Cliodynamics – Numbers that predict revolutions

Interesting article in NEW SCIENTIST looking at Cliodynamics – mathematical modelling of social events that looks for patterns from which to make predictions of the future.

Reminds me of the  section in the film Choice Point that talks about understanding patterns








PETER TURCHIN thinks he can see the future. Unlike the fortune teller you might find at a seaside carnival, he needs no crystal ball. Instead, the tools of his trade are mathematics and testable theories. Armed with these, his goal is nothing less than to revolutionise the study of history, turning it from a mass of anecdotes into a rigorous, predictive science.

Turchin calls his new discipline cliodynamics, after Clio, the classical Greek muse of history, and so far its biggest focus has been the fate of empires. Now Turchin is using patterns he has found underlying their rise and fall to make predictions of political changes to come. His forecast is alarming. If his calculations are correct, the US faces major civil unrest and political violence sometime around the end of this decade.

Critics call Turchin’s approach simplistic and naive, with some arguing that recorded history is too short to provide adequate evidence for his assertions. Turchin, meanwhile, is happy to have his theories tested – and hopes that he is wrong. If things are left to run their course, we should know quite soon. There is another option, however. We could head off any instability, Turchin says – but only if we take some unpalatable remedies now.

Turchin has not always been a bull in the china shop of history. A professor at the University of Connecticut, he is a respected mathematical ecologist with a lengthy list of influential papers on animal movements and populations behind him. “It was a midlife crisis,” he recalls. “I turned 40, and I had achieved tenure and some notoriety in population dynamics. At some point I thought, ‘Where is the challenge?'” So he started looking for a new field where he could put his formidable mathematical chops to work. “It turns out that the only science that didn’t have that already was history. The field was wide open.” So Turchin rolled up his sleeves and began the familiar process of forming hypotheses and testing them.

That was 15 years ago. Since then he has used his analytical approach to address all sorts of historical questions, including how religions spread, why empires tend to arise where steppe meets farmland, and why empires collapse. “The goal is to make history an explanatory science, which means rejecting some theories in favour of others,” he says. There are more than 200 explanations for the fall of the Roman Empire, for example, because historians keep coming up with new ideas but never cull the old ones, he says. Cliodynamics, by contrast, aims to work out which theories best fit the evidence.

Perhaps it is no surprise that his adopted discipline has not greeted him with open arms. Most historians have been trained to dig as deeply as they can into the details of a culture to understand what makes that particular time and place unique. “Historians tend to tell stories,” says Anthony Beavers, a philosopher and cognitive scientist at the University of Evansville, Indiana. “They get a cast of characters and follow them through events.” This focus on the particular understandably leaves many historians deeply suspicious of any analysis that treats individual cultures as mere data points.

A few disciplines traditionally buck this trend for subjectivity, such as economic history, which is intrinsically quantifiable. And the explosion in computing power means that, increasingly, data analysis is being used to address diverse historical questions. Fred Gibbs and Dan Cohen at George Mason University in Fairfax, Virginia, for example, plundered the online library Google Books to test the long-held belief that religious faith declined in England during Victorian times. Tracking the use of religiously charged words in the title of every book published in the UK during the 19th century, they charted a sharp fall in the use of “God” and “Christ” after about 1850, while the more neutral “Jesus”, which can refer to the historical person without the religious baggage, held relatively constant (Victorian StudiesDOI: 10.1353/vic.2011.0146). But Turchin’s approach goes beyond the analysis of trends such as these to try to pick out repeating patterns in history.

Reasoning that the fate of an empire rests ultimately on social cohesion, he has used historical records to track the prevalence of what he calls collective violence – deaths due to political assassinations, riots and civil wars, but not international wars or ordinary crimes – in three major civilisations, the Roman Republic, medieval Europe and Tsarist Russia. Applying mathematical tools borrowed from population biology, he has found that in each case deaths from collective violence follow two superimposed cycles, one spanning two to three centuries and the other about 50 years (Secular Cycles, Princeton University Press, 2009). What’s more, he thinks his data provide enough leverage to understand what drives the longer cycle.

The likeliest explanation, he says, is an idea known as demographic-structural theory, proposed two decades ago by Jack Goldstone at George Mason University in Arlington, Virginia. This argues that in a prosperous culture, population growth or advancing technology eventually leads to an oversupply of labour. That is good news for an expanding upper class who can more easily exploit an increasingly desperate labour force. Eventually, though, the society becomes so top-heavy that even some members of the elite can no longer afford the good life. Factionalism sets in as the upper classes fight among themselves, social cohesion declines, and the state begins to lose control of its citizens. Then, and only then, does widespread violence break out. Anarchy reigns until enough people fall out of the elite classes, at which point growth and prosperity can return.

A tidy story, but is it true? Fortunately for Turchin, the theory makes predictions that can be tested. In particular, it predicts that social collapse and widespread violence do not rear their heads when life first gets grim for the working classes, as you would expect if workers’ misery were the catalyst. Instead, unrest should follow a generation or two later, because it takes that long to accumulate an excess of wealthy, highly educated elites. This is exactly what Turchin found when he compared the timing of collective violence with economic indicators such as wages, social inequality and population growth – a measure of labour supply – in the three civilisations. As a further test, he looked at the dates on coins in hoards unearthed by archaeologists. Coin hoards are an excellent proxy for political unrest, since their owners must have buried them in fear during dangerous times and then experienced some misfortune that prevented them from digging them up later. Again, he found that civil war lagged behind economic hardship by a generation or two. Moreover, the same pattern holds true for the US over the past 200 years, he reports in a new paper (Journal of Peace Research, vol 4, p 577 and see diagram).

Turchin is less certain about the causes of the 50-year cycle. His best guess is that people who grow up in times of strife come to crave stability, while those who grow up in stable times are more willing to rock the boat. This leads, he thinks, to a two-generation cycle of stability and violence. “It’s not as well tested,” he says. “Take it with a grain of salt.”

Oversimplifying reality

That will not be enough to make Turchin’s ideas palatable to his critics. Detractors point out that his data can be a bit shaky even for the longer cycles. “One of the things to keep in mind is how little data we actually have. We’ve only been keeping records for 5000 years,” says Beavers. As a result, Turchin’s analysis tracks just eight complete long cycles in each of the civilisations he looked at. What’s more, there are times when his interpretations of the evidence falls short, says historian Johannes Preiser-Kapeller at the Austrian Academy of Sciences in Vienna. For example, Turchin points to a decline in population in Tang-dynasty China during the 9th century as evidence of his theory, but this decline is more apparent than real, says Preiser-Kapeller. What really happened was that the state’s power decreased so that the census failed to include people in outlying provinces.

Turchin’s cyclic theory of history also seems to leave out any role for unique events such as changes in climate, disease outbreaks or the appearance of a remarkable, history-changing individual. “The patterns are more complex, more chaotic, than the patterns created by his model,” says Preiser-Kapeller.

Turchin readily admits that his broad-brush approach oversimplifies reality, but that doesn’t worry him. “Any model, any theory, has to oversimplify things. That’s how you find the critical variables,” he says. “The question is how good is the theory at explaining and being tested by data. That’s how we know whether we’re focusing on the correct variables.” He does concede that, occasionally, history can turn on the actions of a single person. For example, military historians have calculated that Napoleon’s presence at a battle improved the odds of victory by the same amount as having 30 per cent more troops. On the other hand, he notes, many apparently unique contributions may be less dependent on individuals than they appear. If, for example, a Tunisian fruit vendor had not triggered the Arab Spring by immolating himself, some other factor would likely have precipitated the crisis.

In the end, history will prove Turchin right or wrong – and we won’t have to wait long to find out. From the start, he has argued that the cycles underlying the rise and fall of empires closely resemble the population cycles that he used to study in rodents and insects. If so, it should be possible to predict what those cycles might do, just as any decent biologist can predict whether lemmings or snowshoe hares will be scarce or abundant next year.

Two years ago, Turchin put his reputation on the line by predicting publicly that political instability in the US and western Europe will shoot up in the coming decade (Nature, vol 463, p 608). In his new paper he provides more evidence for an impending crisis in the US, where both cycles look to be approaching a peak in 2020. Allowing for some imprecision in his calculations, Turchin says that if we make it to 2030 without major turmoil he will conclude that his prediction – and hence the underlying theory – is wrong. He doesn’t think that will happen, though, and estimates that he has an 80 per cent chance of being right.

The scale of the potential unrest, although more uncertain, also concerns him. “It is easier to predict timing than the height of the peak. My feeling is that it’s going to be worse than we expect. Hopefully I’m wrong – I have to live through this.”

If Turchin is right about the things that drive the cycles, there may be ways to avert the crisis. For example, increasing tax rates on high earners should help reduce social inequality and slow the growth of elites. Turchin also suggests that the US should reduce rates of immigration – a step he finds unpalatable because he is himself an immigrant. “I’m not a xenophobe,” he says. “It’s because the theory suggests that immigrants are depressing wages and making the problem of inequality worse.”

Turchin’s third prescription may be even more controversial: he says that fewer people should get a university education, since degrees are traditionally a stepping stone into the elite classes. He notes that collective violence in Europe in the early 17th century and in pre-revolutionary Russia was closely correlated with an oversupply of graduates.

Not many historians would be willing to make such bold predictions. And Turchin has the scientist’s openness to being proven wrong. “The most convincing way to show that I’m wrong is to propose an alternative theory that fits the data better,” he says. “That’s what science is all about.”

Taking the long view of history

Many scientists are happy to remain in their ivory towers. Not Peter Turchin. He believes his scientific analysis of history holds lessons for policymakers (see main story). If they want to avert civil upheaval, he argues, they need to understand the evolution of human cooperation and conflict.

That is why he has recently become involved with the Evolution Institute, the world’s first think tank applying evolutionary principles to policy issues. Set up in 2008 by David Sloan Wilson of Binghamton University, New York, the institute is already tackling an ambitious range of areas. Projects so far include redesigning neighbourhoods and building community parks to maximise good social interactions(New Scientist, 29 August 2011, p 28); changing school environments to make better use of children’s innate ability to learn; incorporating an evolutionary perspective on human nature and behaviour into studies of economic decision-making; and studying how cultural evolution has shaped the success and failure of nation states, focusing particularly on the different traditions within the Afghanistan-Pakistan region of Asia.

“Evolution is all about the relationship between an organism and its natural environment,” says Sloan Wilson. That is why it can help us rethink the way we live.


2 thoughts on “Cliodynamics – Numbers that predict revolutions

  1. I was thinking the same thing. It does look like psychodynamics, and whilst it is surely an inexact science, the idea looks plausible. It would be interesting to apply the ideas to the reformation, the rise and fall of Nazi Germany the “great leap forward” of the cultural revolution in China, and the Arab Spring to see if these events have similar underpinnings.

    There are thing however to be aware of which make for a radically situation now to all previous generations.

    1. The widespread use of contraception and modern medicine with consequent change in the age profile of Western civilisation will certainly change the risk taking propensity of the population.

    2. The massive deployment of technology in the last 50 years with the result that the poor are living far more privileged lives in the West than was the case decades ago.

    3. Western democracies have a mechanism for removing unpopular or under-performing politicians without revolution. This was not true of any of the three kinds of society studied in detail.

    There are also potential risks specific to this time – massive global climate change leading to widespread famine, and the risk of specific local collapse of ecosystems – such as may occur if the Amazon Rain Forest were to fail due to changing weather systems (partly driven by deforestation) resulting in rising temperatures and decades long drought, or substantial sea level rise / changes to ocean and jet stream currents due to increased melting of the Greenland Icecap.

    History shows that several civilisations have failed due to natural disaster, climate change or failing ecosystems in the past. (Christmas Island, Ancient Egypt, or the Minoan civilisations to name a few.)

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s