One of the most flattering things that has happened over the past month or so is that a lot of people–most of them strangers–have been asking about my dissertation. This is the Ph.D. candidate’s equivalent to “let me tell you about my grandchildren.” Then, when I tell them, they don’t actually get fish eyes. Although some of them were on the phone and I couldn’t see them. Maybe they were politely asking the question and then rolling their eyes so far back into their sockets that their very retinal strands popped right out of their heads.

For those of you who think you might resort to mere bovine eye-rolling or less, here’s the deal with the dissertation. It’s about violent societies, that is, those societies where the governing body does not hold a monopoly on violence. In other words, if you and I are living in medieval England and you come over to my house, kill a bunch of my guys and take my stuff, the king does not give two hoots. I have to get some more guys, go over to your place, kick your butt and take my stuff back, maybe abscond with some of your women. Hey, you deserve it. The point is, in a violent society, we’re not talking about, say, guns per capita, but about the nature of retribution in the wake of violent crime: if the society has mechanisms in place that handle that sort of thing, and there is a reasonably competent juridical system that enforces it, it is not a violent society. If, however, we’re talking about retribution between equals rather than punishment from above, it is. Make sense?

The main thing I’m focusing on is how societies get that way. I’m using Lonnie Athens’ model of how people become dangerous violent criminals and expanding it to the societal scale. So far, it’s working great. I’ve observed 5 basic patterns: where the governing body is both unwilling and unable to maintain its monopoly on violence; where it is willing but unable; where it is able but unwilling; smaller pockets of violent societies inside larger, stable societies; and historical cases. I’ve picked out a pattern study for each of those cases, and I’m writing that part right now. Each pattern study is selected for geographical distribution and to dispel some common myth about violence: it’s because Africa is a bad neighborhood, it’s because of poverty, it’s because of the drug trade, etc. My contention is that the society has to pass through each of the stages of Athens’ model to become violent, and that while these kinds of explanations are sometimes symptoms or contributing factors, they don’t actually explain how a society becomes violent any more than they explain how individuals become violent. For example, millions of poor people living in “bad neighborhoods,” local or global, never become violent.

The final section of the project then applies some methods of nonlinear dynamics to analyze the patterns: cusp catastrophe; Langevin (differential) equations and fuzzy logic; and agent-based modeling. I operationalize the processed I discuss in the previous section and see how the crazy pictures come out. This was sort of the essence of what I wanted to do from the beginning, but because nonlinear dynamics is still pretty new, I had to approach it in a sort of two-pronged attack and do a more familiar, qualitative piece too. A big part of why I’m in the program I’m in is that I told them from the outset what kind of dissertation I wanted to write, and they said okay.

So the larger point I’m trying to make with this project is that mathematical modeling in the social sciences does not have to suck. Often, there is a sort of tension in international relations between wanting to be “scientific” and reaching toward the kinds of insights mathematics gives us on the one hand; and the fact that it’s hard to quantify human behavior and operationalize such models on the other. Usually what happens is that an author will just do an over-simplified cheesy model that doesn’t really describe real-world behavior all that well, or he will simply reject the mathematical option and do qualitative work. I don’t think we have to choose. I think we just have to use better models that are capable of handling the situations we’re curious about. If linear models don’t really explain something, why not nonlinear ones? If statistical models can’t capture human behavior, why not use ones that can? Like that.

So that’s what I’m working on. I hope your eyeballs are not too icthystic at this point. Let me know which part of this, if any, you’d like to know more about.