Today’s issue of Nature has a Feature Article by Laura Spinney on cliodynamics. Laura interviewed me when we both attended the Frankfurt Forum on Cultural Evolution (about which I wrote in an earlier blog). I think she did a great job capturing the excitement of our new fledgling discipline and explaining in easy-to-understand language some of the fairly complex concepts and results from cliodynamic studies.
In this blog I’d like to amplify on a couple of points raised in the Nature article. Also, if there are any comments or questions about any aspects of this research, I’d like to hear them.
This graph shows the total number of political violence events per five years, as well as numbers of three major types of violence. These types are classified as clashes between groups of people (‘riots’), violence by groups against individuals (‘lynchings’), and violence by individuals against groups (‘terrorism’). The last category, for reasons explained in the article, includes rampage shootings. Note that this graph focuses on the numbers of events, but the JPR article also shows and discusses other views, e.g. the number of people killed in political violence events per one million of population.
As Laura explains, the dynamical pattern is complex, because it has two oscillations: the long-term secular cycle and a shorter 50-year cycle on top of the secular trends. She, understandably, devotes much time to the discussion of the 50-year cycle, because it is a striking feature of these data. However, I personally find the secular cycles much more worthy of attention. Partly it is because I don’t fully understand the mechanisms underlying the 50-year cycles, and partly because, unlike secular cycles, they are not a universal feature of instability dynamics in all societies. As is discussed in my books Historical Dynamics and War and Peace and War, 50-year cycles don’t show up in the Chinese data (and there I also explain why).
On the other hand, longer secular cycles are apparently a universal feature of all complex, state-level societies (at least, they crop up in every large-scale agrarian society for which we have reasonably detailed data on political violence). As I wrote in my previous blog, the chief motivation for the American project was to find out if these cycles continue to operate in post-agrarian modern societies.
I think that the nice progression (going backwards) of 1970, 1920, and 1870 spikes is partly due to ‘luck,’ because in historical data the periodicity is not quite as clean-cut, and periods between successive spikes can vary between 40 and 60 years. Note also that there is a missing peak of 1820, although the previous one, the American Revolutionary War of 1776-83 took place pretty much on schedule (American Revolution falls outside of the time frame of the JPR paper, but we know from the work of Paul Gilje that there was a build up to it in the preceding decade or so – as in the pre-Civil War period the rising frequency of political violence events singaled an impending crisis). The reason why there was no violence spike in 1820 is probably because in this period all structural-demographic conditions were so favorable that they suppressed any interest in making trouble on everybody’s part (this is discussed in the book that I am working on right now).
An important point here is that when we talk about ‘cycles’ in historical dynamics, we do not mean strict cycles in the mathematical sense (as perfectly periodic dynamics). All of these oscillations are produced by nonlinear feedbacks between different interacting parts of the social system, and such oscillations can be quite variable in period due to both external shocks and what is known as the ‘endogenous chaos’ (when an interplay between two or more mechanisms results in very complex non-periodic dynamics).
Oof. Congratulations if you struggled through the previous paragraph. I guess that is why we need good science writers, who are willing to sacrifice the technically correct language to get to the main point…
If we had to rely solely on graphs, such as the one shown above, we wouldn’t even be able to say whether there really is a systematic cycle (or just random fluctuations – “one damn thing after another”). There is one complete secular cycle there (from c.1780 to c.1930) and another incomplete one (from c. 1930 to the present). One and a half oscillations is certainly not enough to characterize them statistically as a systematic cycle.
My confidence that this interpretation is correct rests on much more data (and theory) than what is in the graph. First, the theory (here I am talking about the structural-demographic theory) was developed by analyzing historical societies other than the U.S., so finding the same pattern in the American data represents a successful empirical test passed by the theory. Second, and even more important, the theory makes specific predictions about how other variables should be changing in relation to political instability. Some should go up as instability goes up, and others should go down. The last graph in the JPR paper (which looks like a bunch of spaghettis) demonstrates that this is indeed the case.
Turning now to the question of 2020, the problem for us all (in the United States, that is; I have not done a systematic analysis of other western democracies) is that since the 1970s a variety of structural-demographic variables have been trending in ways that usually indicate a very serious crisis just around the corner. As Laura quite correctly says in the article, the impending crisis is not inevitable. But will we collectively catch on to defuse it in time? Or will we have to experience a revolution, or some very significant period of political violence, which, as Herb Gintis suggests, will force our elites to implement the needed reforms?