Posner and Becker on Preventive War

Richard Posner, the 7th Circuit Chief Judge and University of Chicago Law Professor , and Gary Becker, a Nobel Prize winning economist also at the University of Chicago, launched a blog Monday with two articles, one by each man, on "Preventive War". To say the least, it was an underwhelming debut.

Both gentlemen concluded that preventive war was justified under certain circumstances. Posner, not surprisingly, uses an cost benefit analysis to define those circumstances:

Suppose there is a probability of .5 that the adversary will attack at some future time, when he has completed a military build up, that the attack will, if resisted with only the victim’s current strength, inflict a cost on the victim of 100, so that the expected cost of the attack is 50 (100 x .5), but that the expected cost can be reduced to 20 if the victim incurs additional defense costs of 15. Suppose further that at an additional cost of only 5, the victim can by a preventive strike today eliminate all possibility of the future attack. Since 5 is less than 35 (the sum of injury and defensive costs if the future enemy attack is not prevented), the preventive war is cost-justified.

Wow. Is that all there is to it? The concept is easy. Even the math is easy. Why is everyone struggling so much with this issue?

Well the reason, of course, is that Posner entirely skipped the hard part: in a world of massive uncertainty how does one go about trying to reliably estimate (a) the probability of the attack, (b) the cost of the attack at current states of readiness, (c) how imminent is the attack; (d) what "additional defensive measures" are available, (e) which of the available measures could actually be implemented in the time before the attack, (f) how much would those additional measures cost, and (g) how much would they lower the cost of the attack. Unless you can attach values to all of those with a fair degree of certainty, the formula is worthless. Yet defining any of these variables with precision is extraordinarily difficult -- as has been illustrated so potently and painfully by the complete collapse of the estimates of these variables that ostensibly justified the invasion of Iraq. Moreover, if the cost of enduring the attack and the cost of the preventive war are even in the same ball park, almost any degree of uncertainty in the estimates of the variables will make it impossible to tell whether the preventive war is "justified" or not.

To illustrate, Posner's example assumes that cost of enduring an attack is 20 times higher than the cost of a preventive war (100 vs. 5). Rarely will the real world offer so disproportionate an advantage to a potential victim of an attack. For the sake of argument, then, let's assume that the cost of being attacked is only 4 times the cost of attacking; i.e. 100 vs 25. Still a pretty high ratio, but it will serve to illustrate the point. Also assume, for the sake of simplicity that, after exploring all available options, the victim determines that there are no "additional defensive measures" it can take that would appreciably reduce the cost of the attack. This leaves only three variables to consider: The probability of the attack (Pa), the cost to the victim of enduring that attack (Ca), and the cost to the victim of a preventive attack (Cp). Under Posner's approach, the preventive attack would be "cost-justified" so long as Pa x Ca > Cp. Now, assume we estimate Pa at 0.5, Ca at 100, and Cp at 25. Since, 0.5 x 100 > 25, the war is in Posner's view, cost justified.

But wait. Clearly there is uncertainty around all three of these variables. So, let's assume, that each is only accurate to +/- 30%. That means that Pa is actually somewhere between 0.35 and 0.65; Ca is somewhere between 70 and 130, and CP is somewhere between 17.5 and 32.5. This in turn means that the "true" probability-adjusted cost of waiting to be attacked could be as low as 24.5 (0.35 x 70) while the cost of attacking first could be as high as 32.5. If this is the case, then the war would not be "cost-justified" in Posner's estimation. On the other hand, the cost of waiting to be attacked could be as high as 84.5 (0.65 x 130), while the cost of attacking first could be as low 17.5. In this case, the preventive war clearly would be "cost justified" under Posner's analysis. Now what do we do?

To the extent we add more variables -- such as the costs and benefits of "additional defensive measures -- each of which has it own associated uncertainty, the spread between the possible outcomes become ever greater, at an exponential rate. It is very very, hard to see how such a formula is helpful in making decisions.

To make matters worse, there are (at least) two important variables that Posner does not even mention. First, there is no consideration of the costs to the putative aggressor of either option. Surely there are some, and while the putative victim would not value those costs as highly as he does his own, it is easy to imagine scenarios where it would be impossible for any civilized nation to refuse to take those costs into account. Iraq is, again, an obvious case in point. In deciding whether to invade Iraq, shouldn't the costs to the Iraqi people be at least a factor in the equation?

Second, the probability of an attack is not immutable. Insofar as it depends on human decision-making, that probability could be affected by such things as diplomacy, sanctions, alliances, etc. These sorts of factors work in much the same way as Posner's "additional defensive measures" do, but they cannot be subsumed there because the operate on the probability of an attack rather than its costs.

It is hard to believe, but Professor Becker's analysis is even less helpful than Judge Posner's. Becker takes a much more qualitative approach to the problem, but ends up with something that at best a truism and at worst a truly scary theory. He starts by noting that, historically we have used the same strategy to prevent war as we have to prevent crime: the deterrent effect created by the threat of retaliation/punishment. However, he argues, our enemies today are largely immune to deterrence because they are too widely dispersed for retaliation to be effective, unconvinced (or unaware) that retaliation is likely, or indifferent to the impact of the retaliation. Thus, he argues that we should use a new model, also drawn from the criminal law: We need to be willing to punish intent, much as we do in prosecuting conspiracies. Relying exclusively on this facile analogy, Becker argues that since we punish bad intentions in the criminal law, we are also justified in punishing bad intentions of international actors.

In his only nod to nuance, Becker does acknowledge that "the evidence of intent must be analyzed carefully". What I think (hope) this means is that there must be some reasonable basis for concluding that the requisite intent actually exists. But then he vitiates this by arguing, first, that we have never insisted on 100% certainty regarding intent:

But criminals are convicted too on less than 100% certain evidence. As Posner says in his commentary, it is necessary to consider probabilities, not certainties.

and then arguing that:

[T]he degree of certainty required before preventive actions are justified has been considerably reduced below what it was in the past because the destructive power of weaponry has enormously increased.

Becker's thesis, then, comes down to is this: If we have sufficient evidence that someone intends to attack us, we can take pre-emptive action. Hardly an insight worthy of a Nobel Prize winner. A more interesting article would have started with that principle as a premise (since no one disputes it) and then discussed what "sufficient evidence of intent" means today. As it is though, his article is susceptible to the interpretation that, because of the danger of modern weapons, preventive war is justified even by the vaguest hint of intent. I certainly hope that is not what he intended.

Posner and Becker are both incredibly gifted men, who unquestionably have much to say that is perceptive, informative, and enlightening. Having decided to venture into the blogosphere, let's hope they actually take the venture seriously.