Scientific Prediction ≠ Prophecy

By Peter Turchin April 12, 2013 6 Comments

Yesterday Wired published an article by Klint Finley, Mathematicians Predict the Future With Data From the Past. Apart from a couple of minor details Klint does a good job explaining the goals and the methods of Cliodynamics. However, he (or his editor; it is almost always editors who come up with titles) couldn’t resist injecting a bit of sensationalism by implying that Cliodynamics can predict the future. I don’t blame him – it’s part of the business they are in. But here, in my blog, where I have no editors over me and nothing to sell, I want to make it absolutely clear that


The future is not predictable, except in a most trivial sense (yeah, in 2020 the Earth will be circling around the Sun. If it’s not, we will be part of an expanding radioactive dust cloud, so the last thing I’d care about would be the failure of my ‘prediction’).

Sign up for our newsletters

I wish to receive updates from:

Cliodynamics, instead, is about understanding why and how social systems change. We look for general principles (‘laws’, if you will), and build mathematical models based on these principles. Then comes the most critical part – testing model predictions with historical data so that we can tell which models and theories are correct, and which are not. So prediction is instrumental – it is subordinated to the main goal, that of understanding. The chief purpose of mathematics is to make sure that predictions really follow logically from the premises. Otherwise, we could wrongly reject a theory, if we mistakenly test a prediction that doesn’t follow from it.

It is useful to distinguish this kind of prediction, which is subordinated to the main goal of testing theories (I’ll call it ‘scientific prediction’) from prophecy. A prophecy is an unconditional statement of what will happen in the future. For example,  ‘life on Earth will end in 2012.’

Another one is ‘the United States will collapse in 2020.’ To my great amusement, there are ‘reporters’ out there who claim that I propounded such a prophecy!

For the record: I never said it. It might happen – great empires did collapse in the past – but the probability of such an event in the next 10 years, in my opinion, is pretty low. In any case, the structural-demographic model that I have developed for the United States predicts no such thing (details in my forthcoming book on the structural-demographic analysis of American history; the current series of blogs on the dynamics of real wages describes one of the components of the model).

What the model does predict is that, given the trends of major structural-demographic variables over the past 40 years, we are due for a fairly major wave of sociopolitical violence – unless something changes. This ‘something changes’ sounds weasely, but actually the model says what needs to be done to avoid the outbreak of instability – reverse the trend of growing income inequality, moderate intraelite competition, get the state finance back into balance, and so on.

So the ‘bad news’ is that the future is unpredictable. But, as I said in a previous blog, prediction – or, rather, prophecy – is overrated. What’s the use knowing that doom is upon us, if there is nothing you can do about it? Wouldn’t it be better to understand the causes of the looming danger, so that we could take steps to avoid this undesirable future?

Other blogs on similar issues:

Does History Cycle?

Game of Predictions

Psychohistory and Cliodynamics

Science and the Art of Motorcycle Maintenance

Notes on the margin: Starting today and through the weekend I am hosting the evolutionary biologist Michael Rose. So it will be a few days before I am able to get back to the next installment of the real wages series.

Published On: April 12, 2013

Peter Turchin

Peter Turchin

Curriculum Vitae

Peter Turchin is an evolutionary anthropologist at the University of Connecticut who works in the field of historical social science that he and his colleagues call Cliodynamics. His research interests lie at the intersection of social and cultural evolution, historical macrosociology, economic history and cliometrics, mathematical modeling of long-term social processes, and the construction and analysis of historical databases. Currently he investigates a set of broad and interrelated questions. How do human societies evolve? In particular, what processes explain the evolution of ultrasociality—our capacity to cooperate in huge anonymous societies of millions? Why do we see such a staggering degree of inequality in economic performance and effectiveness of governance among nations? Turchin uses the theoretical framework of cultural multilevel selection to address these questions. Currently his main research effort is directed at coordinating the Seshat Databank project, which builds a massive historical database of cultural evolution that will enable us to empirically test theoretical predictions coming from various social evolution theories.

Turchin has published 200 articles in peer-reviewed journals, including a dozen in Nature, Science, and PNAS. His publications are frequently cited and in 2004 he was designated as “Highly cited researcher” by Turchin has authored seven books. His most recent book is Ultrasociety: How 10,000 Years of War Made Humans the Greatest Cooperators on Earth (Beresta Books, 2016).

Sign up for our newsletters

I wish to receive updates from:


  • karlfrost says:

    As Pete Richerson and Rob Boyd have pointed out in an article that addressed a point of agreement between some postmodernists and some social scientists, one of the ‘predictions’ of cultural evolution models is that future human behavior will in important ways be unpredictable (Boyd & Richerson, 1992).

    Boyd, R., & Richerson, P. J. (1992). How microevolutionary processes give rise to history. History and evolution, 179–209.

  • Peter Turchin says:

    It’s a ‘metaprediction’! (that prediction is impossible)

  • Peter says:

    Thanks for the insightful post and interesting work. Do I understand correctly that the goal of cliodynamics is more to understand what the effects a policy would have in different environments? Not sure if I’m wording it clearly, but something like being able to create a tree of probability distributions for social outcomes where each branch would be a particular government policy conditional on some environmental effect, and this could then be simulated to arbitrary depth (e.g. time from present)? Something like that would be immensely useful to policy makers in this increasingly complex world.

    • Peter Turchin says:

      That’s close to what I envision as the chief goal. An important aspect of it is to be able not only to calculate the direct consequences of a particular policy proposal, but also indirect effects resulting from feedback loops. Those ‘unintended consequences’ can really kill you.

      We are not there yet, by any means, but this at is a worthy – and realistic – goal.

  • Emulator says:

    I wouldn’t subordinate predictions to understanding. Quite the opposite.

    I would argue that science is about making predictions. Understanding is important precisely because it allows us to make better predictions. For example, by understanding how electrical components and circuits work, we can make predictions such as “if I put this and this electrical component together and connect it to a diaphragm, it would convert acoustic waves into electromagnetic waves, and we can do the same thing except in reverse by putting together these electrical components and a diaphragm in so and so arrangement.” In this case, the predictive power gained form our understanding of electrical circuits allowed us to invent radio communication.

    Of course, Cliodynamics is not the same in that it is much harder to directly control society than it is to control electrical engineering. Cliodynamics is more like astrophysics. In astrophysics, we can’t directly use predictions to invent things, because astrophysical inventions are well beyond our engineering capability. However, astrophysics serves to give us back insights into basic physical theories and science, which in turn are more directly applicable to everyday life. I can see Cliodynamics doing the same thing in terms of providing more feedback as to the real life validity of game theoretical and group-centric sociological models.

    Finally, I’d argue that you shortchange yourself with the caveat about “something changes”. I don’t see how something could change. The truth is that the psychology of the elite is to take more and more, and to compete with each other more and more. Sure, if this changed, then we would not have a crisis. But the fact is that it won’t.

  • Peter Turchin says:

    I am with you on most of what you say. My primary beef is about making predictions about the future. I talk about different meanings of ‘prediction’ in this paper:

    To use an analogy, we cannot predict weather more than 7-10 days ahead (and not even that in the part of the world where I live). But we have a pretty good understanding of why global climate has been getting warmer. And what needs to be done about it, if we want to reverse it. That’s the goal that I see for Cliodynamics – useful and realistic (as our theories and modesl get better).

Leave a Reply