December 07, 2004

Consequentialism for beginners

Posted by John Quiggin

Now that, thanks to Kieran and the Medium Lobster, we’ve all had our fun with Richard Posner’s case for pre-emptive war, complete with toy numerical example, it’s time for me to play straight man.

Posner’s starting assumption is consequentialism: that we should evaluate an action based on whether its probable consequences are, on balance, good or bad. I broadly agree with this, so I’ll try to explain why it shouldn’t lead to conclusions like those derived by Posner.

I’ll ignore a range of more complex objections and come straight to the first distinction learned by beginning students of the subject. Should we evaluate the consequences of general rules such as “don’t engage in pre-emptive wars” (rule-consequentialism) or should we evaluate each action on a case by case basis (act-consequentialism)

For perfectly rational decision makers, following the rules of Bayesian decision theory, the answer is easy and, in fact, trivial. It’s best to make the optimal decision on a case by case basis, and an optimal set of rules would be so detailed and precise as to yield the optimal decision in every case. Posner routinely assumes this kind of perfect rationality, which is why he doesn’t see any big problems with toy examples, or with claiming that this kind of reasoning can usefully be applied to improbable catastrophes with incalculable consequences.

There are two objections that can be made here

  • Human beings are not perfectly rational and do not follow the rules of Bayesian decision theory
  • Since war is a negative sum game, rational decision makers do not fight wars

The first point is well known, and has been demonstrated by the work of economists like Allais, and psychologists like Kahneman and Tversky (yes, we’ve got Nobel prizes on our side of the argument as well!).

But it’s the second point that’s really critical. Posner’s one-sided discussion of pre-emptive war, with no consideration of the opponent’s motives, is broadly equivalent to the Medium Lobster’s assumption that the other side of the proposed war consists of malign Moonmen (Anglais Casse has more on this). In this respect, and in his failure to consider the possibility that the pre-emptive war might end in defeat, Posner illustrates two characteristic biases of human beings on this subject.

  • we tend to overestimate the malignity and underestimate the rationality of people different from ourselves
  • we tend to overestimate our own competence, and ignore the likelihood that our plans will fail

So, in relation to war, the case for adhering to rules that would discourage resort to war, even when a Posnerian calculation suggests that it looks like a good idea, is particularly strong.

In particular, experience suggests that the case against pre-emptive war is remarkably strong. Lots of pre-emptive wars have been planned, and many commenced. In many cases where pre-emptive war was suggested but not undertaken, the danger has resolved itself peacefully. In many cases where a pre-emptive strike has been planned, the intended quick knockout has turned into a long grinding war, often ending in defeat.

The example almost invariably cited here by supporters of pre-emptive war, and chosen by Posner, is that of Hitler. Posner suggests that a pre-emptive war to overthrow Hitler would have been an appropriate response to the reoccupation of the Rhineland. For the reasons put forward by many of the commenters on Posner’s blog, I don’t think this is clear even with the benefit of hindsight, and it certainly couldn’t have been justified on the basis of what was known about Hitler in 1936. An invasion of Germany, based on a violation of a treaty regarded by all Germans and many others as grossly unfair, would have been a highly dangerous undertaking and could easily have failed. Much the same would be true of an invasion in response to the annexation of Austria.

Certainly, the British and French governments should have learned more than they did from these events, and should have taken a stand at Munich. But the resulting war would have been one of collective self-defence like the actual one, though with a better starting point.

Posted on December 7, 2004 11:29 PM UTC
Comments
Followups
Post a comment










Remember personal info?






Please only hit the "Post" button once. Although there may be a delay while the page reloads, something is happening, we promise.