Вы находитесь на странице: 1из 23

Unfolding The Future: Bifurcation In Organizing Form And Emergence In Social Systems

E:CO Issue Vol. 13 No. 3 2011 pp. 57-79


Theoretical

Unfolding The Future: Bifurcation In


Organizing Form And Emergence In Social
Systems
James K. Hazy & Allan S. Ashley
School of Business, Adelphi University, USA

This paper presents a complexity science informed theory to describe how


organizing forms emerge and foster innovation. The theory explores the
bidirectional linkages between fine-grained interactions among human
beings in a complex adaptive system and the emergent coarse-grained
properties that characterize qualitatively distinct and yet stable organiz-
ing forms in social systems. By exploiting a mathematical foundation that
has been successfully employed in analogous cases in the natural sciences,
it opens the door to a rigorous theory of performance and adaptation in
human systems by relating changing local rules of interaction to qualita-
tive changes in emerging organizing forms. This process is mediated by
evolving models of the system in the environment developed and shared
among individuals. Finally, the paper explores whether this model can be
used canonically and does so in the context of axiomatic hurdles that must
be overcome if a practical mathematical theory of human organizing is to
be realized.

Introduction

T
his paper begins the development of a practical mathematical theory of
human organizing forms that can be used to inform social innovation. As
a foundation for this undertaking, we further develop a model of social
entrepreneurship recently proposed by Goldstein et al. (2010). The GHS model,
as it will be called throughout the remainder of this paper, purports to describe
how social innovations can emerge as a means to qualitatively change social
processes. Presumably, the unfolding of a game-changing social program such
as successful public health initiatives to combat the spread of HIV/AIDS would
be an example of such a social innovation. The hope is that eventually positive
social change processes like these, and the particulars of their success, can be
modeled rigorously to inform future interventions.
The GHS systems model (based on an entrepreneurship model originally de-
scribed by Goldstein et al., 2010) uses mathematical results from differential to-
pology, and in particular, singularity or bifurcation theory (Arnold, 1989; Thom,
1989; Zeeman, 1977). The model is general in its treatment of topologically dis-

Hazy & Ashley 57


tinct organizing forms (and also presumably those at work within social systems
under appropriate assumptions), how these might come about in a dynamical
system, and under what conditions a change from one stable form to another
might occur.
To advance our argument the paper first provides a discussion of the founda-
tions upon which the theory rests. After that, it describes key modeling con-
structs that define systems’ properties and how these change at critical points,
the singularities. The parameters and variables that determine when stability
arises are discussed and framed in a mathematical context. Next the paper iden-
tifies attributes of human systems that make them much more challenging to
model than comparable physical and biological systems. We will close by look-
ing ahead.

Foundations For A Theory Of Organizing Forms

T
he GHS model describes how changes in the environment combine with
the rules of interaction in the system to change the patterns or forms
wherein organizing unfolds. Swanson & Zhang (this issue) provide exam-
ples of this phenomenon. They describe how various forms of social enterprise
organizing might occur differently in the face of changing legal forms. Likewise,
Shepherd & Woods (this issue) describe the emergence of the ideal of learned
“digital citizenship” as an organizing principle for a childhood education social
enterprise in the face of a changing technology landscape. Baker et al. (this is-
sue) describes the formation of social network hubs or “spaces” as enablers of or-
ganized social action in Australian communities. These examples demonstrate
the salience of the coarse-grained effects that characterize social innovation.
What is missing is a theory to describe the mechanisms that enable organizing
forms to emerge and be selected. This paper addresses this gap.1
The GHS model avers that profound change occurs when certain changes in the
environment or ecology—the “shape” of the container constraining the system
within an ecological niche—apply or relieve pressure on the choices and actions
of individuals that are part of the system, thus changing the rules of interaction
in a particular way. Changes to the rules of interaction are caused by tensions
inside the system, including the implicit impact of each individual’s memories,
preferences, and networks of relationships (Hazy et al., 2007). Other “ingredients”
in the container include available resources and knowledge. This amalgam is

1. To begin to describe how organizing unfolds, the argument first takes as a given, that
human organizations can be represented a complex adaptive systems (CAS) and that the
interactions that occur within these systems can result in emergent properties at the aggre-
gate level (Holland, 1975, 2001). The fine-grained process whereby interactions among indi-
viduals lead to emergence of cooperative action has been discussed elsewhere (Hazy, 2006,
2007, 2008a,b). The fine-grained process is clearly a subject that needs additional study, but
except as reverenced, is beyond the scope of this paper.

58 E:CO Vol. 13 No. 3 2011 pp. 57-79


“mixed” under the container’s ecological pressure until stable properties (as de-
fined below) emerge—or more precisely are intentionally constructed through
experimentation (Goldstein et al., 2010)—within the context of human interac-
tion dynamics (HID).

Topology Of Organizing Forms Defined


A system property is a pattern of choices and actions that can be recognized,
modeled, and used as the basis for prediction. For example, a business plan for
a social enterprise or a quality control chart for a business enterprise could be
such a model that hopefully provides those participating with a reasonably ac-
curate forecast of what is likely to unfold.
Because these properties have structure, and because this structure has
emerged from out of individual interactions that are being guided by rules en-
acted locally, a set of properties is called an organizing form2. Note that a set or
even a subset of a set with a single property can be an organizing form; a weekly
lunch at a local restaurant can be an organizing form with the property of being
predictable as a weekly event. Since government agencies such as the Centers
for Disease Control have many properties, taken as a whole, they are likewise
an organizing form. This larger organizing form may contain within it subsets of
properties which are also organizing forms. These might include various weekly
or monthly meetings at different times within various policy offices. Thus, the
notion of an organizing form is recursive. In fact, sets of sets and subsets of sets
(including the whole set and the null or empty set) can be assumed to define a
topology3 of organizing forms across levels of scale.

Organizing Form Stability And Linear Approximation


The behavior of stable dynamical systems models can often be approximated
by assuming linear relationships (Epstein, 1997; Hirsch et al., 2004)4. Thus sta-
ble properties operating within organizing forms can often be approximated
using linear models because they can also be modeled as dynamical systems

2. The choice of terms is to emphasize the dynamical nature of the form as well as distinguish
this idea for the more traditional concept of an organizational form, which tends to be more
legalistic and institutional in its usage (Hannah & Freeman, 1989).
3. Topology is the field of mathematics that studies properties of geometric forms that re-
main invariant under continuous deformations such as bending and stretching. Differential
topology in particular is concerned with shapes that are differentiable. A topological space
is a set, X, for example the set of all organizing forms, together with what is also called “a
topology” which is the set of subsets, r, of X, that are being considered such that the empty
set and whole set are both included in r, any union of elements of r is also in r, and any finite
intersection of sets in r is also in r. A topological space can have multiple topologies.
4. Interested readers should explore the implications and limitations of the Linearization
Theorem in a dynamical systems text such as Hirsch et al. (2004)

Hazy & Ashley 59


(Hazy, 2011). In the business plan example, adding incremental staff to push
the selling process can be expected to have proportional revenue benefits. Thus,
linear approximation can often work in a business plan because such a plan is
actually a simplified dynamical systems model of what is observed to be taking
place (Hazy et al., 2010). An organizing form is thus more precisely defined as a
collection of individual interactions (what we call fine-grained HID) exhibiting
emergent regularities or patterns at a higher level of scale (what we call coarse-
grained properties) such that these regularities can be modeled and their be-
havior can be predicted with greater accuracy than a random distribution would
imply. When the predictions of the patterns are approximated with linear mod-
els, the organizing form is called linearly scalable since outcomes are approxi-
mately proportional to inputs and thus the organizing form can be scaled up or
down because adjusting inputs proportionally impacts outputs. Similarly, when
the regularities relate to power law relationships, the organizing forms are said
to be logarithmically scalable because in this case the approximate proportion-
ality is observed on the log-log plot.
Although the above definition is simple and general, it is not trivial. Even though
these stable properties and thus organizing forms can frequently be approxi-
mated by linear relationships, the GHS model also demonstrates that profound
change or environmental conditions can engender criticalization in HID, a pe-
riod of instability and uncertainty about the direction of change. These condi-
tions often imply situations where linear approximation is not appropriate. A
recent example of this was the initial inaction to the nuclear disaster after the
2010 Japanese tsunami. Those responsible for the nuclear power plants and the
welfare of the population around those plants were disoriented, an indication
that criticalization was happening within the models being used by these indi-
viduals to organize their action. Their linear approximation models did not ef-
fectively inform action for them.
Important contributions in the area of computational complexity made over the
last few decades can ground these ideas in complexity science. This work ex-
plores the nature and limits inherent in understanding human activity. The next
section describes some important aspects of the field and defines Gell-Mann’s
(2002) notion of effective complexity that was also supported by the work of
Crutchfield and others (Crutchfield & Feldman, 1997, 2003; Feldman & Crutch-
field, 1998; Feldman et al., 2008; Prokopenko et al., 2009).

Modeling The Effective Complexity Of Emergent Properties

T
he statistical complexity (Crutchfield & Feldman, 1997) of an observed phe-
nomenon—or following Gell-Mann (2002) what we will call “effective com-
plexity”—means that a model can predict the statistical behavior of un-
folding events at the coarse-grained or aggregate level. Modeling at this higher
level of analysis avoids the shortcomings of other computational complexity
procedures which would attempt to reproduce the details of fine-grained re-

60 E:CO Vol. 13 No. 3 2011 pp. 57-79


sults as well as coarse grained outcomes5. In contrast, effective complexity pre-
dicts coarse-grained results independent of the fine-grained details, and thus
takes advantage of the aggregated statistical characteristics of events. This is
done by defining random variables, using statistics such as mean, median, vari-
ance, and so forth to define them as distributions. One then uses correlation
measures to identify relationships. The caveat described in the last section aside,
this approach avoids the problem of predicting the highly contingent and inter-
dependent fine-grained events that underlie these coarse-grained regularities.
When properties are recognized as relevant aggregates at the system level, and
when they have effective complexity that is less than would be the case if events
were random, these regularities can be measured and modeled as coarse-
grained phenomena in their own right (McKelvey, 2002). Modeling proceeds
when random variables are assigned to quantities, and relationships among
the random variables are defined and specified. In a business context, examples
of properties of interest that might be modeled in this way might be: quality,
efficiency, employee turnover, or organizational sustainability. These can be
defined as random variables that describe aggregate attributes of the system
rather than attempting to reflect any particular fine-grained event or individual
decision.
Although these properties may have limited practical relevance at the individual
level per se, the outcomes may have feedback effects that operate at the fine-
grained individual level through the models that individuals use to understand
and predict events. If individual rewards or sanctions are made contingent on
performance according to coarse-grained outcomes as reflected in these mod-
els, feedback effects are likely to reinforce the model-in-use, or to distort it ac-
cording to individual interests. Informational feedback loops like these enable
stable configurations of an organizing form to settle into an attractor that is
actualized and reinforced by the model being used to define the underlying
dynamical system (Hazy, 2011). This caveat aside, system properties are aggre-
gate quantities and the relationships among them are also aggregations. Three
elements that make up an organizing form—HID, models used by individuals
to predict what will happen, and emergent properties—are shown in Figure 1,
Panel A.
System properties are distinct from simple aggregations (for example the defini-
tion and measurement of a random variable, as for example, the “mean” of many
individual measurements) because they describe and measure not only what
individuals do or the resources that they use, but also how they work together

5. In this fine-grained approach to computational modeling, systems are more and more
complex as they get more and more difficult to predict, reaching their maximum measured
complexity when one attempts to model purely random systems (Crutchfield & Feldman,
1997; Prokopenko et al., 2009) in which case, the most economical model of the system is
the system itself.

Hazy & Ashley 61


Patterns &
Properties

Model of a Property

Models Variable 1 Variable 2


Relationship

Human Human
Interaction Interaction
Dynamics Dynamics

A B
Figure 1 The emergence of system properties exerts downward influence on
human interaction dynamics, a process that is synchronized through models
built by individuals to inform their choices and actions. Panel A shows the
interacting elements of an organizing form that characterizes an HID system’s
functioning. Patterns emerge from human interaction dynamics; these
are observed, modeled by, and are shared among, individual participants;
as events are predicted by the models, the modeling influences the HID in
anticipation of events and thus impacts the emergent properties in a complex
feedback process. Panel B shows how a property is modeled by defining
random variables and predicting the relationships between their values. These
models inform individual choice and action which in turn impacts HID and the
resulting emergence of system level properties, which are then modeled, and
on and on, recursively. A stable set of properties together with the models that
enable them to emerge constitutes an organizing form.

and exert mutual influence. When the properties of a system are predictable—
implying the variables and relationships among them as indicated in Figure 1
Panel B are predictable—and when a small perturbation or push on the system
is such that the system maintains the same organizing form and therefore re-
mains predictable under the same model, then the system’s properties and its
organizing form also exhibits a degree of stability6.

6. Stability is defined differently in dynamical systems theory (Hirsch et al., 2004): a system is
said to be stable if upon a small perturbation, the system’s new trajectory remains arbitrarily
close to it’s previous one. This definition is narrower than that in the main text because it re-

62 E:CO Vol. 13 No. 3 2011 pp. 57-79


The Uses And Limits Of Linear Approximation Techniques
Because stable dynamical systems models can often be approximated using lin-
ear techniques, one can assume that the behavior of stable organizing forms
can also be approximated using linear methods that rely on the stability char-
acteristics of the underlying dynamical system models. For example, downward
pressure on sales and thus revenue results would require a proportional reduc-
tion in expenses in order to maintain profitability (Hazy et al., 2010). Assuming
feedback about this trend is present, this “model” of the world would be used by
individual actors to trigger real-world actions as a means to maintain profits—
and thus stability—in response.
However, any model is obviously only an approximation of the actual phenom-
enon, and therefore, the efficacy of linear approximation techniques when pre-
dicting real world outcomes is not always assured. Even when the model would
seem to imply stability, the nature of that stability can change qualitatively. This
is what happened to IBM in the 1990s when mainframe margins collapsed. Sim-
ply selling more was not enough because each mainframe sale was losing more
money (Goldstein, Hazy & Lichtenstein, 2010). Changing environmental charac-
teristics can lead to situations of criticalization where linear approximation no
longer works.

Qualitative Change And Topologically Distinct Organizing Forms


When incorporated in a mathematical model, qualitative changes to stability
dynamics occur at what are called bifurcations when a small change to a pa-
rameter implies a qualitative difference in system behavior. When modeling the
environment mathematically, the points where this kind of behavior occurs in
the models are called singularities. Goldstein et al. (2010) in their GHS model
make the argument that changes in organizing form are triggered when condi-
tions in the environment, in interaction with the HID within the system, enter
criticalization near singularities in the models being used to guide HID. Fur-
ther, they argue that the specific shape of the container in which the HID unfold
determines what stable organizing forms can occur—not necessarily what will
happen, but what outcomes are possible. They show that as a parameter—for
example, the level of government grant funds that are flowing to an organiza-
tion and what this implies for rules of interaction among members of the orga-
nization—changes, the ecological-container in which HID unfold also changes.
There is a point of bifurcation in the unfolding of stable organizing forms.

fers to the dynamical system model itself and not to an observed phenomenon or property
in HID. For our purposes, the dynamical systems definition can be interpreted as what is
necessary to imply that the model being used is stable, a necessary but not sufficient condi-
tion for the properties themselves, the subject of the model, being called stable.

Hazy & Ashley 63


Goldstein et al. (2010) also point out that qualitative change can occur quite
suddenly (although bifurcation is related to a change in parameter and is not,
in general, related to the passage of time) even for a small change in the param-
eter. There may be a point as funds are cut when a previously workable, grant-
only funding model becomes unworkable. When this happens, the resulting
discontinuity implies a choice. The individuals involved must either transform
their interaction rules into an earned-revenue organizing form, or dissolve or
liquidate the entity entirely.

Families Of Opportunity Potential Functions Are Related To Organizing Forms


In order to examine the changing environmental characteristics that lead to
situations of criticalization where linear approximations are no longer appro-
priate, the concepts of bifurcation and singularity become a critical part of the
GHS model. With respect to these ideas, the GHS model defines two relevant
parameters that reflect the ecological container within which organizing occurs.
In doing so, the model describes where these points of singularity occur such
that there is a phenomenon of sudden change.
First, opportunity tension measures the organizing potential given the resources
available to the organization and the range of choices available to individuals
to acquire them. Choices are made by autonomous individuals about whether
and to what extent participation in groups might be beneficial, as well as how
best to organize the work that is necessary to gain continued access to these
resources (that is, to energy sources). If fresh water is needed for agriculture, a
dam and reservoir might be built to ensure continued access to it, for example.
Achieving this social value would require collective action and thus organizing.
In contentious environments, defenses against intruders might also be needed
to protect access to the water. The technology that is available for use and how
it is being leveraged to create value (Arthur, 2009) are also relevant to this pa-
rameter (Hazy et al., 2011).
Second, there is a need to acquire and use knowledge about these opportu-
nities and in particular the informational differences that are scattered across
the environment and embedded in the individual experiences of many actors.
These different individuals with their informational differences are in a position
to provide new ideas and approaches to realize the opportunity potential pre-
sented in the environment. If informational differences persist, convergence to
a new organizing form is not possible and asymmetries persist. At the same
time, if informational differences are ignored, adaptation is not possible because
events in the environment are not absorbed into the system. To capitalize on
opportunities when information is distributed, informational differences must
be observed broadly, filtered for relevance, and synthesized into the system’s
functioning though what Goldstein, Hazy & Lichtenstein (2010) call generative
leadership.

64 E:CO Vol. 13 No. 3 2011 pp. 57-79


An organizing system’s ecology holds that system within a container of specific
resources and distributed information about them. The nature of this container
is implicitly defined by the parameters above, and it can change shape when
either opportunity tension or informational differences change. (For an illustra-
tion, see Figure 2 which is described in a later section.) Because the changing
parameters change the ecological container, they are predictors of how stable
organizing forms might unfold.
Each distinct and stable set of interacting properties is an organizing form. In the
earlier example, a group may be organized around government grants, and then
shift to a self-sustaining revenue-based model. Each of these different organiz-
ing forms represent a different stability regime; all of these possibilities taken
together represent a family of opportunity potential functions at a given level
of coarse-graining that are presented to the individuals involved in the ecology.
For a discussion of the continuum of these organizing regimes, see Swanson &
Zhang (this issue) who describe different approaches to organizing as governed
by these factors.

Bifurcation In Opportunity Potential Implies Distinct Organizing Forms


The potential for bifurcation in organizing forms admits the possibility that
multiple regimes of organized stability coexist under certain parametric condi-
tions but not in others. According to the GHS model, the necessary conditions
for innovation in social forms involve situations where the potential to acquire
needed resources through organizing reach a bifurcation threshold. At the same
time, but independently, information must be gathered in the environment and
flow among individuals in ways that are sufficiently resonant so as to synthesize
informational differences to enable effective organizing. In these cases, the sta-
tus quo is exposed as no longer working, and it becomes clear that something
different is needed. Together, these conditions—the need to do something dif-
ferent and the capacity to gather and use the right information—reflect the or-
ganizing potentials presented by the environment (Pask, this issue). It is within
these potentials that human interaction dynamics unfold toward stability.
The GHS model represents organizing potential mathematically as the opportu-
nity potential function (OPF), F(X), describing a measure of unrealized potential
to achieve a sustainable level of fitness as defined by some performance metric,
Y = F(X), for each organizational state X and a given parametric situation (a,b)
such that:
F^ X h = aX + k1 bX2 + k2 X 4 (1)

Where X is the internal state of the system as defined by some random variable
such as profitability, the parameters a and b represent the informational differ-
ences and opportunity tension parameters respectively, and k1 and k2 are con-
stants (needed to simplify equation 2). The particular values of the parameters,

Hazy & Ashley 65


a and b in the OPF function thus define the specific affects of the ecological “con-
tainer”—including the particulars of resource/energy and information distribu-
tion and flows—on the organizing system of HID7. Each specific organization
lives at a unique spot in the ecosystem (that is, at unique values of a and b) and
thus operates along only one OPF as defined by the parameters. However, either
a or b or both can change thus changing the shape of the OPF.

Changing Informational Differences Implies


Stable Organizing Forms Also Change
Differences in the shape of the OPF as parameter a changes are shown in Figure
2. In Panel 1, organizing is stable at the minimal point of the OPF as the informa-
tional differences parameter a is low relative to what is needed for criticalization.
This case implies that even though there might be benefits to organizing (when
the parameter b is past the bifurcation point) there is little information flowing
among individuals and thus little awareness that there is any reason to organize.
As an example, at Hector’s World (see Shepherd & Woods, this issue) the original
organizing principle, Attractor A1 in the Figure, was the belief that providing
lessons to toddlers on whom to “trust” on the Internet should be the core orga-
nizing principle for their bourgeoning online social enterprise. Absent informa-
tion to the contrary, this approach seemed to make sense even though a more
sophisticated organizing principle (around the concept of “digital citizenship”)
was also a possibility. There was little information flow that would enable those
involved to creatively explore this alternative.
With additional unique individual experiences and thus increased informational
differences, however, one can see in Panel 2 that the more sophisticated per-
spective has made an appearance as a nub on the left slope of the opportu-
nity potential function. This is because some individuals begin to recognize that
there is an alternative possibility as they interact with one another and gather
new information from the users. However, at this point, there is still only limited
information and this information exits in disconnected pockets of interaction
within the organization. As a result, the new approach does not yet seem to
have sufficient momentum to compete effectively with the already working or-
ganizing model built around “trust”. Why change what’s working?
In Panel 3, we see that informational differences are now present throughout
the organization and interactions and synthesis has made it clear that there are

7. Note that a 4th degree equation is the lowest degree equation for which two distinct points
of stability can be identified when certain values of the parameters a and b are reached.
This equation known as the Landau-Ginzburg equation (Guastello, 2002) has applicability to
many state-change and phase transition models in the natural sciences. Further, it has been
shown mathematically that many such cases can be modeled in this way simply by choosing
the coordinate system and parameters wisely.

66 E:CO Vol. 13 No. 3 2011 pp. 57-79


indeed two viable ways to organize. By Panel 4 it becomes clear that “digital citi-
zenship” has greater value and only pockets of conservative resistance remain.
By Panel 5 the new way of organizing has come to dominate the team. In this
way, the GHS model can help the analyst understand how flow of informational
differences can lead to a wave of change in an organization.

Opportunity Potential And The Stability Of Organizing Forms


Because an activity usually carries with it both costs and benefits—an extra
status meeting every week for example might help communication or be over-
kill—changing an activity, can slightly improve or slightly degrade overall per-
formance. This change in state X would move the system along the OPF thus
changing F(X). (This is why unambiguous feedback about collective efficacy,
F(X), is so important; feedback enables decisions to be taken about what work
needs to be done to exploit unrealized potential. Effective use of feedback can
therefore move the organizing form along the OPF to a local maximum perfor-
mance (shown as a minimum of unrealized potential in the OPF). Importantly,
there are also points of stability. For example, when any small change in X and
thus F(X) along the OPF degrades performance, the position is at a local maxi-
mum performance, and is self-reinforcing as actions are taken to return to the
stable organizing form.

Panel 1 Panel 2 Panel 3 Panel 4 Panel 5

Initial Beginning Two attractors New attractor New


attractor of a new are in conflict displaces the Attractor
attractor old attractor dominates

A2
A2
A1
A1 A1 A2 A1 A2

Wave of Change in parameter a

a<o a=o a>o

Figure 2 The shape of the opportunity potential function (OPF) and thus the
possibility for qualitative change varies as informational differences (parameter
a) vary. For simplicity, the level of opportunity tension (parameter b) is assumed
to be constant and positive.

Hazy & Ashley 67


Building models of opportunity potential, and thus of stable organizing forms,
serves the needs of those individuals who are acting in concert. Assuming that
the relevant relationships among internal and external factors can be framed
by participants in terms of a model of an opportunity potential function, the
model can be used to share expectations about performance, to identify unreal-
ized potential benefits, and to discuss and evaluate alternative approaches. For
example, a social enterprise might have the potential to pursue a self-sustaining
business model with internally generated earned revenue, or it might pursue
grant-based funding, an organizing approach that is heavily dependent upon
third party largess (Swanson & Zhang, this issue). For a time, during criticaliza-
tion, both possibilities are apparent and can even coexist at once.
This idea is important because it shows not only that deep change is possible,
but also how models of coarse-grained opportunity potentials that are built and
shared by individuals are a key enabler of these changes. By changing the inter-
actions that are occurring in the HID, that is, by changing the rules of interac-
tion as implied by the models, different emergent properties interact producing
different organizing forms that express different levels of performance or “fit-
ness” as reflected in the OPF. In the previous example, a self-sustaining model
is less dependent on governmental largess. As grant opportunities decline, the
“fitness” of a grant-dependent business model also declines relative to that of a
self-sustaining model.
The implications of this model for social change theorists, like Torras (2009), who
hope for a new and different social equilibrium is profound. The model described
by Goldstein et al. (2010)(GHS) provides a concrete roadmap for bringing about
qualitative change across levels of scale.

Bifurcation In Organizing Form—An Illustration

T
he GHS model says that the constraints on social systems that lead to bifur-
cated forms of organizing are not usually physical barriers (like sheer rock
cliffs). Rather the relevant constraints often relate to two broad categories
of barriers to growth: limits to resource availability, what is called opportunity
tension and parameterized by b; and limits to the synthesis of informational dif-
ferences by autonomous participants in the complex adaptive system, and pa-
rameterized by a.
The spread of HIV/AIDS in Africa is an example that can be used to illustrate
the process where a wave of information sharing (overcoming constraints in the
process) can lead to significant social change and innovation. Panel 1 in Figure
3 highlights the emerging realization that there is a raging epidemic (which in
the dynamical systems model described earlier, is equivalent to an infection rate
being in excess of a threshold value that engenders an epidemic outcome). At
first this information sharing is limited. Those who are threatened do not have
access to the information they need to take advantage of resources to prevent

68 E:CO Vol. 13 No. 3 2011 pp. 57-79


infection. This might be due to either poorly working information flows in HID,
legal strictures, or cultural restrictions.
In an effort to reduce the infection rate, people might begin to interact with one
another, advocating for the organization of an effort to promote the wide spread
use of condoms—a new organizing form operating in parallel with the old way
of doing things as is illustrated in Panel 2. Suppose, however, that this pathway
is blocked by legal restrictions and cultural norms. In this case both resources
(access to condoms in the example) and the flow of information (knowledge of
their benefit vis-à-vis prevention of infection and disease) are constrained. These
ecological conditions create criticalization as described by the GHS model.

Panel 1 Panel 2 Panel 3

Organizing Organizing Organizing Organizing Organizing Organizing


Form 1 For m 2 Form 1 For m 2 Form 1 For m 2
Patterns & Patterns & Patterns & Patterns & Patterns & Patterns &
Properties Properties Properties Properties Properties Properties

Models Models Models Models Models Models

Human Human Human


Interaction Interaction Interaction
Dynamics Dynamics Dynamics

a<0 a=0 a>0

Figure 3 Bifurcation of organizing form occurs as human interaction


dynamics are influenced by external conditions impacting the rules and norms
that govern local interactions. Panel 1 shows cases where a < 0 meaning
informational differences are not synthesized and the old way of organizing
persists at aggregating information. Panel 2 where a = 0 shows that both
organizing forms coexists as informational differences are being absorbed,
but prior structures persist, a condition called criticalization. Panel 3 shows
situations where a > 0 meaning information about changing conditions is
driving innovation into a new synthetic organizing form. As new and different
properties emerge from these dynamics, these can be modeled as exerting
downward influence on the human interaction dynamics to stabilize the
emerging organizing form.

Hazy & Ashley 69


During this period of criticalization (shown in Panel 2), information and resourc-
es may be marshaled to innovate successful social, technical or culture programs
that are intended to overcome or work around whatever is constraining the sys-
tem—publicity programs are tried, public awareness programs are initiated, dis-
tribution and availability are improved, or cultural norms are relaxed to the point
where a successful condom use program is implemented—until the epidemic
is lessoned to the point where stable HID with a relatively healthy population
returns, Panel 3.
Either organizing form is possible, and would in the end be a stable result, but
the social value at the point of stability, the quality of life, in each of the two
organizing forms might be quite different. Innovation is needed to achieve the
latter; in a sense, the former represents a failure to successfully innovate, a failure
of what is called generative leadership (Goldstein, Hazy & Lichtenstein, 2010;
Surie & Hazy, 2006).

The Family Of Opportunity Potential Functions Defines A Topology Of


Organizing Forms
We now limit the discussion to the points of stability on opportunity potential
functions where the first derivative is equal to zero. Because they represent
minima of unrealized potential, they represent relative maxima of stable per-
formance. As such, these points of stability on the opportunity potential func-
tions can be mapped to the attractors in the dynamical systems models that
represent stable organizing within HID (Goldstein, Hazy & Lichtenstein, 2010).
We next look at all of the points of stability taken together when considering all
of the possible opportunity potential functions of a given system.
The subset of all stable points within the set of all opportunity potential func-
tions (as the parameters a and b are allowed to vary) is represented by what is
called a response surface consisting of those points where the first derivation of
equation (1) is zero, i.e., F’(X) = 0:
0 = Fl^ X h = a + b2 + X3 (2)
This folded surface, called the fold catastrophe (Thom, 1989), can be assumed to
form a manifold or surface that is defined as a topological space, and is shown in
Figure 4. Because of the mapping to dynamical system attractors described in
the previous paragraph, each point on this surface can be thought to represent
a distinct form of stable organizing for HID when considered at a particular level
of coarse-graining plus a cusp-shaped set of points of singularity. Each point
that is not a singularity, represents a potential stable organizing form for each
ordered pair (a, b). The points are not organizing forms, per se, since organiz-
ing forms are defined to be the emergent properties within the underlying HID,
rather these points are abstract points of potential stability when models are
constructed to describe possible organizing forms at a level of coarse-graining,

70 E:CO Vol. 13 No. 3 2011 pp. 57-79


Figure 4 The GHS model shows how ecological conditions relating to available
resources (opportunity tension, or parameter b) as well as information about
the precise nature of the system properties that define the collective response
(informational difference, or parameter a) are relevant to determining stable
end state responses (measured on the Y-axis as the potential function when the
derivative is zero). Panel A: Since two outcomes with distinctly different fitness
levels (values of Y) are likely in the cusp-shaped area (shown in the figure as ab
projection below the surface) the Response Y is said to bifurcate (as shown in
the bY projection to the right of the surface). Panel B: The fold-shaped surface
represents the stable solutions where the first derivative of the OPF (shown at
the top) is equal to zero. As can be seen, where the system begins with regards
these two parameters is an important predictor of where the system will end
up as conditions change. In particular, path dependency along the information
dimension implies that better information sharing even absent resources
will increase the likelihood that a best case end state will result as resource
availability varies.

possible business planning scenarios, for example. When looking at Figure 4,


one can immediately see that not all points on the response surface are alike
because the surface has singularities. Two observations are relevant.
First, a bifurcation point is apparent on this response surface. This is shown in
Figure 4, Panel A, and can be seen as a projection of the surface onto a Y x b plane
on the right side of the Figure. As b increases beyond the bifurcation point, two
distinct stable states become possible. Second, these two states are only appar-
ent within the cusp-shaped area that is shown as a projection of the a x b plane
below the surface. This shows that in addition to b being greater than the bifur-
cation point, the value of a is also important. The cusp-shaped area indicates the
combinations of parameter values that imply criticalization.

Hazy & Ashley 71


To illustrate this, Figure 4, Panel B shows the points of stability in a series of op-
portunity potential functions as the parameter a is allowed to vary (and such
that the first derivative is equal to zero) and how these are reflected on the re-
sponse surface when b > 0 is held constant. [Note also that when b < 0, there is
no fold in the response surface as a increases.] Thus criticalization is only appar-
ent when there are two stable solutions for a given ordered-pair (a, b) and this
only occurs in the cusp (see Figure 4, Panel A).

Topological Equivalence Crosses Scale

I
n this paper we have defined a representational framework for organizing
forms as a topological space that includes two complementary topologies.
The first defines sets and subsets of organizing forms that cross scale, rang-
ing from one-on-one meetings to large organizations and even institutions. The
second defines sets and subsets of organizing forms at a given level of coarse-
graining, but allowing the relevant environmental conditions to vary as reflect-
ed by system parameters such as opportunity potential, b, and informational
differences, a8.
If one represents organizing forms to meet the requisite mathematical assump-
tions that were described herein (see Arnold, 1989 for more details), then any
mathematical result that is proven for open sets is true given both topologies.
This means that the GHS model and other related extensions can be applied
generally within this topological space. Further, it implies that the GHS model
can be applied at any coarse-grained level of scale, all the way from individual
interactions, to work groups, to institutions and governmental interactions so
long as the requisite assumptions are met. This is the challenge of course, and
some of these issues are taken up in the next section.

Moving Beyond Metaphor:


Caveats For Application To Human Systems

U
p to this points the implicit assumption has been that HID can be mod-
eled in much the same way as other systems such as physical or biologi-
cal ones. Many authors have argued that there is a qualitative difference
in what complexity means in human terms versus physical systems. While we
also recognize that there is a difference, we suggest that the difference is more
one of degree than of category. This next section discusses several of these dif-
ferences and how theory must develop to accommodate them.

8. Note that we are using the single internal variable X and two parameter (a,b) case for illus-
tration. However, the result is true in general and other more involved cases also apply with
no fundamental modifications to the argument being necessary (Arnold, 1989).

72 E:CO Vol. 13 No. 3 2011 pp. 57-79


Problem Domain Framed As A Topological Space
For the approach described herein to have applicability, the problem being
modeled must meet the assumptions of topological space with the two topolo-
gies that were defined in this article. In particular, sets and subsets of organiz-
ing forms at various levels of coarse-grained scale, the whole set and the null
set must be included. Further, the unions and intersections of subsets must be
closed sets.
The same requirements apply to sets and subsets of opportunity potential func-
tions and their points of stability within a family of opportunity potential func-
tions that result as the parameters are allowed to vary. If both of these topologies
are defined the GHS model applies to the problem domain if the parameters and
state variables can be operationalized. Finally, because the response surface is
the derivative of the family of opportunity potential functions, there is a techni-
cal requirement that the manifold be differentiable almost everywhere (Arnold,
1989).

Discontinuous Change Limited To Immediate Adjacent Activity


The mathematical formalisms behind the GHS model are only strictly applicable
arbitrarily close to the singularity being modeled and do not necessarily hold at
a significant distance from it. The unfolding described herein only applies at the
singularity, that is, at the disorientation point or razor’s edge. Thus the GHS mod-
el does not necessarily hold even a short step beyond or away from the singular-
ity but only near the singularity where criticalization is occurring. Beyond this
“closeness”, additional factors might come into play, and these can overwhelm
the basic insight that, locally, the organizing form has indeed bifurcated into
two distinct possibilities.

The Ergodic And Invariance Assumptions


As discussed previously, statistical models are likely to be a methodology that
is useful for the modeling of emergent phenomena, as well as the previously
mentioned opportunity potential function. However, for the statistical model-
ing to be applicable it is necessary to ascertain that the random process charac-
terizing the system properties (emergent phenomena) of interest satisfies what
are called invariance assumptions and the even stronger ergodic assumption. (A
technical discussion of this point is provided as a technical appendix and is ac-
cessible on line.) This is because for the most part, statistical methods implicitly
assume that both of these assumptions are true.
It is not always easy to evaluate whether or not a part of the system or of an
emergent property is ergodic. This poses a challenge to the use of statistical
models as well as the GHS model. However, this is not necessarily a deadly flaw.
In many real world applications judgment is often utilized to determine the rel-

Hazy & Ashley 73


evance of this issue. More importantly, there are techniques that can enrich this
judgment decision although these must be used judiciously to avoid inappro-
priate inferences9.

Future Directions

T
his paper provides a general framework for models of human organizing
and how organizing forms adapt to changing circumstances to innovate
and create new social forms. Although conceptual at present, the model
presented is described in mathematical terms and is consistent with accepted
mathematical results.
Further work is needed to define the parameters that support the model empiri-
cally. In fact one of the most important next steps is to develop an application of
this model to a real life business or social service situation. In doing so, the pa-
rameters defining opportunity tension, b, and informational differences, a, must
be identified, defined and operationalized. If specific numerical values cannot
be directly assessed for “a” and “b” within a given organization it may be possible
to estimate relative values for a and b when compared to other organizations,
and this ordering approach might be used as a surrogate in empirical settings.
For example an organization that has significantly greater access and/or control
over needed resources on an attribute type scale would have a higher rank value
than some other organization that are less well positioned. Research taking this
approach would find support in the considerable literature on dynamic organi-
zational capabilities and the resource based view of the firm (see for example
Helfat, et al., 2007)
This approach is also similar to the Decision Analysis Model that utilizes the
Hurwicz criterion and the coefficient of optimism—a relative value defined be-
tween 0 and 1—to select appropriate decision alternatives. In this case, a to-
tally optimistic individual is assessed the value of 1 while a totally pessimistic
individual is assessed a value of 0. The indifferent individual is assigned a value

9. One of these, called Statistical Process Control, provides a methodology to determine


whether or not the variation within a system is inherently random. Control charts are utilized
to detect non-random patterns resulting from non-invariant causes such as changes over
time due to human action (error or other causes). These charts may also be able to detect
changes in random variation due to other non-ergodic assignable causes such as changes
in the environment over time as well as changes in the leadership function that might lead
to differences in certain rules of interaction between and among the agents within the orga-
nization (Hazy, 2011). Furthermore, a sophisticated analysis and interpretation of this meth-
odology can detect changes in the patterns of data e.g. a normal distribution as well as the
mean, variance, and to some degree the skewness and kurtosis of the data. Although this
methodology cannot guarantee that the process is corrected for being non-ergodic, it can
determine if it is an invariant process or not, and it can enhance judgment about the level of
ergodicity of the overall process.

74 E:CO Vol. 13 No. 3 2011 pp. 57-79


of .5. Individuals with different relative degrees of optimism or pessimism are
assigned values that are either closer to 1 or closer to 0. Something similar can
be done with opportunity potential and informational differences. Future work
in these areas could help to establish the GHS model as a canonical model for
changing organizing form.

References
Arnold, V.I. (1989). Catastrophe Theory, ISBN 9783540548119.

Arthur, W.B. (2009). The Nature of Technology: What It Is and How It Evolves, ISBN
9781416544050.

Crutchfield J.P. and Feldman, D.P. (1997). “Statistical complexity of simple one-dimen-
sional spin systems,” Physics Review E, ISSN 1539-3755, 55 (2): 1239-1242.

Crutchfield J.P. and Feldman, D.P. (2003). “Regularities unseen, randomness observed:
The entropy convergence hierarchy,” Chaos, ISSN 1054-1500, 15: 25-54.

Crutchfield, J. (1994). “Is anything ever new? Considering emergence,” in G. Cowan,


D. Pines and D. Meltzer (eds.), Complexity: Metaphors, Models, and Realty, ISBN
9780201626063, pp. 515-537.

Epstein, J. M. (1997). Nonlinear Dynamics, Mathematical Biology, and Social Science, ISBN
9780201419887.

Feldman D.P. and Crutchfield, J.P. (1998). “Statistical measures of complexity: Why?”
Physics Letter A, ISSN 0375-9601, 238 (415): 244-252.

Feldman, D.P., McTague, C.S. and Crutchfield, J.P. (2008). “The organization of intrinsic
computation; Complexity-entropy diagrams and the diversity of natural information
processing,” Chaos, ISSN 1054-1500, 18: 043106-1.

Gell-Mann, M. (2002). “What is complexity?” in A.Q. Curzio and M. Fortis (eds.), Com-
plexity and Industrial Clusters: Dynamics and Models in Theory and Practice, ISBN
9783790814712, pp. 13-24.

Goldstein, J., Hazy, J.K. and Lichtenstein, B. (2010). Complexity and the Nexus of
Leadership: Leveraging Nonlinear Science to Create Ecologies of Innovation, ISBN
9780230622289.

Goldstein, J.A., Hazy, J.K. and Silberstang, J. (2010). “A complexity model of social inno-
vation in social enterprise,” Journal of Social Entrepreneurship, ISSN 1942-0676, 1(1):
101-125.

Guastello, S.J. (2002). Managing Emergent Phenomena: Nonlinear Dynamics in Work


Organizations, ISBN 9780805831634.

Hannah, M.T. and Freeman, J. (1989). Organizational Ecology, ISBN 9780674643499.

Hazy, J.K. (2006b). “Measuring leadership effectiveness in complex socio-technical


systems,” Emergence: Complexity & Organization, ISSN 1521-3250, 8(3): 58-77.

Hazy & Ashley 75


Hazy, J.K. (2007). “Computer models of leadership: Foundation for a new discipline or
meaningless diversion?” The Leadership Quarterly, ISSN 1048-9843, 18(4): 391-410.

Hazy, J.K. (2008a). “Toward a theory of leadership in complex systems: Computational


modeling explorations,” Nonlinear Dynamics, Psychology, and Life Sciences, ISSN
1090-0578, 12(3): 281-310.

Hazy, J.K. (2008b). “Leadership or luck? The system dynamics of Intel’s shift to micro-
processors in the 1970s and 1980s,” in M. Uhl-Bien and R. Marion (eds.), Complexity &
Leadership Part 1: Conceptual Foundations, ISBN 9781593117955, pp. 347-378.

Hazy, J.K. (2011). “Parsing the ‘influential increment’ in the language of complexity:
Uncovering the systemic mechanisms of leadership influence,” International Journal
of Complexity in Leadership and Management, ISSN 1759-0256, 1(2): 164-191.

Hazy, J.K. (in press). “The unifying function of leadership: Shaping identity, ethics, and
the rules of interactions,” International Journal of Society Systems Science, ISSN 1756-
2511.

Hazy, J.K., Ashley, A., Moskalev, S. and Torras, M. (2011). “Technology leverage and sus-
tainable society systems: A call for technology forecasting that anticipates innova-
tion,” International Journal of Society Systems Science, ISSN 1756-2511, 3(1/2): 5-20.

Hazy, J.K., Goldstein, J.A. and Lichtenstein, B.B. (2007). Complex Systems Leadership
Theory: New Perspectives form Complexity Science on Social and Organizational Effec-
tiveness, ISBN 9780979168864.

Hazy, J.K., Moskalev, S and Torras, M. (2010). “Mechanisms of social value creation:
Extending financial modeling to social entrepreneurship and social innovation,”
International Journal of Society Systems Science, ISSN 1756-2511, 2(2): 134-157.

Helfat, C.E., Finkelstein, S., Mitchell, W., Peteraf, M.A., Singh, H., Teece, D. and Winter, S.G.
(2006). Dynamic Capabilities: Understanding Strategic Change in Organizations, ISBN
9781405135757.

Hirsch, M.W., Smale, S. and Devaney, R.L. (2004). Differential Equations, Dynamical Sys-
tems and an Introduction to Chaos, ISBN 9780123497031.

Holland, J.H. (1975). Adaptation in Natural and Artificial Systems, ISBN 9780262581110.

Holland, J.H. (2001). “Exploring the evolution of complexity in signaling networks,”


http://www.santafe.edu/media/workingpapers/01-10-062.pdf.

McKelvey, B. (2002). “Model-centered organization science epistemology,” in J.A.C.


Baum (ed.), Companion to Organizations, ISBN 9780631216940, pp. 752-780.

Prokopenko, M., Boschetti, F. and Ryan, A.J. (2009). “An information-theoretic primer on
complexity, self-organization and emergence,” Complexity, ISSN 1099-0526, 15(1):
11-28.

Surie, G. and Hazy, J.K. (2006). “Generative leadership: Nurturing innovation in complex
systems,” Emergence: Complexity & Organization, ISSN 1521-3250, 8(4): 13-26.

76 E:CO Vol. 13 No. 3 2011 pp. 57-79


Thom, R. (1989). Structural Stability and Morphogenesis, ISBN 9780201406856.

Torras, M. (2009). “Social entrepreneurship, social value, and the environmental Big
Push: Some remarks,” in J.A. Goldstein, J.K. Hazy and J. Silberstang (eds.), Complexity
Science and Social Entrepreneurship: Adding Social Value through Systems Thinking,
ISBN 9780984216406, pp. 215-230.

Zeeman, E.C. (1977). Catastrophe Theory: Selected Papers, 1972-1977, ISBN


9780201090147.

James K. Hazy is an Associate Professor of Management at the Adelphi University


School of Business in Garden City, NY. He has published over 40 articles and book
chapters and has won numerous awards including a Best Paper Award in Entrepre-
neurship from the Academy of Management. In 2010 he co-wrote the book: Com-
plexity and the Nexus of Leadership: Leveraging nonlinear science to create ecologies of
innovation which achieved Amazon Top-100 status in Management Science in both
hardback and paperback editions. He has also co-edited the books: Complex Sys-
tems Leadership Theory and Complexity Science and Social Entrepreneurship: Adding
Social Value through Systems Thinking. He obtained his Doctoral degree in Human
Resource Development from the George Washington University and his MBA in Fi-
nance from the Wharton School of the University of Pennsylvania. Before joining
Adelphi University, he was a Senior Executive with the multinational firms AT&T and
Ernst & Young.

Allan S. Ashley is Professor of Management and Decision Sciences and Chair of


the Department of Management, Marketing & Decision Sciences at Adelphi Uni-
versity School of Business in Garden City, NY. He has published articles in the ar-
eas of decision sciences, operations, total quality management and ethics. He has
also published two books in the area of mathematics for managers and total quality
management. He obtained his Ph.D. in Operations Research and Management from
Polytechnic Institute of Brooklyn (n/k/a Polytechnic Institute of New York Univer-
sity).

Hazy & Ashley 77


Technical Appendix—Ergodicity
The ergodic assumption or ergodicity requires first that the random process is
stationary (defined below) meaning it doesn’t change with time. There are sta-
tionary processes that are non-ergodic as well as stationary processes that are
ergodic. However if a process is not stationary then it cannot be ergodic. In
other words stationarity is a necessary although not sufficient condition for er-
godicity (Bendat & Pierso. 1966). Consequently we first need to be able to de-
termine whether or not the system properties we are studying are characterized
by the concept of stationarity.
Stationarity is defined as follows: Suppose we compute the mean and/or higher
moments as well as joint moments (auto correlation) at some point in time, t1,
for a collection of N sample functions (ensemble). The mean X(t1) could be cal-
culated as:
N
1/ ^ h
nX ^t1h = lim Xk t1
N"3 N =
k 1

And the auto correlation function RX(t1+) could be calculated as:

1 / 6 ^ h ^ + h@
N

RX ^t1, t1 + x h = lim Xk t1 Xk t1 x
N"3 N =
k 1

It is assumed that each sample function is equally likely.


If X(t1) and RX(t1+) vary as time t1 varies then the random process defined by
the sample functions or ensemble [X(t)] is considered to be non-stationary. If,
however, X(t1) and RX(t1+) do NOT vary as time t1 changes, that is:
X(t1) = X and RX(t1+) = RX()
then the random process is considered to be weakly stationary. For the situa-
tion where all possible moments, e.g. mean, variance, skewness, kurtosis, etc.,
and the joint moments are time invariant, then the random process [X(t)] is con-
sidered to be strongly stationary. Furthermore for numerous applications, weak
stationarity justifies the assumptions of strong stationarity, thus satisfying the
time invariant assumption.
Ergodicity also requires that for a stationary process its properties do not require
computation over multiple sample functions in the ensemble. For example sup-
pose:
T
1
nX ^ k h = lim # Xk ^t1h dt
T"3 T
0

and,
T
1 # ^ h ^ + h
RX ^x, k h = lim Xk t Xk t x dt
T"3 T
0

78 E:CO Vol. 13 No. 3 2011 pp. 57-79


If the random process X(t) is stationary and X(k) and RX(,k) defined above do
not differ where computed over different sample functions, the random process
is considered ergodic. That is:

X(k) = X and RX(,k)= RX(k)

as well as other time-averaged properties are equal to the corresponding en-


semble averages value.
Although stationarity and ergodicity are taken to be true for many processes
that involve human systems, it seems likely that this is not always, nor is it even
usually the case. In numerous systems involving humans and the measurement
of human performance, it is axiomatic that outcomes vary with time. Thus sta-
tionarity and therefore ergodicity are not necessarily valid, especially for longer
time horizons or in different contexts. For example, one of the most measured
system properties is quality, and many measurements of the quality of human
performance are not necessarily stationary, that is the individual events them-
selves vary over time and space and are thus not necessarily comparable. For
example many performance measurements in a hospital such as medication er-
rors are much larger at the end of July and in August, than any other month.
Thus, it is not necessarily appropriate to use statistical methods that depend upon
stationarity (or ergodicity) when comparing medication error samples, or any oth-
ers for that matter, that were taken at various times of the year. There is no guaran-
tee that the error introduced by the lack of stationarity does not impact the relevant
statistics being reported.

Hazy & Ashley 79

Вам также может понравиться