About Rationally Speaking

Rationally Speaking is a blog maintained by Prof. Massimo Pigliucci, a philosopher at the City University of New York. The blog reflects the Enlightenment figure Marquis de Condorcet's idea of what a public intellectual (yes, we know, that's such a bad word) ought to be: someone who devotes himself to "the tracking down of prejudices in the hiding places where priests, the schools, the government, and all long-established institutions had gathered and protected them." You're welcome. Please notice that the contents of this blog can be reprinted under the standard Creative Commons license.

Monday, August 04, 2008

Cliodynamics, a science of history?

My colleague Peter Turchin over at the University of Connecticut (my Alma Mater) has recently published an intriguing short article in Nature (3 July 2008) on what he termed “cliodynamics,” the possibility of turning history into a science. The word comes from Clio, the muse of history for the Greek and Romans, with the “dynamics” part referring to the central concept proposed by Turchin, that history -- contrary to what most historians might think -- is not just one damn thing after another, that there are regular and predictable patterns, from which we can learn and that we can predict.

It’s a big claim, and one that is bound to generate little enthusiasm among scientists and positive distrust among historians. For the first group, history is the quintessential mine field, where contingency and human agency rule the day, unlike the tidy behavior of subatomic particles, always the same under easily imposed identical conditions. As for historians, this will be seen as yet another arrogant attempt by a scientist to colonize their field and push aside the humanities (despite Turchin’s claim of potential unification of science and the humanities).

Turchin complains that there are more than 200 explanations proposed for the collapse of the Roman empire, a situation he finds “as risible as if, in physics, phlogiston theory and thermodynamics coexisted on equal terms.” Hmm, but then again, there are hundreds of different types of string theory, and none of them is, at the moment, empirically testable... Nonetheless, Turchin goes on to explain that there are, in fact, regularities, in human history. For instance, with two of his colleagues, Turchin found a statistically significant trend (statistics applied to history!) across various societies, according to which “the number of instability events per decade is always several times higher when the population was declining then when it was increasing.” This result was obtained by studying societies and time periods as different as the Roman Empire and eight Chinese dynasties.

Of course, this is not the first attempt to inject some science into the study of history. Most famously, another biologist, Jared Diamond, has provided us with empirically-based theories of the rise and fall of civilizations in his books, Guns, Germs and Steel and Collapse. Biologists are particularly well positioned for this sort of cross-disciplinary work because biology itself is largely historical in nature, especially evolutionary biology (Diamond’s specialty) and ecology (Turchin’s).

While I am sympathetic to the idea of moving history from a purely narrative-based discipline to one that relies on some degree of data collection, pattern analysis, and most importantly empirical hypothesis testing, I also find that scientists seem to chronically underestimate the problems posed by historicity in their own discipline, let alone in history itself. Diamond’s scenarios are very compelling, but they are far from the only explanation of why the Fertile Crescent, for instance, gave us so many early advances in civilization while meso-America lagged behind. This is not very different from the continuing debate, say, among paleontologists about why the dinosaurs went extinct: pretty much everyone agrees that the impact of an extraterrestrial body had something to do with it, but it is also clear that the dinosaurs had been in decline for millions of years before, and we may never know exactly why.

Turchin’s own example in the Nature article is really confined to uncovering a pattern, not testing an explanation. Pattern discovery is of course crucial, but Turchin himself admits that “the connection between population dynamics and instability is indirect, mediated by the long-term effects of population growth on social structures,” going on to list a good number of interacting causes that may underlie the pattern he uncovered. But how is one going to test a multi-causal, multi-level hypothesis with a chronic paucity of comparable data? Again, the situation is similar to the discussions among paleontologists about what causes mass extinctions, where people disagree even on the number of such events during earth’s history, and where the number of occurrences is barely sufficient to reach statistical significance as a pattern, let alone to provide enough discriminatory power among complex causal hypotheses.

Nonetheless, the proof is in the pudding, so to speak, so I wish Turchin and his cliodynamics much luck, and I will follow with interest whatever else he, Diamond and colleagues may turn up over the next few years. Now, any historian out there who would like to tell us why this is so wrongheaded as not to be worth even trying?


  1. This is a very intriguing concept indeed. Though, the major problem would be actually finding data that could be tested...It is possible that history is simply too subjective of a field to employ truly scientific research.

  2. Isaac Asimov proposed the idea in his Foundation series of sci-fi novels. I think he termed it psychohistory. It was a central premise of those books.

  3. This is hardly a *new* idea. Philosopher R.G. Collingwood wrote extensively about the ways in which history is scientific. In fact, a comprehensive review/analysis of Collingwood's philosophy of history (published in 1981) was titled, appropriately enough, History as a Science: The Philosophy of R.G. Collingwood.

  4. thinkmonkey,

    yes, but Turchin is actually trying to do science in history, which goes further than a philosophical analysis (as interesting as the latter is).

    Of course, it remains to be seen to what extent he'll succeed.

  5. If the Butterfly Effect can trip up such endeavors, surely all the Black Swans flapping around in our rapidly changing, technological world make such historical prognostication a losing batte?

  6. [This just in from Peter Turchin, the author of said Nature article]


    good to hear from you, and I am glad you found my Nature essay thought-provoking. Yes, the proof is in the pudding. There is a book co-authored with Sergey Nefedov, where we attempt to demonstrate how one could test dynamic theories with historical data. If you are willing to wade through 300 pages of text and c.150 graphs and tables, you can examine yourself the evidential basis for our claims. The book will be published by the Princeton University Press next year, but the final version is already available on my "cliodynamics" web site:


    Peter Turchin

  7. D'oh! I was all set to comment that cliodynamics sounded like a sort of reverse psychohistory, but I see agent moselle already beat me to it.

  8. Jeez Man, Massimo, and the rest of you, where have you been?

    There has been a empirically based science of history, its called anthropological archaeology!

    (As opposed to archaeology as an appendage to Classics and history.)

    "Nonetheless, Turchin goes on to explain that there are, in fact, regularities, in human history."

    Oh no shit! Parallel and divergent paths of socio-political and cultural evolution are issues that are regularly discussed in anthropological archaeology. Turchin has made no big breakthrough there!

    Many of the big questions explored in archaeology have been how and why do some people in some times and places domesticate crops and become agriculturalists, how some come to form village based societies, while others remains nomadic hunter/gatherers or pastoralists, how states and civilizations arise and collapse? What are the origins of social inequality? What are the variations and generalities of socio-political organization etc.?

    Back in the late 60 and 70s the trend in anthro-archaeology was to emphasize so-called "prime movers", to answer those questions. Then the field started to emphasize more multi-variable causality, and to emphasize variation in contrast to generalization.

    And a note about Diamond, while much of his work and focus is appreciated to a certain extent in archaeology, he is generally considered to often oversimplify his case studies, and sometimes he gets important details about prehistoric societies like Easter Island, the Maya, and Anasazi in the U.S. southwest flat out wrong!

    There recently was a conference on Diamond's work by some top scholars in the field, and an edited volume is forthcoming.

    Anyway, as a starting point, you might look into the work of archaeologists like Kent Flannery and his partner Joyce Marcus, who have done alot to develop what they called a cultural-ecology approach to socio political evolution. They have a popular book synthesizing theirs and other's work on the cultural evolution in the valley of Oaxaca Mexico from plant domestication, sedentism, to the formation and collapese of ancient states.

    "...but they are far from the only explanation of why the Fertile Crescent, for instance, gave us so many early advances in civilization while meso-America lagged behind."

    And this is really a non-issue if there ever was one Massimo. There was no real "lag". People were just concentrated in the Old World since much earlier times than the Americas. Aggregating populations, and just getting the process of agriculture, village, and state formation started a little bit earlier.

    Besides, the earliest complex urban societies in the Americas developed on the coast of Peru, and were inhabited at the same time as Egypt's pyramids were being buillt.

    I have heard of this guy Turchin cited through a paper by a Dr. Timothy Kohler and collegues of Crow Canyon Archaeological Center, who are affiliated with what is called "The Village Project" of southwestern Colorado (where I used to work).

    Check it out and learn about what kind of research archaeologists are doing instead of acting like they don't exist and there is no science of human history!

  9. Sheldon,

    thanks for the detailed comment. However, I do think that Turchin is talking about something different: he is attempting to bring cross-historical periods statistical analysis to history, which I don't think is what you mean by anthropological archeology.

    "There was no real "lag". People were just concentrated in the Old World since much earlier times than the Americas. Aggregating populations, and just getting the process of agriculture, village, and state formation started a little bit earlier."

    I don't buy it, by that reasoning the Fertile Crescent should have been in Kenya...

  10. [This is a follow up comment by Peter Turchi]

    In response to the post by Sheldon:

    First, I have a lot of respect for archaeologists and anthropologists, I have co-authored papers with them and cite their work. The format of the Nature Essay did not allow me to express this deep appreciation, but the evidence is in my publications posted at the "cliodynamics" web site.

    Nevertheless, as far as I know, anthropologists have not advanced very far to answering the question I raised in the essay, which of the theories explaining the fall of Rome (and the collapse of territorial empires, in general) should be rejected, and which are supported by data? The main problem, I believe, is that archaeology has not advanced far enough along the mathematization path (yes, there is a volume edited by Renfrew in the 1980s and I know of several current modeling initiatives, but not a whole lot beside that).

    A science cannot mature until it is "mathematized", because mathematics provides a formalized language for precise description and rigor ensuring that conclusions indeed follow from premises. Mathematics is not just about quantities. Some of its branches study qualitative statements (mathematical logic) or structure (abstract algebra). However, if we are interested in understanding the dynamics of such historical processes as population change, territorial expansion and contraction, and the spread of religions, we must get involved with numbers and rates. A human mind unaided by mathematical formalism is a poor tool for predicting dynamical processes characterized by nonlinear feedbacks and capable of such complex behaviors as mathematical chaos. Without mathematics (understood broadly) we are doomed to make vague statements and to arrive at wrong conclusions. How can we test theoretical predictions with data, if we are not even sure that the prediction to be tested in fact follows from the theory's premises?

    So what is at issue is not being empirically based (many historians are very good empiricists), but working at the interface between mathematics and data.

    The paper by Tim Kohler that Sheldon mentions approvingly is a wonderful illustration of the value of bringin math in. The history of it is as follows. Tim read a preprint of a paper by myself and Andrey Korotayev (now published in Social Evolution and History), the text is here:


    In this paper we advanced several mathematical models for the dynamical interactions beween population pressure and warfare. The main result is that we should expect not direct, but lagged correlation between the two variables. Tim and co-workers then tested model predictions with their wonderful data (for example, they quantified the degree of warfare by looking at skeletal injuries). So this is precisely an example of how an abstract mathematical model can motivate empirical research and result in very fruitful advance.

    I understand the exasperation that practitioners in a scientific field must feel when outsiders intrude on their turf. But I propose that instead of quarelling, we can work together. The best results are achieved by truly interdisciplinary (or, as they say at the Santa Fe Institute, "transdisciplinary") research. I have a good model for this. My former field, population ecology, profited enormously from the intrusion by such outsiders as Lotka, Volterra, May, and Levin.

  11. Hmm, I must say that I do not share Peter's confidence that mathematization is the way to go with every science. Indeed, the philosophical literature is full of good reasons to suspect that mathematics has strict limits that derive from it being a branch of logic (and philosophers have long abandoned the hubristic project of being able to derive sure knowledge about the world through deduction).

    This, of course, is not to say that mathematics doesn't play a role in much science, it's just meant as a note of caution. Moreover, there is a distinction between quantification (as in statistical analyses of patterns) and mathematical causal modeling, the latter, I think, being less and less useful when problems become more and more complex.

    Finally, contrary to what Peter seems to imply, the hallmark of science is empirical hypothesis testing, which may be aided by, but it is certainly not confined to, mathematics-driven theory.

  12. For some reason, Turchin's comment above about the Fall of Rome made me think of Kevin Phillips, and the recurring them in his 5 to 6 books about the parallels in the collapses various republics and empires and the comparison to the current state of the U.S. (You can see what I'm talking about in these excerpts.) Of course, I'm still thinking in terms of Asimov's sci-fi take on the reverse concept I suppose.

  13. "... recurring theme in his last 5 to 6 books ..." that should have read.

  14. "I don't buy it, by that reasoning the Fertile Crescent should have been in Kenya..."

    Regarding "Mesoamerican lagging behind Mesopotamia", admittedly my comment was a quik and sloppy response to what I percieved to be a quik and sloppy question. And then you return with this silly response.

    While the evolution of ancient states shouldn't be thought of as a race, it is important to understand where the base lines are for what made that evolution possible. Of course no ancient civilization developed prior the evolution of anatomically modern humans sometime around 150,000 years ago, or prior to the Upper Paleolithic revolution 50,000 years ago, or until after the end of the Pleistocene at around 12,000 years ago. So that basically takes out of consideration the issue of why early hominids didn't organize themselves into complex societies in Kenya.

    Better and more important questions to ask are how and why did the first ancient states develop in particular regions, and not others? I think the answers revolve around many variables of geography and demography that allowed that evolution to take place.. These variables came together in particular ways and particular times in Mesopotamia, Egypt, the Indus Valley, China, Mesoamerica, and Andean South America.

    So again, I really think its an empty and uninteresting question, with false premises and assumptions to ask, why did Mesoamerica "lag" behind Mesopotamia? Its almost like asking why George Washington became president before Ronald Reagan, or why mammals lagged behind early birds in the evolution of flight.

    At the end of the Pleistocene in the Old World, populations in regions amenable to the formation of complex societies, with a higher relative population densities and sedentism were in place earlier. In the New World, there was still two vast continents to settle, thus there naturally was a time "lag" until those many variable conditions were right for the development of complex societies.

  15. Sheldon,

    I really don't think my questions are silly at all, nor are Diamond's answers (they may or may not be correct, but they ain't silly).

    Of course civilization is not a race, but one still has to ask why certain things happened in certain places rather than others, despite ample time available (meso-America was colonized thousands of years ago).

    You say: "Better and more important questions to ask are how and why did the first ancient states develop in particular regions, and not others? I think the answers revolve around many variables of geography and demography that allowed that evolution to take place.. These variables came together in particular ways and particular times in Mesopotamia, Egypt, the Indus Valley, China, Mesoamerica, and Andean South America."

    That sounds to me like the sort of answer that Diamond articulated in his book, no?

  16. Posting this before reading Massimo's latest comment. Maybe I will regret it?

    Massimo and Turchin,
    My initial negative comment was a reaction to what I thought was Massimo's suggestion that there had been no previous scientific research into the patterns of human history. And there has been in the discipline of anthro-archaeology. At least according to your (Massimo's) own definition based on hypothesis testing against empirical data.

    I thought Massimo gives too much credit to Diamond's parasitic, secondary, non-peer reviewed research. For Diamond, the issue is not that he is an "outsider", it is that he takes the rich complex results of archaeologists and reduces it into his rather oversimplified package for public consumption. A good story with some basis in fact for sure, but not necessarily the most accurate one. At best Diamond is a synthesizer for public consumption, which is good on one level, and I think he raises important issues for consideration to a broader public.

    Turchin may indeed have a productive line of research. I checked his web site, and would like to look at some of other work. Specifically " Historical Dynamics: Why States Rise and Fall." My dissertation proposal from years ago, (which probably will never come to fruition, but that is another story), was concerned with the growth and collapse of the Wari empire of Peru. If I could ever get back to that project I might be very interested in looking at and applying Turchin's work. Anyhoo....

    Archaeology has always been an inter-disciplinary field borrowing from all kinds of other disciplines. Archaeology would be no-where without chemistry and physics (C14 dating) geology (stratigraphy) and many contributions from biological related fields (botany, zoology, ecology, osteology etc.). Archaeology also borrows ideas and theories from nearly all the social sciences and humanities. So my comments were not concerned with defending archaeology from encroachment from outsiders, but only in thinking it should be given some credit for its own unique and often ignored contributions to knowledge.

    And regarding various theories of the fall of the Roman Empire. Could it be that there are so many because it has been Classcist historians and archaeologist who have studied the topic of Rome? The archaeologist who primarily study ancient Greece and Rome are much more humanities orienteted than the archaeology found in anthro. departments.

  17. Massimo,

    I didn't say the question of "lagging" was silly, or that Diamond's answers were silly. Just not relevant for comparing Mesoamerica to Mesopotamia in my opinion. I think the question of "lagging" could certainly be relevant for other cases. Such as why Europe industrialized ahead of the rest of much or the world for example.

    Heck, you are much smarter than I, thats why I habitually read your blog! Hello! Long time fan here! :)

    What I said was silly was your first response to my comment about the duration of human settlement and adaptation of modern Homo s.s. between hemispheres. What it seemed to imply was that if I had any point at all, then why didn't Homo habilis or Homo erectus organize themselves into ancient states?

    And I don't think Diamond is completely wrong, probably right on alot of stuff, and definitely not silly.

    But when I read through some of Collapse last year, particularly the chapters dealing with the Maya and the U.S. Southwest, where I know some of the primary literature, I can't help but think he oversimplifies alot.

    And now I go back to read your post, and you suggest that as well. So my apologies.

    I guess what I was reacting to is what seemed to be a near total unawareness on your part about the work and debate that goes on in anthro-archaeology. Such as when you say:

    "While I am sympathetic to the idea of moving history from a purely narrative-based discipline to one that relies on some degree of data collection, pattern analysis, and most importantly empirical hypothesis testing,..."

    This is pretty much what archaeology has been doing since the early 1960s with the "New Archaeology"

    Of course given that you work in both biology and philosophy, which is quite a full plate, you can be forgiven for not knowing about everything! :)

    So apologies for ranting!

  18. Sheldon,

    most definitely no need to apologize, especially from such an old time 'fan'! I for one learned interesting stuff from these exchanges, which is one of the reasons I blog to begin with...

  19. Whenever I hear about the work that Diamond and others are doing in trying to make history and other humanities more scientific I find myself wondering about a couple of things. The first is to what degree the thing that's been stopping these developments has not been the lack of tools but just attitudes. It seems to me quite clearly the case that what moiz says about a lack of data has to be wrong, for example. Historians have been collecting mountains of data but they have not been putting it together with 'a scientist's eye'. This brings me to the second thing I think about and that is that historians etc. are wrong to worry about a scientific history etc. becoming shallow by trying to make history become like physics. The best example to think about is biology. Rather than biology becoming like physics, the scientists who investigated biological phenomena had to change their methods to suit the data they were dealing with. The same will happen when scientists come to fully engage with history. A big part will be the development of the appropriate methodological tools. In other words, history will not become physics but scientists will become historians. Treating data seriously will force them to.

  20. Konrad,

    What do you mean by mountains of data? It would seem to me that of course there would be data, but would it be strong enough to survive scientific testing?

    Take for example the fall of the Roman Empire (since it seems to be a common theme on this post)...what kind of data that has already been collected, that if viewed with a "scientific eye" would bring about such answers? It would seem to me most of it would be at the level of correlation, and not nearly enough to imply causation.

    Science demands a certain kind of evidence that I believe history falls quite short on delivering.

  21. "That sounds to me like the sort of answer that Diamond articulated in his book, no?"

    For Guns, Germs and Steel, yes, which I just went out to pick up from the library, and I have been skimming tonight.

    I think my biggest complaint about Diamond came from Collapse where he vastly oversimplifies the processes of regional abandonment, in the southwest. He basically took the conclusions of the archaeology in the region from the 1980s, and ignored the much more nuanced picture constructed in the last 15 years or so. I don't have Collapse at hand to discuss the specifics.

    I think I agree with your skeptecism about the thinking that that mathmetization is some magic bullet to turning soft social sciences into hard ones. They call it physics envy don't they?

  22. [more comments from Peter Turchin]

    I agree with both points that Konrad makes. A good example of a physicist-turned-biologist is Robert May. The same thing is happenning to natural scientists who turn to history. For example, former physicist Bertrand Roehner (check out his book Repertoire and Pattern).

    As to the mountauins of data that is literally true. Naturally, there are problems - measurement errors, gaps, etc, but nothing qualitatively different from what I had to deal with when I worked in population dynamics. For more recent periods, look at cliometrics as practiced by such titans as Robert Fogel. For medieval and ancient periods we also have lots of data, but there we often need to employ proxies (indirect indicators). In the already mentioned article by Kohler et al, they used the frequency of diagnostic skeletal injuries as a proxy for the intensity of warfare. Another excellent and widely available proxy for political instability is the frequency of coin hoards. (Because people bury treasure in times of danger, and don't recover it if they are killed, or enslaved, etc).

    A great proxy for the population pressure on resources is the average height. An NSF-funded project is measuring and dating the c.2 million of skeletons for Western Europe from the Iron Age to the nineteenth century. And these are just two examples. I invite Moiz to look at the final version of our book with Sergey Nefedov, posted here:


    Just skim through the c.150 figures and table to gain appreciation of what can be done with historical data. Incidentally, there are two chapter on Roman history there.

  23. Very interesting comments from Turchin. As a history major myself, this could very well become an interesting breakthrough.

  24. From chapter 10:

    The main goal of this book has been to determine how well the predictions of the demographic-structural theory map onto empirically observed patterns in the studied historical societies.

    Note the use of the word "prediction," typically associated with scientific endeavors. Note that this word is being applied to the process sometimes known as hindcasting. Note the utter lack of out-of-sample data - at least until we decipher the Inca quipu, or discover the lost chronicles of Atlantis, or something. Professor Turchin is not predicting, he is retrodicting. Somewhat easier.

    Professor Turchin starts by reciting the fact, known to all bright eleven-year-olds since well before Jesus was a little boy (oddly, his brief historiography dotes on Ibn Khaldun but does not even mention the leading classical exponent of anakyklosis, Polybius), that empires in the past seem to, um, rise and fall. Obviously we don't seem to be making our tax checks out to Pharoah Khufu DCCXVII. So something must be going on.

    He then proposes a simplistic, Annales-style and generally Malthusian-Marxish, agricultural-productivity model for this cycle. This enables him to write a 300-page book in which each chapter recounts the rise and fall of an empire. Sure enough, each empire is healthy in its early, rising stages, then goes to shit at the end. Model verified!

    There is nothing in Professor Turchin's methodology that, observing a symptom of imperial cycles, can tell us whether it is a cause of imperial cycles or an effect of imperial cycles. In other words, there is nothing in his methodology that could conceivably validate his model. As far as I can tell, causality is demonstrated by sheer mathematical intimidation.

    This is not science, which makes falsifiable predictions which work for out-of-sample data. It is not history, which tells the story of the past. (Professor Turchin's version of the rise and fall of empires, totaled up as a sort of accounting exercise, makes Carlyle's Dryasdust look like Ernst Gombrich.)

    For grand theories of history, give me Marx, Spengler or Toynbee any day. Even Spengler's Jungian mysticism is a vast improvement on Professor Turchin's differential equations - the former, at least, has the basic intellectual decency to smell like what it is.

  25. Dear Prof. Pugliucci,

    Searching about Cliodynamics, I found your blog, linked directly from Professor Turchin's page.

    It's regrettable that a potentially interesting and useful debate was cut short, due to Mencius Moldbug's insolence. I found the observations by some posters, among whom Konrad Talmont-Kaminski, highly pertinent.

    Although I am personally interested in the subject, I myself am slightly skeptical about the feasibility of Prof. Turchin's ideas, and in fact to a large degree I share your observations about the possibility and usefulness of a mathematical history (as opposed to a scientific history).

    The recent experience with economics and other disciplines that have embraced rational choice acritically, for one, could provide a useful warning to Prof. Turchin.

    Having said that, I do believe Prof. Turchin has interesting things to say and could make a positive contribution.

    In fact, once discounted Moldbug's uncalled-for rudeness, he/she did point to a couple of reasonable questions: it is known that empires rise and fall. What is the "value-added" by Cliodynamics to this? In other words: what additional insight is provided by Cliodynamics?

    More generally, Marx provided a narrative of the dynamics of modes of production, whatever the failings his narrative had. Why should one accept Cliodynamics instead of a Marxian view of history, for example?

  26. As a naïve student of History (i'm in the second semester) I would ask first: What does a discipline needs to deserve been called a science?
    If the answer is: Using the scientific method, then I would say that history is a science. I can't say that something happens in the past, if I don't have sources to support my view points.

    But maybe the necessary thing to be a science is to do good predictions. Then, in order for History to be a science, we historians, and other scientist are of course welcome, need to work hard to develop a testable theory for predictions.

    But probably the answer is way more complicated.

    Just for the record, I love maths, Physics, Biology and Chemistry, so I would love for History to be a science.


Note: Only a member of this blog may post a comment.