Saturday, August 18, 2012

Scienciness and History: Cliodynamics

There seems to be an uptick in historical studies by scientists (henceforth Scientists-Doing-History or SDH). I guess everyone wants to be the next Jared Diamond.  As a recent example, a couple of physicists (Mac Carron and Kenna 2012) have published a study comparing networks in myths (Beowulf, the Iliad, and the Táin Bó Cuailnge) and fiction (Les Misérables, Richard III, The Fellowship of the  Ring, and Harry Potter). It got worked over in detail by Richard Carrier at Freethought Blogs (Bad Science Proves Demigods Exist!) and more generally by Maria Konnikova at Scientific American (Humanities aren’t a science. Stop treating them like one). I feel a little bad for the authors.  It was a small study in a physics journal and could have stayed buried there, but it may have become a lightning rod for irate humanities researchers. 

There have also been some higher profile efforts to bring scienciness to bear on world history. Jared Diamond and Guns, Germs and Steel (1997) is obviously the success story here. Now Steven Pinker (2011) has published a large trade press book on violence and human nature, The Better Angels of our Nature, my encounter with which has left me with a nervous twitch. I’m trying to write up something on that one, but it is hard to know where to begin. Another aspirant is Peter Turchin, another evolutionary biologist who has decided to apply himself to history, even coining a neologism, “cliodynamics.”  I’m a little disappointed he didn’t call it clionomics  (although we do, thank god, have “culturomics,” which, if nothing else, gave us the fun Google N-gram Viewer).

Turchin has been plugging cliodynamics for a while, but he just published a study in the Journal of Peace Research on “The dynamics of political instability in the United States, 1780-2010” (Turchin 2012). Nature had a piece on it by Laura Spinney, and it also broke through into the mass media  with a piece at MSNBC, probably because Turchin excitingly predicts a violent upheaval in 2020.  Maria Konnikova at Literally psyched criticised it also in the aforementioned post, and Rebekah Higgitt at the Guardian and Paul Lay at History Today have also weighed in.  Rebekah Higgitt had the best subtitle, one that perfectly sums up the exasperation of many historians at the Scientists-Doing History” genre--”At least one historical prediction holds true: the regular appearance of claims to have the key to 'scientific history'.”

To be clear, when I refer to history and historians here, I am referring to the disciplines that study the human past, whether they do so from an anthropological or historical perspective, so I am scandalously folding archaeology into history.  I couldn’t keep writing “historical disciplines” and I wasn’t going to write “historical disciplinarians.”

I am not quite sure what cliodynamics is.  At its most general it seems to be synomous with quantitative approaches to “big history” and at its most specific it is Turchin’s particular approach.  Making the assumption that cliodynamics is more than a rebranding exercise, I use the term to refer to Turchin’s particular approach— a combo of using quantitiative methods to empirically identify historical cycles, and explaining those cycles in terms of “structural-demographic theory.”

The big contribution of SDH is the idea one can solve history simply by slapping down the correct method on the nearest datasets. I don't think this works. You can’t just shovel numbers into your statistical/ideological hobbyhorse and hope for something worthwhile to come out the other end.  You need to understand the data you are using. Minimally you need to understand how the numbers were created. If any data is “socially constructed” it is historical data. I imagine understanding one’s data is good practice in science as well as history. It is this point that Richard Carrier emphasized in his critique of Mac Carron and Kenna. This attitude apparently led to at least one howler in Turchin’s book Historical Dynamics: Why States Rise and Fall (Tainter 2004). He used two peaks in archaeological settlement data from Roman Gaul as population oscillations, not realising that the first corresponded to Romanization and the establishment of colonia, and the second to changes in taxation under Diocletian.

These contrasting explanations (a spike in archaeological site frequencies as a result of secular cycles vs. as a result of conquest and settlement) bring us to the “D” word—determinism. If you are a SDH, you’ve got a fever, and the only prescription is deterministic monocausal explanations. Why? The argument seems to be because superficially similar phenomena must have identical explanations.  Maybe only humanities majors worry about things like affirming the consequent, equifinality, and contingency. Logical problems aside, the key divide seems to be one of determinism vs. contingency. In Guns, Germs, and Steel Jared Diamond defined contingency away by arguing that the “proximate” contingent causes were themselves caused by “ultimate” deterministic causes (since he is a biogeographer, his “ultimate causes are geographical ones). The contingent causes exist, but they aren’t important. 

I am not sure if Turchin ever explictly addressed critiques of his work and thus how he deals with historically-specific alternative explanations for his historical cycles, but I would suspect a similar tactic. For example, in the Peace Studies article he sees immigration in to the US as the fundamental cause of political instability during that period (by driving up labor oversupply). Personally, I think you need to not know an awful lot about US history and the dynamics of migration to make an argument like that. But for Turchin, the concrete arguments (boom and bust economic cycles, the growth of industrial capitalism, etc.) are probably just "proximate" causes.  The ultimate cause?  Population growth through immigration.  As an aside, given the current nativist climate, this is probably an argument that will, unfortunately, garner a great deal of attention—science proves the need for immigration crackdowns. 

Turchin’s current article is better than many of this genre in one regard, possibly because he may taking some of the criticism of his previous work seriously, if not addressing it explicitly. Here he deserves credit for using his cliodynamical hammer on a single case, using a relatively consistent data set, and one that he gathered himself. He does seem to be a more aware that there are methodological issues in gathering and classifying data as well as just analysing it.  This stands in contrast to Steven Pinker’s credulous reliance on dimly understood secondary sources in Better Angels. The problems may still be there with Turchin’s work, but but by focusing on a single case with a good dataset he at least allows one to get a handle on the problems, rather than having to ramble backwards and forwards over multiple areas, periods, and specialties.  When the evidence is cherry-picked from the entirety of world history, it is a little harder to engage in substantive critique. At the very worst, Turchin has reduced the problems to a manageable scale.

By taking this more focused approach Turchin may be able to kick his cliodynamical research program out of first gear, which is where it seems to be stuck. I am not sure though. Cliodynamics still seems to have a tight citation circle, which is not a sign of a particularly healthy research program. And by "tight", I mean really tight. The degree to which Turchin cites himself in this article is positively embarrassing, especially given how well-worked his historical ground is.  It does indicate a disturbing insularity and lack of engagement with the relevant historical literature.

Another danger signal is that Turchin’s approach just seems to be to do the same study over and over repeatedly with different datasets (i.e., cultures).  I suppose the ultimate idea is accumulate a big pile of cases where cliodynamics does work, and let everybody generalize from that. Of course, he might still be in the exploratory phase, seeing if it works, rather than trying to confirm it systematically. But in the mean time it does leave us with the problem of trying to assess a deterministic hypothesis using only judgmentally-selected evidence. At some point the much-touted science needs to kick in, and we need to see some systematic testing.

Which brings us to the big question, the one all archaeologists and historians dread about their research—”why should I care?” Does cliodynamics tell us something interesting? What exactly does Turchin (or any other SDH) bring to the table? In Turchin’s case, it really doesn’t look like much— (1) quantification, which identifies (2) cycles, which are explained by (3) “structural-demographic theory.” 

(1) Quantification
There is nothing new here.  The historical disciplines had their big quantitative/scientific awakening back in the70s and early 80s. The polemics, the scientism, the pie-eyed enthusiasm…it’s all there in Turchin. Reading Turchin on quantifying history is like having a mullet again.  Let's get one thing straight--historians use quantitative methods.  The idea that they don’t is sheer nonsense.  Admittedly they are cautious about it, but that is the result of hard-learned lessons. Researchers soon ran headlong into the complexity of the datasets and their subjects (humans and societies). I covered this earlier. You really need to be sure that the numbers you are using refer to the same historical entity or process and are comparable.

(2) Cycles
The observation that there is are cyclical components to human history is certainly true, but I am not sure the observation is that interesting, in the sense of being particularly novel. For example in her book Time and the Shape of History, Penelope Corfield (2207) devotes a chapter (”Shaping History—Time Cycles”) to this topic. We take for granted that there are there are economic cycles for one (Turchin’s 50-year secular cycles brought to mind Kondratieff cycles—I wonder if there is a tradition of cyclical history in the Russian academy).  States are unstable, and have cycles of  expansion and collapse. Environments have cycles.  These things lead to cyclicality in other domains. There nothing new there. The novelty of cliodynamics is that Turchin argues that the cyclical nature of states is chronologically regular and therefore has a unitary explanation.   Which brings us to...

(3) Structual-demographic theory
Structural-demographic theory “postulates that labor oversupply leads to falling living standards and elite overproduction, and those, in turn, cause a wave of prolonged and intense sociopolitical instability." (Turchin 2012:578)” Yes, population pressure.  The distinctive feature here is that instability is driven only by competition between elite factions, rather than by factors such as economic cycles, class struggle, or external military pressures.  The emphasis on population is unsurprising from someone whose previous work was in population ecology.  Before moving into history, Jared Diamond was a biogeographer. Go figure. When you have a hammer, everything looks like a nail. 

I am not sure what it is Turchin is actually pushing—quantitative methods, cycles, or structural demographic theory.  My sense is he has seized on structural-demographic theory as the theory he needs to justify his cycles. It is awkwardly tacked on the end and the logic is questionable.  It boils down to the idea that
(1) the empirical regularity of cyclicality necessitates a single deterministic explanation.
(2) Structural-demographic theory predicts cycles (as, unfortunately do many other theories),
(3) therefore structural-demographic theory.

Not satisfying. 

Right now, the high profile “big history” is being done by people outside the historical disciplines, all of them claiming the mantle of science, all of them with simplistic deterministic explanations, and all of them reaching wildly different conclusions. Peter Turchin proves the ultimate cause is population, Jared Diamond proves it is geography, Steven Pinker proves it is Reason and the Enlightenment.  And all of it is Science.

The relationship between science and history is not simple. I do agree we have reached the stage where we can really start looking at large-scale and global processes, and careful quantification is part of that. The datasets are improving and we are developing the technology and understanding to make this possible.  But it is going to be a long difficult job, and it won’t result in the simplistic, easily digested narratives of the current crop of “scientific history”.

Saturday Morning Breakfast Cereal
Corfield, Penelope J.
2007    Time and the Shape of History. Cambridge, MA, Yale University Press.

Diamond, Jared
1997    Guns, Germs, and Steel: The Fates of Human Societies. New York , NY, W. W. Norton & Company.

Mac Carron, Pádraig, and Ralph Kenna
2012 Universal properties of mythological networks. EPL (Europhysics Letters) 99 28002

Pinker, Steven
2011     The Better Angels of Our Nature: Why Violence Has Declined. New York , NY, Viking.

Tainter, Joseph A.
2004    "Review of Historical Dynamics: Why States Rise and Fall. By Peter Turchin, Princeton University Press, Princeton 2003". Nature (247). February:488–489.

Turchin, Peter
2012     Dynamics of Political Instability in the United States, 1780–2010. Journal of Peace Research 49(4). July 1:577–591.

No comments:

Post a Comment