Michael Walzer suggests that our common beliefs about individual responsibility and liability become largely irrelevant in the conduct of war. In conditions of war, everything is changed. Political realists have claimed that war eliminates morality; Walzer claims that war collectivizes it. I believe that conditions of war change nothing at all; they simply make it more difficult to ascertain relevant facts. This is not to say that the principles and laws that do or should govern the activity of war are identical to those governing relations among individuals. Just as domestic law cannot simply restate the principles of individual morality, because the declaration and enforcement of laws have effects that must be taken into account in the formulation of the law, so too the principles, conventions, and laws of war cannot simply restate the principles of individual or international morality. The rules of war have to accommodate our epistemic limitations and to be formulated with a regard for the ways in which their announcement is likely to affect people’s behavior. But they should otherwise reflect as closely as possible the same principles of justice and liability that govern conduct outside of war.

Walzer offers various challenges to the elements of this individualist approach to war. He suggests, for example, that this position is really only a closet version of collectivism. When I claim that, during the Gulf War, Iraqi units composed of conscripts ought to have been treated differently from the Republican Guard, Walzer contends that I am thereby accepting the erasure of individual moral identity, though only at the divisional level, whereas the traditional view accepts the collectivization of moral identity at the less arbitrary level of the military as a whole. I do not, however, accept collectivization at any level. No Iraqi soldier’s moral status or liability was determined by his membership in the Republican Guard or by his being a conscript. The difference in treatment required by my position is instead grounded in the necessity of acting on the basis of reasonable presumptions. Just as I may reasonably presume that the person pointing a gun in my direction and shouting threats is a potential murderer, even if he is in fact an actor so deeply absorbed in rehearsing his role as a murderer that he has failed to notice my presence, so an American soldier in the Gulf War was entitled to presume that any member of the Republican Guard was culpably defending a wrongful aggression, even if he had in fact been coerced to join the Guard by his family and was privately resolved to fire over the heads of American soldiers, and so posed no threat at all.

What should we say when, as in these cases, a reasonable presumption turns out to be mistaken? I think it is reasonable to accept an objective account of justification and therefore to conclude that I act wrongly, though perhaps with a fully exculpating excuse, if I kill the actor whom I reasonably believe to be a murderer. We should also conclude that if a just combatant uses force against unjust combatants that would be proportionate if they were all culpable, his action may nevertheless be objectively disproportionate if some of the unjust combatants are morally innocent.

It might be thought that how we classify these acts is unimportant given that the knowledge required to act in the objectively right way is unavailable at the time of action. That may be right; but my point in contrasting the Republican Guard with a conscript unit was to show that we sometimes have information that shifts the presumption we are entitled to make. One point of contention between Walzer and me is whether reasonable presumptions about the moral responsibility of enemy combatants for the threat they pose is relevant to how they should be treated. Walzer thinks not. His view is that in war a person is liable to attack by virtue of posing a threat and that a combatant’s status is ‘collectivized’ in that he counts as posing a threat simply by being a member of the military. I claim, by contrast, that moral responsibility for an unjust threat is the criterion of liability to attack in war and that, while virtually all unjust combatants are morally responsible for posing an unjust threat, some are more responsible than others and the degree of their individual liability varies with the degree of their responsibility. I of course concede that reliable information about the degree of an individual’s moral responsibility is virtually never available in conditions of war. But this does not mean that a combatant’s liability is collectivized; it means that soldiers must act on the basis of presumptions of liability. But these presumptions may vary from one context to another.

Walzer goes on, however, to question whether issues of individual responsibility can ever be of practical moral significance in war. His skepticism is directed primarily at the supposition that matters of individual responsibility can be relevant to how combatants may treat other combatants. While I think that judgments of individual responsibility are perhaps of greatest importance when made by individual combatants about their own action, I will follow Walzer in focusing on how a combatant’s moral responsibility affects what other combatants may do to him. Walzer challenges my suggestion that judgments of individual responsibility were relevant to the conduct of the Gulf War. What would I have said, he asks, to American soldiers if I had had to explain to them that the proportionality restriction was more stringent in its application to attacks on conscripts in the regular army than in its application to attacks on the Republican Guard whose flank the regular army was protecting?

Let me make this easier on myself by focusing on an example in which the diminished responsibility of combatants is more salient than it was among Iraqi conscripts. Suppose I were a commander in the field facing an army of child soldiers, children ranging from 10 to 15 years of age who have been abducted from their families, brutalized, and indoctrinated and who are now fully armed, drugged, and therefore utterly fearless. There are such armies. And the child soldiers who compose them have combatant status according to Walzer’s view and are therefore fully liable to be attacked and killed. If they are morally innocent, or lack moral responsibility for the threat they pose, that is, on Walzer’s view, wholly irrelevant to what it is permissible to do to them. But here is what I would say to those combatants under my command.

You’re aware that we sometimes have to adjust the level of our firepower to take proper account of the status of our targets. If, for example, we knew that enemy forces had taken neutral civilians hostage and were using them as shields, we might have to employ lesser force than we otherwise would in order to minimize the harm we would cause to the civilians. We might, indeed, have to fine-tune our level of force in relation to the number or proportion of civilians present among the enemy combatants. We now confront an army of children. They’re as dangerous as adults would be – indeed more so because of the drugs they’ve been given. But they’re also victims of those who’re exploiting them. They’re scarcely responsible at all for what they’re doing. We should do what we can to spare them. Take additional risks, if necessary, to try to drive them back without injuring them, or to incapacitate, subdue, or capture them rather than kill them.

I suspect that a commander who uttered these words to his troops would earn their respect rather than finding his words received with incredulity or scorn. Nor would it be absurd to address similar remarks to combatants confronting an army of conscripts who had been driven from their homes and onto the battlefield by threats against their families.

Walzer also claims that judgments of individual responsibility should make no difference to the treatment of captured soldiers. Again I disagree, though matters are more complicated here for reasons that Walzer implicitly recognizes in his claim that “the surrender convention is a reciprocal agreement that is obviously of benefit to both sides. But it also fits nicely with the ‘traditional’ argument about the moral equality of soldiers.” The suggestion here is that the moral equality of soldiers is more than merely conventional, more than a matter of agreement for mutual advantage. Thus, the treatment of prisoners is governed simultaneously by a convention that has evident utility and by considerations of individual rights and liabilities that are not matters of convention.

While the convention applies without regard to questions of individual responsibility, that is not true, I believe, of nonconventional considerations of rights and liability. Suppose, for example, that we have captured enemy combatants on our own territory. They are mercenaries who were promised booty as a reward for participation in an aggressive campaign to seize our oilfields and other natural resources. We learn that the aggressor is sending a massive force to the area in which we are now located. We must evacuate. If we release our prisoners, they will rejoin the enemy army. But if we take them with us, they will hinder our movement, diminish our already depleted stores of food, and divert our attention to the prevention of their escape, and may attempt to alert the enemy to our location. They will, in short, significantly increase the risks we will face in our efforts to evade the enemy. These prisoners have thus, through their own voluntary and wrongful action, put us in a position in which we must either kill them or expose ourselves to a significantly increased risk of being killed. I believe that considerations of justice would permit us to kill them in these circumstances. They have made themselves liable to be killed by wrongfully making it the case that our sparing them would increase the peril to our own lives. Yet we also, as I have conceded, have moral reason to abide by the conventions governing the treatment of prisoners, both for our own sake in the future and for the sake of any of our fellow soldiers who are being held prisoner by the enemy. How these opposing considerations balance out in the end depends on the details of the case.

But now imagine that our prisoners are not mercenaries who have participated in unjust aggression for reasons of self-interest but are instead morally innocent conscripts who were both lied to by their government and coerced to fight by threats against their families. In that case their responsibility for the dilemma we face would be minimal and we would have decisive conventional and nonconventional moral reasons not to kill them, even in the circumstances I have described.

Walzer contends that international courts are yet another arena in which judgments about the individual moral responsibility of combatants are out of place. He contrasts the individual trials of Dutch citizens accused of collaboration with the Nazis with our refusal to try individual German or Italian soldiers for participation in campaigns that we regard as uncontroversial instances of unjust aggression. His explanation of why the Dutch trials were appropriate while trials for unjust combatants would not have been appeals to a particular form of excuse that was available to the soldiers but not to the collaborators. Each soldier could truthfully have said: “Everyone else was doing just as I did and our conduct had the sanction of the moral authorities in our society: the government, the clergy, and so on. I can’t be held liable in such circumstances for failing to perceive or act on what I can only now perceive to have been the right moral reasons.”

It is true that most unjust combatants have this excuse while collaborators and other traitors do not and indeed cannot. (If collaborators had this excuse, the collaboration would be so widespread as to constitute submission and assimilation and there would be no need for or possibility of later trials. What would remain as the object of treachery if everyone had joined the enemy? Who would there be to try the traitors if virtually everyone were a traitor?) But there had better be more to the explanation of why it was appropriate for the Dutch to try their collaborators while it would not have been appropriate for the Allies to have tried German and Italian soldiers. For when crime is serious enough, the claim that ‘everyone else was doing it’ is very unlikely to be exculpating. I think that is so in the case of the German and Italian soldiers. In the aftermath of the war, most of these soldiers had ample reason to experience not just what Bernard Williams calls “agent-regret” but also guilt. Even when they had restricted their use of force to Allied combatants, they had served political goals that were evil and abhorrent and had killed people simply for attempting to defend their lives, their families, and their political independence.

The reason why it was not appropriate to try these soldiers is not that there were no moral crimes of which they were guilty, nor that they shared a fully exculpating collective excuse. The reasons are instead pragmatic. For example, while there is considerable risk of injustice and vengeance even in domestic trials of collaborators, the risk is far greater when victors try defeated foreign enemies. There are, moreover, far too many of the latter to be tried with any prospect of fairness. There are also pragmatic reasons not to threaten combatants with punishment ex ante: for example, that the prospect of punishment, or vengeance under cover of punishment, would make the termination of war more difficult by giving soldiers in the field an incentive to continue to fight rather than to surrender. Recognition that there are these objections to the post bellum punishment of unjust combatants is fully compatible with the position I have defended. For these objections do not appeal to the moral equality of soldiers or to the collectivization of moral status in war but to considerations of an entirely pragmatic nature.

Walzer concludes his critique by challenging the practical relevance of my notion of noncombatant liability. He concedes that if we had precision-guided warmonger-seeking missiles, some traditional just war theorists might allow their use in residential areas. But since this is fantasy, my view will not, he suggests, diverge from his own in matters of practice, except perhaps with respect to the stringency of the constraints that each of us would recognize on targeted killing.

But again I suspect that there are more substantial divergences. Suppose that there are two military targets in enemy territory that are of equal strategic importance but that we have the resources to destroy only one. Neither can be attacked, however, without civilian casualties and the number of casualties would be roughly the same in either case. One target is located in a government stronghold, where people are in general ethnically, religiously, and politically aligned with the government and exert a powerful influence over its decisions, including the decision to launch this unjust war against us. The other is located in an area populated by an ethnic minority whose members are in open revolt against the government and its war. If moral responsibility for an unjust threat is irrelevant to liability, as the traditional theory asserts, it should make no difference morally which target we attack. Yet I believe that, given that we can attack only one of the targets, it would be wrong to attack the target in the area populated by dissenters – and perhaps that is true even if an attack on that target could be expected to cause fewer casualties than an attack on the other. (Again, this is not because individual liability is collectivized. It is because it is reasonable to presume of each person in the disaffected area that he or she bears less responsibility for the government’s action than a corresponding person in the government stronghold.) There are, moreover, other types of case that raise the same issue – for example, a case in which one possible target is in the center of enemy territory, while the other is on the border. Assuming an attack on each would cause civilian casualties, it does not seem a matter of moral indifference whether the civilians harmed are citizens of the enemy country or citizens of the adjacent neutral country. For it is a reasonable presumption that most citizens of the enemy state bear some degree of responsibility for that state’s action, while most citizens of the neutral state bear none at all.

Walzer believes that what he calls the war convention is just an adaptation, developed over many centuries, of our ordinary morality to the circumstances of war. But again I have doubts. The traditional theory of the just war, as manifested in the war convention, seems much more like an adaptation to the circumstances of war of a Hobbesian vision of the right of self-defense combined with various elements of chivalric morality. Indeed, what my paper sought to demonstrate is precisely the incompatibility between certain central elements of the traditional theory and our ordinary beliefs about moral liability to defensive force. I have argued, for example, that our criteria of liability imply that some civilians may be morally liable to certain forms of harm (though not normally, as I stress in the paper, to killing). In his jocular reference to warmongers, Walzer concedes that some traditional theorists might accept this. But my point is that if they were to accept this, even if only for hypothetical circumstances, they would have ceased to be adherents of the traditional theory, which insists on the immunity of noncombatants. The traditional theory incorporates the principle of noncombatant immunity not because it is a mutually beneficial convention but because it is an implication of the theory’s own criterion of liability to defensive force. According to the traditional theory, people become liable to attack by virtue of actively posing a threat to others; thus, because noncombatants do not pose a threat, they cannot be liable to attack. But this, I argue, is not what we in fact believe – not only about liability in relations among individuals but even in the circumstances of war.

What I have proposed is that we explore with greater care and in greater depth the moral foundations of the just war in, for example, the morality of individual self- and other-defense in order to determine what those foundations properly imply about such matters as liability to attack in war. Walzer is certainly right that in the end we must arrive at codes of conduct and laws of war that are sensitive to the nature of war and that would, if promulgated, mitigate rather than exacerbate the terrible effects of war. But I do not think that we can know, in advance of inquiring, that the principles and laws that would constitute the best compromise between our foundational convictions about justice and liability and the necessity of having workable rules adapted to the complexities of war would turn out to be, with perhaps minor modifications, the prevailing war convention that Walzer has defended.