual (and doesit matter)l

Figure1I. StephanieStrickland'sVniverse.Both a constellationand poetictext appearagainstthe starrysky as the readermanip-

ulatesthe mouse.

In one important sense,these projects, L1keBrqnded.,appearessentially

ant*rypertextual.If one takeshypertextto be an information technologythat
demands readerstake an active role, then these animated texts enforce the
oppositetendency.In contrastto hypertext,they demand the reader assume
a generally passiverole as a member of an audience,rather than someone
who has some sayin what is to be read.They add,in other words, to the power
of the author-or at least to the power of the text-and deny the possibility
of a more empowered reader. Strickland's Vniverserepresents a compara-
tively rare exampleof text-animationhypermediathat strivesto grant readers
control; it is, however,quite unusual (Figure 11).
If one were to arrange print text, hypertext, video, and animated text
along a spectrum, hypertext,perhaps surprisingly, would take its place clos-
est to print. Readingwritten or printed text, one cannot changeits order and
progression,but becausethe text is fixed on the page,one can leaveit, read-
ing another text, taking notes, or simply organizing one's thoughts, and

RECONFIGURINC return to find the text where one left it, unchanged.The characteristicfixity
THE TEXT of writing, therefore, endows the reader with the ability to processit asyn-
chronously-that is, at the convenienceof the reader.nConsider the differ-
enceof such fixed text from video and animated text: if one leavesthe televi-
sion set to answerthe phone or welcome a guest,the program has moved on
and one cannot retrieve it, unless, that is, one has a digital or analoguecopy
ofit and can replay it. The very greatdifferencein degreeofaudience control
betweenvideo as seenon broadcasttelevision and video viewed from storage
media, such as videotape,DVD, or TiVo, suggeststhat they are experienced
asdifferent media. Still, sincevideo,like cinema, is a temporal form-a tech-
nology that presentsits information in necessarysequence-one generally
has to follow long patches of the story or progtam in its original sequence
to find one's place in an interrupted narrative. Animated text, in contrast,
entirelyconlrolsthe reader'saccessto information at the speedand at the time
the author wishes. One could, it is true, replay the entire animated text, but
the nature of the medium demandsthat the minimum chunk that can be ex-
amined takes the form of the entire sequence.
Another form of moving text appearsin the timed links of Stuart Moul-
throp's Hegirascope,linksthat dramaticallyaffect the reader'srelation to text.
The reading experienceproducedby thesetimed links contrastssharplywith
that possiblewith wdting, print, and most hypertext.Sincethe text disappears
at timed intervals outside the reader'scontrol, the characteristicfixity of writing
disappearsasthe document being readis replacedby another.Someof the re-
placementshappenso quickly that this text enforcesrapid reading,preventing
any closereading, much less leisurely contemplation of it. Michael foycefa-
mously assertedthat "hypertextis the revengeof text upon television" (OfTwo
Minds,47, tll), by which I take him to mean that hypertext demands active
readersin contrastto television'srelativelypassiveaudience.Theseexamplesof
animated(or disappearing)text in contrastappearto be extensionsof television
and film to encompassand dominate text, or in Joyce'sterms, the revengeof
television (broadcastmedia) on hypertext.This is not necessarilya bad thing,
anymore than cinemais worsethan print narrative.Animated text,like cinema
and video, existsas an art form with its own criteria. It's just not hypertext.

Not all animated alphanumeric text, it turns out, is nonhyper-

Stretchtext textual. In fact,Ted Nelson'sstretchtext,which he advancesas
a complement to the by-now standard node-andJink form,
producesa truly reader-activatedform.1oExceptfor researchersworking with
spatial hypertext, most students of hypermedia, like all users of the Web,

3.0 work on the assumption that it must take the form of node and link. A good
deal oftheoretical and practical attention has appropriatelybeen paid to the
description,implementation, and categorizationof linking. However,as Noah
Wardop-Fruin has reminded us, Ted Nelson, who did not confine hypertext
to the node-and-link form, also proposed stretchtext.According to Nelson's
ComputerLib/Dream Machines(1974),"this form of hypertext is easyto use
without getting lost . . . Gapsappearbetweenphrases;new words and phrases
pop into the gaps, an item at a time . . . The Stretchtextis stored as a text
streamwith extras,codedto pop in and pop out at the desiredaltitudes" (315).
Compare a reader'sexperienceof stretchtextto that when reading on the
Web.When one follows a link on the World Wide Web,one oftwo things hap-
pens: either the present text disappearsand is replacedby a new one or the
destination text opens in a new window. (On Windows machines, in which
the newly openeddocument obscuresthe previous one becauseit appearson
top of it, only an experienceduser would know that one can move the most
recently openedwindow out of the way.Macintosh machines follow a differ-
ent paradigm, emphasizing a multiple-window presentation.)By and large,
standard web browsers follow the replacementparadigm whereas other
hypertextenvironments,such as Intermedia, Storyspace,and Microcosm,
emphasizemultiple windows. Stretchtext,which takes a different approach
to hypertextuality,doeswhat its name suggestsand stretchesor expandstext
when the readeractivatesa hot area.
For an example,let us look at a single sentenceas it appearsin a docu-
ment basedon passagesfrom this book that I made using Nicholas Friesner's
Web-basedstretchtext.One first encountersthe following:

of criticaltheorynowhavea laboratory
withwhichto test
its ideas.Mostimportant,
perhaps, an experience
of reading
or reading
manyof themostsignificant
ideasof criticaltheory.As
J. DavidBolterpointsout in thecourseof explaining
conceptions in printbecomes
ofthe opentext,"whatis unnatural
naturalin theelectronic
mediumandwillsoonno longerneedsaying at all,because
it canbeshown."(www.cyberartsweb.org/cpace/ht/stretchtext
l gplz.html)

Clicking on "critical theory" produces "critical theory. In fact, some of the

most exciting student proiectstake the form of testing, applyrng,or critiquing
specific points of theory, including notions of the author, text, and multi-
vocality."Nert, clicking on "student proiects" in turn produces"student proi-
ects in Intermedia, Storyspace,html, and Flash, and published examplesof
hypermediatake the form of testing, applyrng,or critiquing specificpoints of

RECONFICURING theory including notions of the author, text, and multivocality." Clicking on
THE TEXT any ofthe four instancesofbold text generatesadditional passages,the last
three of which also contain a standard HTML link. Stretching "text," for ex-
ample,"produces"text-and PearlForss'sWhat Is qn Author aclsas an exper-
iment contrasting reader'sreactions to moving text versus reader-centered
hypertext."Clicking on the title of her project opens it in a new window.
A most important distinguishing characteristicof stretchtextfollows from
the manner in which it makes new text appearframed by the old: stretchtext
doesnot fragment the text like other forms of hypermedia. Instead,it retains
the text on the screenthat provides a contextto an anchor formed by a word
or phraseevenafter it has been activated.Stretchingthe text provides a more
immediate perceptualincorporation of the linked-to text with the tert from
which the link originates.In effect, lrzrtbecomesconturtas new text is added;
or rather, the previouslypresenttext remains while the new text appearsand
sewesas its context.This conversionof tert to contextfor other texts may be
seenmore abstractlyin any textual medium, but stretchtexttakesthis notion
quite literally.
The experienceof using Friesner'sWeb-basedimplementation demon-
stratesthat in certain situations stretchtexthas an advantageover link-and-
nodehypertext;in other usesthe link-and-nodeform works better.One strong
advantageofstretchtext derivesfrom the fact that hidden text is alreadypres-
ent, though not visible, when the web browser loads the HTML file, and it
therefore appearsinstantly when the text expands.The text also contracts
instantly,thus providing two real advantages:first, becausethe newly appear-
ing text appearsin immediate physicalproximity to the text one was reading
before activatingthe stretchtext,the readerexperiencesnone ofthe disorien-
tation that may occur when following a link. Second,the very speed with
which the stretchtextappearsencouragesreadersto check stretchableareas
to seeif they in fact want the additional information on offer with the effect
that readersfeel they havemore control over obtaining information'
Experienceusing stretchtextsuggeststhat it providesa convenientmeans
of obtaining definitions, brief explanations,and glossary-likeannotations.In
addition, a secondor even third layer of stretchtextseemsbetter suited than
the replacement-window paradigm for more detailed information directly
related.to the oiginal anchor.The one disadvantagethe more clearly alomiz-
ing link-and-node hypertext does not have appearsifone expandsancillary
information-say, comments on Morris in an essayabout Ruskin: reading a
successionof increasingly more detailed stretchtext passagesof the main
topic can createreaderdisorientation when he or she returns to the passage's

3.0 original form by shrinking the text. Here the occasionallycriticized atomiz-
ing effectof link-and-notehypertextin fact provesa major advantagebecause
when readersfollow a link, they know they have moved to someplacenew.
The gap that alwaysplays an essentialrole in linked hypermediahere has an
orienting, rather than a disorienting, effect. One obvious way to take advan-
tageof both forms of hypertext,of course,involvesincluding links to exterior
lexiasat appropriatepoints within the stretchedtext ("appropriate"here mean-
ing thoseplacesat which further expansionof the original text makesreturn-
ing to the original contractedtext confusing).
Friesner'sWeb-basedstretchtextalso works well with images and hence
proves itself to be a form of hypermedia that provides authors with new
options.Web-basedhypermediahasthree main waysof incorporatingimages:
(1) placing images at the end of the link whether they appearalone or within
text containingexplanatoryinformation, (2)placingimageswithin a favascript-
createdpop-up window usually smaller than the document it overlays,and
(3) placing thumbnail imageswithin a text, often at the right or left margin,
which can link to larger images (a simple use of align : "left" or "right" plus
hspace= "10" within the image tag provides an easyway to allow tert to flow
around an image, providing an aestheticallypleasing border that separates
them). stretchtextoffers a fourth way to handle images,and like its purely text-
basedform, it has some particularly effectiveapplications.A lexia chiefly de-
votedto discussingone painting most effectivelyincludes the image of it as a
linked thumbnail within the text, but passingmentions of details,sources,or
analogouspaintings work betteras stretchtextpresentations,for they arenon-
intrusive,quickly viewed,and quickly dosed and left behind. Thus, stretchtext-
image presentationseemsparticularly well suited for introducing imagesto
which one may wish to refer briefly. Since one can nest images as weli as
alphanumeric information in Friesner'sstretchtext,it also providesa conven-
ient way for the readerto accessdetailsofpainting, earlierversions,and so on.
When I first read about stretchtext I envisioned it functioning verti-
cally,as it does so effectivelyin Friesner'sversion; that is, I assumedthe text
would move apart aboveand below the stretching section,although I did not
imagine the text would appearinstantaneously.When Ian M. Lyons created
TextStretcheras a Director demonstration of the concept,he experimented
with severalapproachesto make words appearfrom within an alreadypres-
ent text. In one version the text moveshorizontally, and as new words arrive,
they push the old text to the right. In a secondversion the stretchabletext
divides vertically,leaving an empty area.In one of these implementations of

RECONFIGURING stretchtext,the words fiIl in from left to right until the spacedisappears;in
THE TEXT anotherthat Lyonsconsidersespeciallyusefirl for poetry,the first word of the
new text appearsin the centerofthe vacantarea,and words then flow out from
either side. If speedis calibratedwith effectivestopsand starts,word-by-word
presentation may give the impression of speechinscribed on the fly. Letter-
by-letterpresentationmay then give the impression of the text'sbeing typed
out by either a human or an artificial source,depending on speedchanges.
In linear stretch mode, activating the stretchtextanchor makes the sen-
tence or phrase in which it appearsdivide, producing an empty spacethat
new words begin to fill, moving from left to right. Once one begins animat-
ing the presentation of new text, a temporal delay occurs. As Lyons points
out, this delaycan perceptuallycreatethe impression of a certain amount of
resistancethe medium itself presents to the embodiment of the text' fust
how much resistanceis met in turn dependson the relativespeedwith which
words and letters appearon the screen.With linear presentation (here pre-
sumed to be in the direction of normal reading),the potential for a different
reading experiencemultiplies, depending on whether one presentstext one
word or one letter at a time. (Lyonschoseto present one word at a time.)
In the expansion-from-centermode, Lyonsexplains,

Expansionandcontractionmovein twoopposite at once.Again,if prop-

to allowfor readingat bothends
the speedat whichtextappears
on the fly significantly
simultaneously heightens depend-
the feelingof contextual
ency.Withexpansionfromthecenterofthenewlexia,textmovingbackward worksto
completethe newlycreated
hangingfragment thatprecedesthe newinsertion;text
movingforward(inthenormaldirection aimsto givereverse
of reading) iustification
for thewaitingtextthat nowmarksthe latterhalfof the newlyold,contextualizing
Moreover, experience
thecognitive of learning directions
to readin multiple
atoncesoasto realign is at firstratherchallenging.
context Whatis exciting
isthat,froma cognitive
aboutthisapproach standpoint, an alternativeread-
it creates
one that whilepredictably
ing procedure, possible.

This form of stretchtext, which Lyons created for writing poetry', obviously
draws attention to the experienceof text itself, intentionally preventing the
readerfrom readingthroughthe text, from too readily taking the text astrans-
parent. With informational hypertext, however, one does not wish to fore-
ground the linguistic aspectofthe medium, and one thereforeneedsdevices
that enablereading.

3.0 Like all hypertextenvironments, TextStretcherneedsa means of indicat-
ing to the readerthe presenceof an anchor,here defined as a span of text or
other information that, when the proper protocol is followed, activatesthe hy-
pertert functionality; in node-and-linkhypertext such activationinvolvesfol-
lowing a link; in stretchtextit involvesinserting nestedor hidden text within
the present text. Different link-and-node systems have employed different
ways of indicating hot texl Intermedia used static indications of links-a
horizontal arrow within a rectangleat the end of a text span;the World Wide
web usesboth staticand dynamic means-the standardblue underlined text
that authors and users can customize and the changing appearanceof the
user'smouse from an arow to a hand when positioned over an anchor.
In creating TextStretcher,Lyons chosetwo simple symbols:(1) a vertical
line within parenthesesto indicate an anchor and (2) a hyphen between
parenthesesto indicate that one can contractthe stretchedtext to produce its
earlier state.when one clicks on an icon representinghot text in all versions
of TextStretcher,new words appear,and depending on the icon, clicking
again can either make the newly arrived text disappearor expand it further
with new information. Friesner'sweb-basedstretchtextuses a boldfacefont
to indicate an expandableanchor and a colored unboldface font to indicate
contractible text. Some experimenting with his version have used phrases,
such as "fSho{Hide Additional Content]" and icons, and others working
with stretchtextcombine a three- or four-inch gap followed by'.More,' on a
coloredrectangle.None of these approachesseemsentirely satisfactoryand
one possiblesolution might involve revealingthe locationsof stretchabletext
by mouse-overs.Stretchtextcomplementsnote-and-linkhypermedia so valu-
ably that solving these interface issueswarants considerableattention.

At the sametime that the individual hypertextlexia has looser,

The DispersedText or lessdetermining, bonds to other lexiasfrom the samework
(to use a terminology that now threatensto becomeobsolete),
it also associatesitselfwith text createdby other authors. In fact, it associates
with whatevertext links to it, therebydissolvingnotions of the intellectualsep-
aration of one text from others,just as some chemicalsdestroythe cell mem-
brane of an organism. Destroyingthe cell membrane destroysthe cell; it kills.
In contrast,similarly destroyingnow-conventionalnotions of textual separa-
tion may destroycertain attitudes associatedwith text, but it will not neces-
sarily destroytext. It will, however,reconfigure it and our expectationsof it.
Another related effect of electronic linking is that it disperses..the',text

RECONFIGURING into other texts. As an individual lexia losesits physical and intellectual sep-
THE TEXT aration from others when linked electronically to them, it also finds itself
dispersedinto them. The necessarycontextualizationand intertextuality pro-
duced by situating individual reading units within a network of easily navi-
gable pathwaysweavestexts, including those by different authors and those
in nonverbal media, tightly together. One effect is to weaken and perhaps
destroyany senseoftextual uniqueness.
Such notions are hardly novel to contemporary literary theory, but here
as so many other caseshypertext createsan almost literal reification or
embodiment of a principle that had seemedparticularly absffactand difficult
when read from the vantagepoint of print. Since much of the appeal,even
charm, of thesetheoreticalinsights lies in their difficulty and evenprecious-
ness,this more literal presentationpromisesto disturb theoreticians,in part,
of course,becauseit greatly disturbs statusand power relations within their
field ofexpertise.

Hypertext fragments, disperses,or atomizes text in two re-

HypertextualTranslationof lated ways. First, by removing the linearity of print, it frees
the individual passagesfrom one ordering principle-se-
cuence-and threatensto transform the text into chaos.Sec-
ond, hypertext destroysthe notion of a fixed unitary text. Considering the
"entire" text in relation to its component parts producesthe first form of frag-
mentation; considering it in relation to its variant readingsand versionspro-
ducesthe second.
Lossof abelief in unitarytextualitycould producemanychangesinWest-
ern culture, many of them quite costly,when judged from the vantagepoint
of our present print-based attitudes. Not all these changes are necessarily
costlyor damaging,however,particularly to the world of scholarship,where
this conceptual changewould permit us to redress some of the distortions
of naturalizing print culture. Accustomed to the standard scholarly edition
of canonical texts, we conventionally suppressthe fact that such twentieth-
cent.)ryprint versions of works originally createdwithin a manuscript cul-
ture are bizarrely fictional idealizationsthat produce a vastlychangedexperi-
ence of text. To begin with, the printed scholarlyedition of Plato,Vergil, or
Augustine provides a text far easierto negotiateand decipherthan any avail-
ableto theseauthors' contemporaryreaders.They encounteredtextsso differ-
ent from ours that evento suggestthat we sharecommon experiencesof read-
ing misleads.

3.0 Contemporary readersof Plato, Vergil, or Augustine processedtexts
without interword spacing,capitalization,or punctuation. Had you read
theselast two sentencesfifteen hundred yearsago,they would havetaken the
following form:

atwesharecom mon

Suchunbroken streamsof alphabeticcharactersmade evenphonetic literacy

a matter of great skill. Since deciphering such texts heavily favoredreading
aloud, almost all readersexperiencedtexts not only as an occasionfor stren-
uous actsof codebreaking but also as a kind of public performance.
The very fact that the text we would have read fifteen hundred years ago
appearedin a manuscript form also implies that to readit in the first placewe
would first havehad to gain accessto a rare,evenunique object-assuming,
that is, that we could have discoveredthe existenceof the manuscript and
made an inconvenient,expensive,and often dangeroustrip to seeit. Having
gained accessto this manuscript, we would also have approachedit much
differently from the way we today approachthe everydayencounter with a
printed book. We would probably have taken the encounter as a rare, privi
legedopporruniry,and we would alsohaveapproachedthe experienceof read-
ing this unique object with a very different set of assumptions than would a
modem scholar.As Elizabeth Eisensteinhas shown, the first role of the scholar
in a manuscript culture was simply to preservethe text, which doubly threat-
ened to degradewith each reading: each time someone physically handled
the fragile object,it reducedits longevity,and eachtime someonecopiedthe
manuscript to preserve and transmit its text, the copyist inevitably intro-
duced textual drift.
Thus, evenwithout taking into accountthe alien presenceof pagination,
indices, references,title pages,and other devicesofbook technology,the
encounter with and subsequentreading of a manuscript constituted a very
different set of experiencesthan those which we take for granted. Equally
important, whereas the importance of scholarly editions lies precisely in
their appearancein comparativelylarge numbers, each manuscript of our
textsby Plato,Vergil, or Augustine existedasa unique object.We do not know
which particular version ofa text by these authors any reader encountered.
Presentingthe history and relation of texts createdwithin a manuscript cul-

RECONFIGURING ture in terms of the unitary text of modern scholarship certainly fictional-
THE TEXT izes-and falsifies-their intertextual relations.
Modern scholar\ editions and manuscripts combine both uniqueness
and multiplicity, but they do so in different ways.A modern edition of Plato,
Vergil, or Augustine begins by assuming the existenceof a unique, unitary
text, but it is produced in the first place becauseit can disseminatethat text
in a number of identical copies.In contrast,eachancient or medievalmanu-
script, which embodiesonly one of many potential variations of "a text,"
existsas a unique object.A new conceptionoftext is neededby scholarstry-
ing to determine not some probably mythical and certainly long-lost master
text but the ways individual readers actually encountered Plato, Vergil, or
Augustine in a manuscript culture. In fact, we must abandonthe notion of a
unitary text and replaceit with conceptionsof a dispersedtext. We must do,
in other words, what some art historians working with analogousmedieval
problems have done-take the conception of a unique type embodied in a
single objectandreplaceitwith a conceptionof a type asa complex setofvari-
ants. For example,tryingto determine the thematic, iconological, and com-
positional antecedentsof early-fourteenth-centuryivory Madonnas, Robert
Suckaleand other recent students ofthe Court Stylehave abandonedlinear
derivations and the notion of a unitary type. Instead, they emphasize that
sculptors choseamong severalsetsof fundamental forms or "groundplans"
as points of departure. Somesort of changein basic attitudestoward the cre-
ations of manuscript culfure seemsnecessary.
The capacityofhypertext to link all the versions or variants of a particu-
lar text might offer a means of somewhat redressing the balance between
uniqueness and variation in preprint texts. Ofcourse, evenin hypertextpre-
sentations,both modern printing conventionsand scholarly apparatuswill
still infringe on attempts to recreatethe experienceof encountering these
texts, and nothing can restorethe uniqueness and corollary auraof the indi
vidual manuscript. Nonetheless,asthe work of PeterRobinsonshows,hyper-
text offers the possibility ofpresenting a text as a dispersedfield ofvariants
and not as a falselyunitary entity. High-resolution screensand other techno-
logical capacitiesalso increasinglypermit a means of presenting all the indi
vidual manuscripts. The Bodleian Library, Oxford, has already put online
detailed,large-scaleimages of some of its most precious illuminated manu-
scripts. An acquaintancewith hypertext systemsmight by itself sufficiently
changeassumptions about textuality to free students of preprint texts from
someof their biases.
All forms of hypertext, even the most rudimentary, change
A Third Convergence:Hypertext our conceptionsoftext and textuality.The dispersedtextual-
iry characteristicof this information technologythereforecalls
and Theoriesof scholarty Editing
into question some of the most basic assumptions about the
nature of text and scholarly textual editing. The appearanceof the digital
word has the major cultural effect of permitting us, for the first time in cen-
turies, easilyto perceivethe degreeto which we havebecomeso accustomed
to the qualities and cultural effectsof the book that we unconsciouslytrans-
fer them to the productions of oral and manuscript cultures. We so tend to
take print and print-based culture for granted that, as the ;'argonhas it, we
have"naturalized" the book by assuming that habits of mind and manners of
working associatedwith it havenaturally and inevitably alwaysexisted.Eisen-
stein, Mcluhan, Kernan, and other students of the cultural implications of
print technology have demonstrated the ways in which the printed book
formed and informed our intellectual history. They point out, for example,
that a great part ofthese cultural effectsderive from book technology'scre-
ation of multiple copiesof essentiallythesametext. Multiple copiesof a fixed
text in turn produce scholarshipand education as we know it by permitting
readersin different times and placesto consult and refer to the "same" text.
Historians of print technologyalsopoint out that economicfactorsassociated
with book production led to the development of both copyright and related
notions of creativityand originality. My reasonfor once again going overthis
familiar ground lies in the fact that all thesefactorscombine to make a single,
singular unitary text an almost unspoken cultural ideal. They provide, in
other words, the cultural model and justification for scholarlytextual editing
as we haveknown it.
It is particularly ironic or simple poetic justice-take your pick-that
digital technologyso calls into question the assumptionsof print-associated
editorial theory that it forces us to reconceiveediting texts originally pro-
duced for print as well as those createdwithin earlier information regimes.
Print technology'semphasis on the unitary text prompted the notion of a
single perfect version of all texts at preciselythe cultural moment that the
presenceof multiple print editions undercut that emphasis-something not
much recognized,if at all, until the arrival of digitality. As the work of fames
Thorpe, George Bornstein, ferome J. McGann, and others has urged, any
publication during an author's lifetime that in some manner receivedhis or
her approval-if only to the extent that the author later chosenot to correct
changesmade by an editor or printer-is an authentic edition. Looking at
the works of authors such as Ruskin and Yeats,who radically rewrote and

RECONFIGURING rearrangedtheir texts throughout their careers,one recognizesthat the tra-

THE TEXT ditional scholarlyedition generallymakes extremely difficult reconstructing
the version someonereadat a particular date.Indeed,from one point ofview
it may radically distort our experienceof an individual volume of poems by
the very fact that it enforcesan especiallystatic frozen model on what tums
out to havebeen a continually shifting and changing entity.
This new conception of a more fluid, dispersedtext, possibly truer than
conventional editions, raisesthe issire ifone can have a scholarlyedition at
all, or if we must settlefor what McGann terms an archive ("CompleteWrit-
ings")-essentially a collectionoftexh;al fragments (or versions)from which
we assemble,or havethe computer assemble,any particular versionthat suits
a certain reading strategy or scholarly question, such as "What version of
Mod.emPainters,Volume 1, did William Morris read at a particular date and
how did the text he read differ from what American Ruskinians read)"
One doesnot encounter many of theseissueswhen producing print edi
tions becausematters of scaleand economy decide or foreclosethem in
advance.In general,physical and economic limitations shapethe nature of
annotations one attachesto a print edition just as they shapethe basic con-
ception of that edition. Sowhat can we expectto happen when these limita-
tions disappearl Or, to phrasethe question differently, what advantagesand
disadvantages,what new problems and new advantages,will we encounter
with the digital wordl

One answer lies in what hypertext does to the concept of

Hypertext,Scholarly annotation. As I argue at length in the following chapter, this
new information technologyreconfigures not only our expe-
Annotation,and the
rience of textuality but also our conceptions of the author's
ElectronicScholarlyEdition relation to that text, for it inevitably produces severalforms
of asynchronouscollaboration,the first, limited one inevitably
appearingwhen readerschoosetheir own ways through a branching text. A
secondform appearsonly in a fully networked hypertext environment that
permits readersto add links to texts they encounter. In such environments,
which are exemplified by the World Wide Web, the editor, like the author,
inevitably loses a certain amount of power and control. Or, as one of my
friends who createdthe first websitefor a major computer company pointed
out, "If you want to play this game, you haveto give up control of your own
text."Although one could envision a situation in which any readercould com-
ment on another editor'stext, a far more interesting one ariseswhen succes-
sive editors or commentators add to what in the print environment would

T. 0 be an existing edition. In fact, one can envisagea situation in which readers
might ultimately encounter a range of annotations.
An exampletaken from my recent experiencewith having students cre-
ate an annotatedversion-read "edition"-of Carlyle's"Hudson's Statue"on
the World Wide Web illuminates some of the issueshere. I intended the as-
signment in part to introduce undergraduatesto various electronicresources
availableat my university,including the online versionsof the OxfordEnglish
Dictionary and Encyclopoedia
Britannica.l wished to habituatethem to using
electronicreferencetools accessibleoutside the physical precinctsofthe
library both to acquaintthem with thesenew tools and alsoto encouragestu-
dents to move from them to those presentlyavailableonly in print form. For
this project students choseterms or phrasesranging from British political
history ("Lord Ellenborough" and "People'sLeague")to religion and mlth
("Vishnu," "Vedas,""Loki"). They then defined or describedthe items chosen
and then briefly explained Carlyle'sallusion and, where known, his uses of
theseitems in other writings.
This simple undergraduateassignment immediately raised issues cru-
cial to the electronic scholarlyedition. First of all, the absenceof limitations
on scale-or to be more accurate,the absenceof the same limitations on
scale one encounters with physical editions-permits much longer, more
substantialnotes than might seem suitablein a print edition. To some ertent
a hypertext environment alwaysreconfiguresthe relative statusof main text
and subsidiary annotation. It also makes much longer notes possible.Elec-
tronic linking makes information in a note easily available,and therefore
these more substantialnotes convenientlylink to many more placesboth
inside and outside the particular text under considerationthan would be
either possible or convenientlyusable in a print edition. Taking our present
example of "Hudson's Statue,"for instance,we seethat historical materials
on, say,democraticmovementslike Chartism and the People'sInternational
League,can shift positions in relation to the annotated text unlike a print
environment, an electronic one permits perceiving the relation of such
materials in oppositemanners. The historical materials can appearas anno-
tations to the Carlyle text, or conversely"Hudson's Statue" can appear-be
experiencedas-an annotation to the historical materials. Both in other
words exist in a networked textual field in which their relationship depends
solelyon the reader'sneed and purpose.
Such recognitions of what happens to the scholarly text in wide-area-
networked environments, such as those createdbv World Wide Web and
RECONFIGURING HyperG, only complicates matters by forcing us to confront the question,
THE TEXT "What becomesof the conceptand practiceof scholarlyannotationl" Clearly,
linking by itself isnt enough, and neither is text retrieval. At first glance,it
might seem that one could solve many issues of scholarlyannotation in an
electronicenvironment by using sophisticatedtext retrieval.In the caseof my
student-createdannotated edition of "Hudson's Statue,"one could just pro-
vide instructions to use the availablesearchtools, though this do-it-yourself
approachwould probably appealonly to the already-experienced
Our textual experiment quickly turned up another,more basicproblem when
severalbright, hard-working neophpes wrote elegantnotes containing accu-
rate, clearly attributed information that nonethelessreferred to the wrong
person, in two casesproviding material about figures from the Renaissance
rather than about the far-lesser-knownnineteenth-centuryfigures to whom
Carlyle referred. What this simple-minded example suggests,of course, is
nothing more radical than that for the foreseeablefuture scholarship will
alwaysbe needed,or to phrase my point in terms relevantto the present
inquiry, one cannot automatetexfual annotation.Tert retrieval,howevervalu-
able,by itself can't do it all.
Fine, but what about hypertextl The problem, after all, with information
retrievallies in the fact that activereadersmight obtain either nonsignificant
information or information whosevalue they might not be ableto determine.
Hypertext, in contrast, can provide editorially approvedconnecfions in the
form of links, which can move from a passagein the so-calledmain text-
here "Hudson's Statue"-to other passagesin the same text, explanatory
materials relevant to it, and so on. Therefore, assuming that one had per-
mission to createlinks to the various online resources,such asthe OED, one
could do so. If one did not have such permission, one could easilydownload
copies of the materials from them, chooserelevant sections,and put them
back online within a web to which one had access;this secondprocedure
is in essencethe one many students chooseto follow. Although providing
slightly more convenienceto the reader than the text-retrievaldo-it-yourself
model, this model still confronts the reader with problems in the form of
passages(or notes)longer than he or she may wish to read.
One solution lies in creating multilevel orlinked progressiveannotation.
Looking at the valuable,ifoverly long, essayone student had written on Car-
lyle and Hindu deities,I realizedthat a better way of proceedinglay in taking
the brief concluding section on Carlyle'ssatiric use of these materials and
making that the first text or lexia the reader encounters; the first mention

3.0 of, say,Vedasor Vishnu, in that lexia was then linked to the longer essays,
thereby providing conveniently accessibleinformation on demand but not
before it was required.
I have approachedthese questions about scholarlyeditions through the
apparentlyunrelated matters of a student assignment and educationalmate-
rials becausethey remind us that in anything like a fully linked electronic
environment, all texts have variable applicationsand purposes. One conse-
quenceappearsin the variableforms that annotation and editorial apparatus
will almost certainly haveto take: since everyonefrom the advancedscholar
down to the beginning student or readeroutsidethe setting ofan educational
institution might be ableto read such texts,they will require variouslayersor
levels of annotation, something particularly necessarywhen the ultimate
linked text is not a scholarlynote but another literary text.
Thus far I have written only as if the linked material in the hypertext
scholarly edition consists of textual apparatus,explanatorycomment, and
contextualization,but by now it should have become obvious that many of
those comments inevitably lead to other so-calledprimary texts.Thus, in our
putative edition of "Hudson's Statue" one cannot only link it to reference
works, such as the OED,Ihe Britannica, (and possibly in the future) to the
Dictionary of National Biography,but also to entire linguistic corpora and to
other texts by the same author, including working drafts, letters, and other
publications. Why stop therel Evenin the relativelyflat, primitive version of
hypertext offered by the present World Wide Web of the Carlylean text
demands links to works on which he draws, such as fonathan Swift's Taleof
a Tub, and those that draw on him, such as Ruskin's "Trafficl'whose satiric
image of the Goddess-oiGetting-on (or Britannia of the Market) derives
rather obviouslyfrom Carlyle'sruminations on the never-completedstatueof
a stock swindler. Finally, one cannot restrict the text field to literary works,
and "Hudson's Statue" inevitably links not only to the Bible and contempo-
rary guidesto its interpretation but alsoto a wide rangeof primary materials,
including parliamentary documents and contemporary newspapers, to
which Carlyle'stext obviouslyrelates.
Once again, though, Iinking, which reconfiguresour experienceand
expectationsof the text, is not enough, for the scholarlyeditor must decide
how Io link various texts. The need for some form of intermediary lexias
again seemsobvious,the first, say,briefly pointing to a proposedconnection
betweentwo texts,the next in sequenceproviding a summary of complex
relations (the outline, in fact, of what might in the print environment have
been a scholarlyarticle or evenbook), the third an overviewof relevant com-

RECONFIGURING parisons,and the last the actualfull text ofthe other author. At eachstage(or
THE TEXT lexia), the reader should have the power not only to return to the so-called
main text of "Hudson's Statue" but also to reach these linked materials out
of sequence.VannevarBush, who invented the general notion ofhypertext,
thought that chains or trails of links might themselvesconstitute a new form
of scholarlywriting, and annotationsin the form of such guided tours might
conceivablybecomepart of the future scholarly edition. We can be certain,
however,that as constraints of scalelessen,increasing amounts of material
will be summoned to illuminate individual texts and new forms of multiple
annotation will developas a way of turning availability into accessibility.

The fact that a single lexia can function very differently within
Hypertextand the Problemof large networked hyperterts raisesfundamental questions about
the applicability of StandardGeneralizedMarkup Language
(SGML) and its heir, ExtendedMarkup Language(XML), to
eiectronic scholarlyeditions, which increasingly appearin vast electronic in-
formation spacesrather than in the stand-aloneversionswe seetoday in CD-
ROMs, such as PeterRobinson'sChaucerproject and Anne McDermott's edi
tion of JohnsonsDicIionary.Therelation of markup languagesand hypertext
appearsparticularly crucial to scholarlyediting since so many large projects
dependon SGML, XML, and their more specificscholarlyforms specifiedby
the Text Encoding Initiative (TEI).
One of the fundamental strengths of XML, of course,lies in its creation
of a single electronictext that can lend itself to many forms of both print and
electronic presentation. Looking at medieval scripta continua above,we
encounteredtext without any markup, not evenspacesbetweenwords. Later
manuscript and print text contains presentationalmarkup-that is, the
encoding takes the form of specific formatting decisions; one indicates a
paragraphby skipping, say,an ertra line and indenting five or sevenspaces.
Although perfectly suited to physical texts, such forms of encoding appear
particularly inefficient and even harmful in electronic environments, since
they prevent easytransference and manipulation of texts. So-calledproce-
dural markup characterizeshandwritten and printed text; to indicate a para-
graph, authors and scribes,as we haveseen,follow a certain procedure,such
as that describedabove.Electronictext works better when one createsa gen-
eralizedmarkup that simply indicatesthe presenceof a text entity, such as a
paragraph,that is then defined in another place.
Once all aspectsof any particular text havebeen indicatedwith the correct
SGML and XML tags, the text appearsin a wonderfully generalized,poten-

T. 0 tially multiplicitous form. For example,after one has tagged(or "marked up")
eachinstance of a text element, such as titles at the beginning of eachchap-
ter, by placing them betweena particular set of tags-say, <chaptertitle>and
</chaptertitle>-one can easily configure such text elements differently in
different versions of a text. Thus, if printed on my university's mainframe
printer, which permits only a single proporfional font, chapter titles appear
bolded in the larger of two availablesizes.If printed with a typesettingdevice,
however,the same chaptertitles automaticallyappearin a very different font
and size, say,30-point Helvetica.If presentedelectronically,moreover,chap-
ter titles can appear in a color different from that of the main text; in the
DynaTert translation of the version of Hypefiert, for example,they appearin
green whereasthe main text appearsin black. My first point here is that once
one has createdsuch a generalizedtext, one can adaptit to different publica-
tion modes with a single instruction that indicates the specific appearance
of all labeled text elements. My secondpoint here is that such tagged text
recordsits own abstractstructLrre.
In "What Is Text Reallyl," their pioneering essayon SGML, Stephen f .
DeRose,David G. Durand, Elli Mylonas,and Allen H. Renearargue that text
consistsofhierarchically organized contextobjects,such as sentences,para-
graphs, sections,and chapters.Do hypertext and markup languages,there-
fore, conceiveof text in fundamentally opposedwaysl At first glance, this
seems to be the case,since hypertext produces nonhierarchical text struc-
tures whereas SGML and XML record hierarchical book structures. The
question arises, To what extent do such visions of markup languagesand
hypertext conflictl After all, SGML and XML fundamentally assert book
structure. But do they asserta single essentialstructure, however reconfig-
urablel Hypertext subvertshierarchy in text and in so doing might seem
to subvert markup languagesand call into question their basic usefulness.
In electronic space,as we havealreadyobserved,an individual lexia may in-
habit, or contribute to, severaltext structures simultaneously.At first consid-
eration this fact might appearto suggestthat markup languagesfundamen-
tally opposehypertert, but such is hardly the case.
Once again,Ted Nelson provides assistance,for it is he who pointed out
that the problem with classificationsystemslies in the fact not that they are
bad but that different people-and the same person at different times-
require different ones. One of the great strengths of hypertext, after all, lies
in its ability to provide accessto materials regardlessof how they are classi-
fied and (hence)how and where they are stored.From the Nelsonianpoint of
view,hypertextdoesnot so much violate classificationsas supplementthem,
RECONFIGURING making up for their inevitable shortcomings. From the point of view of one
THE TEXT considering either the relation of hypertext to markup languages or the
hypertextualizationof them, the problem becomesone of finding some way
to encodeor signal multiple structures or multiple classificationsof struc-
ture. Ifa scholarlyannotation and main text can exchangeroles, status,and
nature, then one needsa devicethat permits a SGML- or XMl-marked lexia
to present a different appearance,ifso required, on being enteredor opened
from different locations.
Returning to our examplesfrom "Hudson's Statue,"we realize that read-
ers starting from Carlyle'stert will experiencelinked materials on Chartism
and the People'sLeagueas annotations to it, but readers starting with pri
mary or secondarymaterials concerning these political movements will
experience"Hudson's Statue" as an annotation to them. When discussing
writing for electronic spacein chapter 5, I suggestways in which both soft-
ware designersand individual authorshaveto assistreaders.For the moment
I shall point out only that one such means of orienting and hence empower-
ing readerstakesthe form of clearlyindicating the permeableborders of the
provisional text to which any lexia belongs. Using such orientation rhetoric
might require that materials by Carlyle have a different appearancefrom
those of conceivablyrelatedmaterials, such aslexiasabout the English Revo-
lution of the 1640sand Victorian political movements. In such a case,one
needsa way of configuring the text according to the means from which it is
accessed.This textual polyrnorphism in turn suggeststhat in such environ-
ments text is alive, changing, kinetic, open-endedin a new way.

Electronic linking, which gives the reader a far more active

Argumentation,Organization, role than is possible with books, has certain major effects.
Consideredfrom the vantagepoint of a literature intertwined
and Rhetoric
with book technology,these effectsappearharmfirl and dan-
gerous,as indeed they must be to a cultural hegemonybased,as ours is, on a
different technologyof cultural memory. In particular,the numerating linear
rhetoric of "first, second,third" so well suited to print will continue to appear
within individual blocks of text but cannot be used to struchrrearguments in
a medium that encouragesreadersto choosedifferent paths rather than fol-
low a linear one. The shift away from linearization might seem a major
change,and it is, but we should remind ourselvesthat it is not an abandon-
ment of the natural.
"The structuring of books,"Tom McArthur reminds us, "is anything but
'natural'-indeed, it is thoroughly
unnatural and took all of 4,000 years to

3.0 bring about. The achievementof the Scholastics,pre-eminently among the
world's scribal elites, was to conventionalizethe themes, plot and shapesof
booksin a truly rigorous way,asthey alsostructured syllabuses,scripture and
debate" (69).Their conventionsof book stmcture, however,changedfunda-
mentally with the adventof the printing press,which encouragedalphabetic
ordering, a procedurethat had never before caught on. Whyl

to thematically
borea closeresem-
blanceto the "normal"organization
of writtenwork:... Alphabetization
to theglobalScholastic
viewof things.lt musthaveseemeda
meaninglesswayof orderingmaterial
to menwho
in neatframesfor containing
all knowledge.
tion posesproblems
of fragmentation
that maybe lessimmediately

McAr*rur's salutaryremarks,which remind us how we alwaysnaturalize the

social constructions of our world, also suggestthat from one point of view
the Scholastics',the movement from manuscript to print and then to hyper-
text appearsone of increasing fragmentation. As long as a thematic or other
culturally coherent means of ordering is availableto the reader, the frag-
mentation of the hypertextdocument doesnot imply the kind of entropy that
such fragmentation would havein the world of print. Capacitiessuch as full-
text searching,automatic linking, agents,and conceptualfiltering potentially
have the power to retain the benefits of hypertextr-ralitywhile insulating the
readerfrom the ill effectsof abandonins lineariF.

The concepts (and experiences)of beginning and ending

Beginningsin the Open Text imply linearity. What happensto them in a form of texruality
not governedchiefly by linearityl If we assumethat hypertex-
tuality possessesmultiple sequencesrather than that it has an entire absence
of linearity and sequence,then one answerto this query must be that it pro-
vides multiple beginnings and endings rather than single ones. Theorists
whose model of hypertert is the Word Wide Web might disagreewith this
claim. Marie-Laure Ryan, for example, assertsthat "every hypertext has a
fixed entry point-there must be an addressto reach first before the system
of links can be activated"(226).Although many web and other hyperfictions
seemto support that statement,examplesin various hypertextenvironments
show that such is not the case.foyce'safi.emoon,which was published in
the PageReaderformat, doeshave a fixed starting point, but other works

RECONFIGURING in Storyspacedo not. Like most webfictions, PatchworkGirlhas an opening

THE TEXT screen,a drawing of Frankenstein'sfemale monster that is equivalentto the
frontispiece in a book, and it is followed by a secondscreen,which simulta-
neously servesas a title page and sitemap, presenting the reader with five
points atwhichtobeginreading: "agraveyard," "ajournal,""aquilt," "astory,"
and "broken accents."Much, of course,dependson what Ryanintends by "sys-
tem of links"; if one means the narrative,then her claim is not alwaystnre. If
she means "that point of the hypertext that one seesfirst," then the claim is
true but only in a trivial sense,becausein those Storyspacehyperfictions that
do not have an opening screen but present the reader with the software's
graphicrepresentationoffolders and documents,the readercan chooseany
starting point in thesespatialhypertexts.AdarvlsBookstore,
for example,which
arrangesits lexias in a circular pattem, invites readersto begin at any point.
(Observationsuggestsmost readersbegin at the top center or top right.)
Similarly, even if we concentrate on webfiction-and Ryan's use of
"address"suggeststhat she is thinking only about the Web, since a URL is an
address-we encountertwo ways in which readersdo not enter narrativesat
a fixed point. First, search engines can guide readersto any lexia within a
hyperfiction; authors of course do not intend such an apparentlyrandom
starting point, but once they placetheir work on the Internet, they allow it to
happen. (That is the reason I tell my students writing both hyperfiction and
hlpertext essaysto be prepared for readerswho "fall in through the living-
room ceiling rather than entering through the front door," and therefore at
least consider including navigation and orientation devicesthat will give
readerssome idea of where they havelanded-and perhapsencouragethem
to keep reading.)Second,webfictions can also open,like PotchworkGirl,with
a first screenthat providesthe readerwith multiple beginnings. The opening
screen,then, is no more the beginning of the narrativethan arethe tide pages
of Jane Eyreor Waterland.
Drawing on EdwardW Saidkwork on origins and openings,one can sug-
gest that, in contrast to print, hypertext offers at least two different kinds of
beginnings. The first concernsthe individual lexia, the seconda gathering of
them into a metatext.Whenever one has a body of hypertext materials that
stands alone-either becauseit occupiesan entire system or becauseit
exists,howevertransiently,within aframe, the readerhas to begin reading at
somepoint, and for the readerthat point is a beginning. Writing of print, Said
explainsthat "a work's beginning is, practically speaking,the main entrance
to what it offers" (3). But what happens when a work offers many "main"
entrances-in fact, offers as many entrancesas there are linked passagesby

T. 0 means of which one can arrive at the individual lexia (which, from one per-
spective,becomesequivalent to a work)? Said provides materials for an
answerwhen he arguesthat a "'beginning' is designatedin order to indicate,
clarify, or define alaterlime, place, or action. In short, the designation of a
beginning generallyinvolvesalso the designationof a consequenlintention"
(5).In Said'sterms, therefore,evenatomizedtext can make a beginning when
the link site, or point of deparhrre, assumesthe role of the beginning of a
chain or path. According to Said,"we seethat the beginning is the first point
(in time, space,or action) of an accomplishmentor processthat has duration
and meaning. The beginning,then, is thefirst stepin the intentionalprod.uaion
Said'squasihypertextual definition of a beginning here suggeststhat "in
retrospectwe can regard a beginning as the point at which, in a given work,
the writer departsfrom all otherworks; a beginning immediately establishes
relationships with works alreadyexisting, relationships of either continuity
or antagonism or some mixture of both" (3).

If hypertext makes determining the beginning of a text diffi-

Endingsin the Open Text cult becauseit both changesour conception oftext and per-
mits readersto "begin" at many different points, it similarly
changesthe senseof an ending. Readersin read-write systemscannot only
choosedifferent points ofending, they can also continue to addto the text, to
extendit, to make it more than it waswhen they beganto read.As Nelson,one
of the originators ofhypertext, points out: "There is no Final Word. There can
be no final version, no last thought. There is alwaysa new view, a new idea, a
reinterpretation. And literature, which we proposeto electronify,is a system
for preserving continuity in the face of this fact . . . Remember the analogy
betweentext and water.Waterflows freely,ice doesnot. The free-flowing,live
documents on the network are subjectto constant new use and linkage, and
those new links continually become interactively available.Any detached
copy someonekeeps is frozen and dead,lacking accessto the new linkage"
(ComputerLib,216l,48).Here, as in severalother ways,Bakhtin'sconception
of textuality anticipateshypertext. Caryl Emerson, his translator and editor,
explainsthat "for Bakhtin'the whole'is not a finished entity; it is alwaysa re-
lationship . . . Thus, the whole can never be finalized and set aside;when a
whole is realized,it is by definition alreadyopen to change" (Problems,dtxl.
Hypertext blurs the end boundaries of the metatext, and conventional
notions of completion and a finished product do not apply to hypertext,
whose essentialnovelty makes difficult defining and describing it in older

RECONFIGURING terms, since they derive from another educationaland information technol-
THE TEXT ogy and have hidden assumptions inappropriate to hypertext. Particularly
inapplicableare the relatednotions of completion and a finished product. As
Derrida recognizes,a form of texruality that goesbeyond pdnt "forces us to
extend . . . the dominant notion of a 'text,"' so that it "is henceforth no longer
a finished corpus of writing, some content enclosedin a book or its margins
but a differential network, a fabric of tracesreferring endlesslyto something
other than itself, to other differential traces" ("Living On," 83-84).
Hypertexhralmaterials,which by definition are open-ended,expandable,
and incomplete, call such notions into question. If one put a work conven-
tionally considered complete, such as Ulysses,
into a hypertext format, it
would immediately become "incomplete." Electroniclinking, which empha-
sizes making connections, immediately expands a text by provrdinglarge
numbers of points to which other texts can attachthemselves.The fixity and
physicalisolation of book technology,which permits standardizationand rel-
ativelyeasyreproduction, necessarilyclosesoffsuch possibilities. Hypertext
opens*rem up.

Hypertext redefines not only beginnings and endings ofthe

Boundariesofthe Open Text text but also its borders-its sides,as it were. Hypertext thus
Drovides us with a means to escaDewhat G6rard Genette
terms a "sort of idolatry which is no less serious,and todaymore dangerous"
than idealizationof the author,"namely,the fetishism of the work-conceived
of as a closed, complete, absolute object" (Figwres,147).When one moves
from physical to virfual text, and from print to hypertext,boundaries blur-
a blurring that Derrida works so hard to achievein his print publications-
and one thereforeno longer can rely on conceptionsor assumptionsof inside
and out. As Derrida explains, "To keep the outside out . . . is the inaugural
gesture of 'logic' itself, of good 'sense' insofar as it accordswith the self
identity of that which is:being is what it is, the outside is outside and the in-
side inside. Writing must thus return to being what it shouldneverhove
to be:an accessoryan accident,an excess"(Dissemination,I28). Without lin-
earity and sharp bounds betweeninside and out, betweenabsenceand pres-
ence, and between self and other, philosophy will change.Working within
the world of print, Derrida prescientlyargues,using Platonictexts as an
example,that "the terhral chain we must set back in place is thus no longer
simply 'internal' to Plato'slexicon. But in going beyond the bounds of that
lexicon, we are less interested in breaking through certain limits, with or
without cause,than in putting in doubt ttre right to posit such limits in the

3.0 first place.In a word, we do not believethat there exists,in all rigor, a Platonic
text,closedupon itself, completewith its inside and outside"(130).
Derrida furthermore explains, with a fine combination of patience and
wit, that in noticing that texts really do not have insides and outsides, one
does not reduce them to so much mush: "Not that one must then consider
that it [the text] is leaking on all sides and can be drowned confusedlyin the
undifferentiated generalityof the element. Rather,providedthe articulations
are rigorously and prudently recognized,one should simply be able to un-
tangle the hidden forces of attraction linking a present word with an absent
word in the text of Plato"(130).
Another sign of Derrida'sawarenessof the limitations and confinements
of contemporaryattitudes,which arise in associationwith the technologyof
the printed book, is his protohypertextualapproachto tertuality and mean-
ing, an approachthat remains skeptical of "a fundamental or totalizing
principle," since it recognizesthat "the classical system's 'outside' can no
longer take the form of the sort of extra-textwhich would arrest the concate-
nation of writing" (5).
Hypertext both on the Internet and in its read-writeforms thus createsan
open, open-borderedtext, a text that cannot shut out other textsand therefore
embodies the Derridean text that blurs "all those boundaries that form the
running border of what used to be calleda text, of what we once thought this
word could identify, i.e.,the supposedend and beginning of a work, the unity
of the corpus,the title, the margins, the signatures,the referential realm out-
side the frame, and so forth." Hypertext therefore undergoeswhat Derrida
describesas "a sort ofoverrun lddbordement]thatspoils all these boundaries
and divisions" ("Living On," 83). Anyone who believesDerrida is here being
overly dramatic should consider the power of the open hypermedia systems
discussedin chapter 1 to add links to someoneelsek Web document.
In hypertext systems,links within and without a text-intratextual and
intertextual connectionsbetweenpoints of text (including images)-become
equivalent,thus bringing texts closer together and blurring the boundaries
among them (Figure 12). Consider the caseof intertextual links in Milton.
Milton's various descriptions of himself as prophet or inspired poet in Par-
adiseLostand his citations of Genesis3:15provide obvious examples.E*ra-
textual and intratextual links, in contrast,are exemplified by links betweena
particular passagein which Milton mentions prophecyand his other writings
in prose or poetry that make similar or obviously relevant points, as well as
biblical texts,commentariesthroughout the ages,comparableor contrasting

The BorderlessText

Linkingchangesthe experience
of text and
authorshipby renderingthe bordersof atl text

Thelliad echoes,references,
By reifoingattusions, and so
o n ,[ i n k i n g

(1) makes

(2) drawsindividuattexts experientia[[y


Consider, for exampte,a hypertextpresentation

(or "edition")of Milton'sParadise

in ParodiseLost

Bookof Exodus

Figure 12. The BorderlessEledronic Text

poetic statementsby others, and scholarlycomment. Similarly, Miltonic cita-

tions of the biblical text about the heel of man crushing the serpent'shead
and being in turn bruised by the serpentobviouslylink to the biblical passage
and its traditional interpretations as well as to other literary allusions and
scholarlycomment on all thesesubjects.Hypertextlinking simply allowsone
to speed up the usual process of making connections while providing a
meansof graphing such transactions,if one can applythe word sirnplyto such
a radicallytransformative procedure.

3.0 The speedwith which one can move betweenpassagesand points in sets
of texts changesboth the way we read and the way we write, just as the high-
speed number-crunching computing changed various scientific fields by
making possible investigationsthat before had required too much time or
risk. One changecomesfrom the fact that linking permits the readerto move
with equalfacility betweenpoints within a text and thoseoutside it. Once one
can move with equalfacility between,say,the opening sectionof ParadiseLost
and a passagein Book 12 thousandsof lines "away:'andbetweenthat open-
ing section and a particular anterior French text or modern scholarly com-
ment, then, in an important sense,the discretenessof texts,which print cul-
ture creates,has radically changedand possibly disappeared.One may argue
that, in fact,all the hypertextlinking of such textsdoesis embodythe way one
actuallyexperiencestextsin the act ofreading; but ifso, the act ofreading has
in some way gotten much closerto the electronic embodiment of text and in
so doing has begun to changeits nature.
These observations about hypertext suggest that computers bring us
much closerto a culture some of whosequalities havemore in common with
that of preliterateman than evenWalter I. Ong has been willing to admit.
In Orality and Literacyhe argues that computers have brought us into what
he terms an ageof "secondaryorality"thathas "striking resemblances"to the
primary, preliterate orality "in its participatory mystique, its fostering of a
communal sense,its concentrationon the present moment, and even its
useof formulas" (136).Nonetheless,althoughOng finds interestingparallels
between a computer culture and a purely oral one, he mistakenly insists:
"The sequentialprocessingand spatializing of the word, initiated by writing
and raised to a new order of intensity by print, is further intensified by the
computer, which maximizes commitment of the word to spaceand to (elec-
tronic) local motion and optimizes analytic sequentialityby making it virhr-
ally instantaneous" (136).In fact, hypertext systems,which insert everytext
into a web of relations, produce a very different effect,for they allow non- or
multisequential reading and thinking.
One major effect of such nonsequential reading, the weakening of the
boundariesofthe text, can be thought ofeither as correctingthe artificial iso-
lation of a text from its contextsor asviolating one of the chief qualities of the
book. According to Ong, writing and printing produce the effect of discrete,

Byisolating detached
thoughton a writtensurface, making
in thissenseautonomous attack,
andindifferentto writingpresents

RECONFIGURING anceandthoughtas uninvolved

with all else,somehow
THE TEXT Printin thesamewaysituates
andthoughton a surface from
else,but it alsogoesfartherin suggesting

We have alreadyobservedthe way in which hypertext suggestsintegra-

tion rather than selfcontainment. Another possibleresult of such hlpertert
may also be disconcerting. As Ong points out, books, unlike their authors,
cannot really be challenged:

Theauthormightbechallengedifonlyheor shecould bereached, buttheauthorcan-

not bereached
in anybook.Thereis nowayto refutea text.Afterabsolutely
it saysexactly
thesamethingasbefore. Thisisonereason
tantamountto "it istrue."lt isalsoonereason whybooks
havebeenburnt.A textstatingwhatthewholeworldknowsis falsewillstatefalse-
so longasthetextexists.(79)

The question arises,however,Ifhypertext situatestexts in a field ofother

texts, can any individual work that has been addressedby another still speak
so forcefully? One can imagine hypertext presentations of books (or the
equivalent)in which tlle reader can call up all the reviewsand comments on
that book, which would then inevitably exist as part of a complex dialogue
rather than asthe embodiment of a voice or thought that speaksunceasingly.
Hypertext,which links one block of text to myriad others, destroysthat phys-
ical isolation of the text, 1'ustas it also destroysthe attitudes createdby that
isolation. Becausehypertext systemspermit a readerboth to annotatean
individual text and to link it to other, perhapscontradictory,texts, it destroys
one of the most basic characteristicsof the printed text-its separationand
univocality.Whenever one placesa text within a network of other texts, one
forcesit to exist aspart of a complex dialogue.Hypertext linking, which tends
to changethe roles of author and reader,also changesthe limits of the indi
vidual text.
Electroniclinking radically changesthe experienceofa text by changing
its spatialand temporal relation to other texts. Readinga hypertextversion of
Dickens's Great Expectationsor ETiot'sWasteland,for example, one follows
links to predecessortexts,vaiantreadings, criticism, and so on. Following an
electroniclink to an image of say,the desertor a wastelandin a poem by Ten-
nyson, Browning, or Swinburne takesno more time than following one from
a passageearlier in the poem to one near its end. Therefore, readerserperi-
ence these other, earlier terts outside The Wastelandand the passageinside
the work as existing equally distant from the first passage.Hypertext thereby

3.0 blurs the distinction betweenwhat is inside and what is outside a text. It also
makes all the terts connectedto a block of text collaboratewith that text.

Alvin Kernan claims that "Benjamin's general theory of the

The Status ofthe Text,Status in demystification of art through numerous reproductions ex-
plains preciselywhat happenedwhen in the eighteenth cen-
the Text
tury the printing press,with its logic of multiplicity, stripped
the classicaltexts of the old literary order of their aura" (152),and it seems
likely that hipertext will extend this processof demystification evenfurther.
Kernan convincingly argues that by Pope'stime a "flood of books, in its
accumulation both of different texts and identical copies of the same texts,
threatenedto obscurethe few idealizedclassics,both ancient and modern, of
polite letters, and to weaken their aura by making printed copies of them"
(153).Any information medium that encouragesrapid dissemination of terts
and easyaccessto them will increasingly demystify individual texts. But
hypertexthas a secondpotentially demystifying effect by making the borders
of the text (now conceivedasthe individual lexia)permeable,it removessome
of its independenceand uniqueness.
Kernan further addsthat "since printed books were for the most part in
the vernacular,they further desacralizedletters by expanding its canon from
a group of venerabletextswritten in ancientlanguagesknown only to an elite
to include a body of contemporary writing in the natural language under-
stood by all who read" (153-54). Will electronic Web versions of the Bible
accompaniedby commentaries, concordances,and dictionaries, like Naue's
TopicalBible(hnp:l lbible.christiansunite.com/Naves-TopicalBible/),which
seem to be essentiallydemocratizing, similarly desacralizethe scripturesl
They have the potential to do so in two ways. First, by making some of the
scholar's procedures easily available to almost any reader, this electronic
Bible might demystify a text that possessesa talismanic power for many in its
intended audience.
Second,and more fundamental, the very fact that this hypertext Bible
enforcesthe presenceof multiple versionspotentiallyundercuts belief in the
possibility of a unique, unitary text. Certainly,the precedentof Victorian loss
of belief in the doctrine of verbal inspiration of the scriptures suggeststhat
hypertert could havea potentially parallel effect (Landow,VictorianTypes,54-
56). In Victorian England the widescale abandonment of belief that every
word of the Bible was divinely inspired, even in its English translation, fol-
lowed from a variety of causes,including influence of German higher criti-
cism, independent British applicationsof rational approachesby those like

RECONFIGURING Bishop Colenso,and the discoveriesofgeology,philology,and (later)biology.

THE TEXT The discovery for instance,that Hebrew did not possessthe uniquenessas a
languagethat some believers,particularly Evangelicals,long assumedit did,
eroded faith, in large part becausebelievers became aware of unexpected
multiplicity where they had assumed only unity. The discoveryof multiple
manuscripts of scripture had parallel effects.Hypertext, which emphasizes
multiplicity, may causesimilar crisesin belief.
Although the fundamental drive of the printed page is a linear, straight-
aheadthrust that capturesreadersand forces them to read along if they are
to read at all, specializedforms of text have developedthat use secondary
codesto present information difficult or impossible to include in linear
text. The footnote or endnote, which is one of the prime ways that books
createan additionalspace,requires some code,such as a superscriptnum-
ber or one within parentheses,that signals readersto stop reading what is
conventionallytermed the main text or the body of the text and begin read-
ing some peripheral or appendedpatch of text that hangs offthat part of the
main text.
In both scholarly editing and scholarlyprose such divisions oftext par-
take of fixed hierarchiesof statusand power.The smaller size type that pres-
ents footnote and endnote text, like the placement of that text awayfrom the
normal center of the reader'sattention, makes clear that such language is
subsidiary dependent,lessimportant. In scholarlyediting, such rypographic
and other encodingmakes clearthat the editor'sefforts, no matter how lavish
or long suffering, are obviously less important than the words being edited,
for these appearin the main text. That's why Barthes'sS/2, whoseorganiza-
tion makes the readerencounter its many notes before coming to the text on
which they comment, is both such a reconfiguration of conventional schol-
arly editions and an effective parody of them. In scholarly and critical dis-
coursethat employs annotation, these conventionsalso establishthe impor-
tance of the dominant argument in opposition to the author's sources,
scholarly allies, and opponents, and even the work of fiction or poetry on
which the critical text focuses.
One experienceshypertextannotation of a text very differently. In the first
place,electronic linking immediately destroysthe simple binary opposition
of text and note that founds the statusrelations that inhabit the printed book.
Following a link can bring the readerto a later portion of the text or to a text
to which the first one alludes.It may also lead to other works by the same
author, or to a rangeof critical commentary textualvariants,and the like. The
assignment of text and annotation to what Tom Wolfe calls different "statu-

3.0 spheres"therefore becomesvery difficult, and in a fully networked environ-
ment such text hierarchies tend to collapse.
Hypertext linking situatesthe presenttext at the centerofthe textual uni-
verse,thus creating a new kind of hierarchy,in which the power of the cen-
ter dominatesthat of the infinite periphery.But becausein hypertextthat cen-
ter is alwaysa transient, decenterablevirtual center-one created,in other
words, only by one's act of reading that particular text-it never tyrannizes
other aspectsof the network in the way a printed text does.
Barthes, well aware of the political constraints of a text that makes a
readerread in a particular way,himself manipulates the political relations of
text in interesting ways. The entire procedure or construction of S/2, for
example, servesas a commentary on the political relationships among por-
tions of the standardscholarlytext,the problem ofhierarchy. Barthesplaliily
createshis own version of complex footnote systems.Like Derrida in Glas,he
createsa work or metatextthat the readeraccustomedto reading books finds
either abrasivelydifferent or, on rare occasion,a wittily powerful commentary
on the way bookswork-that is, on the way they force readersto seerelation-
ships betweensectionsand therebyendow certain assemblagesof words with
power and value becausethey appearin certain formats rather than others.
Barthes,in other words, comments on the footnote, and a7lof S/Z wrns
out to be a criticism of the power relatronsbetweenportions of text. In a foot-
note or endnote, we recall, that portion of the text conventionallyknown as
the main text has a value for both readerand writer that surpassesany of its
supplementaryportions, which include notes, prefaces,dedications,and so
on, most of which take the form of apparatusesdesignedto aid information
retrieval. These devices,almost all of which derive directly from print tech-
nology,can function only when one has fixed, repeatable,physicallyisolated
texts. They have great advantagesand permit certain kinds ofreading: one
need not, for example,memorize the location of a particular passageif one
has systemfeaturessuch as chaptertitles,tablesofcontents,andindices.So
the reference devicehas enornous value as a means of reader orientation,
navigation,and information retrieval.
It comes at certain costs,coststhat, like most paid by the reader of text,
havebecomeso much a part of our experiencesof reading that we do not
notice them at all. Barthes makes us notice them. Barthes,like most late-
twentieth-century critical theorists, is at his best seeingthe invisible, breath-
ing on it in hopes that the condensatewill illuminate the shadowsof what
others havelong missed and taken to be not there. What, then, doesthe foot-
note imply, and how doesBarthesmanipulate or avoidit? Combinedwith the

RECONFIGURING physical isolation of each text, the division between main text and footnote
THE TEXT establishesthe primary importance of main text in its relation to other texts
even when thinking about the subject instantly revealsthat such a relation-
ship cannotin fact exist.
Takeour scholarlyarticle, the kind of articleswe academicsall write. One
wishes to write an article on some aspectof the Nausicaasection of ]oyce's
Ulysses,a text that by even the crudest quantitative measures appearsto be
more important, more powerful than our note identifying, say,one of the
sourcesof Gerty McDowell's phrasing from a contemporarywomen'smaga-
zine. foyce'snovel, for example,existsin more copiesthan our article can or
will and it thereforehas an enormouslylarger readershipand reputation-all
problematic notions, I admit, all relying on certain ideologies;and yet most
of us, I expect,will accedeto them for they are the valuesby which we work.
Ostensibly,that is. Evendeconstructionistsprivilege the text, the greatwork.
Once,however,one begins to write one'sarticle, the conventionsof print
quickly call thoseassumptionsinto question,since anything in the main text
is clearly more important than anything outside it. The physically isolated
discretetext is very discreetindeed, for as Ong makes clear,it hides obvious
connections of indebtednessand qualification. When one introduces other
authors into the text, they appearas attenuated,often highly distorted shad-
ows of themselves.Part of this is necessarysince one cannot, after all, re-
produce an entire article or book by another author in onet own. Part of this
attenuation comes from authorial inaccuracy,slovenliness,or outright dis-
honesty. Nonetheless,such attenuation is part of the messageof print, an
implication one cannot avoid,or at least one cannot avoid since the adventof
hypertext,which by providing an alternativetextual mode revealsdifferences
that turn out to be, no longer, inevitabilities and invisibilities.
In print when I provide the pagenumber of an indicated or cited passage
from foyce,or even include that passagein text or note, that passage-that
occasionfor my article-clearly exists in a subsidiary comparativelyminor
position in relation to my words, which appear,after all, in the so-calledmain
tert. What would happen, though, if one wrote one's article in hypertextl
Assuming one worked in a fully implemented hypertexrual environment,
one would begin by calling up foyce'snovel and, on one side of the video
screen, opening the passageor passagesinvolved. Next, one would write
one'scomment, but where one would usually cite foyce,one now does so in
a very different way. Now one createsan electronic link between one's own
text and one or more sectionsof the foyceantext. At the same time one also
links one'stext to other aspectsof one'sown text, texts by others, and earlier

3.0 texts by oneself. Severalthings havehappened,things that violate our expec-
tations. First, attachingmy commentary to a passagefrom foycemakes it
exist in a far different, far less powerful, relation to Joyce,the so-calledorigi-
nal text,than it would in the world of physicallyisolatedtexts. Second,as soon
as one attachesmore than one text block or lexia to a single anchor (or block,
or link marker), one destroysall possibility of the bipartite hierarchy of foot-
note and main text. In hypertext,the main text is that which one is presently
reading. So one has a double revaluation: with the dissolution of this hier-
archy,any attachedtext gains an importance it might not havehad before.
In Bakhtin'sterms, the scholarlyarticle, which quotesor cites statements
by others-"some for refutation and others for confirmation and supple-
mentation-is one instance of a dialogic interrelationship among directly
signifying discourseswithin the limits of a single context . . . This is not a
clash of two ultimate semanticauthorities, but rather an objectified (plotted)
clash of two representedpositions, subordinatedwholly to the higher, ulti
mate authority of the author. The monologic context, under these circum-
stances,is neither broken nor weakened"(Problerus,188).
Trying to evadethe
constraints,the logic, of print scholarship,Bakhtin himself takesan approach
to quoting other authors more characteristicof hypertext or postbook tech-
nology than that of the book. According to his editor and translator,Emerson,
when Bakhtin quotes other critics, "he does so at length, and lets eachvoice
sound fully. He understands that the frame is always in the power of the
framer, and that there is an outrageousprivilege in the power to cite others.
Thus Bakhtin's footnotes rarely sewe to narrow down debate by discredit-
ing totally, or (on the other hand) by conferring exclusive authority. They
might identify, expand,illustrate, but they do not pull rank on the body of the
text-and thus more in the nature of a marginal gloss than an authoritative
footnote" (>oorvii).
Derrida also comments on the status relations that cut and divide texts,
but unlike Barthes,he concerns himself with oppositions between preface
and main text and main text and other texts. Recognizingthat varying levels
of statusaccrueto different portions of a text, Derrida examinesthe way each
takes on associationswith power or importance. In discussingHegel'sintro-
duction to the Logic,Denidapoints out, for example,that the prefacemust be
distinguished from the introduction. They do not havethe samefunction, or
eventhe same dignity, in Hegel's eyes(Dissernination,lT).Derrida'snew tex-
tuality, or true textuality (which I have continually likened to hypertextuality),
represents"an entirely other typology where the outlines of the prefaceand
the'main'text are blurred" (39).
One tends to think of text from within the position of the lexia
Hypertextand Decentrality:The under consideration.Accustomed to reading pages of print
on paper,one tends to conceiveoftext from the vantagepoint
ofthe readerexperiencingthat page or passage,and that por-
tion of text assumesa centrality. Hypertext, however,makes such assump-
tions of centrality fundamentally problematic. In contrast,the linked text,the
annotation, exists as the other lert, and it leads to a conception (and experi-
ence)of text as Other.
In hypertextthis annotation, or commentary or appendedtext can be any
linked text, and thereforethe position of any lexia in hypertextresemblesthat
of the Victorian sage.For like the sage,say,Carlyle,Thoreau, or Ruskin, the
lexia stands outside, offcenter, and challenges.In other words, hypertext,
like the sage,thrives on marginality. From that essentialmarginality to which
he stakeshis claim by his skillfirl, aggressiveuse ofpronouns to opposehis
interests and views to those of the reader,he defines his discursiveposition
or vantagepoint.
Hypertext similarly emphasizesthat the marginal has as much to offer as
doesthe central, in part becausehypertext refusesto grant centrality to any-
thing, to any lexia, for more than the time a gazerestson it. In hypertext,cen-
trality, like beauty and relevance,resides in the mind of the beholder. Like
Andy Warhol'smodern person'sfifteen minutes of fame, centrality in hyper-
text only exists as a matter of evanescence.
As one might expectfrom an
information medium that changesour relations to data,thoughts, and selves
so dramatically,that evanescenceof this (ever-migrating)centrality is merely
a given-that's the way things are-rather than an occasionfor complaint or
satire. It is simply the condition under which-or within which-we think,
communicate, or record these thoughts and communications in the hyper-
texfual docuverse.
This hypertextual dissolution of centrality, which makes the medium such
a potentially democratic one, also makes it a model of society of conversa-
tions in which no one conversation,no one discipline or ideology,dominates
or founds the others. It is thus the instantiation of what Richard Rorty terms
edifyingphilosophy,the point of which "is to keep the conversation going
rather than to find objectivetruth." It is a form of philosophy

havingsenseonlyasa protestagainst bypropos-

to closeoffconversation
alsfor universal
through of someprivileged
thehypostatization set
of descriptions.
Thedanger triesto avertisthatsomegiven
somewayin whichpeoplemightcometo thinkof themselves,
vocabulary, will

T. 0 deceive
couldbe,or shouldbe,nor-
maldiscourse. Theresulting freezing-over of culture wouldbe,in theeyesof edifying
p h i l o s o p h et rhse,d e h u m a n i z a toi of hnu m a nb e i n g s( .3 7 7 )

HFpertext,which has a built-in bias against "hypostatization"and prob-

ably againstprivileged descriptionsas well, therefore embodiesthe approach
to philosophy that Rorty urges.The basicexperienceof text, information, and
control, which moves the boundary of power away from the author in the
direction of the reader,models such a postmodern, antihierarchicalmedium
of information, text, philosophy,and society.

Like contemporary critical theory hypertext reconfigures-

Erosionofthe Self rewrites-the author in severalobviousways. First of all, the
figure ofthe hypertext author approaches,evenifit doesnot
merge with, that of the reader; the functions of reader and writer become
more deeplyentwinedwith eachotherthan everbefore.IThis transformatton
and near merging of roles is but the latest stagein the convergenceof what
had once been two very different activities.Although today we assumethat
anyonewho readscan also write, historians of reading point out that for mil-
lennia many peoplecapableof reading could not even sign their own names.
Todaywhen we consider reading and writing, we probably think of them as
serial processesor as procedurescarried out intermittently by the same per-
son: first one reads,then one writes, and then one reads some more. Hyper-
te$, which createsan active,even intrusive reader,carries this convergence
of activitiesone step closerto completion; but in so doing, it infringes on the
power of the writer, removing some of it and granting it to the reader.These
shifts in the relations of author and readerdo not, however,imply that hyper-
text automaticallymakes readersinto authors or co-authors-except, that is,
in hypertext environments that give readersthe ability to add links and texts
to what they read.'z
One clear sign of such transferenceof authorial power appearsin the
reader'sabilities to choosehis or her way through the metatext,to annotate
text written by others, and to create links between documents written by
others. Read-writehypertext like Intermedia or Weblogs that accept com-
ments do not permit the activereaderto changethe text producedby another
person, but it does narrow the phenomenologicaldistancethat separates

T. 0 individual documents from one another in the worlds of print and manu-
script. In reducing the autonomy of the text, hypertextreducesthe autonomy
of the author. In the words of Michael Heim, "as the authoritativenessof text
diminishes, so too does the recognition of the private self of the creative
author" (ElecrricLanguage,22Il. Granted,much of that so-calledautonomy
had been illusory and existed as little more than the difficulty that readers
had in perceivingconnectionsbetweendocuments.Nonetheless,hypertext-
which I am here taking as the convergenceof poststructuralist conceptions
of textuality and electronic embodiments of it-does do awaywith certain
aspectsof the authoritativenessand autonomy of the text, and in so doing
it does reconceive the figure and function of authorship. One powerful
instance of the way hypermedia environments diminish the author'scontrol
over his or her own text appearsin the way so-calledopen systemspermit
readersto insert links into a lexia written by someoneelse.Portal Maximizer,
for example,permits overlaying one author's Web documents with another
author'slinks, although the original document remains unchanged.3
William R. Paulson,who examines literature from the vantagepoint of
information theory arrives at much the same position when he argues that
"to characterizetexts as artificially and imperfectly autonomous is not to
eliminate the role of the author but to deny the reader'sor critic's submission
to any instance ofauthority. This perspectiveleavesroom neither for author-
ial mastery of a communicative object nor for the authority of a textual
coherenceso complete that the readerk (infinite) task would be merely to
receiveits rich and multilayered meaningJ' Beginning from the position of
information theory Paulsonfinds that in "literary communication," as in all
communication, "there is an irreducible element of noise," and therefore
"the reader'stask does not end with reception, for reception is inherently
flawed. What literature solicits of the reader is not simply receptivebut the
active,independent, autonomous construction of meaning" (139).Finding
no reasonto exile the author from the text, Paulsonnonethelessends up by
assigningto the readera small portion of the power that, in earlier views,had
been the prerogativeof the writer.
Hypertext and contemporary theory reconceivethe author in a second
way.As we shall observewhen we examine the notion of collaborativewrit-
ing, both agree in configuring the author of the text as a text. As Barthes
explains in his famous exposition of the idea, "this 'I'which approachesthe
10).Barthes'spoint, which should seem both familiar and unexceptionalto
anyonewho has encounteredfoyce'sweaving of Gerty McDowell out of the

RECONFIGURING texts of her classand culture, appearsmuch clearer and more obvious from
THE AUTHOR the vantagepoint of intertextuality.In this case,as in others at which we have
alreadylooked, contemporarytheory proposesand hypertext disposes;or, to
be less theologically aphoristic, hypertext embodies many of the ideas and
attitudes proposedby Barthes,Derrida, Foucault,and others.
One of the most important of these ideas involves treating the self of
author and reader not simply as (print) text but as a hypertext. For all these
authors the self takes the form of a decentered(or centerless)network of
codesthat, on another level, also sewes as a node within another centerless
network. Jean-FrangoisLyotard,for example,rejects nineteenth-centuryRo-
mantic paradigmsof an islanded self in favor of a model of the self as a node
in an information network: 'A self doesnot amount to much," he assuresus
with fashionablenonchalance,"but no self is an island; each existsin a fab-
ric of relations that is now more complex and mobile than everbefore.Young
or old, man or woman, rich or poor, a person is always located at
points' of specificcommunication circuits, howevertiny thesemay be. Or bet-
ter: one is alwayslocatedat a post through which various kinds of messages
pass"(Postrnodem Condition,15).Lyotard'sanalogybecomeseven stronger if
one realizesthat by "post" he most likely means the modern Europeanpost
office,which is a telecommunicationscentercontaining telephonesand other
networked devices.
Sometheorists find the idea of parricipating in a network to be demean-
ing and depressing,particularly since contemporary conceptionsof texhral-
ity deemphasizeautonomy in favor of pariicipation. Before succumbing to
posthumanist depression,however,one should place Foucault'sstatements
about "the author's disappearance"in the context of recent discussionsof
machine intelligence (Foucault, "What Is an Authorl" 119).According to
Heinz Pagels,machines capableof complex intellectual processingwill "put
an end to much discussionabout the mind-body problem, becauseit will be
very hard not to attribute a consciousmind to them without failing to do so
for morehuman beings. Gradually the popular view will become that con-
sciousnessis simply'what happens'when electroniccomponentsare put
together the right way" (92).Pagels'sthoughts on the eventualelectronic so-
lution to the mind-body problem recall Foucault'sdiscussionof "the singular
relationship that holds betweenan author and a text [as]the manner in which
a text apparentlypoints to this figure who is outside and precedesit" ("What
Is an Authorl" 115).This point of view makes apparentthat literature gener-
atespreciselysuch appearanceofa self, and that, moreover,we havelong read
a self "out" of texts as evidencethat a unified self exists"behind" or "within"

3.0 or "implicit in" it. The problem for anyone who yearns to retain older con-
ceptions of authorship or the author function lies in the fact that radical
changes in terruality produce radical changesin the author figure derived
from that textuality. Lack of textual autonomy, like lack of textual centered-
ness, immediately reverberatesthrough conceptionsof authorship as well.
Similarly, the unboundednessof the new tertuality dispersesthe author as
well. Foucaultopensthis side of the question when he raiseswhat, in another
context,might be a standardproblem in a graduatecourseon the methodol-
ogy ofscholarship:

lfwewishto publish
worksof Nietzsche,
everything butcanweagreeon what,,every-
mustbe published,
thing"means? Wewill,of course,includeeverything that Nietzsche himselfpuO-
l i s h e da, l o n gw i t ht h ed r a f t so f h i sw o r k sh, i sp l a n sf o r a p h o r i s mhsi,sm a r g i n a l i a ;
Butwhatif,in a notebook
a reminder
of anappointment, anaddress,
or a laundrybill,shouldthisbe
includedin hisworks)Whynotl . . . lf somehavefoundit convenientto bypassthe
of thewriteror hisstatusas an authorto concentrate
on a work,they
havefailedto appreciatetheequally problematicnatureof theword,,work"andthe
u n i t yi t d e s i g n a t e( 1
s1. 9 )

Within the context of Foucault'sdiscussion of "the author's disappear-

ance" (119),the illimitable plenitude of Nietzsche'soeuvredemonstratesthat
there'smore than one way to kill an author. One can destroy (what we mean
by) the author, which includes the notion of soleauthorship, by removing the
autonomy of text. One can also achievethe same end by decenteringtext or
by transforming text into a network. Finally, one can remove limits on tex-
tuality, permitting it to expand, until Nietzsche, the edifying philosopher,
becomesequally the author of The Gay Scienceand laundry lists and other
such trivia-as indeed he was. Such illimitable plenitude has truly .trans-
formed" the author, or at least the older conceptionof him, into .,avictim of
his ownwriting" (lU).
Fearsabout the deathof the author, whether in complaint or celebration,
derivefrom claude L6vi-Strauss, whosemythologicalworks demonstratedfor
a generationof critics that works of powerful imagination take form without
an author. In TheRaw and the Cooked(1964),for example,where he showed,
"not how men think in myths, but how myths operatein men's minds with-
out their being awareof the fact," he also suggests"it would perhapsbe bet-
ter to go still further and, disregarding the thinking subject completely,
proceedas if the thinking processwere taking place in the myths, in the

RECONFIGURING reflection upon themselvesand their interrelation" (12)'aL6vi-Strauss'spres-

THE AUTHOR entation of mythological thought as a complex system of transformations
without a centerturns it into a networked text-not surprising, sincethe net-
work servesasone of the main paradigmsof synchronousstnrcture.sEdward
Saidclaims that the "two principal forcesthat haveerodedthe authority of the
human subjectin contemporaryreflection are, on the one hand, the host of
problemsthat arisein defining the subject'sauthenticity and, on the other,the
developmentof disciplines like linguistics and ethnology that dramatize the
subject'sanomalous and unprivileged, evenuntenable, position in thought"
(293).One may add to this observationthat these disciplines' network para-
digms also contribute importantly to this senseof the attenuated,depleted,
eroding, or evenvanishing subject.
Someauthors, such as Saidand Heim, derivethe erosion of the thinking
subject directly from electronic information technology.Said, for example,
claims it is quite possibleto argue "that the proliferation of information (and
what is still more remarkable, a proliferation of the hardware for dissemi-
nating and preserving this information) has hopelesslydiminished the role
apparentlyplayed by the individual" (51).6Michael Heim, who believesloss
of authorial power to be implicit in all electronictert, complains: "Ftagments,
'hypertext,' as Ted Nel-
reused material, the trails and intricate pathwaysof
son terms it, all these advancethe disintegration of the centering voice of
contemplativethought. The arbitrarinessand availabilityof databasesearch-
ing decreasesthe felt senseofan authorial control overwhat is written" (Elec-
tnc Language,22O\.A databasesearch, in other words, permits the active
reader to enter the author's text at any point and not at the point the author
choseas the beginning. of course,as long as we have had indices, scholarly
readershave dipped into specialist publications before or (shame!)instead
of reading them through from beginning to end. In fact, studies of the way
specialistsread periodicalsin their areasof expertiseconfirm that the linear
model of readingis often little more than a pious fiction for many expertread-
ers (McKnight, Richardson,and Dillon, "fournal Articles").
Although Heim here mentions hypertext in relation to the erosion of
authorial prerogative,the chief problem, he argueselsewhere,lies in the way
"digital writing turns the private solitude of reflective reading and writing
into a public network where the personal syrnbolic framework needed for
original authorship is threatened by linkage with the total textuality of
human expressions"(Etectic Language,215).Unlike most writers on hyper-
text, he finds participation in a network a matter for worry rather than cele-
bration, but he describes the same world they do, though with a strange
HYPERTEXT combination of prophecy and myopia. Heim, who seesthis loss of authorial
control in terms of a corollary loss of privacy,arguesthat "anyonewriting on
a firlly equipped computer is, in a sense,directly linked with the totality of
symbolic expressions-more so and essentiallyso than in any previouswrit-
ing element" (215).Pointing out that word processingredefines the related
notions of publishing, making public, and privacy,Heim arguesthat anyone
who writes with a word processorcannot escapethe electronicnetwork: .,Dig-
ital writing, becauseit consists of electronic signals, puts one willy-nilly on
a network where everything is constantlypublished. privacy becomesan
increasinglyfragile notion. Word processingmanifests a world in which
the public itself and its publicity have become omnivorous; to make public
has therefore a different meaning than ever before" (215).Although in 1987
Heim much exaggeratedthe loss of privacy inherent in writing with word-
processing software per se, he turns out, as Weblog diaries prove, to have
been prescient.When he wrote, most people did not in fact do most of their
writing on networks, but the Internet changeseverything: e-mail and per-
sonal blogs blur the boundariesbetweenpublic and private.TAlthough Heim
may possibly overstatethe casefor universal loss of privacy-the results are
not in yet-he has accuratelypresentedboth some implications of hypertext
for writers and the reactionsagainstthem by the print author accustomedto
the fiction of the autonomous text.
The third form ofreconfiguration of selfand author sharedby theory and
hypertext concernsthe decenteredself, an obvious corollary to the network
paradigm. As Said points out, major contemporary theorists reject ..thehu-
man subject as grounding center for human knowledge. Derrida, Foucault,
and Deleuze . . . have spoken of contemporary knowledge (savoir)as decen-
tered; Deleuze'sformulation is that knowledge,insofar as it is intelligible,
is apprehensiblein terms of nomadiccenters,provisional structures that
are neverpernanent, alwaysstrayingfrom one setof information to another"
(376). These three contemporary thinkers advance a conceptualization of
thought bestunderstood,like their views oftext, in an electronic,virtual, hy-
Before mourning too readily for this vanished or much diminished self
we would do well to remind ourselvesthat, although Western thought long
held such notions of the unitary self in a privileged position, texts from
Homer to Freud have steadily argued the contrary position. Divine or de-
monic possession, inspiration,humors, moods,dreams,the unconscious-
all thesedevicesthat serveto explain how human beings act better,worse, or
just different from their usual behaviorargue againstthe unitary conception

RECONFIGURING of the self so central to moral, criminal, and copyright law. The editor of the
THE AUTHOR Soncino edition of the Hebrew Bible reminds us that

is an old enigma,whichhasbaffledthe skillofcommentators.. .

true Prophet,and
in Scriptureas at the sametime heathensorcerer,
He is represented
abhorrentmeansofbringingaboutthe ruin
a peculiarly
the perverterwho suggested
of lsrael.Becauseofthese fundamentalcontradictionsin character,BibleCriticsas-
sume,that the Scripturalaccountof Balaamis a combinationof two or threevarying
traditionsbelongingto differentperiods. . . Sucha view betraysa slightknowledge
of thefearfulcomplexityof themind and soul of man. lt is only in the realm of Fable
that men and womendisplay,as it werein a singleflashof light,someone aspectof
human nature.lt is otherwisein reallife.(568)

Given such long observed multiplicities of the self,' we are forced to real-
ize that notions of the unitary author or self cannot authenticate the unity of
a text.8 The instance of Balaam also reminds us that we have access to him
only in Scriptures and that it is the biblical text, after all, which figures the
unwilling prophet as a fractured self.

Authors who have experienced writing within a hypertext en-

How the PrintAuthor Differs vironment often encounter certain predictable frustrations
when returning to write for the linear world of print. Such
from the HypertextAuthor
frustrations derive from repeatedrecognitions that effective
argument requires closing off connectionsand abandoning lines of investi-
gationthat hypertextualitywould havemade available.Here are two examples
of what I mean. Near the opening of this chapter,in the midst of discussing
the importance of L6vi-suaussto recent discussionsof authorship, I made
the following statement:"L6vi-strauss'spresentationof mythologicalthought
as a complex system of transformations without a center turns it into a net-
worked text-not surprising, sincethe network servesasone of the main par-
adigms of synchronousstructure"; and to this text I appendeda note, point-
ing out that in TheScopeofAnthropology"L6vi-Straussalsoemploysthis model
'Our sociery a particular instance in a much vaster
for societiesas a whole:
family of societies,depends,like all others, for its coherenceand its very
existenceon a network-grown infinitely unstable and complicatedamong
us-of ties betweenconsanguinealfamilies"" At this point in the main text,
I had originally planned to place Foucault'sremark that "we can easilyimag-
ine a culture where discoursewould circulatewithout any needfor an author"
("What Is an Authorl" 138),and to this lemark I had consideredadding the
observationthat, yes,we can easily "imagine" such a culture, but we do not

3.0 haveto do so, since L6vi-strauss'smythographic works have provided abun-
dant examplesof it. Although the diachronic relationship betweenthesetwo
influential thinkers seemedworthy of notice, I could not add the passage
from Foucaultand my comment becauseit disturbed my planned line of ar-
gument, which next required Said'srelation of ethnology and linguistics to
the erosionof "the authority ofthe human subl'ect"in contemporarythought.
I did not want to veer offin yet another direction. I then consideredputting
this observationin note T,buL again,it also seemedout of placethere.
Had I written this chapter within a hypertext environment, the need to
maintain a linear thrust would not haverequired this kind of choice.It would
haverequired choices,but not this kind, and I could havelinked two or more
passagesto this point in the main text, thereby creating multiple contexts
both for my argument and for the quoted passagethat servedas my point of
deparhrre.I am not urging, ofcourse, that in its print form this chapterhas
lost something of major importance becauseI could not easilyappendmul-
tiple connectionswithout confusing the reader.(Had my abandonedremark
seemedimportant enough to my overall argument, I could havemanagedto
include it in several obvious ways, such as adding another paragraph or
rewriting the main text to provide a point from which to hang another note.)
No, I make this point to remind us that, as Derrida emphasizes,the linear
habits of thought associatedwith print technologyoften influence us to think
in particular ways that require narowness, decontextualization,and intel-
lectual attenuation, if not downright impoverishment. Linear argument, in
other words, forcesone to cut offa quoted passagefrom other, apparently
irrelevant contextsthat in fact contribute to its meaning. The linearity of print
alsoprovidesthe passagewith an illusory centerwhoseforce is intensified by
such selection.
A secondexamplepoints to another kind of exclusionassociatedwith lin-
ear writing. During the course of composing the first three chaptersof this
book, severalpassages,such as Barthes'sdescription ofthe writerly text and
Derrida's exposition of borders, boundaries, and d4bordement, forced them-
selvesinto the line of argument and hence deservedinclusion sevenor eight
times. One can repeatedlyrefer to a particular passage,of course, by com-
bining firll quotation, selections,and skillfr.rlparaphrase,but in generalthe
writer can concentrateon a quoted sectionof text in this manner only when
it servesas the center,or one of the centers,of the argument. If I wished to
write a chapteror an entire book about Derridean d.6bord.ement,I could return
repeatedlyto it in different contexts,thereby revealing its richness of impli
cation. But that is not the book I wished to write in L991,or wish to write now

RECONFIGURING nor is that the argument I wish to pursue here, and so I suppressthat text
THE AUTHOR and argument, which henceforth exist only in potentia. After careful consid-
eration, I decide which of the many places in the text would most benefit
from introduction of the quotation and then at the appropriatemoment, I
trundle it forward. As a result, I necessarilycloseoffall but a few of its obvi-
ous points of connection.
As an experiencedwriter accustomedto making such choices,I realize
that selectionis one of the principles of effectiveargument. But why doesone
have to write texts in this wayl If I were writing a hypertext version of this
text-and the versionswould exist so differently that one has to placequota-
tion marks around "version" and "text," and probably "I" as well-I would
not have to chooseto write a single text. I could, instead, produce one that
contained a plurality of ways through it. For example, after preparing the
readerfor Derrida'sdiscussionof ddbordemenf, I could then link my prepara-
tory remarks either to the passageitself or to the entire text of "Living Oni
and I could provide temporary markings that would indicate the beginning
and end of the passageI wished to emphasize.At the sametime, my hyper-
text would link the samepassageto other points in my argument. How would
I go about creating such linksl
To answer this question, let me return to my first and simpler example,
which involved linking passagesfrom L6vi-Strauss'sScopeof Anthropology
and Foucault's"What Is an Authorl" to a remark about the anthropologist's
use of the network model. Let'slook at how one makes a link in three differ-
ent hypermedia environments, Intermedia, Storyspace,and HTML (for the
Web).Unlike creatinglinks in HTML, linking in Microcosm, Storyspace,and
Intermedia follows the now common cut-and-pasteparadigm found in word
processors,graphics editors, and spread sheets.Using the mouse or other
pointing device,one placesthe cursor immediately before the first letter of
the first word in the passagein question, the sign ofwhich is that the tert ap-
pearshighlighted-that is, it appearswithin a black rectangle,and the black
type against a white background now appearsin reversevideo, white letter-
ing against a black background. With the text highlighted, one moves the
mouse until the point of the arrow-shapedcursor coversany part of the word
"Intermedia" that appearsin a horizontal list of words at the top of the screen
("File,""Edit," "lntermedia," and so on). Holding down the mousebutton, one
drawsthe cursor down, therebyproducing the Intermedia menu, which con-
tains choices.Placing the pointer over "Start Link," one releasesthe mouse
button, proceedsto the secondtext, and carries out the same operation until
one opensthe Intermedia menu, at which point one chooses"CompleteLink."

3.0 The systemthen producesa panel containing placesto tFpe any desired
labelsfor the linked passages;it automaticallyaddsthe title of the entire texr,
and the writer can describethe linked passagewithin that text. For example,
if I createda link betweenthe hypermedia equivalentof my text for the pre-
vious section of this chapter and a passagein The Scopeof Anthropology,
Intermedia would automaticallyadd the title of that text, "The Erosion of
the Author," to which I would add a phrase,say,"L6viStrauss & myth as net-
work." At the other end of the link, the systemwould furnish "Claude L6vi
Strauss,The Scopeof Anthropology,"and I would add something like "L6vi-
Strauss& societyas network."When a readeractivatesthe link marker in the
main text, the new entry appearsas an option: "Claude L6vi-Strauss,7he
Scopeof Anthropology:(L6vi-strauss& societyas network)J'Storyspacelink-
ing involvesa roughly similar, if simpler, procedure:to link from a phraseto
another document, one highlights the phrase,moves the cursor to a palette
containing an arrow clicks on it, and then clicks on the other document, at
which point a panel appearsin which one can placea description.To make a
link in HTML (which only permits one link per anchor), one has to tFpe
somethinglike <a href : ' ../../levistrauss/1.html'target: "blank">how men
think in myths</>, or use a handy html editor like BBEdit or Dreamweaver,
which would add the HTML tag (<a href :></a>) after I typed in the infor-
mation betweenquotationmarks (../..ilevistrauss/1.html).
In Storyspace,Intermedia, and similar programs,linking the secondtext,
the passagefrom Foucault,follows the identical procedurewith the single ex-
ception that one no longer has to provide a label for the lexia in the main text,
since it alreadyhas one. In HTML one has to sacrificethis secondlink or find
another appropriatephraseto which one could add a link.
If instead of linking these two brief passagesof quotation, documenta-
tion, and commentary, I createda more complex document set, focused on
Derridean dfbordement,one would follow the sameprocedureto createlinks.
In addition, one would also createkinds of documents not found in printed
text, some of which would be primarily visual or hieroglyphic. One, for
example, might take the form of a concept map showing, among other
things, uses of the term ddbordement
in "Living On," other works by Derrida
in which it appears,and its relation to a range of contexts and disciplines
from cartographyand histology to etyrnology and French military history.
Current hypermedia systems,including popular World Wide Web browsers,
permit linking to interactivevideo, music, and animation as well as diction-
aries, text, time lines, and static graphics.In the future these links will take
more dynamic forms, and following them will animate some procedure,say,

RECONFIGURING a search through a French thesaurus, or a reader-determinedtracking of

THE AUTHOR createdafter I had completedmy document would automancally
My brief description of how I would go about producing this text were I
writing it in something like a completehypertextenvironment might trouble
some readersbecauseit suggeststhat I have sacrificeda certain amount of
authorial control, ceding some of it to the reader.The act of writing has also
changedto some extent. Electronichypertext and contemporarydiscussions
of critical theory particularly those of the poststructuralists,display many
points of convergence,but one point on which they differ is tone. Whereas
most writings on theory with the notable exceptionof Derrida, are models of
scholarly solemnity, records of disillusionment and brave sacrifice of hu-
manistic positions, writers on hypertext are downright celebratory.Whereas
terms like deoth,vanish,loss,and expressionsof depletion and impoverish-
ment color critical theory,the vocabularyof freedom, energy,and empower-
ment marks writings on hypertextuality.One reasonfor thesedifferent tones
may lie in the different intellectual traditions, national and disciplinary from
which they spring. A more important reason,I propose,is that critical theo-
rists, as I have tried to show, continually confront the limitation-indeed,
what they somewhat prematurely take to be the exhaustion-of the culture
of print. They write from an awarenessof limitation and shortcoming, and
from a moody nostalgia,often before the fact, at the lossestheir disillusion-
ment has brought and will bring. Writers on hypertext, in contrast, glory in
possibility, excited by the future of textuality, knowledge, and writing. An-
other way of putting this opposingtone and mood is that most writers on crit-
ical theory howeverbrilliantly they may theorize a much-desirednew textu-
ality,nonethelesswrite from within daily experienceof the old and only of the
old. Many writers on hypertext, on the other hand, have alreadyhad some
experienceof hypertext systems,and they therefore write from a different
experientialvantagepoint. Most poststructuralistswrite from within the twi
light of a wished-for coming day;most writers of hlpertext write about many
of the samethings from within the dawn.

Many featuresof hypermedia derive from its creating the vir-

Virtual Presence tual presenceof all the authors who contribute to its materi-
als. Computer scientistsdraw on optics for an analogywhen
they speakof "virtual machines" createdby an operating systemthat provides
individual users sharing a systemwith the senseof working on their own in-
dividual machines.In the first chapter,when discussingelectronictextuality,

3.0 I pointed to another kind of"virtual" existence,the virtual text: all texts that
one encounterson the computer screenareyirtual, rather than real. In a sim-
ilar manner, the readerexperiencesthe virhral presenceof other contributors.
Suchvirtual presenceis ofcourse a characteristicofall technologyofcul-
tural memory basedon writing and syrnbolsystems.Sincewe all manipulate
cultural codes-particularly languagebut also mathematics and other syrn-
bols-in slightly different ways,eachrecord of an utteranceconveysa sense
of the one who makesthat utterance.Hypermedia differs from print technol-
ogy,however,in severalcrucial ways that ampllfy this notion of virhral pres-
ence.Becausethe essentialconnectivityof hypermediaremovesthe physical
isolation ofindividual texts characteristicofprint technology,the presenceof
individual authors becomesboth more availableand more important. The
characteristicflexibility of this reader-centered
information technologymeans,
quite simply, that writers have a much greaterpresencein the system,as
potential contributors and collaborativeparticipants but also as readerswho
choosetheir own paths through the materials.

The virtual presenceof other texts and other authors contrib-

Writing, utes importantly to the radical reconception of authorship,
authorial property, and collaboration associatedwith hyper-
text. Within a hypertext environment all writing becomescol-
laborativewriting, doubly so.The first element of collaborationappearswhen
one comparesthe roles of writer and reader,since the active reader neces-
sarily collaborateswith the author in producing the particular version of the
text she or he readsby the choicesshe or he makes-a fact much more obvi
ous in very large hypertextsthan in smaller hyperfictions.The secondaspect
of collaborationappearswhen one comparesthe writer with other writers-
that is, the author who is writing now with the virtual presenceof all writers
"on the system"who wrote then but whosewritings are still present.
The w ord collaboration,which derives from the Lattn for working plus that
for with or together,conveysthe suggestion,among others,of working side by
side on the same endeavor.Most people'sconceptionsof collaborativework
take the form of two or more scientists,songwriters, or the like continually
conferring as they pursue a project in the sameplaceat the sametime. I have
worked on an essaywith a fellow scholarin this manner. One of us would a
type a sentence,at which point the other would approve,qualify,or rewrite it,
and then we would proceedto the next sentence.Far more common a form
of collaboration,I suspect,is that secondmode describedas "versioning,"in

RECONFICURINC which one worker produces a draft that another person then edits by modi
THE AUTHOR fying and adding. The first and the secondforms of collaborativeauthorship
tend to blur, but the distinguishing factor here is that versioning takes place
out of the presenceof the other collaboratorand at a later time.
Both of these models require considerableability to work productively
with other people,and evidencesuggeststhat many peopleeither do not have
such ability or do not enjoy putting it into practice.In fact, accordingto those
who have carried out experiments in collaborativework, a third form proves
more common than the first two-the assembly-lineor segmentationmodel
of working together,accordingto which individual workers divide the overall
task and work entirely independently.This last mode is the form that most
people engagedin collaborativework choosewhen they work on projects
ranging from programming to art exhibitions.
Networked hypertext systemslike the World Wide Web, Hyper-G, Sepia,
and Intermedia offer a fourth model of collaborativework that combines as-
pects of the previous ones. By emphasizing the presenceof other texts and
their cooperativeinteraction, networked hypertext makes all additions to a
systemsimultaneouslya matter of versioningand of the assembly-linemodel.
Once ensconcedwithin a network of electronic links, a document no longer
existsby itself. It alwaysexistsin relation to other documents in a way that a
book or printed document neverdoesand nevercan. From this crucial shift in
the way texts exist in relation to others derivetwo principles that, in turn, pro-
duce this fourth form of collaboration:first, any document placedon any net-
worked systemthat supportselectronicallylinked materialspotentially exists
in collaborationwith any and all other documentson that system;second,any
document electronicallylinked to any other document collaborateswith it.
According to the AmericanHeritageDictionaryofthe EnglishLanguage,the
verb to collaboratecan mean either "to work together, especially in a ioint
intellectual effort," or "to cooperatetreasonably,aswith an enemy occupying
one'scountryJ'Thecombination of labor, political power, and aggressiveness
that appearsin this dictionary definition well indicatessome of the problems
that arise when one discussescollaborativework. On the one hand, the no-
tion of collaboration embracesnotions of working together with others, of
fotminga community of action. This meaning recognizes,as it were, thatwe
all exist within social groups, and it obviously placesvalue on contributions
to that group. On the other hand, collaborationalsoincludes a deepsuspicion
of working with others, something aestheticallyas well as emotionally en-
grained since the adventof Romanticism, which exaltsthe idea of individual
T. 0 effort to such a degreethat it, like copyright law often fails to recognize,or
even suppresses,the fact that artists and writers work collaborativelywith
texts createdby others.
Most of our intellectual endeavorsinvolve collaboration. but we do not
alwaysrecognizethat fact for two reasons.The rules of our intellectual cul-
ture, particularly those that define intellectual properfy and authorship, do
not encouragesuch recognitions, and furthermore, information technology
from Gutenbergto the present-the technologyof the book-systematically
hinders full recognition of collaborativeauthorship.
Throughout the past century the physicaland biologicalscienceshave
increasingly conceivedofscientific research,authorship, and publication as
group endeavors.The conditions of scientific research,according to which
many researchprojects require the cooperatingservicesof a number of spe-
cialists in the sameor (often) different fields, bear some resemblancesto the
medieval guild system in which apprentices,journeymen, and masters all
worked on a single complex project. Nonetheless,"collaborationsdiffer
depending on whether the substanceofthe researchinvolves a theoretical
science,such asmathematics,or an empirical science,such asbiology or psy-
chology.The former are characterizedbycollaborationsamong equals,with
little division of labor, whereasthe latter are characterizedby more explicit
exchangeof services,and more substantialdivision of labor" (Galegher,Egido,
and Kraut, 151).The financing of scientific research,which supports the
individual project, the institution at which it is carried out, and the costsof
educating new members of the discipline all nurture such group endeavors
and consequentconceptionsof group authorship.e
In general, the scientific disciplines rely on an inclusive conception of
authorship: anyonewho has made a major contribution to finding particular
results, occasionallyincluding specializedtechnicians and those who de-
velop techniques necessaryto carry out a course ofresearch, can appearas
authors of scientific papers,and similarly, those in whoselaboratoriesa proj-
ect is carried out may receiveauthorial credit if an individual project and the
publication of its results depend intimately on their generalresearch.In the
course of a graduatestudent'sresearchfor a dissertation,he or she may re-
ceivecontinual adviceand evaluation.When the student'sproject bearsfruit
and appearsin the form of one or more publications,the advisor'sname
often appearsas co-author.
Not so in the humanities, where graduatestudent researchis supported
largely by teaching assistantshipsand not, as in the sciences,by research
funding. Although an advisor of a student in English or art history often

RECONFIGURING acts in ways closelyparalleling the advisor of the student in physics, chem-
THE AUTHOR istry, or biology,explicit acknowledgmentsof cooperativework rarely appear.
Evenwhen a senior scholarprovidesthe student with a fairly preciseresearch
project, continual guidance, and accessto crucial materials that the senior
scholar has discoveredor assembled.the student does not include the advi-
sor as co-author.
The marked differences between conceptions of authorship in the sci
encesand the humanities demonstratethe validity of Michel Foucault's
observationthat "the 'author-function is tied to the legal and institutional
systemsthat circumscribe,determine, and articulatethe realm of discourses;
it doesnot operatein a uniform manner in all discourses,at all times, and in
any given culture it is not defined by the spontaneousattribution of a text to
its creator,but through a series ofprecise and complex procedures;it does
not refer, purely and simply, to an actual individual" ("What Is an Authorl"
131).One reason for the different conceptionsof authorship and authorial
property in the humanities and the scienceslies in the different conditions of
funding and the different discipline-politics that result.
Another corollary reason is that the humanistic disciplines, which tradi
tionally applyhistorical approachesto the areasthey study,considertheir own
assumptions about authorship, authorial ownership, creativity,and original-
ity to be eternal verities.loIn particular, literary studies and literary institu-
tions, such as departmentsof English,which still bathethemselvesin the
afterglow of Romanticism, uncritically inflate Romantic notions of creativity
and originality to the point of absurdity.An example comes readily to hand
from the prefaceof Lisa Ede and Andrea Lunsford'srecent study of collabo-
rativewriting, the production of which they discoveredto haveinvolved "acts
of subversion and of liberatory significance": "We began collaborating in
spite of concerned warnings of friends and colleagues,including those of
EdwardP.J. Corbett,the person in whosehonor we first wrote collaboratively.
Weknew that our collaborationrepresenteda challengeto traditional research
conventions in the humanities . Andrea's colleagues(atthe University of British
Columbia) said so when they declined to consider any of her coauthoredor
coeditedworks as part of a review for promotion" (ix-x).
Edeand Lunsford, whose interest in their subjectgrew out of the "differ-
ence between our personal experienceas coauthors and the responsesof
many of our friends and colleagues"(5),set the issue of collaborativewriting
within the contextsof actualpracticein the worlds of businessand academia,
the history of theoriesof creativeindividualism and copyright in recentWest-
ern culture, contemporaryand feminist analysesof many of theseother con-
T. 0 texts.They producea wide rangeof evidencein convincinglyarguing that "the
pervasivecommonsenseassumption that writing is inherently and necessar-
ily a solitary,individual act" (5) supportsa traditional patriarchalconstruction
of authorship and authority. After arguing against "univocal psychological
theoriesof the self" (132)and associatednotions of an isolatedindividualism,
Edeand Lunsford call for a more Bakhtinian reconceptionof the self and for
rather than a hierarchical,mode of collaboration.
what they terma d.ialogic,
I shall return to their ideaswhen I discussthe role of hypertextin collab-
orativelearning, but now I wish to point out that as scholarsfrom Mcluhan
and Eisensteinto Ede and Lunsford havelong argued,book technology and
the attitudes it supports are the institutions most responsiblefor maintain-
ing exaggeratednotions of authorial individuality, uniqueness, and owner-
ship that often drastically falsify the conception of original contributions in
the humanities and conveydistorted pictures of research.The sciencestake
a relativelyexpansive,inclusive view ofauthorship and consequentlyoftext
ownership.ll The humanities take a far more restrictedview that emphasizes
individuality, separation,and uniqueness-often creating a vastly distorted
view of the connection of a particular text to those that precededit. Neither
view possessesan obviousrightness. Eachis obviouslya socialconstruction,
and each has on occasionproved to distort actual conditions ofintellectual
work carried out in a particular field.
Whatever the political, economic, and other discipline-specificfactors
that maintain the conception of noncooperativeauthorship in the humani-
ties, print technologyhas also contributed to the senseof a separate,unique
text that is the product-and hencethe property-of one person,the author.
Hypertext changesall this, in large part becauseit doesawaywith the isola-
tion of the individual text that characterizesthe book. As Mcluhan and other
studentsof the cultural influence of print technologyhavepointed out, mod-
ern conceptions of intellectual properry derive both from the organization
and financing of book production and from the uniformity and fixity of text
that characterizesthe printed book. |. David Bolter explains that book tech-
nology itselfcreated new conceptionsofauthorship and publication:

Because a bookisa costlyandlaborious
nityto become
An authoris a personwhosewordsarefaithfully
meantlessin the ageof manuscripts,
to crossoverandbecome

RECONFIGURING or put one'snotesin aformforothersto

onesimplysatdownandwrotea treatise
THE AUTHOR read.Oncethetreatisewaswritten,therewasno difference it andtheworks
of other"published"writers,exceptthatthe morefamousworksexistedin more

Printing a book requires a considerableexpenditureofcapital and labor, and

the needto protectthat investmentcontributesto notions of intellectualprop-
erty. But these notions would not be possible in the first place without the
physically separate,fixed text ofthe printed book. fust as the need to finance
printing of books led to a searchfor the large audiencesthat in turn stimu-
lated the ultimate triumph of the vernacular and fixed spelling, so, too, the
fixed nature of the individual text made possible the idea that each author
producessomething unique and identifiable as property.
The needs of the marketplace,as least as they are conceivedby editors
and publishing houses,reinforce all the worst effectsof theseconceptionsof
authorship in both academicand popular books.Alleen PaceNilsen reports
that Nancy Mitford and her husband wrote the best-selling The High Costof
Deathlogerher,but only her name appearsbecausethe publisher urged that
multiple authors would cut sales.Another common solution involvesresort-
ing to a pseudonym: Perri O'Shaunessyis the pen-name of Pam and Mary
O'shaunessy,createdwhen their editors would not permit a double-author
byline, and John Case,"author" of The GenesisCode,"is really husband-and-
wife team Jim and Carolyn Hougan."12In another case,to make a book more
marketable a publisher replacedthe chief editor of a major psychiatric text-
book with the name of a prestigious contributor who had not edited the vol-
ume at all (citedby Edeand Lunsford, 3-4). I am sure everyonehas examples
ofsuch distortion ofauthorial practiceby what a publisher believesto be good
business. I have mine: a number of years ago after an exercisein collabora-
tive work and writing with three graduatestudents produced a publishable
manuscript, we decided by mutual agreement on the ordering of authors'
names on the title page. By the time the volume appeared,the three former
graduatestudentsall held teachingpositions,and its appearance,one expects,
might havehelped them professionally.Unfortunately,the publisher insisted
on including only the first editor's name in all notices, advertisements,and
catalogues.Such an action,ofcourse, doesnot haveso seriousan effectas
removing the editors' names from the title page, but it certainly discrimi
natesunfairly betweenthe first two editors,who did equal amounts of work,
and it certainly conveysa strong messageto beginning humanists about the
culturally assignedvalue of cooperationand collaboration.

3.0 Even though print technology is not entirely or even largely responsible
for current attitudes in the humanities toward authorship and collaboration,
a shift to hypertextsystemswould changethem by emphasizingelements
of collaboration.As Tora K. Bikson and ]. D. Evelandpoint out in relation to
other, nonhumanities work, "The electronic environment is a rich context
in which doing work and sharing work becomesvirrually indistinguishable"
(286).If we can make ourselvesawareof the new possibilitiescreatedby these
changes,we can at the very leasttake advantageofthe characteristicqualities
of this new form of information technology.
One relevant characteristic quality of networked hypertext systems is
that they produce a sense ofauthorship, authorial property, and creativity
that differs markedly from those associatedwith book technology.Hypertext
changesour senseof authorship and creativity(or originality) by moving away
from the constrictions of page-boundtechnology.In so doing, it promises to
have an effect on cultural and intellectual disciplines as important as those
producedby earlier shifts in the technologyof cultural memory that followed
the invention of writing and printing (seeBolter, Mcluhan, and Eisenstein).

Collaborativework in hypertexttakes many forms, one of the

Examplesof Collaboration most interesting of which illustrates the principle that one al-
most inevitably works collaborativelywhenever creating doc-
in Hypertext
uments on a multiauthor hypertext system. I discoveredthe
inevitably collaborative nature of hypermedia authorship in the old Inter-
media days.While linking materials to the overview(or sitemap)for Graham
Swift's Waterland(19831,IobservedNicole Yankelovich,project coordinator
of the Intermedia project at IRIS, working on materials for a coursein arms
control and disarmament offered by Richard Smoke of Brown University's
Center for Foreign PolicyDevelopment.Thosematerials,which were created
by someone fiom a discipline very different from mine for a very different
kind of course,fil1eda major gap in a project I was working on. Although my
co-authorsand I had createdmaterials about technology,including graphic
and text documents on canalsand railroads,to attachto the scienceand tech-
nology sectionof the Woterlandoverviewwe did not havethe expertiseto cre-
ate parallel documents about nuclear technology and the antinuclear move-
ment, two subjectsthat play a significant part in Swift'snovel. Creatinga brief
introduction to the subject of Waterlandand nuclear disarmament, I linked
it first to the scienceand technology section in the Waterlandovewiew and
then to the time line that the nuclear arms course materials employ as a
directory file. A brief document and a few links enablestudents in the intro-

RECONFIGURING ductory survey of English literature to explore the materials created for a
THEAUTHOR coursein another discipline. Similarly, students from that coursecould now
encounter materials showing the effectson contemporaryfiction of the con-
cerns coveredin their political sciencecourse.Hypertext thus allows and en-
couragescollaborativework, and at the same time it encouragesinterdisci-
plinary approachesby making materials createdby specialistsin different
disciplines work together-collaborate.
This kind of collaboration-by-linkoccurs all the time on the World Wide
Web. Each time a student or faculty member from another institution has
one of their documents addedto The VicrorianWeb-say, on characterization
or race,class,and genderinJane Eyre-they automaticallyjoin in a discussion
on these topics. Similarly, Phil Gyford'stranslation of Pepys'sDiaies into a
Weblog,at which we looked in the previous chapter,exemplifiesyet another
approachto collaborationon the Web.
The important point here is that hypermedia linking automaticallypro-
ducescollaboration.Looking at the way the arms control materials joined to
those supporting the four English courses,one encountersa typical example
of how the connectivity that characterizeshypertext transforms independ-
entlyproduced documentsinto collaborativeonesand authorsworking alone
into collaborativeauthors. When one considers the arms control materials
from the point ofview oftheir originator, they exist as part ofa discretebody
of materials. When one considersthem from the vantagepoint of a reader,
their statuschanges:as soon as they appearwithin a hypertext environment,
these and all other documents then exist as part of a larger system and in
relation therefore to other materials on that system. By forming electronic
pathwaysbetween blocks of texts, links actualizethe potential relations be-
tween them. fust as hypertext as an educational medium transforms the
teacherfrom a leaderinto a kind of coachor companion, hypertext as a writ-
ing medium metamorphosesthe author into an editor or developer.Hyper-
media, like cinema and video or opera,is a team production.

Sincewriting hypermedia successfullyinvolvesfinding ways

The ProblematicConcept to prevent readersfrom becoming confusedand discouraged
when they encounter text in e-space,let us examine this
of Disorientation
notion of disorientation before considering some of the meth-
ods used to prevent it. Crucial as disorientation might seemto discussionsof
hypertext authoring, this term remains unexamined and inadequatelyde-
fined. Sucha claim might appearparticularly odd becausewriters on the sub-
ject since feffConklin have apparentlyprovided fairly precise statementsof
what they mean by what Conklin himself term ed the d.isonentation
According to his initial statementof the issue,disorientation seemsto inhere
in the medium itself; "Along with the power of being able to organize infor-
mation muchmore complexlycomesthe problem ofhavingtoknow (1)where
you are in the network and (2) how to get to some other place that you know
(orthink) existsinthe network. I callthis the disoientationproblem.Of course,
one has a disorientation in traditional linear text documents, but in a linear
text the readerhas only two options: He can searchfor the desiredtext earlier
in the text or later in the tert" (38). Kenneth Utting and Nicole Yankelovich,
who similarly point out that "hypermedia . . . has the potential to dramatically
confuse and confound readers,writers, teachers,and learners,"quote Con-
klin's definition of disorientation as "the tendencyto lose one'ssenseof loca-
tion and direction in a nonlinear document" (58), and in their example of
three aspectsof disorientation, they mention "confusion about where to go
or, having decidedon a destination,how to get there,"and also disorientation
in the senseof not knowing "the boundaries of the information space"(61)
one is exploring.

RECONFIGURING Three points here demand notice. First, the conceptof disorientation re-
WRITING latescloselyto the tendencyto use spatial,geographical,and travelmetaphors
to describethe way users experiencehypertext. Such uses are obviously
appropriateto dictionary definitions of disorient.Neither TheAmeican Her-
itageDiaionary nor Webster's
CollegiateDicNionarydefines disorientation,but
accordingto TheAmeican HeitageDictionory,Iodisorientis "to causeto lose
one'ssenseof direction or location, as by removing from a familiar environ-
ment," and Webster'soffers three definitions of d.isoient:(1) "to causeto lose
one'sbearings: displacefrom normal position or relationship"; (2) "to cause
to lose the senseof time, place,or identity"; and (3) "to confuse."
In general, authors writing about hypertext seem to mean confuseand
specificallylosebeoingswhenthey use the term, and this usagederivesfrom
commonplaceapplicationof spatialmetaphorsto describethe reader'sbehav-
ior in a hypertextenvironment. Thus, in "The Art of Navigatingthrough Hy-
pertext," ]akob Nielsen points out in the usual formulation that "one of the
major usability problems with hypertext is the user's risk of disorientation
while navigating the information space.For example, our studies showed
that 56 percent of the readersof a document written in one of the most pop-
ular commercial hypertext systems agreed fully or parnally with the state-
menl I wosoftenconfusedaboutwhereI was" (298).Nielsen believesthat "true
hypertext should also make usersfeelthat they can move freely through the
information accordingto their own needs" (298).
Second,as Conklin and others wdting in this field statethe problem of
disorientation, it obviously concernsthe design of the information technol-
ogy alone. In other words, the related conceptsofdisorientation and confu-
sion appear,in their terms, to havenothing to do with the materials,the con-
tent, on the hlpertext system. Nonetheless,we all know that readers often
experienceconfusion and disorientation simply becausethey fail to graspthe
logic or even meaning of a particular argument. Even if the works of Kant,
Einstein, and Heideggerwere to appearon the finest hypertextand informa-
tion retrieval system in the world, they would still disorient many readers.
Although Conklin and other studentsof hypertexthavenot naivelyor incom-
pletely defined what they mean by disorientation, their restriction of this
term to system-generated
disorientationin practicedoesnot takeinto account
a large portion of the actual reading experience-and its implications for
hypertext authors. The issue has a bearing on a third point about the notion
of disorientation.
Third, disorientation, as these comments make clear, is conceivedby
these authors as crippling and disenabling, as something, in other words,

3.0 that blockscompletionofa task one has set for oneselfor that has been set
for one by others. Disorientation, furthermore, is presentedas such a mas-
sive, monolithic problem that these authors pay little or no attention to
how people actuallycopewith this experience.Is it, in fact, crippling, and do
users of hypertext systemssimply give up or fail in whatevertasks they have
engagedthemselveswhen they meet disorientation) As we shall see,expert
users ofhypertext do not alwaysfind the experienceofdisorientation to be
particularly stressful, much less p ar alyzing.
The role of disorientation in literature suggestssome reasonswhy this
might be the case.Readersof literature in fact often describethe experience
here presented as disorientation as pleasurable,even exciting, and some
forms of literature, particularly those that emphasize either allegory or sty,
listic and narrative experimentation, rely on disorienting the readeras a pri
mary effect.Although the kind of pleasurabledisorientation that one finds in
Dante's Divine Comedy,Browning's Ring ond the Book,and Eliotk Wasteland
derivesfrom what we havetermed the content and not from the information
technologythat presentsit, this effect has one important parallel to that en-
countered in some forms of hypertext: in each casethe neophyte or inexpe-
rienced readerfinds unpleasantlyconfusing materials that more expert ones
find a sourceofpleasure.

The reasons for the radically different ways people in the

The ConceptofDisorientationin humanities and technologicaldisciplines regard disorien-
tation become particularly clear in three areas-aesthetic
the Humanities
theories of disorientation, conceptions of modernism and
postmodernism as cultural movements, and the related conceptionsof
hypertext fiction.
The classicstatement of the positive value of cognitive and other disori
entation in aestheticworks appearsin Morse Peckham'sMonisRageforChaos:
Biology,Behavior,and the Arrs $967), which arguesthat "art offers not order
but the opportunity to experiencemore disorder than any other human arti
fact, and . . . artistic experience,therefore, is characterized. . . by disorienta-
tion" (41).According to him, "the artist'srole is to createoccasionsfor disori-
entation, and of the perceiverkrole to experienceit. The distinguishing mark
of the perceiver'stransaction with the work of art is discontinuity of experi-
ence, not continuity; disorder, not order; emotional disturbance, not emo-
tional catharsis,even though some works have a cadentialclose" (254).Hu-
man beings so "passionately"want "a predictable and ordered world" that
"only in protected situations, characterizedby high walls ofpsychic insula-

RECONFIGURING tion," can they permit themselvesto perceivethe gapbetween "expectancyor

WRITINC set or orientation, and the data . . . interaction with the environment acrually
produces.. . Art offers preciselythis kind ofexperience" (313).
Peckham argues finally that aft is "an adaptational mechanism" that
reinforces our ability to survive:

Art is rehearsal
in whichit is vitalfor oursurvival
to endure
cognitivetension,to refusethe comfortsof validationby affective
is inappropriate
because areat stake;art is
too vitalinterests
to enduredisorientation
sothata realandsignifi-
cantproblemcanemerge. to the tensionsandproblems
Art is the exposure of a
himselfto andproblems
thetensions of
a realworld.(314)

Peckhamt positiveviews of aestheticdisorientation,which seemto grow out

of the arts and literature of modernism, dearly present it as a matter of free-
dom and human development.
Studentsof literature and the arts havelong emphasizedthe role of dis-
orientation in both modernism and postmodernism. Like the works of the
cubists,expressionists,and other movementsof twentieth-centuryart, Iames
T. S. Eliot's Wasteland,and William Faulkner's Soundond the
Fury-to cite three classicsof literary modernism-all make disorientation
a central aestheticexperience.Similarly, as recent writers on postmodernist
fiction point out, it is characterizedby a range ofqualities that produce cog-
nitive disorientation:"contradiction,discontinuity,randomness,""inffactable
epistemologicaluncertainty,"and "cognitiveestrangement"(McHale,7, L1,,59).
These attitudes,which students of the past century's culture almost univer-
sally view positively,appearthroughout discussions of hypertext fiction as
well. RobertCoover,for example,makesquite clearthe relationsbetweendis-
orientation,hypertext,and the traditions of the avantgardewhen he describes
the way hypertext fiction promises to fulfill the liberating functions of the
experimentaltradition in fiction.1He alsoemphasizesthe effecton writers of
this disorienting freedom. Discussingthe conservatismof writing students,
he claims that

themto consider or i nnovative
tryingoutalternative thantalk-
ingthemintochastityasa life-style.
Butsuddenly, they
with hyperspace,
haveno choice:
allthecomforting havebeenerased.
structures or go
home.Somefranticallyrebuild somejustgetlostanddriftout
of sight,mostleapin fearlessly
withoutevenaskinghowdeepit is (infnitelydeep),

3.0 admitting, to paddle
evenastheycontinue fordearlife,thatthisnewarenais indeed
a potentially
to trans-
formtheveryartoffiction.("Endof Books,"

Michael foyce describes potentially disorienting qualities of hypertext

fiction in terms that praise the necessaryactivism required of readers:"Con-
structive hlpertexts require a capabilityto act: to create,to changeand to re-
coverparticular encounterswithin the developingbody of knowledge.These
encounters. . . are maintained asversions,i.e., trails, paths,webs,notebooks,
etc.;but they areversionsof what they arebecoming,a structurefor what does
not yet exist" (Of Two Mind,s,42).In m:uchthe samevein StuartMoulthrop, like
Coover,relatesthe experienceofencountering gaps and disorientation that
characterizethe reader'sexperiencein hlpertext as potentially liberating. "ln
a world where the 'global variables' of power and knowledge tend to orient
themselvestoward singular, hegemonic world orders, it becomes increas-
ingly difficult to jump outside 'the system.'And asThomas Pynchonreminds
us: 'Living inside the Systemis like riding acrossthe country in a bus driven
by a maniac bent on suicide' (Gravity'sRainbow,4l2)"("Beyondthe Electronic
Book,"76).Given the fact that many humanities users of hypertext,like those
specificallyconcernedwith hypertextfiction, associatethe generalexperience
ofdisorientation with avant-garde,liberating,and culturally approvedaesthetic
experience,it should be no surprise that they treat the issue ofdisorientation
far differently than do almost all who consider it in the technical disciplines.

In experimentsthat Paul Kahn and I conductedin 1991expe-

The Loveof Possibilities rienced student-usersofhypertext showeda love ofbrowsing
and of the serendipity it occasionsvery much at oddswith by-
now conventionalattitudes toward disorientation in hypertext. For example,
one user explained that by "accidentally clicking" on a parficular link he
found that he had made a "delightful detour,"since it led to an answerto one
of the problems in information retrieval. "Although I guessthis mistake has
an analog in book technology,it would be the improbable act of being in the
wrong section of the library the wrong row of books, the wrong shelf, pick-
ing up the wrong book, and opening up magically to the correct page."
Some of these responseswere disconcertingly unexpectedand for that
reason potentially quite valuable to anyone considering the design, imple-
mentation, and educationalapplicationof hypertext.In two casesvery experi-
encedprogrammershad more difficulties with certain aspectsof information-

RECONFIGURING retrieval tasks than did comparative neophytes. It would appearthat their
WRITING expectationsof systems and retrieval mechanisms servedto hinder rather
than to assisttheir explorations.Accustomedto using full-text searchmech-
anisms in other kinds of computer systems,one of these students spent
some fifteen minutes searchingfor one in Intermedia-the version used
did not havethe system'slater searchtools-and then gaveup on the assign-
ment, assuming that no other methods of locating the information existed.
In contrast,a relatively unsophisticateduser solvedthe first problem of
locating works by one scholar in a matter of moments. As he explained: "I
found these referencesby opening the Critics Quoted Document in the Bib-
liographical Folder in the Dickens Folder . . . Total Time: 6 minl' Another
similarly responded,"I answeredthe first question of the assignment using
fthe Intermedia folder system].Sincethe folders were labeledwell, I found it
quick and easyto first find the 'Bibliography' folder, and then open the 'Crit-
ics Quoted' document. There I found the names of the three authors in the
question. Sincethe web was alreadyengaged,I could activatethe link mark-
ers and seeall the destination documents connectedto a particular author
(if in fact the web was well linked). Thus, I approachedthe web from an odd
angle,from the actualdocument folders, but it was the one which I felt to be
the easiestand quickest for this question. This same information could be
found in the Bibliography Overview. . . If I had never come acrossthe Victo-
rianism Overview,for whateverreason, I might never have come acrossthe
sought after bibliographic information. But I did find the information, out-
sidethe system's(few)attemptsat organization. I felt so comfortablewith the
sight of the Macintosh document icon that I felt there was no 'violation,' as
Intermedia depictsno structure to be violated-documents seemto be either
autonomous or within the web" (MF).
This student's narrative forcefully restatesthe truism that people who
want to find information will find it as much by what they know about that
information as by systemfeaturesalone. In other words, odentation by con-
tent seems able to solvepotential problems of disorientation causedby the
system design consideredin isolation. In this case,some experiencedusers
of computer tended to conceivethis task as a means of testing systemcapac-
ities whereasthose who were content experts,or who took the approochof a
contentexpert,conceivedthe task in terms that made the desiredinformation
the center of the task.
One important lessonfor both designersof hypertext softwareand those
who teach or write with hypertext appearsin the problems encounteredby
the students with more computer skills. In relying too heavily on system

T. 0 features,they implicitly made the assumption that the system, rather than
the author, does most of the work. In doing so, they tended to ignore the
stylistic and other author-createddevicesthat made the searchquick and easy
for a majority of users.
We should also note that a preferencefor browsing up to and including
the senseof "disorientation" can createdisconcerting results for hypertext
designers,despitethe fact that hypertert theorists often praisethis approach
to wandering through a database.For example,one user criticized one of the
systemspreciselybecauseit proved "more difficult to becomedisoriented in
the good way that Intermedia and Storyspacetend to facilitate. I found that
links continually brought me back to crossroadsor overviews,rather than to
other documents. For this reason I felt less like an activereader.Orientation
devicessuch as these explained and categorizedlinks rather than allowing
me to make my own connections and categories"(AM). To those who find
disorientation a negativequality, these comments might seem puzzling,
becauseapparentlynegativequalities here come in for praise.In fact,this stu-
dent specificallymentions "the good way" Intermedia and Storyspacecreate
a senseof disorientation, which she takes to be a condition that empowers
hypertextusers becauseit placesthem in an activerole-one particularly ap-
propriate to this new information medium.
The reactionsofthese student-evaluatorssuggestsix points about reader
disorientation. First, although it representsa potentially significant problem
in some systems,a priori concernsabout it may well arise from lack of expe-
rience with hypertext systems, specifically from attempting to apply read-
ing and information-retrieval protocolsappropriateto booktechnologyto this
Second,what one readerexperiencesas disorientation, another may find
Third, disorientation has quite different connotations in the writings of
those basedin technologicalas opposedto literary disciplines. The techno-
logically basedconceptionofdisorientation relatesto a conceptionofeduca-
tion essentially limited to factual information. Literary or humanistic as-
sumptions about disorientation seem relatedto a conceptionof educationin
which studentslearn to deal with complex matters of interpretation.
Fourth, disorientation-let me emphasize this point yet again-arises
both in the normal act of reading difficult material and in pooily designed
systems.Knowledgeof content, as some of our evaluatorsdemonstrate,has
to be consideredas part of any solution to issuesof system-generated

RECONFIGURING Fifth, since for the foreseeablefuture, book and electronic technologies
WRITING will exist together,in some applicationssupplementing,in others competing
with, eachother, designersof hypertext systemswill continue to find them-
selvesin a terribly difficult situation. Systemsthey design will almost cer-
tainly encounter a heterogeneouspool of users, some still trying to read
accordingto the rules of books,others,increasinglysophisticatedin electronic
media, who find the specific qualities of hypertext reading and exploration,
including occasional"disorientation,"as pleasurable,desirablequalities.
Sixth-and most important-writing, asmuch assystemdesign,asmuch
assoftware,preventsthe lesspleasantforms of disorientation.We must there-
fore developa rhetoric and stylistics ofhypertert writing.

Linking, by itself, is not enough. The hypermedia author can-

The Rhetoricand Stylisticsof not realize the enormous potential of the medium simply by
linking one passageor image to others.The act of connecting
Writingfor E-Space;
or, How
one text to another fails to achieveall the expectedbenefits of
Should We Write Hypertextl hypermedia and can even alienate the user. On the briefest
consideration,such a recognition will hardly surprise, since
authors of print essays,poems, narratives,and books do not expecllo write
merely by stringing together sentencesand paragraphswithout the assis-
tance of stylistic devicesand rhetorical conventions.If to communicate effec-
tively,hypermedia authors must employ devicessuited to their medium, two
questions arise. First, what are the defining characteristicsor qualities of
hypertext as reading and writing mediuml Second,to what extent do they
depend on specific hardware and softwarel What effect, for example, does
the presenceor absenceof color, size of onek monitor, and the speedof one's
computer haveon reading hypertextl
Then there are questions less immediately derived from the hardware.
Assuming that writing at the levelof phrase,sentence,and paragraphwill not
change in some fundamental way-and this, I admit, may be too large an
assumption to make at this stage-what new forms of organization, rheto-
ric, and structure must we developto communicate effectivelyin electronic
spacel In other words, if hypertext demandsa new rhetoric and a new stylis-
tics, of what do they consist,and how if at all, do they relateto issueslike sys-
tem speedand the likel
To begin, let us look once again at the nature of the medium. I havejust
written that "hypertextchangesthe way textsexistand the way we readthem,"

3.0 and in earlier chapterswe have observedmany examplesof such difference
from chirographic and print textuality.Whether or not it is true that the digi
tal word producesa secondaryor new kind of orality, many of the devices
required by hypertext appearin oral speech,just as they do in its written ver-
sions or dialects.Many of thesedevicesto which I wish to direct our attention
fall into a single category:they announcea changeof direction and often also
provide some indication of what that new direction will be. For example,
and on the other hand give
words and phrases like in contrast,nevertheless,
advancenotice to listeners and readers of something, say,an instance or
asserLion,is coming contraryto what has comebefore. Forexampleannounces
a categoryshift asthe discourseswitches,most likely, from generalor abstract
statementto proposedinstancesof it. Causalor temporal terms, such as
or afi.er,similarly readylisteners for changesof intellectual direction.
In both print and oral communication, they are means, in other words, of
preparing us for breaksin a linear stream of language.One must take carein
using this termlinear since, as we have alreadyseenwhen looking at hyper-
text narrative, all experienceof listening or reading in whatever medium is
in an important senselinear, unidirectional. Thus, although readers-or, to
be precise, readings-take different paths through afiemoon,PatchworkGirI,
or Quibbling,each path-each experienceof reading-takes the form of a
sequence.It is the tert that is multisequential, not a particular reading path
through it. I emphasizethis obvious point becausethe problem of preparing
for change of direction (and openings and closings are also such changes)
has been with us since the beginnings of human language.
Sincehypertextand hypermedia are chiefly defined by the link, a writing
devicethat offers potential changesofdirection, the rhetoric and stylisticsof
this new information technology generally involve such change-potential
or actual change of place, relation, or direction. Before determining which
techniques best accommodatesuch change,we must realize that, together,
they attempt to answer severalrelated questions: First, what must one do to
orient readersand help them readefficiently and with pleasurel Second,how
can one help readersretracethe stepsin their reading pathl Third, how can
one inform thosereading a document where the links in that document leadl
Finally,how can one assistreaderswho havejust entereda new document to
feel at home therel
Drawing on the analogyof travel, we can say that the first problem con-
cerns oientationinformatron necessaryfor finding one'splacewithin a body
of interlinked texts. The secondconcerns navigationinformation necessary
for making one's way through the materials. The third concerns exit or
RECONFIGURING departureinformationand the fourth aftivolor entranceinformation.In each
WRITING case, creators of hypermedia materials must decide what readers need to
know at either end of a hypermedia link in order to make use of what they
find there. The generalissue here is one of interpretation. More specifically,
to enablevisitors to this new kind of text to read it pleasurably,comfortably,
and efficiently how much interpretation must the designer-authorattachto
the systemasa whole, to link pathways,and to documentsat the end of linksl
Unfortunately, no analogymaps reality with complete accuracy.Naviga-
tion, the art ofcontrolling the courseofa plane or ship, presupposesa spatial
world, but one does not entirelyexperiencehypertext as such. In navigation,
we remember, one must determine one'sspatialposition in relation to land-
marks or astrallocationsand then decideon a means of moving toward one's
goal,which lies out of sight at some spatial distance.Becauseit takestime to
move acrossthe separatingdistance, one also experiencesthat distance as
time: one's ship lies so many nautical miles and therefore so many daysand
hours from oneb goal.The reader,however,doesnot experiencehypertextin
this way.The readerof ParadiseLost,forexample,experiencesas equallyclose
the linked parts of Homer and Vergil to which the poem's opening section
allude and linked lines on the next pageor in the next book (seeFigwe 72).
Becausehypertert linking takes relatively the same amount of time to tra-
verse,all linked terts are experiencedaslying at the same "distance"from the
point of deparhrre. Thus, whereas navigation presupposesthat one finds
oneself at the center of a spatial world in which desired items lie at varying
distancesfrom one'sown location,hypertext (andother forms of addressable,
digital textuality) presupposesan experiential world in which the goal is
alwayspotentially but one jump or link away.

GeneralObservations.Hlpermedia as a medium conveysthe strong impres-

sion that its links signify coherent,purposefirl, and above all usefulrelation-
ships. From which follows that the very existenceof links conditions the
readerto expectpurposeful, important relationships betweenlinked materials.
One of the presuppositionsin hypertext,particularly when appliedto educa-
tional uses,is that linking materials encourageshabits of relational thinking
in the reader. Such intrinsichypermedia emphasis on interconnectedness
(or connectivity)providesa powerful means of teaching sophisticatedcritical
thinking, particularly that which builds on multicausal analysesand relating
different kinds of data. But since hypermedia systems predisposeusers to
expectsignificant relationshipsamong lexias,those that disappointthese
expectationstend to appearparticularly incoherent and without significance.

3.0 When users follow links and encounter materials that do not appearto pos-
sessa significant relation to the document from which the link pathwayorig-
inated,they feel confusedand resentful. In reading materialson the Web,the
delays encountered by users tend to exaggeratethis effect, thus providing
anotherreasonfor avoidingtime-consuming graphicor other elementswhen-
everpossibleif one wishes to include an audiencewithout high-speedaccess
to the Internet.

System-Cenerated Meansof ReaderOrientation.Devicesof orientationper-

mit readers(1) to determine their present location, (2) to have some idea of
that location'srelation to other materials, (3) to return to their starting point,
and (4) to explore materials not directly linked to those in which they
presentlyfind themselves.
The graphic presentation of information embodied in the useful, if lim-
ited, desktopmetaphor provesan especiallyeffectivemeans of reader orien-
tation in the systemsthat use them, but World Wide Web browsers,in which
the risks ofdisorientation are particularlygrave,do not. Ofcourse, the "Show
Location"window in IE, Firefox, Safari,Netscape,and other HTML browsers
doesprovide the exactaddressof a lexia, and as I write today,following a link
to one student'sessayon PatchworkGirI, say,Lars Hubrich's "Stitched lden-
tity," would produce the following information in the location window:
http: / /www.cyberspaceweb. ory lhtI pg llhpatch.html. Such information not
only appearsin a form daunting to most readers,it fails to be very helpful on
two counts:first, the need to createeconomicallybrief directory names often
rendersthe file incomprehensibleto all but the person who maintains a web-
site and, second,it provides very little information about the relation of this
particular lexia to the information spaceit inhabits.
One way of providing the benefits of graphic presentation of a folder
structure takes the form of fava applets,such as those createdby Dynamic
Diagrams for IBM's website (Figure 13),which generatean animated three-
dimensional image of an individual lexia's location within a file structure.
Where such softwaresolutions are not available,authorshave employedtwo
solutions. One involves organizingan entire site accordingto what is essen-
tially a folder structure and then making that organization clear.Thus, Susan
Farrell'sArt-Crimes,a sile containing graffiti from around the world, presents
its information in terms of country, city, and additional collectionsfor each
city. Of course,this beautiful site, which provides a visual archive,has little
intrinsically hypertextual about it and therefore cannot serveas an example
for other kinds of webs.
lha Afis inMetarian Britdn
stg.brown. edulprajects

Figure13. A View of lhe VictorianWeb.Thismap was createdby DynamicDiagramsMAPAfrom the vantagepoint ofthis web's

homepage(main sitemapor overview).In this screenshot, a user has activateda pop-upwindow displayingthe URLand title of
uTheArts in Victorian Britain," a second-level
overview.By using a computer mouse to move the cursor farther awayfrom the top
level,userscan also learnthe titles of lexiaslinkedto this and other overviews.Doubleclickingon the icon for any lexiaopens it.

DynamicDiagrams.Used by permission.)

Keeping(the)Track:Where'veI Been,What Did I ReadlIn additionto helping

readers discern their general location within an information spaceat any
moment, systemsalso haveto provide both some means of informing them
from whence they came and a means of allowing them to return. As one of
its functions, the Roadmapin Storyspaceprovides a sequentiallist of lexias
that one has visited and provides a slightly cumbersomemeans of returning

3.0 to any one. Web browsersalso havemeans of providing current reading his-
tory: mousing down on the "Go" or "History" menu at the top of the screen
produces a chronologically ordered list oflexias one has opened.2Current
versionsofwebviewers,induding Internet Explorerand Safari,solvethis prob-
lem by retaining lists of the web documentsone hasvisited; Explorerpermits
the user to specifyhow many sites should be retained.
Another valuable orientation tool takes the form of permanent bookmarks
in World Wide Web browsers,a feature anticipatedby HyperCard-basedsys-
tems like Voyager ExpandedBook, Keyboard,and Toolbook. Such a book-
mark function permits readersto record placesto which they might like to re-
turn at some future time, and when designing large websites,most of whose
Iinks remain internal, authors can advisereaderswho contemplatefollowing
links to materials offsite that they might first wish to use their viewerk book-
mark facility, thus making return easier,particularly in a complex session.
One activatesa bookmark simply by choosingit from a list availableat the top
of the screen.Although thesedevicesplay an important role in allowing read-
ers to customize their own experiencesof the webs they read, they do not
compensatefor the absenceof long-term readinghistories in very large,com-
plex corpora,such as one finds on the Web.

Dynamicand StaticTablesof Contents.One often encountersthe tableof con-

tents, a devicedirectly transferredfrom book design,in hypertextdocuments,
often to very goodeffect.Readersof materialson the Webwill havefrequently
encounteredit since a good many homepagesand title screensconsistessen-
tially of linked tablesof contents.I haveused it mysel{,particularly when cre-
ating hypertextversionsof print materials,a subject I shall discussat greater
length in a separatesectionbelow.The World Wide Webversion of the open-
ing chapter of the first version of this book, for instance, employs two such
contents screens,one for the entire volume and a secondfor the first chapter,
the only one availableon the web. Although such a table of contentsprovides
a familiar, often effectivemeans of presenting a work's organization, in its
static form it often overemphasizesthe element of the electronicbook to the
detriment of its hypertexfuality.
ElectronicBookTechnologies'DynaText,which featuresa dynamic, auto-
matically generatedcontents screen,offers a much more powerful version of
this device.DynaTextusestext in the form of SGML, a far richer, more pow-
erful older relation of the Web'sHTML. Since SGML requires that one begin
and end everychaptertitle, sectionheading,and all other text structureswith
specific tags (markup), DynaText employs this information to produce an

RECONFICURING automatically generatedcontents screen,which authors and designers can

WRITING arrangeto appeara particular placeon the screen.ln Hypefiert in Hyperturt,
the electronic version of this book's first version, this contents section ap-
pearedto the left of the main text.
This electronictable ofcontents differs in severalwaysfrom the staticver-
sionsone encountersin the printed book and on the Web.First, clicking on an
icon near the title of a chapterimmediately causesthe sectionheadingsnext
level down to appear,and clicking on them in turn displayssubheadings,and
so on. Essentially,this dynamic table of contentsactsvery much like Nelson-
ian stretchtext.SinceI had addedadditional subdivisionsto almost everysec-
tion betterto suit readingin an electronicenvironment,this featurepermitted
Hyperturtin Hypefiertb displayboth the book'soriginaTorganizatlonand the
added elements as readersneededthem. The secondpoint at which Dyna-
Text'sdynamic contentsscreendiffers from static onesis that clicking on any
sectionimmediately brings up the relevantsectionin the text window imme-
diately to the right of the sectiontitle (Figure 141.Finally,becausethe design-
ers of this system combined this feature with its full-text searchengine, the
results ofa searchappearin the contents screenas well as in the text itself.
Searchingfor "Deridai one learnsthatthis name appearsseventy-fourtimes
in the entire book, forty-threetimes in the first chapter,and five times in the
first sectionof that chapter.This dynamic listing provesparticularly valuable
when a DynaTextweb is configured as an electronicbook, for then following
a link from one point in the text to anothercausesthe destinationtext to replace
the departure one. The systemworks so quickly-nearly instantaneously-
that without the contents listing at the side, readersbecomedisoriented.
Tablesof contents,whether static or dynamic, certainly have their uses,
particularly when hypertextualizing material originally conceivedfor print
presentation.Linked statictablesare alreadycommon in HTML, but one can
also createsome of the effectsof the DynaTextform by using HTML frames,
placing the contents at the left and text at the right.

SupposeYouCouldHaveEverything?-TheIntermedia WebViewand Some

PartialAnalogues.The most important Intermedia feature that current sys-
tems, particularly web browsers,lack is its system-generateddynamic track-
ing map, whosebasic idea evolvedthrough three stages.The first, the Global
Tracking Map, provided graphic information about all links and documents
in a particular body of linked documents.Clicking twice on the icon for a par-
ticular hypertext corpus, such as Contert32,NucleorArrns,or Biology,simul-
taneouslyactivatedthat hypertext web-that is, openedit-and generateda
l: Ionuerqenre - fullteHl

i4E H)rperteld Hype rte xt ual Derrida, Po sl slructuralist Nelso n? 1pp.

,r - Hur,Er+avt on,l

Critical Theofy
When designereof compuier soflware examine lhe pagesof Glaar L7t
5 HypertextualDerrida, Cftmmatolottlhsy encounter a di8italized, hypertertual Derida; and when
P5dtstructuralist Iitenry theorists examine lilenty A'facltlnes they encounter a deconsiruclionisl
or postslruclunlist Nelson, Thce shqks of recognition can occurbecauseover
I B The Definition of the past sevenl decadeslitenrylheory and computer h;lperlext,apparently
Hypertext and Its unconnecled areasofinquiry, have increasingly converged.Stalementsby
Flisloryasa Concept
theorirls concemdwith litenture, Iike those by theoristr concemedwilh
ttEl Other Convergencesi computin& show a remarkable convergence.Working often, but nol alwalp, in
Intertextualitv] ignonnce of each other, writeF in these areasofferevidence ihal prcvides us a
Multivocality, and
way inlo lhe conlemponry episteman the midst of major changes.A pandign
shift, I suggest,hasbegun to take placein the wrilin8s oflacques Derida and
El Vannevar Bushand
the Memex
1G) Virtual Text, Virtual
Authors, and Uterary B The Johns Hopkins
and Misunderstanding
Understanding JacquesDerrida
a^--"+i-- Guideto Literary
Theory and Criticism Rebate
3El The Nonlinear Model
of the Networkin Gl M. M. Bakhtin The difficulty of intrcducing a major contemponry philosopher such as
Current Critical |acquesDerida @.1930)in a referencework pre*nlinE centnl issuesof
El RolandBarthes
litenry criticism is double, and this dan8er,this heritation on ihe
1l El Causeor
CDDeconstruction threshold, has alreadybeenslFtematicallythematized in the writing of
Convergence, E Tacques
Derrida the philosopher himself. FiFt, there is ihe dan8erof ovenimplifying of
InfluencEor pigeonholinp of reducing of defining artificialboundaries, when facing
Confluence? a movemenl of thoutloi that ronstantly evolves so as delibentely to
Analoguesto the defeatand baffle all preordained categories,Then, therc iE lhe danger of
GutenbergRevo Contextualizing being merely mimetir, of just rpeatint slnteties and Besturesthat
1Gl ftedictions uerrloa have been identified with a sitnature, wilh an author (and may well
El Discourse have been ahticipatedby other writer), snd that tend lobe singular,
leEf Rec0nfiguring
the unrepealable,yetendowed with univesal validity. However, the
E Poststructuralist
Text Feminisms possibility ofbypassingsuch an initial aporia exists,and it consistsin
considering the fundamentallyrffimative nature of Drida's thoutbl
iE ReconfiUuring
the El Materialist Feminisms and writintntherthan in rtressintthe "playhrl" or "negative" element
tD Michel Foucault ofhis texlual Fn(lices.
El Claude L5vi-Strauss

Figure14. The DynamicTableof Contentsin ElectronicBookTechnologies'DynaText.This system,which combinesthe features

of an electronicbook with hypertextlinking, automaticallygeneratesa reconfigurable,linked table of contentsfrom the SGML

codesusedto mark elementsof a text, such as chapterand sectiontitles. In this examplefrom Hypertert in Hvpefiert, mousing

down on the plus signs to the left of items in the table ofcontents immediatelydisplaystitles ofsubsections.Clickingon the sub-
sectiontitle immediatelybrings up the relevantsectionin the right-handpanel.The table of contentsalso reinforcedDynaText,s

full-text retrievalfunctions: in this case,after a reader has typed in "derrida" (the searchtool is not case-sensitive),DynaTextboth

highfightsall occurrencesofthe word throughoutthe text (iop cente4 and lists the number ofoccurrencesnext to eachchapter

and sectionheading(leJt).Havingobservedthat uDerrida"appearsfive times in the book'sopeningsection,the readerhas moved

that sectioninto view; noticingthat "Derridanappearsin red ink, the sign ofa link, the readerhas then clickedonce on that link

and oPeneda secondDynaTextubook"at fean-MichelRabat6's"Understandingand MisunderstandingDerrida" fromTheJohns

Hopkins Guide to Literary Theoryond Criticism.

RECONFIGURING document in which icons representing each document in the web were
WRITING joined by lines representingall links betweendocuments.This GlobalTrack-
ing Map, which functioned only during early stagesof Intermedia'sdevelop-
ment, immediately demonstratedthat such a devicewas virtr-rallyuselessfor
all but the smallest document sets or webs. (Although pictures of it have
appearedin articles on hypertext, the Global Tracking Map was never used
educationallyand was never part of any releasedversion of Intermedia.) It is
worth noting this failed approachbecause,accordingto the computer science
literature, it seemsto reappearagain and again as a solution to orientation
problems for the Web.
I RI Snext developedthe LocalTiacking Map, a dynamic hypergraph whose
icons representedthe destinations of all the links in the current document.
Upon opening a newlexia or activatinga previouslyopenedone, this graphic
navigational tool morphed, informing readerswhere links in the new lexia
would bring them. This dynamic hypergraph, which did much to prevent dis-
orientation, becameeven more useful in its third and final version, the Web
View with the addition of two more features:a graphic representationof the
reader'shistory and transformation of the icons into links. Double clicking
on any icon in the Web View opened the document it represented,thereby
adding anotherway of making one'sway through webs. (For illustrations, see
Utting and Yankelovich.;
Although this feature succeededwell in orienting the reader, it worked
evenbetter when combined with author-generatedconceptmaps, such asthe
overviews(sitemaps)I have employed on systemsincluding Intermedia,
Interleaf World View, Storyspace,Microcosm, MacWeb,and the World Wide
Web. One basic form of these overviewssurrounds a single concept (Vic-
torianism, Darwinism, Gender Matters) or entity (Gaskell'sNorth ond South,
Dickens) by a series of others (literary relations, cultural context, economic
background),to eachof which many documentslink. Whereasthe Web View
presentedall documents attachedto the entire overview,the overviewhas a
hierarchical organization but doesnot revealthe nature or number of docu-
ments linked to eachblock. Intermedia provided two ways of obtaining ttris
information-a menu that following links from a particular link marker
activatedthe Web View (seeFigure 8). Clicking on a particular link and thus
activating it darkened all the links attachedto that block in the Web View.
Thus, working together,individual documentsand the Web View continually
informed the readerwhat information was one jump awayfrom the current
text. This combination of materials generatedby authors and system fea-
tures well exemplifies the way hypertext authors employ what are essentially

3.0 stylistic and rhetorical devicesto supplement systemdesign and work syner-
gistically with it.
These features ofthis no-longer availablesystem solvedthe basic prob-
lem of orienting readers.Unforfunately, current World Wide Web browsers
are very disorienting becausethey provide no overall view of materials and
neither do they indicate to readerswhere links will take them. The use of site-
maps, HTML documentsthat list or graphicallydisplay destinationsof links,
have greatly contributed to web usability, and many sites now include them
(seeKahn and Krzysztof, MappingWebsites\.
Various research and commercially available systems have had partial
analoguesto the Intermedia Web View. One researchversion of the Univer-
sity of SouthamptonMicrocosmsystem,for example,had somethingverylike
the local tracking map, but it was not implemented in the releasedversion,
and Storyspace,a commercially available system, has its Roadmap,which
has many of the Web View's functions (Figure 15). Like the Web View the
Roadmaprecords one's reading path, shows linked lexias, and permits one
to open them; unlike the Web View, the Storyspacedevicealso lists all links
coming into the current lexia. Unfotunately, the Roadmap, which takes the
form of a menu containing scrollable lists, lacks the Web View's dynamic
quality, for it does not run continuously and has to be opened from a menu
or by means of a key combination for eachindividual document.
Intermedia's dynamic hypergraph proved so valuable as a means of ori-
entation and navigation that I hope someonewill developan equivalent
application either as part of widely used World Wide Web viewers or as an
add-onthat will function with them. Certain stepshavealreadybeen taken in
that direction. The University of Heidelburg'sHyper-Tree,for example,offers
graphic representationsofthe file structure ofindividual servers,but, unfor-
tunately,like the first Intermedia attempt to graph links, it providestoo much
information, thus rendering it of litde practicaluse.The TouchgraphGoogle
Browser (Figure 16),whichis more selective,draws on search results from
Googleto map what it takesto be the most popular connectionsbetweenthe
lexia (or entire website)whose URL the user providesand other lexiaswithin
and without the site, producing results quite different than a sitemapor over-
view The Touchgraph for TheVictoianWeb's"Religion in England" sitemap
(Figure 16)has only a few of the dozensof the documents on religion, such
as "High Church: Tractarianism" and "The BroadChurch Party,"but it omits
perhapsthe most important, heavilylinked document on religion in The Vic-
toian Web,the essayon the Church of England. Interestingly, it revealsa
close connection between religion and British art and, rather unexpectedly,
stic *elatians
Section7 Yi,r*latts

D*rk hnuse, by whieh onee nsr l stnnd, Cort{xt5

H*re in fhe bng unklvlv strs*! VktOriarlbm
Ooorc, rvhere nq he*rt rti,s u;ed lo Phihs*phy
b*al Ptl!*icai SESrxial Coniext
$* quick$" waiting lor a harxi. Scienceand T*chnology
A htnd that {:anbc ela*P'd n$ rr$re -
SeLokl nref*r I eann$t $lesp, kMlrnoniaru
And like a gurltv t}ing I creep e*nwr*nta4v t* t}t* {tvhi:b} fo*m
ltlfetr.di^e la ihg ll$eRl
,{t earlirst fi$a{ng t$ tln d*or.
*.'---"-.----. --.J Seetbns
Hc ii nut herei but far awa_l'
Tlre no!r{: of lifu be.ginsagain
And ghasth'thnr the druzlnrgrrlr'r C0ntsxt
On fhr: baH strct break$ tht,hnk day. ennvson
{Li*ks tn p*eeding and f*ll*rving seefirrn$,
t$ {nngrxntilry a$d to ln Mtnrrrkm$Y}

]6e!d tdqlsBd

Figure 5. The StoryspaceRoadmap Feature.The Roadmap (upper lefil representsa static analogueto the Intermedia Web View.

At the top center ofthe Roadmapappearsone's reading history and, immediately below it, the first few lines ofthe currently active

lexia. Like the Web View, the Roadmapinforms readersof possible destination lexias and permits readersto oPen them directly'

but unlike the Intermediatool, which displayedonly destinationlexias,the Roadmapdisplaysall links in and out of the current

lexia. Unfortunately,whereasthe Web View always remaind in sight and automatically reconfigureditselfas each new document

opened,the Roadmapappearsonly on demand and has to oPeneachdocumentseparately.

Lewis Carroll has a major presence.This diagram also contains oflsite lex-
essayon D. G. Rossettimight be expected,
ias,and whereas the Artcycloped.io
since the Pre-Raphaelitesappear repeatedly,I'm intrigued by the outlying
"Virtual Tour of BrasenoseCollegel' A similar Touchgraph for Charlotte
Bront, which is much denser, includes a substantial number of obvious
externalsiteson the Bront6s,Derbyshire,and other women writers, but I am
mystified by the presence (occupyingthe diagram's entire upper-left quad'
rant) of a concentrationof materials on the Harlem Renaissance,including
the RedHot I azz Archive.The Touchgraphfor TheVicToia'nWeb,whicbistoo
denseto reproduceeffectively,makes many obvious connectionsto extemal
sites and quite a few unexpectedones,too. The very fact that so many unex-
pectedconnectionsappearin these diagrams makes them quite fascinating,
"tnrfdts l4t@*t


aii{qrikrbii ittir'.;' 6liidd ''lt-,-. 'l;.,.

Slrtntri*.*1i!i,. t*? ;:rri

"iedi.&t3gs S{r*t librsy " Crnsixre ot

d 6e turm of S<torir bn ild W. holns
:t|ai{ii ilii&l tdiii.indois
rChd*is ndrr*i sd *e

Figure'16.The TouchgraphGoogleBrowser.This softwaredraws upon searchresultsfrom Googleto map the most popular
nectionsbetweenan individuallexiaor entirewebsitewhose URLthe user providesand other lexiaswithin and without
the site.lts
results,which differ markedlyfrom an author-creatd sitemapor overvieq often revealunexpectedrelationshipsamong

and I can see a value in either including a few screenshotsto show readers
some interesting connectionsor linking to the Touchgraphsite, so they can
explorefor themselves.The Touchgraphapproachshows,however,the short-
comings for our pu{poses of mapping a website according to popularity
(most visited and linked-to lexias): the resulting diagram omits one of the
most valuable characteristicsof hypermedia-its capacityto support indi-
vidual, evenidiosyncratic,approachesto information.

Author-createdorientation Devices:overviews.As the web view and Road-

map show,readersneed effectivelyorganizedpreviewfunctions-what Mark
Bernsteinterms "airlocks"-that show them what lies one jump away.In the
next section I shall suggeststylistic, rhetorical techniquesthat hypertext
authors can employ in the absenceof such software tools. some hypertext
RECONFIGURING systemslike Microcosm, Storyspace,and Intermedia provide severalmeans
WRITING of helping orient the reader;othersprovide little built-in assistanceto solving
basic problems of orientation. But whatever system authors employ, they
should use overview and gateway documents, which are devices entirely
under their own control. Overviewsor sitemaps,which can take many forms,
are author-created(asopposedto system-generated) documentsthat serveas
directoriesto aid in navigatingthe materials.Overviewsassistreadersto gain
convenientaccessto all the materials in many documents or to a broad topic
that cuts acrossseveraldisciplines.
Overviewsor sitemapstake six forms, the first of whic-his a graphic concept
map that suggeststhat variousideasrelateto somecentralphenomenonor im-
pinge on it. This center,the subjectof the overview can be an author (Tennyson,
Darwin), chronologicalor period term (eighteenthcentury Victorian),idea or
movement (realism, feminism), or other concept (biblical typology,cyborg).
The implied and often reinforced messageof such arrangementsis simply
that any idea that the readermakesthe center of his or her investigationsex-
ists situatedwithin a field of other phenomena,which may or may not relate
to it causally.Such graphic presentation of materials depicts one informing
idea or hidden agendaof hypermediamaterials,namely,that one proceedsin
understanding any particular phenomenon by relating it to other contexts.
These kinds of overviewlexias,which I have used since the first daysof
Intermedia, have particular value for the World Wide Web, which tends to-
ward a flattened form of hypertext.Their emphasis on multiple approaches
simultaneously provides a way of breaking out of the implied page format
that confines the Web and also of creating what Paul Kahn has termed a
"crossroadsdocument,"a point to which the readercan retum repeatedlyand
before departing in new directions. The various websites I maintain use
different kinds of overviews. The VictorianWebsunotnds a central image by
a range of related topics. In that for Elizabeth Gaskell'sNorth and South,for
example,a linked icon for political and socialcontext appearsat the top cen-
ter, and immediately beneathcomes those forbiography, other works by the
same author, Victorianism, and women's lives (Figure 17).The icons for lit-
erary relations and visual arts flank the image representingthe novel. In the
line below are five icons representingaspectsof technique-setting, syrnbol-
ism, characteizarion, narration, and genre; centeredbeneaththem appears
that for religion and philosophy.
Although one could use a single image map for such an overview on a
website,using separateicons has some advantages,the first of which is that
'Alt" option in HTML that permits one to include a text label,
by using the
Figure 17. Two World Wide Web overviews. These examples show two different approachesto creating
overview lexias for the
Worfd Wide Web. That on the left, the overview for ElizabethGaskell's North ond Southin The Victorion Web,
representsthe lat-
est versionofthe lntermedia-style
overviews,which emphasizethat readerscan approacha subjectfrom multiple points ofview.
The A' S' Byatt overview,in contrast,Presentsa similarly nonhierarchicalapproachto organizinginformation,
by arrangingits
linked headingsin a series of horizontal rows. This approachto creating overviewswith HTML
(text) documents has several
advantagesover image maps: (1) this text-basedoverviewloads (opens)fasterthan server-basedimage
maps; (2) since Internet
Explorer,Safari, Netscape,and other web browsers retain images in a cache, building different overviews
with the same elements
createsdocumentsthat loadvery quickly;(3) such overviewsare easilymodifiedby addingor subtracting
combinations; (4) these overviewshave the advantageof employing the same files for both overviews
and footer icons. thus
reducingstoragespaceand accesstime.

these kinds of overviewswill work with old-fashionedbrowsers that do nor

have graphics capacities-an important consideration when porrions of
one'sintended audiencemay not havethe kind of computer accessor equip-
ment needed to handle large images. Moreover, one may create standard
templatesfor all the overviewsin a particular web, thus producing a kind of
visual consistency,and yet one can easilymodify appropriateelements.In the
Gaskelloverview,for instance,"works" replaces"other works," and in those
for other texts other icons appear,including those for themes, bibliography,
related'WorldWide Web materials, and so on.
Figure 18. Gunnar Liestol's Kon-Tiki lnteractive: The Introductory Overview This interactive overview surounds an image of lhe

globewith sevencircularimages,representingThor Heyerdahland six of his expeditions.Theseimagesserveas icons,previews,

and conceptualoverviews.Clickingon any one ofthem halts sound and animationand opens an interviewfor the subiectit rep-
resents.(Usedby permissionofGyldendalPublishers.)

Gunnar Liestol'sKon-Tiki Interactive,whichI shall discussin more detail

below, surrounds an image of the globe by seven circular images, each of
which animates in turn, Thor Heyerdahl and six of his expeditions (Figure
18).Not all the overviewsthat wish to avoidhierarchy or linearity needto have
this kind of circular format. Unlike the Liestsl and Victoian Weboverviews,
those for the hypertext sectionof The Cyberspece Weband all materials in the
PostcolonialLiterature Web do not emphasizecentrality. Instead,taking the

3.0 A. S. ByattOveryiewas an example,we find the topic of the document above
three rows of five squareicons each(seeFigure 17).This arrangement,which
also avoidsthe linearity of a table of contents,has the advantageof permit-
ting one to employ some of the same icons both in overviewsand at the foot
ofeach screen.
Chronologiesrepresentanother form of sitemap or overview,which one
can easily createusing two-column tables in HTML. They offer a means of
clearlyorganizing materialsor evenentire coursesthat havea strong chrono-
logical orientation. Any timeline with links in fact servesas an overviewfor
the materials it joins. Although timelines provide a means of organization
particularly convenientto authors,rememberthat they may simplify complex
relationships and do little to compel the interest of a reader unacquainted
with the subject.
Images of natural objects,like the photograph of a cell or maps, provide
a kind of naturally occurring concept map that authors can easily apply.
Attaching links to labels in technical diagrams similarly provides an obvi-
ous way of enriching conventional information technology.These kinds of
overviews,incidentally, exemplify a perfect use for World Wide Web image
maps. Perhapsmy favorite is a map of Italy showing major Italian websites:
click on the tiny squarerepresentinga particular city, and a link takesyou to
its website.
If hypermedia is characterizedby connectivity,to realize its potential one
must employ devicesthat emphasizethat quality.Lists,tablesof contents,and
indices,though still of significantuse,do notworkinthis manner,butone may
wish to use them in addition to other kinds of graphic organizingdevices,as
does the elegant Kon-Tiki Interac-tiveCD-ROM, which parallels its circular
overviewwith an interactive outline. Mousing down on its individual elements,
say,that for the Kon-Tiki itself, producesa list of elevenitems (Figures19-20).
When converting text documents originally createdfor book technology
for presentationon hypermedia,one may occasionallyuse the document
itself as its own overview.Any document in a hypermedia systemwith more
than a fewlinks in essenceseryesasa sitemapsince,onceopened,itprovides
the immediate center and referencepoint for the reader'snext act of explo-
ration. The author of educationalmaterials,particularly those involving liter-
ary texts or those that place primary emphasis on the details of a text, may
therefore wish to take advantageof this quality of hypermedia. Section7 of
In Memoiam (seeFigure 9) exemplifies a brief tert document that functions
as its own overview or sitemap in a Storyspaceweb, and each section of
the heavily annotated HTML version of "Hudson's Statue" by Carlyle func-

Figure'19.The Help Functionin the Kon-TikiCD-ROM.Movingthe mousenearthe raystoneshieldhaltsall sound and videowhile

simultaneouslyrevealinga menu of choices.(Usedby permissionof GyldendalPublishers.)

tions in the same manner. One must take care not to overdo this kind of
heavily linked text document on the Web, which has few orienting devices,
sincelinked text alone doesnot alwaysprovidevery clearindications of where
its links take the reader.In a scholarlyor critical presentationof a text, such
as the Pepys'sDiary Weblog or "Hudson's Statue,"in which the links clearly
take one to annotations and commentary these heavily linked lexias can
function in this way,in large part becausethe nature of the document indi-
catesthe kind of links that will attach to it. In contrast,some heavily linked
opening screensof personal siteson the Web,though occasionallyamusing,
often appearcompletelychaotic.

.I :r:I


Figure 20. Kon-TikiOverview.Selectingany item in this interactive overview produces a sublist of items. (Used by permission of


Whatever kind of overviews or local sitemaps one chooses,one should

accommodate-and encourage-different stylesof hypertextreadingby pro-
viding as many as is convenientfor eachsubject,and one should also expect
that individual lexias,particularly in information hypertexts,will link to mul-
tiple overviews.Thus, an essaycomparing women'sissuesin Graham Swift's
Waterlandand A. S. Byatt'sPossasslon
would link to the literary relations doc-
ument for eachwork but also to those for themes, gender matters, and tech-
niques as well.
Closelyrelated to overviewsand directories are those documents that serve
as gatewaysbetweencoursesor bodies of materials in separatedisciplines.

RECONFIGURING Such gatewaylexiasare particularly useful on the World Wide Webwhen

WRITING a link brings the readerfrom the present website to another.The most com-
mon form of such gatewaysappearsin the separatedocuments containing
lists of links to other websites.Another example of such a transitional lexia
is The Victoian Web'sbrief introduction to the University ScholarsProgram
(USP), National University of Singapore,which funded severalpostdoctoral
and senior researchfellows whose work appearson the site. As a means of
identifying their work, icons representing the USP appearthroughout The
Viaorian Web.Insteadof linking these icons directly to the USP site, which
would confuse readers-who might wonder, "Why am I reading about Sin-
gaporel"-I havelinked them to a lexia that describesthe USP and its role in
supporting the site; links within that lexia then bring anyone who wants to
know more about the USP to its homepage.Evenwithin a small sectionof a
single website,such as that formed by materials on a single author, concept,
or event, such introductory transitional lexias prove usefirl. For example,
when creatinglinks from an icon or subjectheading for one author'srelation
to other writers, one can either link to a local sitemaplisting all relevantessays
or one can link first to a generalintroduction; this latter approachworks par-
ticularly well as a means of introducing complex relations not evident from a
sitemap or of indicating a specialconcentrationof materials in one area.

Cleamware.In addition to describing some effective software solutions to

meet the needsof the hypertextauthor, permit me to proposesomething like
a wish list. Computer users often refer to promised projects as so much
vaporware,meaning that aproduct or researchproject that someonehas pre-
sentedas alreadyexisting is in fact little closerto reality than a plan or a prom-
ise. Let's go even farther back-from promise to desire. When I was much
younger, I remember hearing the expressionthat mentioned a time when
someone "was just a gleam in" their eyesof their parents. Let us consider
gleamwareor wishware.
Such an example of gleamwarewould be semiautomaticallygenerated
sitemaps and crossroadsdocuments in HTML that would permit reader-
authors on the Web to produce such intermediary documents by combining
complex searcheswith elegant templates. At the moment of writing, no
World Wide Web browser has the one-to-manylinking that I believeso cru-
cial to creating a fully multiple hypertext. Therefore, to translate materials
originally created in systems that have such linking or to emulate them,
authors find themselvesforced to expendan enonnous amount of time and
effort manually creating-and maintaining-link menus. The implications

3.0 of such difficulty will becomeclearwhen I report that the multiple links that
required twenty to thirty minutes to createfor an Intermedia or Storyspace
overview-and less than half that time for Microcosm, using its more so-
phisticatedgeneralizedlink-options-can take severaldayswhen translating
such materials for the Web: one must go through the subset of documents
that will link to the overview and manually createseparateones for literary
relations,themes, biographicalmaterials, and so on. Evenif one alreadyhas
an earlier version of one's materials in another software environment to
remind one of possiblelinks, it still takeshours of repetitivework-with the
result that authors inevitably tend to avoid as much of it as possible and
thereby produce a relativelyflattenedhypertext.
Sohere'smy first two gleamwareproposals,the first of which may already
exist as a proprietary researchtool in some large corporations in *re com-
puter industry. Imagine combining Macintosh OS 10.43Spotlight feature
or a commercially availablesearchtool, such as On Location,with some
C-programming, and a set of templates that would permit one to generate
with minimum expenditure of time and effort a subovewiew entitled, say,
"Political Themes in Dickens" simply by calling up a menu and typing "Dick-
ens,""themes,"and "politics."An evenbetterversion-one that I havespoken
about longingly since the last few yearsof the Intermedia project-involves
automatically generatedgraphic representationsof literary and other com-
plex relations. In this exampleof gleamware,one would simply choosefrom
a menu a literary relations (or similar) option, and one'sauthoring (readingl)
system would combine a graphics engine, search tools, resulting indices,
glossaries,chronologies, and synonym lists to produce automatically the
kind of conceptmaps Paul Kahn createdin the early DickensWeb:usingsyn-
onym lists and chronologies,this Relations Map Generator-let's give it a
properly stuffy name-places the chronologically earlier authors or texts
toward one end of a chosenaxis;earlier onescould appear,for example,at the
left, at the top, or, in a three-dimensional representation,farther away.Au-
thors or texts that I consideredmore important-either for reasonsof some
relation to the author in question (Hallam
cultural standard (Shakespeare),
to Tennyson,the Brownings to eachother), or quantity of availablecommen-
tary could be made to appearlarger or in brighter colors. You get the idea.
Let'stake the gleam one step further: if one could produce such documents
quickly enough-something that probably assumed preexistent indices-
such documentscould exist only dynamically,createdeachtime one followed
a link from an overview,and hence alwayscurrent, alwaysup-to-date.

OrientationDevices:Markingthe Edges.In the absenceof
WRITING such tools, what kinds of techniquescan one use to assist readersl One
deviceespeciallyimportant to those creatingmaterials in HTML involves
using visual indications of a lexia'sidentity, location, and relation to others.
Thesesignals can take the form ofheader icons, color schemes,background
texhrres,linked icons that appearat the foot of lexias, or all in combination.
Such devicesplay a crucial role on the Web, where readersmay arrive at any
document via a search engine, entering at what could be the middle of a
planned sequenceor setof documents.Without some such deviceevenread-
ers who find that a particular arrival lexia meetstheir needsand tastebecome
frustrated becausethey cannot convenientlydetermine whether it forms part
of a larger structure.
One of the most commonly used such devicesis the header icon, which
immediately informs the reader that a lexia belongs to a particular web or
subweb.For example,lnTheVictoianWeb a blue-and-whiteheaderelement
appearsimmediately following the lines providing title and author.At the left
of this icon, which is a third of an inch high and 7 inches wide, appearsa
black-and-whiteimage of Queen Victoria followed by the words "The Victo-
rian Web" and a white line extending the remaining length of the header.
Using an editor, such as Dreamweaveror BBEdit, which permits easilymak-
ing global changes-that is, changing all occurrencesof a word or phrasein
an entire set of documents rather than having to do them one at a time-
makes inserting such elements exkemely easyto do. Whereas The Victoian
Web empToysa single header icon, some of the other websites I maintain,
such as that on recent postcolonial literature, uses a different one for each
major division or subweb.This secondweb has separatesectionsfor anglo-
phone literature of Great Britain, the Indian subcontinent, Afica, and Aus-
tralia and New Zealand, and therefore employs different headersas well as
other devicesfor eachsection.Similarly,Adam Kenney'sMuseum,a Web ver-
sion of anthology-fictionlike TheDecameron,organizesitself around a series
of individual narrators and uses an image at the top center of each lexia to
identify different narrative arcs.
Other devicesinclude color schemesaswell as backgroundtexhrres,and
combined with footer icons they make an efFectivemeans of simultaneously
orienting the readerwhile indicating the permeableborders of both the lexia
and the larger units to which it belongs. For example,an essayfrom The Vic-
torian Webthat comparesthe railway swindlers in Trollope'sThe WayWeLive
Now and Carlyle's"Hudson's Statue"has five footer icons, one for the main

3.0 Victorian overviewfollowed by one eachfor Trollope,his novel, Carlyle,and
his text (Figure 21).The first three icons denoteincreasingspecificity,thus in-
dicating that the document contributes to the web as a whole, to those mate-
rials concerningTrollope,and to thoseaboutthis particular novel.In contrast,
the five icons, taken together, indicate that the lexia in question simulta-
neouslyparticipatesin two subwebsor directories.Theseicons thus serveto
orient readersby clearly stating how the lexia in which they find themselves
relatesto one or more large categories-in this example,five separateones.
Furthermore,becauselinks attachto eachof theseicons,clicking on them
brings readersto a sitemap for theselarger categories.Attaching links to the
icons, in other words, makes them devicesof navigation as well as orienta-
tion. In one sensethesedevicesmark the edgesof one or more groups or cat-
egoriesto which the lexia belongs or with which it associates,but the most
important function involvesnot so much delimiting an edge or border of a
document as indicating its relation to, or membership in, one or more sub-
webs. The effect ofthis congeriesofdevices, therefore, is to orient readers
who find themselvesin a particular lexia by clearly indicating its relation to
others, its (intellectual)placewithin a web.
This combination of headers, color schemes, and linked footer icons
works particularly well in large or parricularly complex collectionsof inter-
linked lexias, such as those created by participants in courses or depart-
ments. The Cyberspaceand Citic al Theoryw eb,for example,containsnot only
course materials and links to many websitesedernal to Brown but also to a
collection of elaborateindividual student projects, some consisting of more
than one hundred lexiasand graphic elements.In this situation such identi
fication and bordering schemesproveparticularlyuseful becausethey inform
readersthat they have arrived at a discrete document set. Following a link
from the print sectionof the information technologyoverviewbrings one to
Amanda Griscom's Trendsof Anarchy and Hierarchy:Comparingthe Cultural
of Print and Digital Media,herWorld Wide Web translation of a
substantialhonors thesis comparing the seventeenth-centurypamphlet wars
in England and the periodical press that succeededthem with the situation
on the Internet today.In contrastto the black background and white and yel-
low text of the Infotech overview,Trend.s
of Anarchyconfronts the readerwith
a pale yellow background, black text, and a light pastel header announcing
the title and author of the entire piece.Like many studentcontributions to the
cyberspace,this one contains a link to the web overviewonly on its contents
page;the other lexiascontain only footer icons to the contents,next page,and
works cited. Sincethe readercan enter portions of this subwebfrom various
1. Firstthreeiconsindicate
increasing As one
movesfrom left to right, one

2. In contrast,the five icons

indjcatethat the present
document simuttaneous[y
in two subwebs
or directori


Figure 2I. Footer lcons in World Wide Web Documents, This example from The Victorian Web shows how linked icons at the bot-
tom of a lexiacan indicateits simultaneousparticipationin severalsubwebsor documentsets,therebyorientingthe reader.

HYPERTEXT overview headings that indicate discussions of Mcluhan, scribal culture,
media in the seventeenthcentury and so on, the reader needs to know the
separateness-as well as the entire scope-of this subweb.
In concluding this section, I have to emphasizethat on the World Wide
Web the borders and limits of these hypertext documents, their edges,as it
were, clearlyhaveto be understood only as fictions, as agreed-onconvention,
sinceboth links and searchengineseasilycrosstheseproposedmargins. The
header graphic, for example,announcesboth the existenceof a web (or ma-
jor sectionof one) and implicitly proclaims its boundaries-only documents
containing that headerbelong to this web-but on the Internet these claims
are at best provisional, at worst an obvious fiction, since links to or from the
lexiason any site make any assertionof boundaries into a gestureor wish or
hope, particularly when, as in any large and complex web (site),the docu-
ments do not possessthe kinship endowedby author function, for they have
been producedby more than one author or entity. In setsof lexiascreatedby
a single author one can posit limits-that is, pretend they exist-more eas-
ily than one can for webs that both draw extensivelyon quoted passagesand
images createdby others and also link to other sitesas well. Nonetheless,we
need such classificationsin order to read. But the crossing of such textual
(non)bordersis one of the characteristicsof hypertexfuality,one completely
analogousto the way links both permit one to employ a folder structure and
yet not be confined by it.

This Text ls Hot. Readersof hypermedia need some indication of where they
can find links and then, after they havefound them, where those links lead;
finally, after they havearrived at a new lexia, they need some justification why
they havebeen led there. All these issuesraise the question ofthe degreeto
which specific systems,authors, or both working together require an active,
even aggressivereader.In examining a range of solutions to the first prob-
lem-how to indicate the presenceof hot (or linked) text-l shall follow my
usual procedure and begin by surveying the means thus far employedto do
so and then suggestways authors can write with and againsttheir systems.
As always,a ma;'ortheme will be to suggesthow World Wide Web authorscan
benefit from lessonsprovided by other forms of hypermedia.
In examining some ways existing hypertext systemssignal the presence
of linked text or images,I shall begin with least successfirlexamplesand pro-
ceedto better ones. Let us begin with Intermedia, though this time because
its solution, though clearand unambiguous, provedtoo visually intrusive. All
versions of this system employed a link marker in the form of a small hori-

RECONFIGURING zontal rectangle within which appears an arrow. This icon appeared auto-
WRITING matically aboveand to the left of any section of linked text and could not be
moved;in graphicsdocuments,however,the authors could placeit wherever
they wished. Like many elementsin Intermedia, the link marker worked syn-
ergisticallywith other systemfeatures.Clicking onceon it, for example,high-
lighted the icons representing destination lexiasin the Web View and by go-
ing to the main menu, readerscould learn both anchor extent (the extent of
linked tert) and anchor description (the label the author attachedto it). Un-
fortunately,placing the icon abovelinked text proved too intrusive, for it dis-
torted the documentk leading-the spacingbetweenlines-a parti cttlarly an-
noying effectwhen one placedlinks in a print text, say,a poem by Tennyson.
World Wide Web browsersoffer a slightly better solution to the problem
of how to inform readers about the location of links. As is well known,
browsersconventionallyindicate the presenceof links with color and under-
scoredtext: Mosiac and Netscapeestablishedthe convention that unlinked
text appearsin black against a light gray background, blue underlined text
indicates the presence of a link, and magenta underlined text indicates a
link that one has previously followed. I am of course describing the default
version-that, in other words, which one receivesif neither rcader nor au-
thor customizesthese elements; authors can chooseentirely different color
schemesor chooseto make regular and linked text the same color, thus
employingonly underlining to indicate link presence.
Although the conventionalHTML approachseemslessvisually intrusive
than Intermedia's annoying link marker, its manner of signaling link pres-
ence,like that of the earlier system,producesa visual hodgepodgein alpha-
numeric text. The simple fact is that the links and the colors are alwayspres-
ent, and for certain purposestheir presencebecomestoo annoying.Although
annoyng in written text, such permanently displayedlink markings some-
times work well with graphicelements,since a coloredoutline doesnt always
intrude on an icon the way a combination of color and underlining do on text.
Many intenselygraphicsites,however,employ HTML options that permit au-
thors to turn offsuch color variation, providing no visual cuesto the location of
links at all, becausethey know that in current browsers,when a mouse moves
over a link, it turns into the image of a hand. As thosewho use the Web have
becomemore sophisticated,designersincreasinglyassumethat when users
encounter a screenwithout any obvious links, they will move their cursors over
imagesand other screenelementsthat they havelearnedare likely to serveas
link anchors.In contrast,earlierwebsitesoften usedlinked icons and, unsure
whether readerswould know what to do, they addedlinked text,usually in very

T. 0 small type,immediately below them. An increasinglysophisticatedaudience,
in other words, no longer finds the absenceof obvious clues disorienting.
In contrastto basic HTML and Intermedia, Microcosm,like severalother
hypertext systems,doesnot permanently display indications of links. As we
havealreadyobservedwhen examining Microcosm'srich assortmentof link
types,this systeminvites active,even aggressivereaderswho interrogatethe
text they encounter (see"Forms of Linking," chapter1).This approach,whi&
removes any possibility of visually marring the appearanceof literary texts,
appearsin the way it provides information about the presenceand extent of
linked material. In keeping with Microcosmk encouragementof the active
reader,userswho come upon a word or phrasethat they believelikely to serve
as an a link anchor have severalchoices:they can double click on it, perform
the equivalentaction by choosing "Follow Link" from a menu, ask about link
extent from a menu, or create their own links with the "Compute Links"
function. For many applications,particularly educationaland informational
ones, the Microcosm approachstrikes me as wonderfully appropriateto the
medium. My only suggestionwould be to follow Hypercard and Storyspace
and make a simple key combination show both the presenceand extent of
author-created links. Storyspace'suse of frames that surround hot text
when readers hold down Option and Apple (or Command) keys simulta-
neouslyprovesa particularly valuablefeature and one that I would like to see
both Microcosm and HTML viewers emulate. Changing the cursor from an
arrow to a hand to indicate that readershaveencounteredhot areasin a doc-
ument has many advantages,yet it still requires in many casesthat readers
grope blindly around the screen,and in large text documents omitting any
signs of existing links createsconfusion; transient indications of links that
readerscould activateby simple operations,such as a key combination,would
be usefirl as an additional feature.

andthe Rhetoricof Departure.
All thesesystem-
based devicesthat we have just observedconstitute the first, and simplest,
part of any rhetoric of departure, for they inform readersthat they can leave
a text stream for somewhereor something else.Not surprisingly, most read-
ers do not feel comfortablejumping offinto limbo. Although much hypertext
fiction and poetry playswith surprise and disorientation as desiredaesthetic
effects,other kinds of hypertext writing require some way of giving readers
an idea of what links will do.
Such preview functions-what Mark Bernsteinhas calledan "airlock"-
serveto both inform and reassurereaders,and when systemsdo not provide

RECONFIGURING adequateinformation, authors must find their own ways to obtain it. The
WRITING just-discussedHTML conventionof changing the color of anchorsindicating
links that havealreadybeen followed exemplifiesone kind of valuablesystem
support.When one mousesoverhot text (an anchor),most web browsersalso
show the destination URL, though unfortunately not the title aswell, in a panel
at the bottom of the viewer window-though this feature does not seem to
workwith documentsusing frames.3
The point is that readersneed a generalidea ofwhat to expectbefore they
launch themselvesinto e-space.Help them, therefore, by making text serve
as its own preview: phrasestatementsor posequestionsthat provide obvious
occasionsfor following links. For example,when an essayon Graham Swift or
SalmanRushdieaddslinks to phraseslike "World War I" or "selireflexive nar-
rators,"readerswho follow them should encountermaterial on thesesubjects.
In addition, wheneverpossibleprovide specificinformation about a link
destination by directly drawing attention to it, such as one does by creating
text- or icon-basedfooter links. Another preciseuse of text to specify a link
destination takes the form of specific directions. For example,in The Victo-
ian Webto which student-authors contributed differing interpretations of the
sametopic, say,labor unreslin Northand Southor genderissuesin GreatEx-
peclations,functioning as an editor, I have added notifications of that fact.
Thus, at the close of essaysquoting and summarizing different contempo-
rary opinions about strikes and labor unrest, I haveadded"Follow for another
contemporary view," a devicethat should be used sparingly, and lists of re-
lated materials,which are particularly useful when indicating bibliographical
information and documents on the same sub;'ect.
Such careful linking becomesespeciallyimportant in writing hypertext
for the World Wide Web, since current browsers lack one-to-many-linking
(seeFigure 3), and it doesnot seemlikely after more than a decadethat they
will ever incorporate it. This apparently minor lack has devastatingconse-
quencesfor authors, who haveto createmanually the link menus that other
ers lose the crucial preview function they provide. I find that the effect of
being reminded of branching possibilities producesa different way of think-
ing about text and reading than does encountering a series of one-to-one
links sprinkled through a text.

The Rhetoricof Arrival.Many non-Webhypertextsystemsusevariousmeans

to highlight the reader'spoint of arrival, thus permitting links into portions
of longer lexias.Intermedia, for example,surrounded the destinationanchor

3.0 with a marqueeor moving dottedline that traverseda rectangularpath around
the intended point of arrival; a single mouse click turned offthe marquee.
Storyspace,in contrast,employs a rectangular block of reversevideo around
arrival anchors. Unfortunately, thus far, although World Wide Web authors
can use the <A NAME> anchor feature to bring the readerto a particular por-
tion of a document, no browser showsthe exactextent of the arrival anchor.
Instead,HTML just opensthe arrival lexia at the line in which the anchor ap-
pears,somethingextremelyuseful for bibliographicalcitationsand other lists.
The difficulty in the World Wide Web is exacerbatedby the fact one often
links to documents over which one has no control, and hence cannot insert
an anchor.One way of accommodatingthose who link from outside involves
using the identifying color schemesand headersdescribedearlier. If one can
obtain permission from the document'sauthor or owner, one could placean
anchor there. Similarly,if one can obtain permission to do so,one could copy
and incorporate the arrival lexia within one's own web. Although such an
approach,which I haveusedinTheVictoianWeb, occasionally provesuseful,
it strikes me as basically inefficient and contrary to the spirit of the World
Wide Web'sdispersedtextuality.

ConvertingPrint Textsto Hypertext.Before consideringthe best ways to

h)?ertextualize printed matter, we might wish to ask why one would want to
bother. After all, a somewhat sy'rnpatheticdevil's advocatemight begin, it's
one thing to expend time and effort developing new modes of reading and
writing, but why modify the book, which is in so many ways a perfectly good
text-deliverymachinel For many nonliterary uses, the answer seems obvi-
ous, sincelinked digital text permits an adaptability,speedof dissemination,
and economyof scalesimply not possiblewith print. Maintenancemanuals
for large,complexmachines,like airplanes,parts catalogues,and many other
uses of the codexform of text presentation seem better servedin electronic
form. For thesereasonsin some scholarlyor scientificfields, such as high
energy physics,the digital word has increasingly replacedthe physical,and
the most important form of publication takes place online. This movement
awayfrom printed text has certainlyhappenedin the workplace,where people
who formerly used printed schedules,parts catalogues,tax and real estate
information, and delivery forms now read on screen;evencourier and pack-
age delivery services now have customers sign electronic pads, as do the
check-out girls at many supermarkets. Those of us who work with books
every day and who enjoy doing so may not have noticed that for a growing
number of people,printed matter, not just books, plays an ever smaller role

RECONFICURING in their work day.The implications of this changeseem obvious:in coming

WRITINC years,the printed book will eventuallybecome,for many people,an increas-
ingly exotic object in much the same way that beautifirl illuminated manu-
scripts eventuallyappearedexoticto many raisedon reading books. Perhaps,
since as Bolter and Grushin argue, no information medium ever dies out
completely,the codex book will be reservedfor classic novels or recreated
with on-demandpublication; in an increasinglydigital world, whateverhap-
pens, future readerswill not experiencebooks as we do.
But why in literature and in humanities education, our devil's advocate
might continue, would we want to take works originally conceivedfor print
and translatethem into hypertextl Particularly given the comparablyprimi-
tive state of on screentypography,why take a high-resolution object like a
book and transform it into blurrier words on flickering screensl Now that the
World Wide Web promises to make all of us selipublishers, these questions
becomeparticularly important. I would answer:we translateprint into digi-
tal text and then hypertextualizeit for severalobviousreasons:for accessibil-
ity, for convenience,and for intellectual, experiential, or aesthetic enrich-
ment impractical or impossible with print.
When I beganto work with hypermediaa decadeago,the combination of
a desire to creatematerials best suited for reading in an electronic environ-
ment and the need to avoid possiblecopyright infringement led my team to
createall materials from scratch,but soon enough teaching needs drove us
to include hypertext translations of print works. These needswill be famil-
iar to anyone teaching today: works around which I had planned portions
of a course suddenlywent out of print. Placing otherwise unavailabledocu-
ments within a hypertext environment allowed us to createan economical,
convenientelectronicversion ofa reservereadingroom, one that nevercloses
and one in which all materials always remain availableto all readerswho
need them.
Now that the World Wide Web can link togetherlexiaswhose sourcecode
resideson different continents-the texts of many classicVictorian novels,
for example,resideon a serverin fapan-such accessibilityprovidesan even
stronger impetus to hypertextualize otherwise unavailable materials. This
easeof accessibilityfrom a great distance means that more readerscan use
one'stext-and, in return, that one can hope to find texts translatedby oth-
ers for one'sown use.
Books and articles on the Web take two very different forms, the first of
which is the Webversion that closelyfollows the print paradigm and usesthe
Internet solely as a means of making texts available*hardly an unworthy

3.0 goal. In contrast,the secondapproachattemptswith varying degreesof suc-
cessto translateworks createdfor print into hypertext,thereby exploring the
possible modes of scholarly publication in a digital world. By far the most
common approachthus far is the Web version that preservessome format-
ting from the original print version but ignores many characteristicadvan-
tagesof the new medium. The many PDF versionsof scholarlyand scientific
articles downloadablefrom the Web exemplify this approach,as doesProject
Gutenberg,which tries to provide as many digitized versions of printed
texts as is possible.ProjectGutenbergembracesboth primary and secondary
texts,but its mission doesnt allow distinguishing betweenthem. More schol-
arly textbases,like the Women Writers Project,necessarilyhave as their
chief goalthe preservationof as much information as is practicableabout the
physical form of often rare and usually inaccessibleprinted books; text
encoding, rather than hypertextualization, understandably has the highest
priority. Other projects,like Mitsuharu Matsuoka'sVictoian Literary Studies
Archiveat NagoyaUniversity take texts by 150 British and American authors
of the nineteenth century-it includes, for example, two dozen books by
Dickens and another half dozen about him-and join them to Masahiro
Komatsu'sHyper-Concordance;although the Archivedoes not createhyper-
text translationsof theseworks, it takes advantageof their digitization to
createon-demand corpus-widesearches.
As valuable as these print preservationprojects are, they do not help us
answer the question, What will happen, and what has alreadyhappened,to
the scholarlyor critical book on the Webl Phil Gyford'stranslation of Samuel
Pepys'sDiary into a Weblog, at which we looked in chapter 3, exemplifies a
new scholarly genre that took form outside the academy.Siteslike Slashdot
and many smaller oneson technicalsubjects,such as softwarefor Weblogsor
digital photography,are understandable,given the nature ofearly adopters:
Slashdot'smotto is "News for Nerds. Stuffthat Matters."One doesnot expect
scholarshipin the humanities, with its long-establishedhostility to collabora-
tive publication,to comeonline in sucha radicallycollaborativeform, and,yes,
it turns out it that did not take placewith any support from academicinstitu-
tions, most of the membersof which, I'm sure,do not know it exists.In con-
trast, new media studies (not surprisingly) has embracedthe blog as a schol-
arly genre, attaching them to interviews and book reviewspublished online.
TheseWeblogsproducea kind of scholarshipand criticism, perfectlyvalid,
which centerson short forms-the essayand review What about the schol-
arly book or monographl In an attempt to answer this question, which has

RECONFIGURINC been nagging at me for a decade,I decidedto carry out an experiment, test-
WRITINC ing my proposition that hypermediaprovidesliterary scholarswith the equiv-
alent of a scientific laboratory.Therefore, as some of my own books on Vic-
torian art, literature, and religion havegone out ofprint, I retrievedcopyright
from my publishers, translated them into HTML, and placed them on The
Victoian Web; since the appearanceof Hypertert 2.0 more than a dozen
scholarshavecontributed one or more of their works to this enterprise.As I
did so, I began to take advantageofcharacteristics ofhypertext that justify
translating a book into a web, the most basic of which is the synergythat
derivesfrom linking materials together. Reasonsof economy and scalehad
prevented including illustrations of the many paintings mentioned in the
original print version of my edition of the letters of fohn Ruskin and the Vic-
torian artist, W. Holman Hunt, but once I had createdweb versions of this
correspondence,my book on that artist, and articles from Art Bulletin and
other iournals, I found that these texts all worked together better than they
had alone.A footnote providing the sourceof a letter could now, for example,
lead to the text of the letter itself. Evenmore important, if an illustration was
availableanywherein the text, itbecame availableeverywhere.Textsneeding
illustrations particularly benefit from electronic presentation,since a digital
image,which is a matter of codesrather than physicalmarks on physicalsur-
faces,multiplies while taking up no additional space.Using an image fifty
times within one text or set of texts requires no more storagespaceor other
resourcesthan doesusing it once. Digitization thus permits the reuse of the
image at severalfixed placesin a text; hypertexfualizationpermits the image
to be calledup from numerous points as the readerfinds its presenceof use.
In World Wide Web viewers, which temporarily store images downloaded
from a network in a cache,reusing the sameimage takesmuch lesstime than
it did obtaining it in the first place.
A final reason for translating a book into an electronic environment
involvesadding capacitiesnot possible in print. Hypermedia translations of
print texts in mathematics,sciences,music, history and the arts havealready
appeared,employing sound, animation, video,and simulation environments.
Let us look at someinstancesofthese when proposing some generalrules for
employing sound and motion within alphanumeric text.
Assuming that you have a text that demands hypertextualization,how
should you go about carrying it out? Since I have thus far convertedseveral
dozen books into various hypermedia systems,I believe the best way to
answer that question would be to summarize my experienceand use it to

3.0 draw some general guidelines. Furthermore, since some of these electroni
fied books exist in two or more different hypertext environments, we can
observe the degree to which minor differences in hardware and software
influence hypertextuali zation.
First, one has to obtain an accuratedigital version ofthe text to be con-
verted. For my earlier books and articles,written before I beganto work with
computing, I used OminPage Professional,software for scanning text and
then interpreting the resultant image into alphabeticcharacters-an often
time-consuming process.Since I had written the first version of this book in
Microsoft Word, working with it proved to be fairly easy.Converting the text
for Intermedia required only saving it in a particular format (RTF)and then
creating links within Intermedia. Working in Storyspaceproved even easier
because this system imports Word documents, automatically translating
footnotes into linked lexias; the one chapter translated in HTML used the
Storyspaceexport function to create a basic linked working text to which I
then addedheader and footer icons. The DynaTextversion required adding
SGML tags and manually adding coding for links.
In adapting the printed text for all four kinds of hlpertext systems, I
found I had to make decisionsabout the appropriatelength of lexias.In each
case,I took chaptersalreadydivided into sectionsand createdadditional sub-
divisions. Whereasprint technology emphasizesthe capacityof languageto
form a linear stream of text that moves unrelentingly forward, hlpermedia
encouragesbranching and creatingmultiple routesto the samepoint. Hyper-
texhralizing a document therefore involves producing a text composed of
individual segmentsjoined to others in multiple waysand by multiple routes.
Hypermedia encouragesconceiving documents in terms of separatebrief
readingunits. Whereasorganizingone'sdataand interpretation for presenta-
tion in a print medium necessarilyleadsto a linear arrangement,hypermedia,
which permits linear linking, nonethelessencouragesparallel, rather than
linear, arguments. Such structures necessarilyrequire a more activereader.
Sincea major sourceof all these characteristicsof hypermedia derivesfrom
these linked reading units, one has to createhypermedia with that fact in
mind. Therefore,when creatingwebs,conceivethe text units asbrief passages
in order to take maximum advantageof the linking capacitiesof hypermedia.
Whateverits ultimate effecton scholarlyand creativewriting, hypermedia
todayfrequently contains so-calledlegacyte)ct-texts, like the 1992version of
Hyperturt,originally createdfor deliveryto the readerin the form of a printed
book. Such materials combine the two technologiesof writing by attaching
linked documents,which may contain images,to a fixed steam of text. Any-

RECONFIGURING one preparing such materials confronts the problem of how bestto preserve
WRITING the integrity of the older text, which may be a literary, philosophical, or other
work whose overall structure plays an important role in its effect. The basic
question that someonepresenting text createdfor print technologyin hyper-
media must answeris, Can one divide the original into reading units shorter
than those in which it appearedin a book, or does such presentationviolate
its integrityl Someliterary works, such as sonnet sequencesor Pascal'sPen'
sdes,seemeasilyadaptedto hypermedia sincethey originally havethe form of
brief sections,but other works do not seemadaptablewithout doing violence
to the original. Therefore, when adapting documents createdfor book tech-
nology, do not violate the original organization, though one should take
advantageof the presenceof discrete subsectionsand other elements that
tend to benefit from hypertextualization.However,when the text natura$ di
vides into sections,these provide the basis of text blocks. The hypermedia
version must contain linkagesbetweenprevious and following sectionsto re-
tain a senseofthe original organization.

ConvertingFootnotesand Endnotes.The treatment of notes in the four

hypertext versions ofthe first version ofthis book provides an object lesson
about the complexities of working in a new kind of writing environment- It
reminds us, in particular, how specificwriting strategiesdependon a combi-
nation of equipment and often apparentlytrivial features of individual sys-
tems, some of which militate against what seem to be intrinsic qualities of
hypertext.For example,as we havealreadyobservedin "Reconfiguring Text,"
hypertextualizationtends to destroythe rigid opposition between main and
subsidiarytext, thereby potentially either removing notes as a form of text or
else demanding that we createmultiple forms of them. Certainly,in hyper-
textualizing some of my own works, longer footnotesand endnotescontain-
ing substantivediscussionsbecamelexias in their own right.
Briefer notes that contain bibliographical citations embody a more com-
plex problem that has severaldifferent solutions, each of which createsa
different kind ofhypertext. If one wishes to produce a hypertextversion of a
printed original that remains as close as possibleto it, then simply convert-
ing endnotesinto a single list of them makes sense;using the <A Name>tag,
eachlink will leadto the appropriateitem. Another approach,which produces
an axially structured hypertext, involvesplacing eachbibliographical note in
its own lexia and linking to it. Using simple HTML, authors can make this
lexia open in a number of ways. Following standard HTML procedure,the
note replacesthe text in which the link to the note appears;one can createa

3.0 return link to the main text, or rely on knowledgeablereaders to use the
browser's"Back" button. In addition, one can leavethe body of the document
on screenby using the "target: -blank' ' option in the link (A HREF),which
makes the note open in a new window; authors who take this route, opening
annotationsin a separatewindow often include instructions to closethe win-
dow to return. Unfortunately, both ofthese convenient approachesproduce
an unattractivedocument in which a sentenceor two appearsat the very top
of a large, otherwise empty window A third approachuses HTML frames to
bring up the text of the note in a column next to the main text, and yet a fourth
usesHTML tablesto placethe text in the margin, therebyrecreatingthe effect
of some eighteenth-centurybooks; tables are, however,very difficult to use
when one of the columns has large blank areas.Perhapsthe most elegant
solution employs fava scripts to createsmall pop-up windows for eachnote.
The problem here involvesthe nature ofone's audience:fava scripts notori-
ously do not work in all browsers or even in all versions of the same one. If
you createyour materials for the widest possible audience,which includes
many userswith old versionsof Netscapeand Internet Explorer,you will have
to forgo using this elegant,if time-consuming, solution, but if you direct your
materialsat a singleeducationalor commercialinstitution that has announced
standards for supported hardware and software,you can use anything that
works there, though you may lose peopleworking at home.
Whateveroption you choose(and all haveadvantagesand disadvantages),
try not to use superscriptnumbers to indicate the presenceof links. Not only
doesit prove difficult to mouse down on the small target provided by a single
tiny character,but, more important, numbered notes only make sensewhen
readersconsult them in a list, and placing notes in separatelexias destroys
this list. I find that one can almost alwaysuse a relevantphraseas an anchor
for a note to a link. When the note contains bibliographical information, one
can link to a relevant phrase, such as "(source)"or "(bibliographical materi-
als),"placedat the end of the sentence.
Whereverpossible,the best and most obvious solution to the problem of
representing annotations in Web documents involves converting all biblio-
graphicalnotesto the current Modern LanguageAssociation(MtA) in-text
citation form, whether one links all such citations to a list of referencesor
just includes the relevantbibliographicalitems in eachlexia; I prefer the
latter approach.
In addition to dividing a print text into sections,adapting notes and bib-
liography,and adding headerand footer icons, creatinga hypermediaversion

RECONFICURING requires adding featuresand materialsthat would be impracticableor impos-

WRITING sibleto havein a printed version.Thus, the Storyspace,Microcosm,and World
Wide Web versions of both my Pre-Raphaelitematerials and Hypertert con-
tain a great many links that serveas cross-referencesand that provide addi
tional pathsthrough the text, and they haveadditionalimages,too. Thesewebs
also have elements not found in books, such as multiple overviewsthat per-
mit traversingthem in waysdifficult, or impossible,in a print version.A1lthe
hypermediatranslationsof Hypertert,for example,contain overviewsfor both
critical theory and hypertext, and various versions add ones for information
technology,scribal culture, and individual theorists.
Perhapsthe most obviousdifference-in addition to links-between the
hypertext and print versions lies in its size: the way links produce an open-
ended,changing,multiply authoredVelcro-textappearsnowheremore clearly
than in the fact that so much new material appearsin Hypefiert in Hypefiert
than in the print version.As one might expectfrom what I've alreadywritten,
once I createdIntermedia and Storyspaceversions for my course on hyper-
text and literary theory my students read them as wreaders-as active,even
aggressivereaderswho can and did add links, comments, and their own sub-
webs to the larger web into which the print has version has transformed it-
self.Within a few yearsthe classroomversion containedfive hundred of their
interventions, criticizing, erpanding, and commenting on the text, often in
waysthat take it in very different directions than I had intended. In addition
to some fifty of these new lexias, Hypefiert in Hyperturt contains entries on
individual theories and theorists from Thelohns Hopkins Guide to Literary
Theoryand Citicism, edited by Michael Groden and Martin Kreiswirth, as
well as materialsby Gregory Ulmer and facquesDerrida. To thesematerials,
we added,with permission, some of Malcolm Bradbury'sparodiesof critical
theory and all the reviewsthe book had receivedby the time we went into pro-
duction. These new lexias,which constitute a subweb of their own, serveto
insert other voices,not always in agreementwith mine, into the expanded
text. Throughout, the principle of selectionwas be a cardinal mle of hyper-
text adaptation-use materials only when they servea purpose and not just
becauseyou havethem. Hypertext writing, in other words, should be driven
by needsand not by technology.

Rulesfor DynamicData in Hypermedia.The precedingpageshavefocused

on writing hypertext with essentially static forms of data-words, images,
diagrams, and their combinations. Kinetic or dynamic information, which

3.0 includes animation and sound (animal cries,human speech,or music),raises
additional problems becauseit imposes a linear experienceon the reader.
This strong element of linearity itself presentsno fundamental difficulty, or
evennovelty,since as Nelson pointed out long ago,individual blocks oftext,
particularly those with few links, offer a linear reading path.
The differencebetweendynamic and static datalies in the fact that it is so
importantly time-bound. Speechor visual movement potentially immerses
readersin a linear processor progressionoverwhich they haverelativelyless
control than they do when encountering a staticdocument, such as a passage
of writing. One can stop and start one's reading at any point-when the
phone rings, a child cries, or a thought strikes one. Turning one's attention
awayfrom time-bound, linear media, in contrast,throws one out of one's
position within a linear stream, and this place cannot be recovered,as it can
with writing, simply by turning attention back to the tert, since one'spoint,
or location, or place within the text has moved on. Therefore,when one fol-
lows a link from a text discussing, say,mitosis to digitized animation of
a cell dividing or from a work ofcriticism to a scenefrom Shakespeare,one
cuesthe beginning of a process.Such dynamic dataplacethe readerin a rel-
ativelypassiverole and turns hypermediainto a broadcast,rather than an
Many websitesand CD-ROMs employ Quicktime movies accompanied
by sound to present materials often more efficiently and more enjoyably
encountered as text. Using the talking heads approach in which someone
filmed from the chest or neck upward looks out of the screenand talks to us
can occasionallyprove an effectivestrategy,particularly when establishing a
person'sappearanceand characterseemsimportant, but it expendstwo kinds
of valuable resources: storage capacity and, becauseit occupies time, the
reader'spatience, or at least forbearance.Too many creatorsof HTML and
CD-ROM materials seem grossly unaware of the fact that one of the key
advantagesofwritten languagelies preciselyin abstractnessand indirection
that permit communicating important information with great economy.
Systemsdesignersand hypermedia authors haveto empower readersin
at leasttwo ways.First, they must permit readersto stop the processand exit
the environment easily.Second,they must indicate that particular links lead
to dynamic data. One may use labels or icons for this purpose, and one may
also connectthe link-indicator to an additional document, such as a menu or
command box, that gives users precisely the kind of process-information
they can activate.Such documents in the form of a control panel, which per-

RECONFIGURINC mits the readerto manipulate the processto the extentof replaying all or part
WRITING of a sequence,make the readermore active.If one employs talking headsor
voiceovers,one must, of course,allow readersto stop them in midsequence.
(Voyager'sFreakShowCD-ROM makes a witty play on these featuresby hav-
ing its Master of Ceremoniesor Ringmaster respond with different expres-
sions of annoyanceeachtime we cut him offin midsentence.)
The one digital form that doesnot createproblems for hypermedia is the
fundamentally controllablemultimedia document createdby Apple'sQuick-
time VR (Virtual Reality) or rival software like lpix and Live Picture. These
kinds of software, whose creations are easily inserted into HTML, pro-
duce two different kinds of manipulatable three-dimensional images, the
so-calledspherical and cylinder panoramas. In the first, one finds oneself
placedin a three-dimensionalspacewithin which one can rotate 360 degrees
by using the mouse to stop, start, and control the speedof rotation or, using
a zoormfunction, move closer or farther away.The World BookEncyclopedia's
two CD-ROMs, for example, include dozens of Ipix scenesin the format
they call "bubble views," including St. Mark's Square,Venice; Stonehenge;
the Maya ruins at Palenque;the Coliseum in Rome; and the Zojoi Temple
in fapan. Or if one wants a Web example, go to NASA's Mars Pathfinder
site for a panorama of the SaganMemorial Station,or A Wrinklein Time:
q collaborativesynchronizedtfort by QTVR producersqround the globeIo
createa hundred panoramic views at the same time on December 21, 7997
The other kind of Quicktime VR (the cylinder panorama)takesthe form
of a virtual objectthat users can turn 360 degrees,examining it as they wish,
as well as zooming in and out. This kind of image, possible only with com-
puting, has greatvalue when representingthree-dimensionalobjectsonline.
The Victoian Web,for example,contains a Quicktime VR image of an unat-
tributed bronze statueof a young woman that I believewas createdby Alfred
Drury (1856-1944).While carrying out researchthat might lead to an attri-
bution, I visited various English photo-archivesand collectionsof sculpture
without finding any particularly convincing evidence. Severalyears later,
after I had createda rotatableQuicktime VR image of the statue,I cameupon
a photograph of Drury's 1896 bronze, Griseldq,in a 1980 catalogue from
Christopher Wood Gallery in London. Observingthe way in which the sculp-
ture depicted the pleats and folds around the shoulder of the Giseldo, I
openedthe Quicktime VR image of the unatlributed statue,using my mouse
to rotate it until I positioned the statue as seenfrom the samevantagepoint

3.0 as the photograph of Giseldo. At that point, it became obvious that both
sculptureshad been executedby the same person. Obviously,anyone inter-
estedin Victorian and Edwardiansculpture would prefer to haveboth works
in the same physical spaceand be able to move them about until one could
look at them from the same vantage point; anyone who has ever studied
sculpture of this period immediately realizeshow unlikely would be the
opporfunity to make such a comparison. Instead,the researcherusually has
to use published photographs,each ofwhich ofnecessity is taken from one
position, and even if one visits major museums, one often finds that sculp-
tures are displayedin such a way (often in a corner or againsta wall) that one
cannot obtain the view one wants. With Quicktime VR, one can. Anyone in-
terestedin sculpture clearly caresabout the materiality, the sheer mass and
surfaceofthe object in question, and so one really wants accessto the origi
nal objectswith the ability to move and touch them and gazeat them for a
long time under different kinds of light conditions. If one cannot have the
actual object,then Quicktime VR is the next best thing; it is certainly su-
perior to my 35 mm slides,which make TheGhentAltorpieceandanny wood-
engravingappearto be on the same scale.

Most current examples of hypertext take the form of texts

Hypertextas CollageWriting originally producedby the hypertextauthor in and for another
medium, generally that of print. In contrast,this section on
collage writing derives from a hypertext, though it incorporates materials
ultimately derived from printed books, too. On Tuesday, June 7, 1994,at
17:.07:54Eastern StandardTime, Pierre Joris, a faculty member at the State
University of New York, posted some materials about collageon a electronic
discussion group called Technoculture. (I have discussedthe first year of
Technoculture'sexistencein "Electronic Conferencesand SamiszdatTexru-
ality: The Example of Technoculture,"in the 1993MIT volume, The Digitol
Word,which I edited with Paul Delany.)foris wished to sharewith readersof
this e-conferencea gathering of texts on the sub;'ecthe had delivered as a
combination of an academicpaper and performance art while in graduate
school.His materials seemedto cry out for a hypertext presentation,and so
after moving them from my mailbox to a file on the Brown University IBM
mainframe, I transferred them-in the jargon, "downloaded them"-in a
single document via a phone line to a Macintosh whirring awayin my study
at home. Next, I opened them in Microsoft Word, and, passageby passage,
quickly copied the individual elements of "Collage between Writing and
Painting," pasting eachinto a separatewriting spaceor lexia in a new Story-

RECONFIGURING spaceweb and then linked them together.Along the way, I createdthe fol'
WRITING lowing opening screen(or analogueto a book'stitle page):




b e i n ga n a s s e m b l a gset a r r i n g

Kurt Schwitters
& Tristanf zara
withspecial by
Ceorges &

to. . .

This opening screen,which also servesas a combination overview,informa-

tion map, contentspage,and index, containslinks from the obviousplaces-
such as,for example,all the proper names it lists. Clicking on "Collage"takes
one either to one possibleterminal point of the web or to a definition of the
term from Le Petit Robert.Since this dictionary definition, which mentions
Picassoand Braque,servesas a another ready-madeoverview or crossroads
document, I linked variouswords in it to permit readersto traverseforis'sma-
terials in multiple ways."COLIAGE," for example,leadsto a dozen and a half
mentions of the term. and the names of the artistslink to illustrations of their
work. BecauseI createdthis web largely as an experiment and not for publi
cation, I did not have to worry at the moment about copyright issues and
therefore scannedmonochrome imagesof Braque'sLe Counier and Picasso's
Still Lifewith Chair Coningandlinked them to the names of the artists.At the
same time I added H. W. Janson'sdiscussions of collage,linking them as
well. Finally, I createda list of thirty authorswhose statementsforis included
in "Collage between Writing and Painting," linking this list to the phrase
"also featuring" on the title screen.
At this point, some of the similarities betweenhypertext and collagewill
have become obvious. Having first appropriatedforis's materials by placing
them in a web and then adding materials to it that they seemedto demand, I
found that, like all hlpertexts, it had become open-ended,a kind of Velcro-
text to which various kinds of materials began attaching themselves.First,
I included a discussion of Derrida and appropriation, after which I added

T. 0 definitions of hypertextand a list of qualities that it shareswith collage.Next,
I added severaldozen screenshots,or pictures of how the screen appears
while reading, of various hypertextwebs;these came from a since-published
web that servedas an introduction to the hypertext anthology,Witing at the
Ed.ge.Then,Iaddeda dozenphotographs,eachinvolving issuesof represen-
tation, illusion, simulation, or subjectand ground. Finally, I addeda newtitle
pagefor Hyperturtand Collage:beinginpart,an appropiotion of "collagebetween
writing and pointing."
After using this web to deliver my contribution to the August 1995Digital
Dialecticconferenceat the Art Center Collegeof Design, I discoveredI would
haveto transform it into a more or lesstraditional essayif it were to be part of
the plannedvolume. Thesepagesthus representa translationof the Hypefiert
and CollogeWeb.When I write "translation," I cannot help thinking of the
Italian maxim "traddutore= traditore"or "translator: traitor." Converfingthe
essayfrom one information technologyto another,I continually encountered
the kind of reduction that one encounters translating-or representing-
somethingin three (or more) dimensions within a two-dimensionalmedium.
An examinationof the differencesbetweenthe two versionswill takeus a way
into understanding the reasonsfor describing hypertext as collagewriting.
The online version of the OxfordEnglishDictionary definescollage,which
it tracesto the French words for pasting and gluing, as an "abstractform of
art in which photographs,pieces of paper, newspapercuttings, string, etc.,
are placedin juxtaposition and glued to the pictorial surface;such a work of
art." The Bitannica Online more amply describesit as the

of applying
or "found"materials,
bitsof newspaper,fabric,wallpaper,
etc.,to a panelor canvas,
in combrna-
tionwith painting.Inthe lgth century,
to formdecorative
In aboutI 912-l 3 PabloPicasso
andCeorgesBraque extended
combiningfragments of paper,
linoleum, andnewspapers
withoil painton canvas
to formsubtleandinteresting
stractor semiabstract
compositions. Thedevelopment of thecollage
to thetransition
to Synthetic

This referencework, which addsthat the term was first used to refer to dada
and surrealistworks, lists Max Ernst, Kurt Schwitters,Henri Matisse,foseph
Cornell, and RobertRauschenbergasarListswho haveemployedthe medium.
ln The History of WorldArt, H. W. fanson, who explains the imporrance
of collageby locating it within the history of cubism, begins by describing
Picasso'sStill Life of 19I1-t2: "Beneaththe still life emergesa piece of imita-

RECONFIGURING tion chair caning, which has been pastedonto the canvas,and the picture is
'framed' by a piece of rope. This intrusion of alien materials has a most

remarkable effect the abstractstill life appearsto rest on a real surface (the
chair caning) as on a tray, and the substantiality of this tray is further em-
phasizedby the rope.'According to )anson,Picassoand Braqueturned from
brush and paint to "contents of the wastepaperbasket" becausecollageper-
mitted them to explorerepresentationand signification by contrasting what
we in the digital agewould call the real and virtual. They did so becausethey
discoveredthat the items that make up a collage,"'outsiders' in the world of
art," work in two manners, or produce two contrary effects.First, "they have
been shapedand combined, then drawn or painted upon to give them a rep-
resentationalmeaning, but they do not lose their original identity as scraps
'outsiders'in the world of art. Thus their function is both to rep-
of material,
resent (to be part of an image) and to present (to be themselves)"(522-23).
Hypertext writing shares many key characteristicswith these works of
Picasso,Braque,and other cubists,particularly their qualities of juxtaposition
and appropriation. Some of these qualities appearwhen one comparesthe
hypertextand print versionsof my discussion.First of all, despitemy division
of this essayinto severalsectionsand the use of figures that a reader might
inspect in different orders, this essayreally only allows one efficient way of
proceeding through it. In contrast, the original hypertext version permits
different readersto traverseit accordingto their needs and interests' Thus,
someonewell versed in twentieth-century art history might wish to glance
only briefly at the materials on collagebefore concentratingfirst on the ma-
terials about hypertext. Someoneelsemore acquaintedwith hypertext could
concentrateon the materials about collage.Others might wish to begin with
one portion of the discussion,and then, using availablelinks, return repeat-
edly to the sameexamples,which often gathermeaning accordingto the con-
texts in which they appear.
Another differencebetweenthe two forms of "my" discussionof this sub-
ject involvesthe length of quoted material and the way the surrounding texts
relate it to the argument as a whole. Thke,for example,the passageI quoted
abovefrom fanson'sHistoryofWorld Art.In the Storyspaceversionthe passage
is severaltimes longer than in the print one, and it appearswithout any intro-
duction. The object here is to let the quoted, appropriatedauthor speakfor
himself,, or, rather, to permit his text to speakfor itself without being sum-
marized,translated,distorted by an intermediary voice.To write in this man-
ner-that is to say,to copy,to appropriate-seems suited to an electronic
environment, an environment in which text can be reproduced,reconfigured,

3.0 and moved with very little expenditure of effort. In this environment, fur-
thermore, such a manner of proceedingalso seemsmore honest: the text of
the Other may butt up againstthat by someoneelse;it may evencrashagainst
it. But it does seem to retain more of its own voice. In print, on the other
hand, one feels constrained to summarize large portions of another's text,
if only to demonstrate one's command (understanding) of it and to avoid
giving the appearancethat one has infringed copyright.
These two differences suggest some of the ways in which even a rudi-
mentary form of hypertextrevealsthe qualities of collage.By permitting one
to make connections between texts and text and images so easily,the elec-
tronic link encouragesone thus to think in terms of connections.To statethe
obvious: one cannot make connections without having things to connect.
Thoselinkable items must not only havesome qualities that make the writer
want to connect them, they must also exist in separation,apart, divided. As
Terence Harpold has pointed out in "Threnody,"most writers on hypertext
concentrateon the link, but all links simultaneously both bridge and main-
tain separation(I74). This double effect of linking appearsin the way it
inevitably producesjuxtaposition, concatenation,and assemblage.If part of
the pleasureof linking arisesin the act of joining two different things, then
this aestheticof juxtaposition inevitably tends toward catachresisand differ-
ence for their own end, for the effect of surprise, and sometimes surprised
pleasure,they produce.
On this level, then, all hypertext webs, no matter how simple, how lim-
ited, inevitably take the form of textual collage,for they inevitably work by
juxtaposingdifferent textsand often appropriatingthem aswell. Sucheffects
appearfrequently in hypertextfiction. foshua Rappaport's"The Hero's Face"
(one of the webs included in Witing at the Edge)uses links, for example,to
replacewhat in earlier literary writing would have been an element internal
to the text; that is, the link establishesa symbolic as well as a literal relation-
ship betweentwo elements in a document. In "The Hero's Face,"after mak-
ing onek way through a seriesof lexias about the members of a rock band,
their experienceson tour, and their musical rivalry-all of which might seem
Iittle more than matters of contemporarybanality-the readerfollows a link
from a discussionof the narrator'sseizing the lead during one performance
and finds herself or himself in what at first appearsto be a different literary
world, that of the Finnish epic, the Kalelava.
Following Rappaport'slink has severaleffects. First, readersfind them-
selvesin a different, more heroic ageof gods and myth, and then, as they re-
alize that the gods are engagedin a musical contest that parallels the rock

RECONFICURING group's,they alsoseethat the contemporaryaction resonateswith the ancient

WRITINC one, thereby acquiring greatersignificanceas it appearsepic and archetypal'
This single link in Hero'sFace,in other words, functions as a new form of
both allusion and recontextualization. Juxtaposingtwo apparently uncon-
nectedand unconnectabletexts producesthe pleasureofrecognition.
Such combinationsof literary homageto a predecessortext and claims to
rival it have been a part of literature in the West at least since the ancient
Greeks. But the physical separationbetween terts characteristicof earlier,
nonelectronic information technologies required that their forms of link-
ing-allusion and contextualization-employ indicatorswithin the text, such
as verbal echoing or the elaborateuse of parallel structural patterns (such as
invocations or catalogues).Hypertext, which permits authors to use tradi-
tional methods,also allowsthem to createtheseeffectssimply by connecting
textswith links. David Goldberg'sweb essay,"New PerverseLogic:The Inter-
face of Technologyand Eroticism in f . G. Ballard's Crashand william Gib-
sonlsNeuromancer"(I996\,uses HTML frames to accomplisha similar form
of juxtaposition without links. Clicking on various topics from the opening
screen opens a two-column document, in one half of which appeardiscus-
sions of Baudrillard and Gibson, virtual textuality and Ballard,and two pas-
sagesfrom the novelist.
Hypertext here appears as textual collage_lartual tefering to alpha-
numeric information-but more sophisticatedforms of this medium also
produce visual collageas well. Any hypertext system (or, for that matter, any
computer program or environment) that displays multiple windows pro-
ducessuch collageeffects.Multiwindow systems,such as Microcosm, Story-
space,Intermedia, Sepia,and the like, havethe capacityto savethe size and
position of individual windows. This capacityleadsto the discoveryof what
seemsa universal rule at this early stageof e-writing: authorswill employ any
feature or capacitythat can be varied and controlled to conveymeaning. All
elements in a hypertext system that can be manipulated are potentially sig-
nifying elements. Controlled variation inevitably becomessemiosis. Hyper-
text authors like Stuart Moulthrop have thus far written poems in the inter-
stices of their writing environments, creating sonnets in link menus, and
sentencesin the arrangementsof titles of lexias in the Storyspaceview'
Inevitably,therefore, authors make use of screenlayout, tiled windows,
and other factors to .. . write. For example, in an informational hypertext,
such as The In Memoriam Web,tiling of documents constructsa kinetic col-
lage whose juxtaposition and assemblingof different elementspermits easy
referenceto large amounts of information without becoming intrusive (see
1 .l :;:., i i-i1.1;{.,
l i:li i-"1 f-1 fjfty ,_ll Ar+gtirir:*

i tilr rr'c('ll,r mt,un trt.$(r.\'.lllr/ril{.

Clang {;1 i: i,n .{ei*ts lliqfu.;;r;

a*l;ss the rv*p/r *i,g/rrr,rl.r,

._ 5 r' ..

: lt was starting"
I r i# * * briv*$ir
" l , t ^ I . l : t - - l - - . . , : , t

i ri rr

Figure22. Digital Collagein Hypertext Narrative: Nathan Marsh's Breoth ofsighs and FallingForever.Marsh has arrangedthe texts

that makeup his web so that some lexiasshow in their entirety,othersonly in part. In makingtheir way through this fiction, read.

ers encountermultiple narrativelines.The web continuallychangesthe juxtapositionoftexts as the narratornavigatesin the text.
In the courseofreading,one is repeatedlyreturnedto the lexia"Clang!'which openswith the sound ofan explosion,butthe mean-
ing ofthe word changesaccording to the lexia that one has read immediately before encountering it.

Figure 11).In addition to employing the set placementof the windows, read-
ers can also move windows to compare two, three, or more poems that refer
back and forth among themselvesin this protohypertextualpoem.
Turning now to anotherwork of hypertextfiction, one seesthat in Nathan
Marsh's Breathof Sighsand Folling Foreverlexias
placethemselvesaround the
surfaceof the computer monitor, making the screenlayout support the nar-
rative as one crossesand recrossesthe tale at severalpoints (Figure22).ln
TheIn Memoiam Webthe collage-effectof tiling, separatewindows, and jux-
taposedtext arisesin an attempt to use hypertexttechnologyto shedlight on
qualities of a work createdfor the world of print (seeFigure 9.) Here this story
arisesout of the medium itself. In making their way through this fiction read-

RECONFIGURING ers encounter multiple narative lines and corollary narrative worlds both
WRITING ioined and separatedby ambiguous eventsor phenomena.At certain
readerscannot tell, for example,if one of the charactershas experiencedan
earthquaketremor, a dmg reaction,or a powerful illumination. Has the floor
actuallyfallen, or are we supposedto take a character'sexperienceas figura-
tivel Certainly,one ofthe first lexiasreadersencountercould suggestany and
'Andy pausedfor a secondand let his sensesadjust
all of these possibilities:
to the shock.The floor had been dropping all week now. As he satby the open
window and the frozen night air embraced the room, he realized that it
was all part of the long slide down." Clicking on this brief lexia leads one to
"Clangl," which opens with a loud sound and displaysits single word in 80-
point type. As one readsone'sway through "Breath of Sighs" one repeatedly
returns to "Clang!" but finds that it changesits meaning according to the
lexia that one has read immediately before encountering it.
Marsh has arrangedeach of the texts that make up his web so some lex-
ias show in their entirety,others only in part. As one readsthrough this web,
one encounters a continually changing collage of juxtaposed texts' Two
points about hypertext writing appearin Marsh's web. First, we realize that
such collage writingproduces a new kind of readingin which we must take
into account not only the main text but also those that surround it. Second,
this emphasis on the increasing importance of the spatial arrangement of
individual lerias leads to the recognition that writing has become visual as
well as alphanumeric; or since visual layout has alwayshad a major impact
on the way we read printed texts, perhaps it would be more accurateto say
that in hypertert (where the author controls more of the layout) hypertext
writing requires visual as well as alphanumeric writing. Marsh'sweb exem'
plifies a form of hlpertext fiction that draws on the collagequalities of a multi'
window systemto generatemuch of its effect.
Despiteinteresting, evencompelling, similarities, hypertextcollageobvi'
ously differs crucially from that createdby Picassoand Braque.Hypertext and
hypermedia alwaysexist as virhral, rather than physical, texts. Until digital
computing, all writing consistedof making physical marks on physical sur'
faces.Digital words and images,in contrast,take the form of semiotic codes,
and that fundamental fact about them leads to the characteristic,defining
qualities ofdigital infotech: virtuality, fluidity, adaptability,openness(or ex-
isting without borders),processability,infinite duplicability, capacityfor be-
ing moved about rapidly,and finally, networkability. Digital text is virtual be-
causewe alwaysencounter a virhtal image, the simulacrum, of something
stored in memory rather than encounter any so-calledtext itself or physical

3.0 instantiation of it. Digital text is fluid because,taking the form of codes,it can
alwaysbe reconfigured,reformatted,rewritten. Digital text is henceinfinitely
adaptableto different needsand uses,and since it consistsofcodes that other
codescan search,rearrange,and otherwisemanipulate,digital text is also
always open, unbordered, unfinished and unfinishable, capableofinfinite
extension. Furthermore, since it takes the form of digital coding, it can be
easilyreplicatedwithout in any way disturbing the original codeor otherwise
affecting it. Such replicability in turn permits it to be moved rapidly across
great spacesand in so doing createsboth other versions of old communica-
tion, such as the bulletin board, and entirely new forms of communication.
Finally-at leastfor now-all theseother qualities of digital textuality enable
different terts (or lexias)to 1'ointogether by means of electroniclinking. Dig-
itality, in other words, permits hypertextuality.
The connection of the fundamental virnrality of hypertextto the issue of
collagebecomesimmediately clear as soon as one recalls the history of col-
lage and the reasonsfor its importance to picasso,Braque, Schwitters,and
other painters.As |anson explains,collagearosewithin the contextof cubism
and had powerful effectsbecauseit offered a new approachto picture space.
Facetcubism, its first form, still retained ,,acertain kind of depth,,,and hence
continued Renaissanceperspectivalpicture space."In collage cubism, on
the contrary the picture spacelies in front of the plane of the 'tray'; spaceis
not createdby illusionistic devices,such as modeling and foreshortening,but
by the actualoverlappingof layersof pastedmaterials" (522-23).Theeffectof
collagecubism comes from the way it denies much of the recent history
of western painting, particularly that concernedwith creating the effect of
three-dimensionalspaceon a two-dimensional surface.It does so by insert-
ing some physicallyexisting object,such as picasso,schair-caningand news_
papercuttings, onto and into a painted surface.Although that actof inclusion
certainly redefines the function and effect of the three-dimensional object,
the object nonetheless resists becoming a purely semiotic code and abra-
sively insists on its own physicality.
The collageofcollage cubism therefore dependsfor its effect on a kind of
juxtaposition not possible (or relevant) in the digital world-that between
physical and semiotic. Both hypertext and painterly collagemake use of
appropriation and jurtaposition, but for better or worse one cannot directly
invoke the physical within the digital information regime, for everything is
The final lexia in this grouping, however, moves this more traditional
form of virtuality to that found in the world of digital information technology,

RECONFIGURING for it both repeatssectionsof all the images one may have seen (in whatever
WRITING order), blending them with multiply repeatedportions of a photograph of a
Donegal,Ireland, sunset,and it alsoinsists on the absenceofany solid, phys-
ical ground: not only do different-sizedversionsof the sameimage appearto
overlayone another but in the upper center a squarepanel has moved aside,
thus revealing what the eye reads as colored background or empty space.
In this photographiccollageor montage,appropriationand juxtapositionrule,
but, sinceall the elementsand imagesconsistof virnral images,this lexia,like
the entire web to which it contributes, does not permit us to distinguish
(in the manner of cubist collage)betweenvirtual and real,illusion and reality.
This last-mentioned lexia bears the title "sunset Montage," drawing on
the secondarymeaning of montageas photographicassemblage,pastiche,or,
as the OED puts it, "the act or processof producing a composite picture by
combining severaldifferent pictures or pictorial elements so that they blend
with or into one another; a picture so produced."I titled this lexia "Sunset
Montage" to distinguish the effect of photographicjuxtaposition and assem-
blage from the painterly one, for in photography,as in computing, *re con-
trast of physicalsurfaceand overlayingimage doesnot appear.Upon hearing
my assertion that hypertext should be thought of as collage writing, Lars
Hubrich, a student in my hypertext and literary theory course,remarked that
he thought rnontagemight be a better term than collage.He had in mind
sometlring like the first OED definition of montage as the "selection and
afiangement of separatecinematographicshots as a consecutivewhole; the
blending (by superimposition) of separateshotsto form a single picture; the
sequenceor picture resultingfrom such a process."Hubrich is correctin that
whereas collageemphasizesthe stage effect of a multiwindowed hypertext
systemon a computer screenat any particular moment, montage, at least in
its original cinematic meaning, placesimportant emphasison sequence,and
in hypertext one has to take into account the fact that one reads-one con-
structs-one's reading of a hypertextin time. Eventhough one can backtrack,
take different routes through a web, and come upon the samelexia multiple
times and in different orders,one nonethelessalwaysexperiencesa hypertext
as a changeablemontage.
Hypertext writing, of course,doesnot coincidefirlly with either montage
or collage. I draw upon them chiefly not to extend their history to digital
realms, and, similarly, I am not much concernedto allay potential fears of
this new form of writing by deriving it from earlier avant-gardework, though
in another time and place either goal might provide the axis for a potentially
interesting essay.Here I am more interested in helping us understand this

3.0 new kind of hypertext writing as a mode that both emphasizesand bridges
gaps and that thereby inevitably becomesan art of assemblagein which
appropriation and catachresisrule. This is a new writing that brings with it
implications for our conceptionsof tert as well as reader and author. It is a
text in which a new kind of connectionhas becomepossible.

What is quality in hypertextl How in other words, do we judge

ls This HypenextAny Goodl a hypertext collection of documents (or web) to be successful
or unsuccessful,to be good or bad as hypertert?How can we
Or, How Do We Evaluate
judge ifa particular hypertextachieveseleganceor neverrises
Qualityin Hypermedial abovemediocrityl Those questions lead to another: What izr.
particularis good about hypertext?What qualities doeshyper-
text havein addition to those possessedby nonhypertexhralforms of writing,
which at their best can boast clarity, energy,rhphm, force, complexity,and
nuancel What qualities, in other words, derive from a form of writing that is
defined to a large extentby electroniclinkingl What good things, what desir-
able qualities, come with linking, since the link is the defining characteristic
of hypertextl As I have argued earlier, the defining qualities of the medium
include multilinearity, consequentpotential multivocality, conceptual rich-
ness,and-especially where informational hlpertext is concerned-reader-
centerednessor control by the reader.Obviously,works in a hypertext envi-
ronment that firlfill some or all of thesepotential qualities exemplify quality
in hypermedia.Are there other perhapsless obvious sourcesof qualityl
One question we must raise while trying to identify sourcesof quality in
hypermedia is, To what extent do literary and informational hlpermedia
differl In the following pages,I shall proposeseveralpossiblewaysto answer
these questions,eachofwhich itselfinvolves a central issue concerning this
information technology.

IndividualLexiasShouldHavean AdequateNumberof Links.Sincethe link

is the characteristicfeaturethat defineshypertextuality,one naturally assumes
that lexiascontaining a larger number of valuablelinks are better than those
that have fewer. Of course,the emphasishere must be on "valuable."In the
early days of the Web, one would often come upon personal homepagesin
which virfually every word other than the articles the, a, and an had links,
many of which led to externalsitesonly generallyconnectedto the discussion
at hand. Obviously,overlinking, like choosing poor link destinations,is bad
linking. As Peter Brusilovskyand RiccardoRizzo havepointed out in .,Map-
BasedHorizontal Navigation in EducationalHypertext," the opposite prob-

RECONFIGURING lem-a lack of linking precisely in those places one would erpect it to ap-
WRITING pear-characterizes much recent World Wide Web hypertext. Part of the
problem here may come directly from the World Wide Web'suse of unsuitable
terminology derived from print technology,such as homepoge, which locks
neophyte users into an inappropriate paradigm. Brusilovsky and Rizzo cor-
rectly note that much hlpertext today takesthe form of passagesof unlinked
text surrounded by navigation links. Encountering these kinds oflexia, one
receivesthe impression that the authors, who have dropped digitized ver-
sions of printed pagesinto an electronic environment, dont seem to gnsp
the defining qualities of hypermedia and use HTML chiefly as a text format-
ting system.They are still working, in other words, with and within the par-
adigm ofthe printed page and book.
The Victoriqn Web(victorianweb.org),an academicsite I managethat re-
ceivesas many as fifteen million hits a month, contains four basic kinds of
documents: (1) overviews(sitemaps),(2) link lists, (3) simple two'column
tablesused primarily for chronologiesaswell as art works and text describing
them, and, (4)lexiascontaining primarily text,though somemay alsoinclude
thumbnail images linked to larger plates. Most text documents on this site
contain two to four navigationlinks in the form oflinked icons that appearat
the bottom of each lexia plus multiple text links that weavethe lexia into a
miniature hypertext network. Although I find myself unable to formulate any
rule as to proper number of text links, I have observedtwo things: (1) lexias
approximatelyone to two screensin length tend to have at least three text
links, and (2) as new documents arrive, older lexias receiveadditional links.
The comparativelack of text links observedin much web-basedhyper-
media alsoappearsin much hyperfiction, as many authorsseemuninterested
in using more than singlelinks, which createan essentiallylinear flow Caitlin
Fisher'sWavesof Girls, aweb narrativethatwon the 2001ElectronicLiterature
Organization (ELO)prize for electronicfiction, exemplifiesthe comparatively
rare literary hypertextthat includes both framing navigationallinks and oth-
ers in the body of the text. Thus, in the following brief example,the phrases"I
was so sad,""our principali' "grade5 boys . . . ," "making out really meant . . ."
all leadto-that is, produce-new text (Figure 23). In addition to the naviga-
tion links that appearat the left of the screen,the main text also contains fre-
quent opportunities to follow links, which lead to other narrative arcs.

Linking ininfot-
Followingthe Link Should Providea SatisfringExperience.
mational hypermedia obviouslyhas to work in a clear,coherentmanner, but
In other words, what should appear
what producesthis requisite coherence?a
talked about nr*king oul nn the senior
pl*ygrcund equipment -- ddiceted h)
thc memory of trro ela*smatesrun
over by a chickenJieken delivery
truck k was s0 gld, I remember lhe
day rhey dedicatedthe playground
equipment:"Now children, this
$quipmfi1t shnuld rEmind us all of Ben
and Th*rese who ars now angel*. fu-
pri,ol]pal wondered wby nr: one {q
wante{, r0 play therc mucb.

Crade five girl* w{ruld pin b*ys dn&xl

on lhe
lnseb*ll diamand_aadFrenchgis*
then:. T*mmy $tvens s*w ills $nce
aod told n* t }}le$ $othiilg sbout
Iove. Mr:stly the gr&de *ixes dared
qach olher to g$ to ths liltls house at
the klp *f the slid* and meke out,
Meking $ut reBlll m*a$t litiir&up
l'our shin. ifJsu w*re * gilt. a*d rl

Figure23. A Link-intensiveHyperfiction:CaitlinFisher'sWovesof Girls.In additionto the navigationlinks that appearat the left of

the screen,the main text alsocontainsfrequentopportunitiesto follow links.

at the end of a link to satisfy the intellectual and aesthetic needs of the
reader?sLet'stake as an examplewhat happenswhen one comesupon linked
text in the midst of the following sentencein a lexia from The Victoiqn Web
about the prose fantasiesof William Morris: "Like fohn Ruskin, Morris cre-
ates prose fantasies perneated by his beliefs about political economics."
What should one find atthe endof the link attachedto thenameJohn Ruskin?
For the reader of the present lexia, which discussesfantastic literature by
Morris, the most useful link would produce a discussionof fantastic fiction
by Ruskin, and in fact TheVictorianWebhassuch a relevantdocument, "fohn
Ruskin and the Literary Fairy Tale,"one section of which explains the rela-
tions of his early fantasyto his later political writings. One might evenhope
that such a link would bring one to a comparison of the distinctive qualities
of each author's writings in this mode, which this existing document does

RECONFICURING not. All these desired link-destinations, one notes, are implied by the word-
WRITING ing of the sentencein which the linked text appears.
What happens,however,when such discussionsare unavailable?What
usually happensboth in the websites I've examined and those I manage is
that the link of the compared author-here Ruskin-goes to very basic or
general information about that figure. Notice that such a link to general
information, which may provide a kind of basic identification of the figure
for neophytes in the field, is not necessarily a bad link. In fact, for cer-
tain users, particularly those new to a certain field or subject, such a link
destination might prove very useful. Still, most users of documents about
quite specific topics require information that directly illuminates the main
subjectat hand (in this case,Ruskin'sfairy tale).The fact is, though, that such
specificlink destinations are far more rare than the more general,glossary-
type ones.
Obviously,one would prefer to give readersa choice of information, in this
caseproviding both general and very specific information, in part because
such a choice offers a richer, more user-centeredembodiment of hyper-
textuality. Unfortunately, the World Wide Web, which at present allows only
links from a word or phraseto a single destination,doesnot offer one of the
most useful kinds of linking-the one-to-manyor branching link that offers
the reader a choiceof destinationsat the point of departure.One solution is
to link the anchor-here "lohn Ruskin"-to another document, which has
to be manually created,that offers multiple choices.Depending on the sub-
ject of the lexia in which this name appears,the link list or areasitemapat the
end of such a link can take the form of lists of links to biographical informa-
tion about "fohn Ruskin," those leading to his influence on various authors,
and so on. Another approachto handling links to severaldestinations, not
always possible to implement, requires adding phrases that might proride
multiple anchors in the deparfure sentence.Thus, one could link general
information to the figurek name (fohn Ruskin) and specificinformation only
to phrases,such as "permeatedby his beliefs,"that leadthe readerto expecta
very specificdiscussionat the destination lexia.

The Pleasuresof FollowingLinks in Hyperfictionand Poetry.Since much

hyperfiction and poetry aims to produce readerdisorientation, howevertran-
sient, the informational hypertextfeaturesof readerempowerment, multiple
approaches,and clarity might not appearparticularly important to it. Instead,
the qualities ofsurprise and delight characterizesuch success,for with hyper-
( (again (history ends)
) (something irr
)( the lines (could(not) be resl (erdless
really) lines) is needing(
(ofesurse)))(alwaysforward (
the alwaysmethod {building
bloekgbuilt in lines))nature{s}
{ )evohs (to better box)) (park
(city {headlorqg,
(one milliors (
)) sraight on {and on)) walls
around) is (not sur*)round (even
pavehent crscks afe {lat i
good foundation) and ansles(tall
x ( ) ( sophist{icated) (methodsfor i (
t t
(econocyelo sehemes)
) ( Oou)i to quickerbetter i) hard eyes))
))]) i with weedsfingering out) {so
box thm too (

hax( ) ( (polfernaturs
( ) { powered down
D))))) to powcrhuman (
{ber( ) ))) (persistent
power sprouts {powerlern})
) (
D )(tuck (powerfirck) porer (lurt live
forevar (the cancer(the infinite cell
) as the fountain (

Figure24. Two StagesofReading and lan M. Lyons's (box(ing)).At the left appearsthe screenone encounters early in one's read-
ing; at right, the screenafterthe readerhas broughtfo*h words and phrasesby clickinghis or her mouse.

fiction and poetry the question must be, not does following the link chiefly
satisfyan intellectual need but doesfollowing the link produce surprise and
delightl Instances of such pleasing results of following links appear in
StephanieStrickland'sVniverse(seeFigure 11)and Ian M. Lyons's(box(ing)),
both of which produce text ex nihilo. When one moves one's mouse over a
predetermined area (near a parenthesisin (box(ing)) and within the night
sl<yin Vniverse)and then clicks, text appears.6Thus, when the reader opens
(box(ing)),little appearson the screenother than multiple gray parentheses
scatteredacrossa white background (Figure 24). Lyons explains, "The plac-
ing of the parentheses"was intended to "conveynested levelsof associative
meaning . . . arrangedhierarchically;that is, if I opened one parenthetic set
and then opened a second,this secondset I alwaysmade to closebefore the
first. For example:(, ... (, ... )r...)rl'Lyons explainsthat "the piece'sparen-
thetically obsessivesyntax closely resemblesthat used in the entirely out-
moded programming language,LisP (more recently reincarnatedunder the
name Scheme)l'Clicking on the screenwithin someparenthesesand outside

RECONFIGURING others incrementally produces text. Lyons'spoem, which he implemented

WRITING in HTML, Storyspace,and Visual basic,was, he tells us, originally written to
be read on paper with the intention of questioning "hierarchical modes of
organization" found in post-Chomskianlinguistics and implicitly con-
founded by hypertext, since, as Nelson has pointed out, the shortcomingsof
classificationsystems,all of which require hierarchies, explain the need of
hypermediainthe firstplace.TThe pleasuresof reading(box(ing)),Ipropose,
come from the discoveriesof text the reader producesand of the meanings
of that quite difficult text.
Stephanie Strickland's Vniverse(see Figure 11), a much more complex
project than (box(ing)), representsa comparativelyrare example of literary
hypermedia that aims at producing both delighted surprise and the virtues
associatedwith information hypermedia-reader empowerment and multi-
vocality,or multiple approachesto a single general subject.8Upon opening
Vniverse,one encounters a night sky-a black screen speckledwith stars-
in which the central portion rotates. A small circle appearsat top right and
a slightly smaller one appears diagonally opposite at lower left. Moving
one's mouse acrossthe sky halts the rotation and revealsvarious constella-
tions. Meanwhile instructions scroll acrossthe bottom of the screen:"Scan
the stars . . . click once or click tvrice . . . click the darkness."Clicking on
darkness brings forth a constellation, a particular star with its assigned
number, and text that appearswhen one keepsonek mouse over the point at
which one clicked.Typing a number in the top right-hand circle producesthe
star with that number and its surrounding constellation. Like many hyper-
media projects that employ Flash and similar software, Vniverseboastsani-
mated text. Unlike many such projects, it also emphasizesa high degreeof

Coherence.Rich linking, plus a substantial degree of reader control, thus

appearto characterize successin both informational and literary hypermedia.
Another necessaryquality, I propose,is some sort of crucial coherence.
Since hypertext fiction and poetry often employ disorientation effects
for aestheticpu{poses, coherent and relevant linking might not seem to be
necessary,but I suspect it's simply that coherencenot take as obvious
forms as it does in information hypermedia. For example, our experience
of reading pioneering hyperfiction, such as Michael foyce's aftemoon,
proves definitively that much of what we have assumed about the relations
ofcoherence to textuality, fixed sequence,and the act ofreading as sense-
making is simply false. Reading afiemoonand other fictional narratives

3.0 shows, in other words, that we can make senseof-that is, perceiveas
coherent-a group of lexias even when we encounter them in varying
order. This inherent human ability to constructmeanings out of the kind
ofdiscrete blocks oftext found in an assemblageoflinked lexiasdoesnot
imply either that text can (or should) be entirely random, or that coherence,
relevance,and multiplicity do not contribute to the pleasuresof hypertext
reading. Movement rn af.emoonfrom a lexia containing, say,the conversa-
tion of two men to one containing that of one of their wives may at first
appearabrupt (and hence random or without any relevance),but continued
reading establishes the essential coherence of the link between the two
lexias:the movement betweenthe one containing the men speakingand the
secondcontaining the women can be repeated,thus establishing a pattern
like cinematic cross-cutting.Similarly,the next lexia one encounterscan
revealthat the words ofone pair ofspeakers serveas the context,the back-
story,for the others.

Coherenceas PerceivedAnalogy.In linking, this necessarycoherencecan

alsotake the form of perceivedanalogy-that is, the link, the;'ump acrossthe
textual gap, to some extent reifies the implied connection (implied link)
found in allusions,similes, and metaphors.For an example,let us look at
another early Storyspacenarrative, Joshua Rappaport'sHero'sFace,which
showshowlinking can serveas a new form of textual allusion. In Hero'sFace,
which relatesthe struggles for musical supremacyin a rock band, one par-
ticular link transports one from adolescentrock'n'roll to an entirely different,
and very unexpected,world of ancient epic. Most of the story consistsof
lexias about the people in the band and the relationships among them. In
one crucial lexia the narrator describesthe first time he "climbed serious
lead"-seized control of the music in midperformance-and realized that
the experienceresemblesthe feelings he has had while mountain climbing:
"Therecomesa momentwhen all of a suddenyoulookbehindyouandyou're
out eight or ten feet from your last piece,which addsup to a twenty-foot fall
onto the dubious support of some quickly-wedged chunk of metal in a
crack-you look behind you, and it's iust straight down, eighty or a hundred
feet,and your belayerbarelyvisible there at the bottom waiting for you to peel
off-every muscle pumped up to bursting, as you realizethat it is the mere
strength of fingers and arms and your innate senseof balancekeeping you
up in the air." After readersencounter this comparison of musical improvi-
sation to mountaineering, they come upon a link that functions as a second

RECONFIGURING analogy,for following this link brings one to the world of the Finnish epic,
WRITING the Kalelava:

T h eo l dV a i n a m o i nseann g :
thecliffsflewin two
therockscracked uoontheshores.
H es a n gy o u n gJ o u k a h a i n e n -
s a p l i n gosn h i sc o l l a r - b o w
a w i l l o ws h r u bo n h i sh a m e s
goatwillowson histrace-tip
sangit to treetrunks
in pools
sanghiswhipknottedwith beads
to reedson a shore

Following Rappaport'slink has severaleffects.First, readersfind themselves

in a different, more heroic ageof godsand myth, and then, asthey realizethat
the gods are engagedin a musical contestthat parallelsthe rock group's,they
also seethat the contemporaryaction resonateswith the ancient one,thereby
acquiring greatersignificancesince it now appearsepic and archetypal.This
single link in Hero'sFace,in other words, functions as a new form of both
allusion and recontextualization.
In hyperfiction, Michael foyce invented this form of reified comparison
or allusion when he had links transport readersfrom his story to passages
from Plato'sPhaed.o,Vico'sNaw Science,Basho'sTheNqrrou)Roadthrougltthe
Provinces,and poems by Robert Creeley and others. Perhapsthe ultimate
sourcehere is Julio Cort6zals Hopscotch(to which Joycealludes in the lexia
entitled "Hop Scotch").Frequently used, such juxtapositions-byJinking
produce the kind ofcollage writing that appearsto be very typical ofhyper-
fiction and poetry.
Suchcombinationsof literary homageto a predecessortext and claims to
rival it have been a part of literature in the West at least since the ancient
Greeks. But the physical separationbetween terts characteristicof earlier,
nonelectronic information technologies required that their forms of link-
ing-allusion and contextualization*employ indicatorswithin the text, such
as verbal echoing or the elaborateuse of parallel structural pattems (such as

3.0 invocationsor catalogues).Hypertext,which permits authorsto use traditional
methods,alsopermits them to createtheseeffectssimply by connectingtexts.
When successfirl,suchlinking-as-allusioncreatesa pleasurableshockof recog-
nition as the reader'sunderstanding of the fictional world suddenlyshifts.

Formof MetaphoricOrgani-
or Necessary
DoesHypertextHavea Characteristic
zationl The creationof coherencein linking via implied analogycan character-
ize not justthe relation betweentwo lexiasbut also an entire hypertext.The kind
of textuality createdby linking encouragescertain forms of metaphor and anal-
ogythat help organizethe reader'sexperiencein a pleasurableway.Someof the
most successfirlhyperfictions, such as Shelley)acksonsPatchworkGirl, employ
powerful organizing motifs, in this casescarsand stitching togetherthat func-
tion as commentaries on gender, identity, and hypertexfuality.Stitchesand
scars,which haveobvious relevancein a tale involving Dr. Frankensteinand
one of his monsters, become metaphorical and createunity and coherence
for the entire assemblageof lexias.At an early crux in the narrative ("Sight"),
facksoncreatesa branching point at which the readermust choosebetween
two lexias,both ofwhich emphasizethe analogousrelationshipsamong writ-
ing, reading a hypertext, and sewing up a monster ("written," "sewn"). Jack-
son'switfy plays on thesetopics all have a role in a hyperfiction that exposes
the way we createand experiencetexts,hlpertexts, gender,and identity.
One can also createunifying metaphors or analogiesthat do not refer to
hypertext,the medium itself. David Yunls SubwayStoryis a work of hyperfic-
tion that employs metaphors that inform the narrative in "nonreflexive" modes.
SubwaySlory employs the organizational metaphor of the map for the New
York subwaysystem:it includes both a map of that systemand a lexia for each
ofits stations.Yun has createda lexia for everystop on the subway,and he has
used the paths ofthe individual trains as link paths that createnarrativearcs.
As StefaniePankepointed out to me when I askedher why she thought it an
example of good hlpertext fi ctton, " SubwayStoryis an extraordinary hypertext
becauseof the application of a spatial metaphor that allows a navigation that
is somehow'linked' to the story itself. It is a beautifi.rlexamplefor a metaphor
that works becauseit is a part of (and not apart from) the storytelling."

Caps. As should be obvious by now, good hypertext-quality in hypertext-

dependsnot only on appropriateand effectivelinks but also on appropriate
and effectivebreaksor gapsbetweenand among lexias.TerenceHarpold long
ago pointed out that Derridean gaps,the presenceof which requires linking
in the first place,havejust as much importance in hypertextas do links them-

RECONFIGURING selves.N. Katherine Hayles has more recently explained that "analogy as a
WRITING figure draws its force from the boundaries it leapfrogs across. Without
boundaries,the links createdby analogywould ceaseto have revolutionary
impact" (93),and the same is true of the hypertext link. Without good-by
which I mean effectiveand appropriate-separations one cannot have good
links. Like the epic hero who requires an adequateantagonistto demonstrate
his superiority,linking requires a suitable gapthat must be bridged. We have
all readhypertextsin which foilowing a link producesa text that seemsto fol-
lowwhat camebefore in such obvious sequence,the readerwonderswhythe
author simply didnt join the two. We've all encountered relatively poor or
ineffectualgapsby which I mean thosebreaksin an apparentlylinear text that
appeararbitrary:the gap,the division betweentwo texts,appearsunnecessary
when the link does nothing more than put back together two passagesthat
belong together when no other paths are possible.
Hyperfiction and poetry can have two very different kinds of gaps,the
first being thosebridged or surmounted by links, the secondbeing thosethat
remain, well, gaps becausenothing in the software environment joins the
two texts or lexias. Whereasthe first kind of gap,that joined by links, seems
obvious becausewe encounter it everytime we follow a link, the other is not.
As an exampleof the secondI am thinking of entire sectionsor narrativearcs
in works like PatchworkGirl that remain separateand separatedin the reader's
experienceand yet may be joined by allusion or thematic parallels.Thus, in
PqtchworkGirl gatheringsof lexias about the stitched-togethernature of the
female Frankenstein monster reside in a different folder or directory than
those comprising Shelley Jackson'scollage of lexias composed of various
textsfrom facquesDerrida, L. Frank Baum, and Mary Shelley.Thesediscrete
sectionsjoin in variations on the themes of text, stitched-together-ness,
coherence,ofigins, and identity. As this example of gaps unjoined by links
makes clear,not all connectionsin effectivehypertextrequire electroniccon-
nections-like nonhypertextualprose and poetry,hypertext also makes use
of allusions, metaphors,and implicit parallels.The real question turns out to
be, then, How doesone decidewhen to make the potential connection,rela-
tion, or parallel explicit by means of an electroniclink and when to leavecon-
nections,relations, or parallels implicitl

IndividualLexiasShouldSatisfyReaders andYetPromptThemto Wantto Fol-

low AdditionalLinks.Hypertextis, afterall, still text, still writing, and we find
it difficult to distinguish many of the qualities of other good writing from
writing with links. In other words, excellencein hypertext does not depend

3.0 solely on the link. To an important extent, the text that surrounds the link
matters, too, becausethe quality of writing and images within an individual
lexia relate to one key hypertextual quality-its ability to make the reader
simultaneously satisfied enough with the contents of a particular lexia to
want to follow a link from that lexia to another.The problem that any writer
faces-whether the writer of hyperfiction or of stories intended for print-
can be defined simply as how to keep the reader reading. Making readers
want to continue reading seemsmuch easierin print text for a variety of rea-
sons: knowing the genre signals, readers know what to expec| looking at
their place in a physical text, they know how much more they have to read;
without choicesdemandedby linking, readershave essentiallyone choice-
to continue reading or to put down the story novel, or poem.
Particularly in these early days of the history of these new technologies
and associatedmedia, readershavea more difficult time deciding whether to
keep reading. The text they read must persuadethem to go on by the essen-
tial, traditional, conventionalmeans-that is, by intriguing, tantalizing, sat-
isfying, and aboveall entertaining them. In a hypertextlexia the readermust
encounter tert that is simultaneously,perhapsparadoxically,both satisfying
and just unsatisfying enough: in other words, the current lexia readersen-
counter has to have enough interest, like any text, to convincethem to keep
reading, and yet at the sametime it must also leaveenough questionsunan-
sweredthat the reader feels driven to follow links in order to continue read-
ing. In the terms of Roland Barthes,the lexia must include sufficient plot
enigmas or hermeneutic codesto drive the reader forward. This demand
explains why the opening lexia of Michael foyce'sclassicafi.emoon,perhaps
the first and still one of the most interesting hyperfictions, takes the form of
such ornate metaphoricalprose. Here, for example,is the secondparagraph
in aftemoonisopening lexia ("begin"):

octopiandpalmsof ice-riversandcontinents
besetbyfear,andwewalkoutto the
in a series
on thehorizon,
theshrapnel likerelics,
settling theechoing
is air.By
meltfreezes intocrystal
againacrossthe blacktop

The rich, sensualmetaphoric style of this lexia promises readersa lush read-
ing experienceand therefore makes them want to keep reading, but this sec-
tion is also self-containedenough to cohere as a separatelexia. As anyone
who has read afiemoor,r
knows, not all its lexias havethis richness-some
r fsb* on a face ts n tek

{}dt r$r rixr*risay* "r*fid* Same*rcn rhouldbe ttthere

lltnvn an soe
+ q#

ffi"-'d " T

1j i ?lmi'f;ri it*"i! .r4i.jiJ}"Jf;f:, l*rrvhilitr lkrs lt$ rilrr* Sntlrffy fs lsxxs i;s grtp*

,:1i:;l;i;ag .!ir$*ii;r:.{ *.q{' Jl*{ #jr}*i$,- ff**rt onyo*c,s*rssrytrdf * &ctwrto frna eowr

Figure 25. A Hyperfiction Sitemap: JackieCraven'sln the Changing Room.

are quite bare and brief-but foyce does employ this style elsewhere, for
example,in "Staghorn and starthistle."

The ReaderCanEasilyLocateand Moveto a Sitemap,Introduction,

or Other
Starting Point. Can the readereasilyreturn to documents or images encoun-
tered in a previous sessionl Such a requirement obviously pertains more to
informational or discursivehypertext than to hyperfiction or poetry,though
some fictions, such as JackieCravenlsIn the ChangingRoozn(Figure 25), em-
ploy a sitemap, in this case,consisting of the names of eachof eight charac-
ters.WhereasCraven'ssitemaptakesthe form ofa typical HTML setoflabeled
links, Deena Larsen'sStainedWord Window (1999;Figure 26) uses an active
(or "hot") sitemapat screenleft (on a black background)to bring up text at the
right. Simply mousing over a word, such as "faces,""in," "understanding,"or
"windows," producesbrief patchesof free versethat contain links, and one
can alwaysreturn to the beginning or opening lexia becauseLarsenprovides
a linked footer icon that brings one back to it. Textsthat invite a more active,
Figure25. An ActiveSitemap.DeenaLarsen'sStainedWord Window usesan activeor "hot" sitemap:simply mousingover differ-

ent portions of it makes poems appear at screen right.

evenaggressivereaderneed,like informational hypertext,such devices,since

the reader'sorientation, rather than disorientation, plays a major role.

The DocumentShouldExempli! TrueHypertextuality

by ProvidingMultiple
Linesof Organization.In a hypertext,whether fiction, poetry,or informa-
tional, one generallydoes not expectindividual lexias to follow one another
in linear fashion. True, linear sequencesdo havetheir use: VannevarBush-
styletrails require linear sequences,and authors offiction use them to create
a main (or default) axis for a narrativefrom which one can easilydepart. Per-
haps surprisingly, much hypertext narrativethus far takes the form of narra-
tive loops or paths in which most of the lexiasfollow one another in a linear
fashion, thus creating a series of self-containedstories. Of course,an elec-
tronic document may work quite well and yet not work hlpertextually in any

RECONFIGURING complex or interesting way. One can, for example,havehypertextsin which

WRITING linking only servesto ioin an index to individual sections.To be clear, let's
remind ourselvesthat hypertextuality-or excellencein hypertext,whatever
we decide that might be-obviously is important in judging a hypertext as
hypertert, but it neednot necessarilyplay an important role in other forms of
digital arts and literature. Here I'm concernedonly with the problem of qual-
ity in hypertext.
SteveCook's"Inf(l)ections" and fef Pack's"Growing up Digerate"exem-
plify successful,richlylinked discursivehypertexts.Cook'sstandsas experi-
ment in new forms of academicwriting whereas Pack'sexperiment autobi-
ography provides three different kinds of organizalton that the reader can
follow: (1) a linear path arrangedchronologically,(2\ a topic-driven reading
facilitated by a sitemap in the form of an alphabeticallist, and (3) a multilin-
ear narrativeprovidedby links scatteredthroughout the text ofindividual iex-
ias. fackie Craven'sIn the ChangingRoomsimilarly allows both linear narra-
tive, permitting the reader to follow the story of a single character,or move
among the eight characters,eachone in effectbeing defined as a storyline, a
narrativearc.As the introduction says,"Click on an underlined word, and the
stories will merge and take new form. Yourpath will not bestraight.Here in
the Changing Room, all things arelinked-and everyoneis a reflection . . . of
a reflection . . . of a reflectioni'

The Hyperdocument
ShouldFullyEngagethe Hypertextual of the
ParticularSoftwareEnvironmentEmployed.In askingif an individualhyper-
media project pushesthe limits of the softwareit employs,one entersa mine-
field. In the first place,such a question implicitly assumesthat the new the
experimental,has major value in itself, and evenif one acceptsthis hypothe-
sis, it might havevalidity only in the early stagesof a genre or media form. Of
course,at the present moment, all writing in hypertextis experimental since
the medium is taking form as we read and write. Electronic linking, one of
the defining features of this technology,can reconfigure notions of author,
text, reader,writer, intellectual property,and other matters of immediate con-
cern to thosewho design hypertext systemsor author documents with them.
Becausehypertext fiction-writing at and over the edge-sets out to probe
the limits of the medium itself, it acts as a laboratory to test our paradigms
and our fundamental assumptions.A sample of hypertextsshows the ways
they illuminate issuesranging from readerdisorientationand authorial prop-
erty to the nature of hypertext genresand the rules of electronicwriting.
Within this project of writing-as-discovery,all elements in a hypertext

T. 0 systemthat can be manipulated can function as signifying elements.To pro-
vide an exampleof the creativeuse of systemfeatures,letus turn to a fewvery
early examplesfrom Writing at the Edge(199a),all of which were createdin
EastgateSystems'Storyspace,a stand-alonehypertextenvironment available
for both Windows and Macintosh platforms.
In addition to containing traditional elements such as fonts, graphics,
sound, and color, Storyspacealso supportsthe creativeutilization of "screen
real estate"-the tiling of windows and the order in which they appearand
arrangethemselves.Nathan Marsh'slexiasin Breathof Sighsplacethemseives
around the screen, making the screen layout support the narrative as one
crossesand recrossesthe tale at severalpoints (seeFigure 22). Marsh'swork,
which datesfrom 1993,provided an early demonstration that writing had be-
come visual aswell as alphanumeric. It also revealsthat a single softwarefea-
ture, such as the ability to control window size and location, leadsdirectly to
a particular mode of writing-here writing as collageand montage in which
the multiple-window format permits readersto move back and forth among
overlappinglexias.This feature alsoencouragesactivereaders,sincethey can
easilymove about among lexias,thus creating a kind of spatial hypertext.
Severalother hypertexts from Witing at the Edgeshow the imaginative
deployment of another systemfeature of the software-the Storyspaceview,
a dynamic graphic presentation of the arrangement of document organiza-
tion. Storyspace,a hypertext environment that also functions as a conceptual
organizer,permits authors to nest individual spaces(lexias)inside others, or
to rearrangethe hypertext'sorganization by moving lexiaswithout breaking
links. Someworks, like ShelleyJackson'sPatchworkGirl (seeFigure 28),take
advantageofthis graphic organizational feature to structure hyperfiction by
means of separatefolders or directories. Others, like Ho Lins NicelyDone,
arrange all lexias on a single level and indicate discretenarrative lines. This
hypertext novel, which links a murder story and the eventsof a professional
football championship game,suggestsits organizationby arranging its lexias,
all of which appearon the top level, in four parallel lines. Timothy Taylor's
LBJ-Lazarus + Barabbas+ fudas-takes graphicindicationsof narrativeand
conceptualorganizationfarther than Ho Lin's NicelyDone,arranging its lexias
in the form of three crosses,the central one of which has a circle (halo)) over
it. Here, rather than indicating the narrative structure, Taylor implies graph-
ically something about the subjectand theme of his fiction. Like similar proj-
ects that Michael foyce reproduces in O/Two Minds (38), Adam Wenger's
AdamisBookstore, which I discussin the following chapter,usesa circular de-
ployment of the graphic elements representing lexias in Storyspaceview to

RECONFIGURING indicate that his document can be entered-and left-at any point (seeFig-
WRITING ure 32). One of the most bravura examples of arranging lexia-iconsin the
Storyspaceview appearsin Marc A. Zbysznskit playfi.rluse of hundreds of
them to createan image of a human face beneath a recyding symbol. Even
the naming of lexias can provide opporrunities for unexpectedsignification.
Andrew Durden'splayful arrangement of lexias in SatynconRandomlyGen-
eratedforms a grammatical sentence.Reading the titles of the upper-level
folders reveals the following playful comment: "l / think this / lexia / is a
good / start place."Stuart Moulthrop famously carried this playful use of sys-
tem featuresmuch farther, creating sonnetswithin a menu of links!
As the previous examples suggest,hypertext environments have, if not
preciselyMcluhan's messagein the medium, at leastcertain tendenciesthat
derive from specificfeaturesofthe software.The capacityto conffol size and
location of multiple windows encouragescollage-likewriting that employs
these features, just as the presenceof one-to-manylinking and menus of
links that have a preview function encouragecertain forms of branching.
Both features and the limitations or constraints of these featuresencourage
certain ways of writing, just as the fourteen-line sonnet encouragescertain
kinds ofpoetry
Turning from Storyspaceto HTML and the World Wide Web, by far the
most widespreadform of hypermedia today,one wonders if it, like other hyper-
media environments, encouragescertain modes of writing. HTML, which is
basically an extremely simple text-formatting language that works on the
Internet, has two defining features-first, the ability to insert links between
lexiasand, second,the ability to insefi other media into individual lexias,orig-
inally just imagesbut soon after sound, video, and animation createdby fava
scripts or Flash.The rapid spreadof accessto broadbandconnectionsto the
Intemet has transformed the World Wide Webfrom a simple systemfor link-
ing text-representationsinto a multimedia platform. The implications of this
changefor anyonetrying to determine the messagein the medium are obvi-
ous: whereas earlier proprietary systems,such as Intermedia, Microcosm,
HyperCard, Storyspace,Guide,and so on, had built-in, clearlydefined charac-
teristics, some of which provided clearlimitations, the World Wide Web does
not. Anyone working with basic HTML encounterscertain obviousfeatures,
which may act as imitations. Theseinclude the absenceof one-to-manylink-
ing, preview features,and preview functions, as well as the inability to place
and control the size of windows. Anyone using Flash or favain HTML docu-
ments,however,doesnot necessarilyconfront any ofthese limitations, though
they may confront others, such as incompatibility with particular versionsof

3.0 browsers.Suchfreedom,suchabsenceof limitations,brings with it the rela-
tive absenceof those restraints that often both limit and inspire creativiW.

Conclusion. A11forms of writing at their best can boast clarity, energy,

rhythm, force, complexity, and nuance. Hypertext and hypermedia, forms
of writing largely definedby electroniclinking, are media that possessthe
potential qualities of multilinearity, consequentpotential multivocality, con-
ceptual richness, and-especially where informational hypertext is con-
cerned-some degree of reader-centerednessor control. Obviously,hyper-
texts that build on the chief characteristics of the medium succeed. In
addition, as we have seen, examplesof hyperfiction and hyperpoetry reveal
other sourcesof quality: individual links and entire webs that appearcoher-
ent, appropriate gaps among lexia, effective navigation and reader orienta-
tion, pervasivemetaphoricity,and the exploration-and testing-of the lim-
its of the medium.
ring Narrative

Every digital narralive, we must remind ourselves,does not

Approachesto Hypertext necessarilytakethe form ofhypertext. A casein point appears
in Christy Sheffield Sanford'svisually elegant World Wide
Web fiction, Saforain the Beginning(1996),which the author
OpeningRemarks describesas "a web-novel" written in the spirit of classical
tragedyabout "a young African princesstaken as a slavefrom
Senegalto Martinique." The opening screen,the first of twenty-one succes-
sivelexias,explainsthat to the left of the "main texh:al body,Bible quotations
and natural history descriptions echo Old Testament Christianity and Ani-
mistic traditions at their point of contact: mythopoetization of naturel' Es-
sentially,Sanfordworks with the powersof digital information technologyto
add colors,images,and some motion to narrative,but the HTML links func-
tion solely to provide sequence.Among her twenty-onelexias, she includes
what she describesas "five filmic scenesusing the close-up,time-lapseand
other cinematic effects.Thesetechniquesenablethe conflict-crisis-resolution
model to haveconcisenessand scope."Thesecinematic effectsappear,not as
firll-motion video but as a kind of a film script, though Sanford'sromantic
tragedydoesuse occasionalanimations aboveand to the left of the main text
spaces.I cite this elegantproject not to criticize its lack ofhypertextuality but
to remind us that the digital word and image, even on the World Wide Web,
doesnot inevitably produce hypertextualnarrative.
The examplesof hlpertext fiction at which we shall look in the following
pagesand have alreadyexamined in the earlier discussion of writing hyper-
media suggestthat even in this early stagehypertexthas taken many forms,
few of which grant readersthe kind of power one expectsin informational

3.0 hypertext. As Michael foyce, our first major author of hypertext fiction, has ex-
plained, the desire to create multiple stories out of a relatively small amount
of alphanumeric text provided a major force driving in writing afiemoon:

I wanted,quitesimply,to writea novelthat wouldchangein successive

to makethosechangingversionsaccordingto the connectionsthat I had for some
time naturallydiscoveredin the processof writing and that I wantedmy readersto
share.In my eyes,paragraphs
on manydifferentpagescouldiust aswellgo with para-
graphs on many other pages,although with different effectsand for different pur-
poses.All that keptme from doingso wasthe factthat,in printat least,one paragraph
inevitablyfollowsanother.lt seemedto me that if l, as author,couldusea computer
to move paragraphsabout,it wouldn'ttake much to let readersdo so accordingto
some schemeI had predetermined.(Of TwoMinds,31)

From one point of view then, such an approach merely intensifies the agenda
of high modernism, using linking to grant the author even more power.
Other authors take a self-consciously postmodern approach, using the
multiplicity offered by branching links to create a combinatorial fiction that in
some ways seems the electronic fulfillment of the French group, Oulipo. For
example, in its forty-nine fictional lexias Tom McHarg's The Late-Nite Maneu-
vers ofthe Utramundane, one of the hypertexts included in Witingatthe Edge,
attempt to "veer toward a narrative . . . not entirely dependent upon linearity,
causality, and probable characterization" ("On Ultramundane"). McHarg cre-
ates seven lexias for each day of the week, each a variation or tansformation
of the other. Choosing the first Monday, for example, the reader encounters
the following narrative (which represents the first half of the lexia):

Dwightawokeat 3:l5 a.m. to find his girlfriend,Johnette,attemptingto conceala

bomb under his pillow."You'vewakenme up," he said. 'And l've discoveredyour
"Theonlytreacheryis yours,"saidJohnette.
"l'm onlysleeping,"said Dwight."You'rethe one plantingbombs."
"Perhapsyou deserveit," saidJohnette.
"But I loveyou,"said Dwight.
"Thenwhy do you accuseme oftreacheryl"saidJohnette.
"ltl obvious,"said Dwight."Youplannedto murderme as I lie heredreamingof
our sex."
"Youweren'tdreaming,"saidJohnette."l waswatchingyour eyes."
"Perhapsnot, but at leastI wasn'ttryingto murderyou,"said Dwight.
"l only meantto scareyou into lovingme more,"saidJohnette.

RECONFIGURING "Witha bombl" saidDwight.

NARRATIVE "Youneedto lovemea lot more,"saidJohnette.

And so it goes.Eachvariation introduces a different weapon,a different be-

trayal, as Utrarnundane explores how "a fictional text must be stretched,
skewered,and sliced if it is to exploit the freedoms and acceptthe responsi-
bilities offered by hypertext technology and its new writing spaces."Thus, on
the first Tuesdaya friend standsat the foot of their bed with a gun, on the first
Wednesdaythe hero'shair has all fallen out, and on Sunday Dwight returns
to find their home on fire. Like afi.emoon,TheLate-NiteManeuversofthe Utra-
mundanecombines into different naratives, many about sex and violence,
producing effects according to the route one follows through it. Otherwise,
this web, whosetone and content suggestthe influence of Coover'sprint fic-
tion, contrastsentirely with ]oycekwork. My point here is not that one should
prefer either the crystalline richness and emotional intensity of aftemoon
to McHarg's PoMo playful, removed senseof literature as its own laboratory
or that Ultramundaneis in some way more hypertextual. Rather, I wish to
emphasizethat, like print fiction, that produced in the form of linked lexias
can take many forms.
In some the author compounds her or his power; in others, such as that
exemplified by Carolyn Guyerk Quibbling,the author willingly sharesit with
readers. Similarly, in some fictional webs, such as ofiemoon,readers con-
struct or discoveressentiallyone main narrative; in others, like Semio-Surf,
FreakShow,or Ultramundqne,one comesupon either a cluster of entirely sep-
arate stories, or else one finds narrative segmentsout of which one weaves
one or more of them. A third opposition appearsbetween those storiesthat,
however allusive, consist largely of fictional lexias,and those like Potchwork
Girl and Semio-Surtthat weavetogethertheory and nonfiction materialswith
the narrative.A fourth such opposition separatesfictional hypertext entirely
derived from the author's "own" writing and those, like PatchworkGirl and
Stuart Moulthrop's ForkingPaths,thattheir authors wrote to varying degrees
in the intersticesof other works.
Hypertext narrative clearly takes a wide range of forms best understood
in terms of a number of axes,including those formed by degreesor ratios of
(1) reader choice,intervention, and empowerment, (2) inclusion of extralin-
guistic texts (images, motion, sound), (3) complexity of network stmcture,
and (4) degreesof multiplicity and variation in literary elements,such asplot,
characterization,setfing, and so forth. Following the lead of Deleuze and
Guattari, I prefer to think of the organizing strucLuresin terms of ranges,

3.0 spectra,or axesalong which one can array the phenomena I discuss rather
than in terms of diametric oppositions,such asmale/female or alphanumeric
versusmultimedia text. I avoidsuchpolaritiesbecause,particularlyin the case
of hypertext fiction and poetry',they hinder analysisby exaggeratingdiffer-
ence,overratinguniformity, and suppressingour abilities to perceivecomplex
mixtures of qualities or tendencies.Another reasonfor emphasizing a spec-
trum of possibilitieswhen discussinghypertextliesin the factthat neither end
of any particular spectrum is necessarilysuperior to the other. For example,
hyperfiction that demands intewention by readers,or otherwise empowers
them, will not on those grounds alone turn out to surpasshyperfiction that
employs links to solidify-indeed amplify-the power of the author.

Hypertext, which challengesnarrative and all literary form

Hypertextand the Aristotelian basedon linearity, calls into question ideas ofplot and story
cunent sinceAristotle. Looking atlhe Poeticsinthe contextof
Conceptionof Plot
a discussion of hypertext suggestsone of two things: either
one simply cannot write hypertext fiction (and the Poeticsshows why that
could be the case)or elseAristotelian definitions and descriptionsof plot do
not apply to stories read and written within a hypertext environment. At the
beginning of this study,I proposedthat hypertextpermits a particularly effec-
tive means of testing literary and cultural theory. Here is a casein point.
Although hypertext fiction is quite new the examplesof it that I have seen
alreadycall into question some of Aristotle'smost basicpoints aboutplot and
story.In the seventhchapterof the Poetics,Aristotleoffers a definition of plot
in which fixed sequenceplays a central role:

Nowa wholeisthatwhichhasbeginning, middle,

andend.A beginning
is notitselfnecessarily
something else
afterit;an endis thatwhichis naturally aftersomething itselfeitherasits necessary
o r u s u acl o n s e q u e natn, dw i t hn o t h i n g
e l s ea f t e ri t ;a n da m i d d l et,h a tw h i c hi s b y
natureafteronethingandalsohasanotherafterit. (1462)

Furthermore, Aristotle concludes,"a well-constructedPlot,therefore,cannot

either begin or end at any point one likes; beginning and end in it must be of
the forms just described.Again: to be beautiful, a living creature,and every
whole made up of parts, must not only present a certain order in its arrange-
ment of parts, but also be of a certain definite magnitude" (1462).Hypertext
therefore calls into question (1) fixed sequence,(2) definite beginning and
ending, (3) a story's "certain definite magnitude," and (4) the conception of
unity or wholenessassociatedwith all these other concepts.In hypertext

RECONFIGURING fiction, therefore, one can expectindividual forms, such as plot, characteri-
NARRATIVE zation, and setting, to change, as will genres or literary kinds produced by
congeriesof thesetechniques.
When I first discussedhyperfiction in the earlier versions of this book,
the novelty,the radicalnewness,of the subjectappearedin the fact that almost
all the sourcescited were unpublished, in the processofbeing published, or
published in nontraditional electronic forms: these sourcesinclude unpub-