Jacob Levy, with the concurrence of "Oxblog"'s David Adesnik, argues for an exception to my stated principle that enemy soldiers are not subject to the same protections as enemy non-combatants. Levy suggests that when one side has an overwhelming advantage over the other, it has a responsibility not to inflict gratuitously large casualties on the vastly inferior force.
The problem with this claim is that "overwhelming advantage" can really only be determined in retrospect. Consider, for instance, the recent battle near An-Najaf, where US Marines may have killed as many as 500 Iraqi soldiers without losing a single man. That sounds like a pretty overwhelming advantage to me. Can we conclude, then, that the Marines should have noticed this imbalance during the battle, and cut the attacking Iraqis a little more slack--perhaps letting them get a little closer before firing on them?
Now, Levy might mean to distinguish between, say, attacking soldiers and fleeing ones. But, again, how are the putative victors to know that the vanquished are not fleeing to join a much larger, more threatening horde? No, if they are truly beleaguered and fleeing for their lives, they always have the option of surrender. Otherwise, they simply cannot be assumed to be at an overwhelming disadvantage--until, of course, they've become casualties.