Via The Dish, which I haven’t read in ages: Bryan Caplan distills pacifism into a comparison of E[benefits of war] and E[costs of war]. That is, we know wars are expensive and awful, but we have much less assurance that they’re going to be worth it:

Of course, “Fight when it’s a good idea, make peace when it’s a good idea” counts as a philosophy. And you might think that this case-by-case approach has to yield better results than pacifism. But that’s only true with perfect foresight. In the real world of uncertainty, case-by-case optimization is often inferior to simple rules.

Which is why I tend to think of pacifism as a small-c conservative position: simple risk-benefit analysis, and a little honest evaluation of history.