After two days at the International Studies Association Annual Conference (ISA) in Montreal, several interesting themes have emerged. Rather than mention panel by panel what I’ve seen, I figured I’d summarize my general impressions and provide some highlights at the mid-point of the conference. I’ll caveat everything by noting that the conference is a huge gathering and any attendee can only see a small portion of the total offerings. So, my observations are necessarily constrained by my choices of panels and side-discussions – I could easily have missed something.
I’m amazed by the methodological and theoretical commitment to rationality. While disaggregation and ‘complexity’ are themes that are becoming more common, no one except those doing historical case studies, appears to be taking the necessary steps of incorporating how people actually think into their work. Time again, a presentation starts by posing a great question, suggests a novel theory, and then appears to fall into a standard set of methods that all preserve rationality and produce less than satisfactory results from my opinion – outcomes seem unjustified, and “success” seem like post-hoc reconstructions of utilities that subsequent provide the “correct” answer. This created some strange inconsistencies in some models, for example one that emphasized the use of frames but simultaneously arguing that because actors are rational, they will consider all possible options and use all information – thus reducing the frame to a utility function. I think the use of frames would be more useful if it suggested that actors actually perceive their available options differently and searched the space of options differently as a result.
I was surprised by several of the intelligence analysis panels. There was a good focus on methods, but there is a surprising gap between what is being done in the academic world and what is being applied in the work presented at the conference. Obviously what appears at an academic conference is not the totality of what the community is doing, but there is a notable gap between what can be done and what was shown. One of the methodological presentations made a crucial point however, emphasizing a search for lightweight, fast, and simple techniques for coping with complexity that any analyst could perform at almost zero cost. This suggests that more advanced techniques may only be feasible on problems that warrant strong and sustained resource commitments, likely limiting the extent to which they might be widely adopted. One presentation was on complexity and intelligence problems. It was interesting but far behind the state of the science, ironically demonstrating the continued challenge of operationalizing complexity theory – particularly under the resource constraints noted before.
Another interesting observation was the invisibility of neorealist approaches to the international systems. While I don’t think neorealism is dead, I think current events have severely limited the extent to which analyzing the strategic choices of unified states are of much interest. Perhaps this is just a function of what I’ve attended, so I don’t think too much should be taken from this. Although, I did see a new book by John Mearsheimer on why leaders lie. I haven’t gotten it yet, but suspect this is quite important to him since he argues that every leader (or at least every responsible leader) is a realist only concerned with military power and economic production, but must lie to couch their policies in moral terms. My suspicion is that his book attempts to dissect the moral arguments and justifications of leaders and show that underneath they are all doing nothing more than playing power politics. At some point I’ll get the book, find the time to read it, and then see if my guess about its content is on point or off the mark. If this is so, I’d suggest that Mearsheimer is overly kind in assuming that leaders are hyper rational, and that I worry in that they may actually believe what they say, creating ideological commitments that are then rationalized on the grounds of national interest. Now I want to get his book.
Leaving aside questions of theory, none of the intelligence studies, particularly the historical case studies, provide much evidence to support neorealism’s unitary or rational actor assumptions. This would seem to suggest that even if leaders talk in terms differently than they think, this doesn’t mean that they process information and options as the neorealists argue.
Lastly, I’m surprised by the gaps between scholarly practice and engaged support to policy-making. This is not a matter of focus, at least directly, but more a matter of understanding how questions are asked and answered. I’m worried that much of scientific practice is essentially “driving while looking in the rearview mirror” as an old boss of mine used to say. Supporting policy-makers requires looking ahead and helping them choose between alternative futures. By comparison, the academic studies are largely focused on the model, not the decision that it informs. This was best summarized by a comment on methodology when one of the panel members thought the idea of testing the model’s assumptions was novel because they were used to accepting assumptions and only looking at outcomes. By contrast the applied work in policy and intelligence analysis methods are obsessed with testing and justifying assumptions – if only to bound the extent to which a conclusion or judgment can be made with confidence. The lack of closing the loop by re-examining model assumptions (whether it fits or fails to fit the data) suggests that the inductive/deductive loop may be broken or frayed and if scholarly research consistently descends into a strictly deductive exercise, it will cease to be social science in any meaningful way and simply become social mathematics.
On a totally unrelated note, Montreal is a wonderful city (at least the parts that I have walked)… you have to work hard to find a bad meal, and every block has a pizza joint or steak house.