Continuing the debate about preventive war begun by Judge Richard Posner (the discussion was begun by him, I mean, not the war) the Medium Lobster presents a competing analysis:

[T]the probability of an attack from the moon is less than one – indeed, it is miniscule. However, the potential offensive capabilities of a possible moon man invasion could be theoretically staggering. … The Medium Lobster has calculated this probability to be 5×10^9^. … the resulting costs would include the end of civilization, the extinction of the human race, the eradication of all terrestrial life, the physical obliteration of the planet, and the widespread pollution of the solar system with a mass of potentially radioactive space debris. The Medium Lobster conservatively values these costs at 3×10^12^, bringing the expected cost of the moon man attack on earth to 1500 (5×10^9^ x 3×10^12^), a truly massive sum. Even after factoring in the cost of exhausting earth’s nuclear stockpile and the ensuing rain of moon wreckage upon the earth (200 and 800, respectively), the numbers simply don’t lie: our one rational course of action is to preventively annihilate the moon.

I’m a bit sorry to break it to the Medium Lobster, but Judge Posner considers scenarios of precisely this kind, and uses pretty much this methodology, in his new book, Catastrophe: Risk and Response. Cases treated include the nanotechnology gray-goo apocalypse, the rise of superintelligent robots, and a strangelet disaster at Brookhaven Labs that would annihilate a substantial chunk of spacetime in the vicinity of our solar system. A recent review of the book raises most of the relevant critical points about the approach Posner takes. In essence, it’s all good geeky fun to apply the methods to cases like these but it’s a stretch to pretend we’re learning anything decisive about what we should do, as opposed to gaining insights on the scope and limits of some techniques for assessing alternatives.

In the case of cost-benefit arguments about preventive war, there are other objections. One, pointed out by Chris in the comments to my earlier post, is the the worry that your adversaries are thinking about preemptive attacks in much the same way you are, and so will move to preempt your preemptive attack with one of their own. You can still be committed to weigh up the costs and benefits as best you can, but it would be foolish to think one had a straightforward technique that cut through the difficulty and reduced it to a matter of simple calculation. One of the most important contributions of game theory is the way it reorients your thinking from a parametric to a strategic point of view. That is, you stop thinking of other agents as passive bits of the world and realize that they, like you, are searching for the smartest decision given what they anticipate their opponents will do. If you’re inspired by rational choice theory, as Judge Posner is, the simple application of cost-benefit methods should not look very plausibly as a way of reaching a strategic decision about the use of preventive action in cases like Iraq.

More broadly, though, the reason I find it disturbing that these elementary cost-benefit methods are presented by Posner as a serious way to resolve decision problems in international politics it that it’s clear he’s aware of their shortcomings and limits. In the Catastrophe book, for instance, he acknowledges that monetizing the costs of extremely unlikely disasters, for which little in the way of good evidence about their likelihood exists, is an essentially arbitrary process. (In more run-of-the-mill cases, of course, “arbitrary” does not mean “random.” The development and application of these pricing technologies has more to do with the push to quantify many kinds of risk than the rationality of the methods themselves. But that’s an argument for another day.) At a minimum, if we’re thinking of things like the invasion of another country it just won’t do to acknowledge the arbitrariness of the numbers being assigned to the variables, and then press on regardless. Nor is it satisfactory to sketch a cost benefit approach as if it represented how a rational agent would assess a situation in practice, and then jump in to specific cases like Hitler invading the Sudetenland. It’s not that these methods aren’t powerful, it’s that they’re being misapplied. Abba Lerner once commented that

Thousands of habits of behavior and of enforced laws had to be developed over millennia to establish the nature and the minutiae of property rights before we could have buying and selling, instead of each man just taking what he wanted if only he was strong enough. … Each set of rights begins as a conflict about what somebody is doing or wants to do which affects others … An economic transaction is a solved political problem. Economics has gained the title of queen of the social sciences by choosing solved political problems as its domain.

When the parameters of a decision are settled—to buy or not buy some commodity, for instance—then a tool like cost benefit analysis can be enormously powerful, especially when there are a lot of parameters. Even when aspects of the decision become increasingly uncertain or subjective, these approaches retain much of their power as heuristics for decision making. But eventually you’ll get to cases which should provoke reflection on the limits and applicability of the methods, rather than a perverse desire to rely on them all the more just because the choice is hard and you want a definitive answer. Whatever you thought of the pros and cons of invading Iraq, the last thing it could be called was a solved political problem. In such cases, wheeling out the cost-benefit machinery in the way Judge Posner does isn’t a way to make the political choice easier, it’s a rhetorical move to make the politics of the choice magically disappear.