UK United Kingdom

Cliodynamics: can science decode the laws of history?

They say history always repeats itself - empires rise and fall, economies boom and bust - but is there a way to map and predict the dynamical processes of history? The new and highly controversial discipline…

Revolution and war: is it all just a little bit of history repeating? blprnt_van

They say history always repeats itself - empires rise and fall, economies boom and bust - but is there a way to map and predict the dynamical processes of history? The new and highly controversial discipline cliodynamics is the most recent attempt to transform history into science.

When the French Assembly of Notables frustrated attempts by the royal government to fix the state fiscal crisis in 1788, because they did not want to pay taxes, these aristocrats did not intend to trigger the French Revolution, during which many of them ended up guillotined or exiled. Yet this is precisely what happened.

When the slave-owning elites of South Carolina declared their secession from the Federal Union in December 1860, they did not intend to trigger a bloody civil war that caused more than 600,000 deaths, killed one quarter of military-aged white Southerners, and resulted in the loss of most of their own wealth, when their slaves were freed. Yet this is precisely what happened.

Protesters in Foley Square for Occupy Wall Street. bogieharmond

Now, when the radical Tea Party Republicans refuse to negotiate with the Democrats to achieve a compromise, they probably don’t intend to push the United States into default, trigger a massive economic crisis, widespread urban riots, political assassinations and terrorism, and bloody clashes between the Tea Party and the Occupy Movement. Yet - well, this hasn’t happened but cliodynamics indicates that during the next decade the United States will be unusually vulnerable to an outbreak of serious political violence.

A scientific method

Clio, muse of history. Jastrow (2006)

Cliodynamics originates from Clio, the Greek muse of history. Despite a common misconception, the cliodynamics approach is not to argue by analogy with past events. This is an incorrect way of learning lessons from history – the historical record is very rich and varied, so finding historical examples to support almost any argument is easy.

Instead, the adherents of cliodynamics treat historical record just as, say, evolutionary biologists treat the palaeontological record. Theories are constructed and based on general principles and tested empirically with comprehensive databases. In short, we use the standard scientific method that worked so well in physics, biology, and many social sciences.

My colleague Herbert Gintis compares cliodynamics to aviation. You cannot predict when a plane will crash, but you can study the black box data to determine the causes of crash and figure out how to fix them.

You certainly don’t fly an aeroplane that has known problems. Similarly, you don’t put additional stresses on the social system that is already in fragile equilibrium; rather you should try to fix the underlying causes.

Clearly our knowledge about why states collapse and civil wars break out is nowhere near the state of aeroplane design. But recently we’ve made a lot of progress.

Theory of revolution and war

Consider the “structural-demographic theory” that was first proposed by the sociologist Jack Goldstone and subsequently developed and tested with data by others, including myself.

The theory explains major outbreaks of political violence, such as the French Revolution or American Civil War, by focusing on several interrelated processes. One is the falling or stagnating living standards of the general population. But contrary to the widely held view, popular discontent by itself is not a sufficient cause of a civil war or a revolution.

A more important factor is what has been called “elite overproduction” – that is, the appearance of too many elite candidates vying for a limited supply of power positions within the government and the economy.

As written about in my book War and Peace and War, elite overproduction results in intense intra-elite competition, polarisation, and conflict that ultimately takes violent forms.

It is important to stress the limitations of this approach. The theory does not predict when and how actual violence will break out, only structural conditions that make such an outcome likely.

The Battle of Gettysburg, fought July 1—3, 1863. The battle was part of the American Civil War and won by the North. Nathaniel Currier and James Merritt Ives

Think of a forest in which deadwood has been accumulating for many years. We don’t know what will start the fire – it could be a lightning strike during a storm, or a careless match thrown away. But sooner or later such a precipitating spark will arrive, and there will be a massive conflagration.

Nobody could predict that Mohamed Bouazizi, a fruit vendor in Tunisia, would decide to publicly immolate himself. But this act of a desperate individual had huge consequences because the structural conditions in Tunisia were ripe for revolution.

The structural-demographic theory has been tested by several investigators on many historical societies. The theory predicts very long-term cycles in which periods when societies are internally at peace are succeeded by waves of unrest. Both of these “integrative” and “disintegrative” phases are about a century long.

The theory focuses entirely on the dynamics of political instability within states as external wars have a logic of their own (in fact, it is typically societies which are in their integrative phases that prosecute successful wars of external conquest).

Our empirical investigations of a variety of historical societies confirm that they go through structural-demographic cycles. But on top of the long cycles are often superimposed shorter oscillations with periods of roughly 50 years. It appears that people eventually tire of incessant fighting, so during the disintegrative phases human generations experiencing a lot of fighting tend to alternate with relatively peaceful ones.

US political violence

Recently the Journal of Peace Research published my article in which I tested the predictions of the theory on American data. Constructing and analysing a database on US political violence (between 1780 and 2010), I found that the dynamics of violent incidences were just as predicted by the theory: a long structural-demographic cycle with a 50-year cycle superimposed on it:

US Population Violence Database. Peter Turchin

Additionally, I found that other structural-demographic indicators moved in a cyclical fashion in ways that were correlated with the waxing and waning of political violence.

In the last three or four decades real wages of unskilled workers stagnated. The incomes of the top one percent, on the other hand, grew explosively, leading to ever increasing economic inequality. Signs of elite overproduction include growing demand for educational credentials: tuition rates at elite colleges that rise much faster than inflation and the exploding numbers of new MBAs and JDs.

Intra-elite competition and conflict are indicated by rampant polarisation within the US Congress and increasing legislative deadlock.

Finally, the declining health of government finances can be traced by the growth of federal debt.

Each of these trends has been noted and commented upon. But what is not broadly appreciated is that each did not develop in isolation; they are actually interconnected at a fundamental level. Moreover, our historical research shows that this combination of trends is typical of historical societies that are in the pre-crisis phase.

An outbreak of political violence comparable to French Revolution or Civil War is not inevitable, or even likely. American society is much more resilient than France of Ancien Régime. Still, we should not forget that Antebellum America was a reasonable, if imperfect democracy.

Nevertheless, its elites were unable to contain their conflicts within constitutional bounds. American political elites today need to take this historical lesson to heart.

Sign in to Favourite

Join the conversation

23 Comments sorted by

  1. Derek Bolton

    Retired s/w engineer

    Interesting article, but I do have a concern about the graph. Bundling acts of terrorism with riots and lynchings entails an assumption. Perhaps that is justified by other research, but it is noteworthy that in the graph there is little correlation. There was a small one in 1870, but otherwise they appear independent. The present rise in 'all' is thus quite different from the earlier spikes.

    1. Dennis Alexander

      logged in via LinkedIn

      In reply to Derek Bolton

      Derek, I rather think that what the graph shows is the relative rise of terrorism and the relative decline of lynchings in the twentieth century, at least, the latter probably due to it being progressively more likely to be prosecuted as murder since 1945. This may indicate a common underlying factor between the two (lynchings and terrorism) as manifestations of civil unrest. Thus, I suggest that the 1920s correlation of lynchings and riots qualifies as a spike the same as the 1870s. So, while either or both (terrorism or lynchings) correlates with riots, I don't think your interpretation is correct. My caveat is about what qualifies as terrorism and is it specifically domestic and home-grown: a little definitional work would have been nice in an otherwise interesting and thought provoking article.

  2. Ken McLeod

    logged in via Facebook

    Beautifully done. It's a concept I have been thinking of but using only intuition rather than study. This article should be published in The Economist, which has been runnning a series of articles along the lines of "is the American political system broken?"
    (And answering the question with cutting criticism of the Republicans.)

  3. Gregory Melleuish

    Associate Professor, School of History and Politics at University of Wollongong

    Peter Turchin's work is fascinating and very important. We should take very seriously what he says about eite overproduction especially at a time when the government is actively seeking to increase the number of university graduates. There are only a limited number of jobs for which these newly minted graduates can compete. Competition will become quite fierce. Turchin notes that an increase in duelling in seventeenth century France was a response to too many people competing for too few opportunities. Following Schumpeter, the modern equivalent may be the production of an educated class that looks for ways of 'changing the system' through whatever means possible. Those of us working in universities can see what happens when we produce too many PhDs for a very limited number of academic positions. It does not make for peace and love.

    1. David Arthur

      resistance gnome

      In reply to Gregory Melleuish

      It depends on what you want. If you want a society and economy of no advances, no progress, just perpetual stasis, then engineering societies to not grow and develop is the way to go.

      Various versions of this has been imagined been imagined in Aldous Huxley's "Brave New World" and Frank Herbert's "Through the Eyes of Heisenberg".

      The more educated people we have, the more likely it is that bright ideas will emerge. You see,

      Most of all, there is no such thing as an educated class that looks for ways to change the system. Youthful graduates are just following the biological drive that occurs in all young adult animals, to go out into the world and prosper.

    2. John Harland

      bicycle technician

      In reply to David Arthur

      Educated people would suppose that bright ideas originate primarily with educated people.

      it may be the converse. Education channels thinking and can make the thinking of new ideas more difficult, rather than easier.

      Development of a society does not rely on constant growth any more than individual learning and wisdom rely on growing fatter or taller.

  4. Iain Stuart


    Hmmm does this sound like the sort of thing that Asimov invented in his "Foundation series'?

    Surely the "laws of history" and the "application of the scientific method" has been thoroughly debunked by generations of historians (after attempts to use the scientific method in the 1950s) not to mention the whole post-modernist movement which surely undermines the notion of such laws.

    1. Gregory Melleuish

      Associate Professor, School of History and Politics at University of Wollongong

      In reply to Iain Stuart

      But surely historical development makes more sense if look at it in the framework of long term cycles that may not cause events but help to explain the circumstances in which certain things happened. This means climate cycles and also what Turchin has discussed in another his books Secular Cycles. Hence what happened in the fourteenth century in both Europe and China does not make a lot of sense if we do not appreciate that there was a cooling in the climate. It is worthwhile plotting the economic cycles in relation to political and social change. This does not necessarily mean some sort of rigid causal mechanism. People make their own history but we need to understand the circumstances in which they made it.

    2. Dennis Alexander

      logged in via LinkedIn

      In reply to Iain Stuart

      Iain, what is available now is not what Sari Heldon had, nowhere near. It is, however, significantly greater and much more digitally accessible and analyzable now than in the 50s or anytime much before 2000. As such, it is possible to mine correspondences between events and a large number of factors that may or may not contribute. Constructing and testing hypotheses to establish predictive validity of models then becomes possible. But it doesn't replace or diminish the descriptive and interpretive functions and roles of historians in any way.

    3. Kelly O'Neill


      In reply to Sherman Dorn

      I read your post. Didn't find your attempt to rope in Krugman to your view convincing, though you could be right. Are you pomo? Is Krugman pomo?

  5. Richard Helmer

    REsearch Engineer

    thanks for this interesting article and approach and the comments it has evoked. i was unaware of this approach til now and am grateful for the concise teaching

    science and enginering oft start with imprecise measurement and poorly defined units...and so struggle to quickly realise grand unified theories. Whilst this can mean lines of inquiry are open to easy attack, its perhaps a genuine indicator that 'more investigation of models and mechanisms is required' (researchers love that!)...especially if it captures imaginations and enables predictive capability and some level of systemic control that avoids catastrophic failure

    i look forward to hearing about how such a model can integrate with other aspects of science


  6. Ian Donald Lowe

    Seeker of Truth

    I send this message to the scientists, the money changers, the empire builders and the political manipulators pursuing this dream.

    You cannot alter the path of history and you cannot prevent your own system from collapsing because of the corruption that lays at it's core. You can and already do great damage in attempting to manipulate people and force them to perform your tasks against their best interests. You are offered a chance to reject the lies and the pressure and find a way off this path and back to the path you intended to pursue. There will be no elite and enlightened era of technological omniscience. That is a lie. Stop hurting people with your acts, stop telling lies to support harm and damage. Stop working for a system that is working toward a form of enslavement for humanity. Remember to breathe, remember you are human. Give yourself a chance before it's too late.

  7. Kelly O'Neill


    I'd love to believe it and I do think that one day science will be used this way, a la Edward O Wilson's Consilience perhaps. But am yet to be convinced. How does the model go with other datasets? China? India (talk to Jim Masselos)? Cambodia? etc etc?

  8. Dianna Arthur


    Fascinating article - one that deserves a reread.

    "Each of these trends has been noted and commented upon. But what is not broadly appreciated is that each did not develop in isolation; they are actually interconnected at a fundamental level."

    Of course these trends are interconnected. Nothing acts in total isolation. What we do at an individual level has impacts around us. A shame that many conservatives don't understand this.

  9. David Arthur

    resistance gnome

    History is not cyclic; at best, it is chaotic in the mathematical sense, never quite repeating and often bursting out into completley new directions. Further, it is chaotic in a space of so many dimensions that it is highly unlikely to quite repeat itself in the entire duration of the universe.

    "Decoding the laws of history" is unlikely to be the same as detailed prediction of future events. After all, we're still having trouble with economic modelling.

  10. Sebastian Poeckes


    Is there a place for Kondratieff Cycles in all this? Surely some sort of economic cycle would be a major factor. I find it unconvincing that social factors alone would drive the system. Alongside the social violence trends illustrated in the graph above there were major technological and economic changes taking place across the American economy.

    1. Ian Donald Lowe

      Seeker of Truth

      In reply to Peter Turchin

      How desperate must a system be that it has to prevent it's imminent collapse by massive acts of violence and enslavement of it's own people?
      Have you read any Orwell novels?
      Did you learn nothing except the techniques employed to manipulate people?
      Do you really think it's a good idea to maintain a system when it has been corrupted beyond redemption?

  11. James Hill

    Industrial Designer

    R.I. Moore in his "The First European Revolution" describes "elite overproduction" in the aristocratic families of France in the immediate centuries after Charlemagne. Competition for a "place in the sun" among these elites put pressure on Church lands which in turn stimulated the rule of celibacy to ensure that church lands would not become alienated. The fear of the progeny of Eloise and Abelard leading to his forced castration.
    The Papal crusade against The Anglo-Saxons in 1066 and the enslaving…

    Read more
  12. Ian Colditz

    Research Scientist in Livestock Health and Welfare at CSIRO

    Interesting article.

    Picking up on the theme of some other comments, scientific analysis starts with categorization, moves on to quantification with standardized units of measure then proceeds to statistical interrogation of the data, modeling of relationships and so on. The study of epidemiology provides a good analogy for what appears to be attempted in cliodynamics. For a defined base population, epidemiologists measure characteristics of the population and its environment, and analyze the…

    Read more
  13. Yoron Hamber


    Quite interesting ideas, although somehow reminding me of Asimov's 'universe'. Still, I would presume that there is more inbuilt unpredictability to what might happen, as we nowhere see any predictions for what 'should have happened' coming absolutely true. If you're thinking statistics I would ask what/where your 'cut offs' are, as they will define your answer.