OPINION NATURE|Vol 454|3 July 2008

ESSAY
Arise ‘cliodynamics’
If we are to learn how to develop a healthy society, we must transform history into an analytical, predictive
science, argues Peter Turchin. He has identified intriguing patterns across vastly different times and places.
hat caused the collapse of the Let histor y continue to focus on the

W Roman Empire? More than 200

explanations have been proposed1,
but there is no consensus about which expla-
particular. Cliodynamics, meanwhile, will
develop unifying theories and test them with
data generated by history, archaeology and
nations are plausible and which should be specialized disciplines such as numismatics

D. PARKINS
rejected. This situation is as risible as if, in (the study of ancient coins).
physics, phlogiston theory and thermody- Is this proposal feasible? The most compel-
namics coexisted on equal terms. ling argument against the possibility of scien-
This state of affairs is holding us back. tific history goes like this. Human societies
We invest in medical science to preserve the are extremely complex. They consist of many
health of our bodies, and in environmental different kinds of individuals and groups that
science to maintain the health of ecosystems. interact in complex ways. People have
Yet our understanding of what makes socie- free will and are therefore unpredict-
ties healthy is in the pre-scientific stage. able. Moreover, the mechanisms that
Sociology that focuses on the past few years underlie social dynamics vary with
or decades is important. In addition, we need historical period and geographical
a historical social science, because processes region. Medieval France clearly dif-
that operate over long timescales can affect fered in significant ways from Roman
the health of societies. It is time for history Gaul, and both were very different to ancient
to become an analytical, and even a predic- China. It is all too messy, argue the naysayers,
tive, science. for there to be a unifying theory.
If this argument were correct, there would
Splitters and lumpers be no empirical regularities. Any relation-
Every scientific discipline has its share of ships between important variables would be
splitters, who emphasize the differences contingent on time, space and culture.
between things, and lumpers, who stress
similarities in search of organizing princi- Empirical empires
ples. Lumpers dominate physics. In biology, In fact, several patterns cut across periods
splitters, who care most for the private life of and regions3. For example, agrarian, prein-
warblers or the intricate details of a chosen dustrial states have seen recurrent waves of
signalling molecule, are roughly matched in political instability — not interstate warfare,
numbers by lumpers, who try to find fun- but lethal collective violence occurring
damental laws. Social sciences such as eco- within states, ranging from small-scale
nomics and sociology are rich in lumpers. urban riots, in which just a few people are
Sadly, few are interested in applying analyti- killed, to a full-blown civil war. This is just
cal approaches to the past. History has an the sort of violence we need to understand:
alarmingly small proportion of lumpers. many more people are killed today in terror-
Rather than trying to reform the histori- ist campaigns, civil wars and genocides than
cal profession, perhaps we need an entirely in wars between nations4.
new discipline: theoretical historical social Recent comparative research shows that
science. We could call this ‘cliodynamics’, agrarian societies experience periods of
from Clio, the muse of history, and dynam- instability about a century long every two or
ics, the study of temporally varying processes three centuries. These waves of instability fol-
and the search for causal mechanisms2,3 . low periods of sustained population growth.
For example, in Western Europe, rapid popu-
lation growth during the thirteenth century
was followed by the ‘late-medieval crisis’,
comprising the Hundred Years War in France,
the Hussite Wars in the German Empire, and
the Wars of the Roses in England. Popula-
tion increase in the sixteenth century was
followed by the ‘crisis of the sev-
enteenth century’ — the wars
of religion and the Fronde in
34
© 2008 Macmillan Publishers Limited. All rights reserved
NATURE|Vol 454|3 July 2008 OPINION

France, the Thirty Years War in Germany, and This explanation — the ‘demographic– the life-blood of historical disciplines such as
the English Civil War and Glorious Revolu- structural’ theory — is a work-in-progress. astrophysics and evolutionary biology.
tion. Similarly, population growth during the Our tests with the eight case studies6 sup- Cliodynamic theories will not be able to
eighteenth century was followed by the ‘age of port some of its predictions: for example, predict the future, even after they have passed
revolutions’, ranging from the French Revolu- élite overproduction preceded the crisis in empirical tests. Accurate forecasts are often
tion of 1789 to the pan-European revolutions every case. The tests also identify areas where impossible because of phenomena such as
of 1848–49 (ref. 5). the theory needs to be modified. Perhaps we mathematical chaos, free will and the self-
Such oscillations between population need an entirely new theory to explain the defeating prophecy. But we should be able to
growth and instability have been termed observed patterns and predict new ones, but use theories in other, perhaps more helpful,
‘secular cycles’6. Given the limitations of his- that is the business of science. The important ways: to calculate the consequences of our
torical data, we need an appropriately coarse- thing is that societies as different as medi- social choices, to encourage the development
grained method to determine the statistical eval France, the Roman Empire and China of social systems in desired directions, and to
significance, and the generality, of the pattern. under the Han dynasty share dynamics, when avoid unintended consequences.
The basic idea is to demarcate population viewed in an appropriately coarse-grained Like other systems with nonlinear feed-
growth and decline phases, way. Not everything in back, societies often respond to interventions
and to count the instability history is contingent and in surprising ways. When the Assembly of
incidents (such as peasant
“Societies as different particular. Notables refused to approve a new land tax in
uprisings and civil wars) that as medieval France, the Even so, theories devel- 1787, they did not intend to start the French
occur during each phase. Roman Empire and China oped and tested on prein- Revolution, in which many of them lost their
Wi t h my c o l l e a g u e s dustrial data must be heads. When Tony Blair was Britain’s prime
Sergey Nefedov and Andrey
under the Han dynasty modified before they can minister, he set out to increase the proportion
Korotayev, I have collected share dynamics.” be applied to contemporary of youth getting higher education to 50%. He
quantitative data on demo- social dynamics. Happily, was presumably unaware that the overabun-
graphic, social and political variables for there are indications that our theories will not dance of young people with advanced educa-
several historical societies. Applying the need to be replaced wholesale. Rapid demo- tion preceded the political crises of the age
above approach to eight secular cycles in graphic change and élite overproduction were of revolutions in Western Europe12, in late
medieval and early modern England, France, still important factors in twentieth-century Tokugawa Japan and in modern Iran and the
the Roman Empire and Russia, we find that revolutions10. Soviet Union9,10.
the number of instability events per dec- Furthermore, over the past 200 years, It is time we heeded the old adage that those
ade is always several times higher when the political instability in the United States has who do not learn from history are doomed to
population was declining than when it was waxed and waned in a pattern reminiscent of repeat it. We must collect quantitative data,
increasing6. The probability of this happen- that in preindustrial societies. Political violence construct general explanations and test them
ing by chance is vanishingly small. The same — urban riots, lynchings, violent labour dis- empirically on all the data, rather than on
pattern holds for the eight dynasties that putes and so on — was almost absent in the instances carefully selected to prove our pet
unified China, from the Western Han to the early nineteenth century, increased from the narratives. To truly learn from history, we
Qing7, and for Egypt from the Hellenistic to 1830s and reached a peak in around 1900. must transform it into a science. ■
the Ottoman periods8. The American Civil War occurred during this Peter Turchin is professor of ecology and
period of growing unrest. The instability then mathematics at the University of Connecticut,
Making waves subsided during the 1930s, and the following Storrs, Connecticut 06269, USA. He is the author
Such strong regularity points to the presence two decades were remarkably calm. Finally, in of War and Peace and War: The Life-Cycles of
of some fundamental principles. Population the 1960s, political violence increased again11. Imperial Nations (Pi Press, 2006).
growth beyond the means of subsistence leads It remains to be seen whether a modi-
to declining levels of consumption and popular fied version of the demographic–structural 1. Demandt, A. Der Fall Roms: die Auflösung des Römischen
discontent, but this is not enough to destabilize theory can explain this pattern. The point is Reiches im Urteil der Nachwelt (Beck, Munich, 1984).
2. Turchin, P. Historical Dynamics: Why States Rise and Fall
agrarian societies. Peasant uprisings have little that the study of such slow-moving processes (Princeton Univ. Press, 2003).
chance of success when the governing élites are requires a long-term view and an explicitly 3. Turchin, P. War and Peace and War: The Life Cycles of
unified and the state is strong9. historical approach. Imperial Nations (Pi Press, 2006).
The connection between population 4. Mack, A. (ed.) Human Security Report 2005: War and Peace
in the 21st Century (Oxford Univ. Press, 2005).
dynamics and instability is indirect, medi- Learning lessons 5. Fischer, D. H. The Great Wave: Price Revolutions and the
ated by the long-term effects of population Any claim that history can become a predic- Rhythm of History (Oxford Univ. Press, 1996).
growth on social structures. One effect is the tive science raises eyebrows. But scientific 6. Turchin, P. & Nefedov, S. Secular Cycles (Princeton Univ.
Press, 2008).
increasing number of aspirants for élite posi- prediction is a broader concept than merely 7. Nefedov, S. PhD dissertation [in Russian] (Ekaterinburg
tions, resulting in rivalry and factionalism. forecasting the future. It can be used to test Univ., 1999).
Another consequence is persistent inflation, theories. For example, two rival theories may 8. Korotayev, A. & Khaltourina, D. Introduction to Social
Macrodynamics: Secular Cycles and Millennial Trends in Africa
which causes a decline in real revenues and a make different predictions about the behav- (URSS, 2006).
developing fiscal crisis of the state. As these iour of some variable, such as birth rate, 9. Goldstone, J. A. Revolution and Rebellion in the Early Modern
trends intensify, they result in state bank- under certain social conditions. We then ask World (Univ. California Press, 1991).
ruptcy and a loss of military control; conflict historians to explore the archives, or archae- 10. Goldstone, J. A. J. Int. Affairs 56, 3–21 (2002).
11. Levy, S. G. Political Violence in the United States, 1819-1968.
among élite factions; and a combination of ologists to dig up data, and determine which (Computer file, Inter-University Consortium for Political
élite-mobilized and popular uprisings, lead- theory’s predictions best fit the data. Such and Social Research, Ann Arbor, 1991).
ing to the breakdown of central authority3,9. retrospective prediction, or ‘retrodiction’, is 12. O’Boyle, L. J. Modern Hist. 42, 471–495 (1970).

35
© 2008 Macmillan Publishers Limited. All rights reserved