It is a privilege to be able to add a footnote to Richard Miller's illuminating moral analysis of the war in Afghanistan. I am in substantial agreement with his argument and share his evident frustration with the Obama administration's failure to provide a cogent justification for the continued killing. My contribution to the discussion will be largely theoretical rather than political. Miller rightly observes that “current just war theory does not provide sufficient guidance” in thinking through many of the moral issues raised by the war (p. 103). Some of the questions he addresses are concerned with proportionality, a notion whose complexities are only beginning to be appreciated. My modest ambition in this comment is to try to sharpen these questions and provide some assistance in thinking about them, though I am far from understanding them fully myself.

Forms of Proportionality

My remarks will focus mainly on the issues of proportionality considered in the section of Miller's essay titled “Moral Costs.” There Miller draws various distinctions among the harms that result from U.S. military action in Afghanistan. He first distinguishes between harms to noncombatants and harms to combatants. Among harms to noncombatants, there are those that the United States causes and those that are caused by its adversaries. Among the harms to noncombatants that are caused by the United States, Miller distinguishes between those it causes intentionally and those it causes unintentionally, as side effects of attacks on military targets. And among those it causes unintentionally, he further distinguishes between those it causes directly and those it causes indirectly. An example of direct causation is the killing of a noncombatant by an explosion intended to kill only enemy combatants. The example that Miller gives of indirect causation is the continuing high rate of under-five mortality attributable to the disruption of domestic life by conditions of war. Among harms to combatants, Miller distinguishes those that are caused to U.S. and allied forces by the Taliban from those that U.S. and allied forces cause to the Taliban.

It has usually been assumed that whether a war fought by a state is disproportionate depends only on whether the harms that that state causes are excessive in relation to the importance of its just cause for war, which is usually the prevention or correction of some serious wrong for which its adversaries are responsible. Miller seems to broaden the class of relevant harms to include all the harms that occur if a state goes to war but would not occur if it were not to go to war, whether they are caused directly or indirectly by that state, or even caused proximately by its adversaries.

Another standard assumption that Miller rejects is that harms to combatants do not count in assessments of proportionality. Many just war theorists assume that harms to a state's own combatants do not count in determining whether its war is proportionate for the reason just given—namely, that they are not attributable to its action but are caused instead by its adversaries. To see why this traditional assumption may seem sensible, consider individual third-party defense of others. Suppose that a culpable attacker is about to harm two innocent people. I can intervene to stop him, but it is predictable that if I do, he will break my arm. It may seem that the prospect of my broken arm does not count in the proportionality assessment; for while my intervening might be imprudent, it does not seem that the prospect of the broken arm could make my action disproportionate and therefore impermissible.

Other theorists make a different assumption that also implies that harms to a state's own combatants do not count in assessing the proportionality of its action. This assumption is that because all combatants are morally liable to attack, the harms they suffer cannot make war disproportionate. For if a person is morally liable to be killed, that fact provides a full justification for killing him. There is no remainder of harm or wrong that has to be justified as necessary for and proportionate in relation to the attainment of one's just ends. If there were—that is, if killing a person who is liable to be killed had to count as a bad effect in assessing proportionality—there would be a limit to the number of evil aggressors it would be permissible to kill to prevent oneself from being killed by them. If, for example, there were enough evil murderers, each of whom would kill me unless I killed him, I would be obliged, on this view, to allow the first one to kill me. For it would, by hypothesis, be disproportionate for me to kill all of them and therefore pointless (because it would only be postponing the inevitable) to kill even one. But most of us believe that if each potential murderer is liable to be killed, it would be permissible, and therefore proportionate, to kill them all, no matter how many of them there were.

Traditional just war theorists have seldom addressed these issues, but to the extent that they have, they have tended to assume, as I noted, that harms that people are morally liable to suffer do not count in assessments of proportionality. If this assumption is combined with the traditional assumptions that one becomes liable to attack in war simply by posing a threat and that all combatants pose a threat, it follows that harms suffered by combatants in war are irrelevant to proportionality.

On the basis of reasoning of this sort, traditional just war theorists have tended to assume that proportionality in war is concerned only with harm to innocent people—that is, people who are not liable to attack and thus are not legitimate targets. Given the further assumption that all noncombatants are innocent in this sense, it follows that proportionality in war is concerned only with whether the aims of a war or act of war are sufficiently important to outweigh the expected harm to noncombatants.

Miller rejects these assumptions. He assumes instead that expected casualties among U.S. forces can contribute to making the American war in Afghanistan disproportionate. (Virtually everyone of course agrees that these casualties matter; it is just that most people think that while they may make the U.S. war imprudent, they do not make it disproportionate.) And he also claims, more controversially, that casualties among enemy combatants can also contribute to making the war in Afghanistan disproportionate.

I believe that Miller is right, but that more must be said. First, it is necessary to distinguish explicitly between proportionality in harms to people who are potentially liable to be harmed and proportionality in harms to people who are entirely innocent—that is, not liable to be harmed. I refer to these as narrow and wide proportionality, respectively. Narrow proportionality is concerned with whether harm to a person is proportionate in relation to his liability. If the harm one inflicts on a person exceeds that to which he is liable, one's action is disproportionate in the narrow sense. Any harm beyond that to which he is liable counts, in effect, as harm to an innocent person.

The justification for harming an innocent person in war must be a necessity or lesser evil justification. When, for example, one's action prevents a wrongful harm but also harms an innocent person to a degree that is excessive in relation to the harm averted, it is disproportionate in the wide sense. In general, harms to the innocent that are intended as a means have greater weight in wide proportionality than harms that occur as unintended side effects. Wide proportionality, in other words, is sensitive to intention.

Narrow and wide proportionality differ in various respects. First, narrow proportionality is a constraint on liability-based justifications for harming, while wide proportionality is a constraint on justifications of necessity or lesser evil. Second, the assessment of narrow proportionality requires both a retrospective judgment about the individual's responsibility for a threat of wrongful harm and a prospective judgment about whether harming him will prevent that harm. If an act that harms a person would have no effect in preventing or rectifying a wrong for which he is responsible, he cannot be liable to be harmed by that act. He may deserve to be harmed, but that is a different matter. The assessment of wide proportionality, by contrast, requires only the prospective judgment. A retrospective judgment might be necessary only to confirm that there is no basis for liability. Third, while it can be proportionate in the narrow sense to inflict a harm on a wrongdoer that is greater than the harm one thereby prevents him from inflicting, it can never be proportionate in the wide sense to cause harm to innocent people, even as a side effect, that is greater than the harm one thereby averts.

Miller does not indicate whether he thinks that harms to different people might count in different ways in assessing proportionality. But once the distinction between narrow and wide proportionality is drawn, it becomes evident that they do. Miller also does not address the question of whether some noncombatants, such as those who provide various forms of support for Taliban forces, could be liable to attack. For our purposes, we can assume that noncombatants in Afghanistan are not liable. In that case, any justification for harming them must be a necessity justification, which is subject to the more restrictive wide proportionality constraint. Miller's argument is unorthodox in that it also takes narrow proportionality seriously. He assumes that even though the Taliban's aims are unjust, so that Taliban fighters are what I call “unjust combatants,” the harms we inflict on them can nevertheless count against the permissibility of our military action.

Where do harms that U.S. forces suffer fit in this taxonomy? I have argued elsewhere, and will here assume, that fighting for a just cause is not a ground of liability to attack.Footnote ¹ I will further assume, at least for the sake of argument, that the aim of preventing the Taliban from returning to political power by force, against the will of the great majority of the Afghan population, is a just cause for war. If that is right, forces whose mission is restricted to the achievement of that aim are “just combatants” who are not liable to attack, so that the harms they suffer are, like harms to noncombatants, a matter of wide proportionality.Footnote ² But it does not follow that these harms count in the same way that harms caused to noncombatants by U.S. forces do. For in the assessment of whether U.S. action in Afghanistan is proportionate, it may make a difference that, while this action exposes U.S. and allied combatants to a risk of harm, it does not itself harm them.

Harms We Indirectly Cause to Noncombatants

Harms that our military action directly inflicts on noncombatants, either intentionally or unintentionally, are relatively well understood. The intentional infliction of serious harms on innocent noncombatants as a means of achieving U.S. aims in Afghanistan would be terrorism and is too unlikely ever to be justifiable to be worth discussing. As I noted, it is widely and, in my view, correctly believed that the foreseeable but unintended harming of innocent people as a side effect of military action is easier to justify on grounds of necessity, as the lesser evil. It is sometimes suggested that a useful heuristic, at least in a democratic society, for determining whether foreseeable harms to noncombatants would be proportionate in the wide sense is to consider whether one would think it permissible to inflict them if the victims were instead one's own fellow citizens.Footnote ³ The application of this test to the Afghanistan war is complicated by the fact that Afghan civilians are among the intended beneficiaries of the U.S. intervention. I nevertheless think that if one would find the ratio of sacrifice to benefit too high if the civilian victims and beneficiaries were one's own fellow citizens, one should draw the same conclusion when they are Afghan civilians instead.

It is more difficult to understand the relevance to wide proportionality of harms that one's action causes to noncombatants, though only indirectly. Consider the example Miller cites: under-five mortality. When infants die for lack of better medical care it is not because U.S. military action kills them; rather, it is at most because military action prevents them from being saved. This may not seem an important difference, but arguably it is. Suppose that it would be militarily advantageous for just combatants to kill ten unjust combatants who serve alternately as fighters and as medics. Each is scheduled to perform a lifesaving surgery on an innocent noncombatant that only he can perform; therefore, if just combatants kill them, they will prevent ten innocent people from being saved. Yet killing the ten combatants might be proportionate in the wide sense even if it would be disproportionate to kill them in different circumstances in which doing so would kill ten innocent bystanders as a side effect. Wide proportionality, therefore, is not just a matter of numbers. How bad side effects are caused is also relevant.

The cogency of Miller's claims also depends on whether his assumptions about causation are true. For it is only if U.S. involvement in the war in Afghanistan is a significant causal factor for high under-five morality that one could reasonably expect the rate to decrease if the United States were to withdraw from the fighting. One admittedly imperfect test would be to determine whether the rate of under-five morality has been higher since the war began in 2001 than it was during the period between the withdrawal of Soviet forces from Afghanistan in 1989 and the renewal of war in 2001. Miller answers this question when he notes that the rate has been “essentially unchanged since 1990” (p. 110). That observation is evidence against his assumption that it is the current war that is responsible for the continued high rate. One could, of course, respond by claiming that rule by the Taliban also kept the rate high. But if this is right, intervention by the United States simply replaced one cause of a high rate with a different cause, so that the same high rate would probably have continued even in the absence of the U.S. war. Thus, if the United States were to withdraw but civil war were to continue, or if the Taliban were again to seize control, there is little reason to suppose that the rate would decrease. Indeed, it is arguable that the best prospect for reducing the under-five mortality rate would be by means of improvements in public health that might come only through a U.S. victory, assuming, perhaps overoptimistically, that the Obama administration would be less indifferent to the well-being of Afghans than the Bush administration was.

Intervening Agency

In his discussion of harms to noncombatants caused by the Taliban, Miller concentrates on those caused as a side effect of action against U.S. and allied forces. There is a related phenomenon that raises similar issues—namely, when one's own action predictably provokes one's adversaries to harm innocent people intentionally. Examples include Nazi reprisals against French civilians in response to attacks by the Resistance and the intensification of efforts by Serbs to rid Kosovo of ethnic Albanians in response to the NATO intervention in 1999. In all such cases, it is unclear how harms that will be inflicted, whether intentionally or unintentionally, by the voluntary action of others affect the proportionality of one's own action.

The significance of this issue in Afghanistan now is different from what it was prior to the U.S. invasion. While the Taliban were treating many people brutally prior to 2001, they were not then killing noncombatants as a side effect of military action. That began only as a result of the U.S. intervention. Now, however, if the United States were to withdraw without ending the war, the Taliban would continue to fight and, thus, to kill innocent people as a side effect. So the relevant question now is whether more noncombatants (and if so, how many more), will be killed by the Taliban as a side effect if the United States continues to fight than would be killed by them in the continued fighting if the United States were to withdraw.

As the example of the combatants who are also medics suggests, wide proportionality seems to be sensitive to the ways in which harms to innocent people are caused. It is therefore possible that harms that our action provokes our adversaries to cause have less weight in assessments of wide proportionality than equivalent harms that we ourselves inflict on innocent people. Some people, indeed, think that harms caused through the intervening agency of others have no weight in determining the proportionality of our action, at least if our action would otherwise be permissible.

Others think that, insofar as our action is a precondition of the occurrence of such harms, they have the same weight they would have if they were caused directly by our action. To the extent that they are predictable, on this view, harmful acts that our adversaries will do (though only in response to our action), are no different from harms caused by natural forces that are triggered by our action. They should be discounted only for the uncertainty of their occurrence. A third view, intermediate between these extremes, is that harms proximately caused by the intervening agency of others must be discounted, not just for the uncertainty of their occurrence, but also for their dependence on the free action of others.

The first of these views implies that if the only action I can take to prevent a mugger from stealing my wallet will provoke him to react in a way that will kill innocent bystanders as a side effect, the prospect of those killings does not count against the permissibility of my action. Similarly, it implies that the Christian fanatics in Florida whose burning of the Koran predictably provoked Muslim fanatics to go on violent rampages are justified in believing, as they do, that the many killings incited by their action do not count against the permissibility of that action.Footnote ⁴ That cannot be right.

The second view is also open to objections. If an evil person could kill millions of innocent people by pressing a button and was prepared to press it whenever his will was opposed, he could, on this second view, make it disproportionate for anyone to resist his wrongful action, assuming they could not deprive him of access to his button. A similar claim applies in practice to anyone who is willing and able to harm innocent people to a degree sufficient to make defensive action against him disproportionate. He can shield his wrongdoing against permissible intervention simply by credibly threatening even worse wrongdoing, thereby reducing decent people to moral paralysis. Many people find these implications of the second view intolerable.

It is, however, unavoidable that evil people can have this power over those who are morally scrupulous. Only the first of the three views precludes our vulnerability to this form of moral blackmail, and that view is unacceptable. The third view, which discounts harms for intervening agency, reduces our vulnerability but cannot eliminate it. Theists may take up the matter with the deity, but for the rest of us the responses are limited. One response is, of course, to try to design and create legal and political institutions that limit the power of evil people. Another is to recognize that even if defiance of a moral blackmailer may seem disproportionate, it can have, beyond its immediate effects, a further good effect—that is, the enhanced deterrence of future blackmail—that may render it proportionate.

There is, perhaps surprisingly, much to be said in favor of the second view, that harms that our action will predictably provoke others to inflict on innocent people have the same weight in determining the proportionality of our action as equivalent harms that are immediate side effects of that action. Suppose that there are two equally effective means of achieving our just aim. One would predictably kill 100 innocent people as an immediate side effect, but the just aim is sufficiently important that these killings would be proportionate. The other option would have no immediate side effects but would be certain to provoke a response from our adversary that would kill 101 different innocent people as a side effect. Unless harms that we provoke our adversaries to cause have as much weight in wide proportionality as harms we ourselves cause, we ought to choose the second means. Because I think we ought in this case to do what will cause the fewest killings of innocent people, this example seems to me to support the second view about intervening agency. This is important for thinking about proportionality in Afghanistan, where the proximate cause of the great majority of civilian casualties is action by the Taliban rather than the United States.Footnote ⁵

Harms Caused to U.S. and Allied Forces by the Taliban

What role, if any, do the expected harms that the Taliban will predictably inflict on U.S. and allied combatants have in determining the proportionality of military action by the United States and its allies? If we assume that anti-Taliban forces are just combatants, harms they suffer are a matter of wide rather than narrow proportionality, if they are relevant to proportionality at all. But even if the Taliban have the power to make U.S. and allied military action disproportionate by killing innocent bystanders in response, it may seem odd to suppose that they could make action by U.S. and allied forces disproportionate, and therefore impermissible, by killing too many U.S. and allied combatants. I think, however, that the oddness of this idea derives from the assumption that these forces are fighting voluntarily, in a strict sense of that term. If individuals willingly risk their lives to protect others from the Taliban, the Taliban cannot, it seems, make it impermissible for them to do that by credibly threatening to kill them. This suggests that harms suffered by those who freely choose to fight for a specific just cause do not count in determining whether their action is proportionate in the wide sense.

Most U.S. combatants in Afghanistan do not, however, fight voluntarily in this sense. Few joined the military specifically to fight the Taliban. U.S. soldiers joined for a great variety of reasons and would now face punishment if they were to refuse to fight. If the harms they suffer are excessive in relation to the importance of preventing the Taliban from regaining political control, it could be disproportionate and therefore impermissible for U.S. leaders to require them to continue to fight, thereby making continued U.S. war in Afghanistan disproportionate.

There seems, moreover, to be no reason to discount the weight of these harms relative to that of harms to other innocent or nonliable people that U.S. and allied action provokes the Taliban to cause. And vice versa: if the harms that U.S. and allied soldiers will suffer at the hands of the Taliban count fully in determining whether continued war in Afghanistan would be proportionate, the harms that innocent civilians will suffer from acts that the Taliban will commit, though only if U.S. and allied forces continue to fight, should count fully as well. The conclusion of this section thus reinforces the conclusion of the previous section.

Harms that U.S. and Allied Forces Inflict on the Taliban

Next consider Miller's most controversial theoretical claim, that harms that military action by U.S. and allied forces inflicts on Taliban fighters can count against the permissibility of their continued war in Afghanistan. He cites three ways in which these harms are relevant to proportionality. The first two are that they sometimes arouse sympathy and support for the Taliban and that they indirectly harm those who are specially related to the victims. I agree that these considerations are relevant, though because they are matters of wide proportionality, I will not discuss them here. Miller's third reason is that the mere fact that these are harms to people constitutes a reason not to inflict them. Again I agree, but I think this reason is less important in the context than Miller may suppose.

As I noted, I accept, in contrast to the dominant view in just war theory, that both war and individual acts of war can be disproportionate in the narrow sense—that is, objectionable solely because of their effects on unjust combatants. In the Falklands War, for example, there were 1,750 Argentine casualties but only 1,800 inhabitants of the islands. Although Britain's aim was just, it can be argued that its war violated the narrow ad bellum proportionality requirement.Footnote ⁶ Similarly, it could be a violation of the narrow in bello requirement to kill a large number of unjust combatants who do not threaten anyone's life and together make only a small contribution to their side's unjust cause. Yet, as I also noted earlier, if a person is morally liable to be harmed in a certain way, harming him in that way does not count toward making the act that harms him disproportionate. So if a Taliban fighter acts in a way that makes him morally liable to be killed, killing him is proportionate in the narrow sense, and thus the effect on him is morally fully justified and accounted for. This kind of judgment is, at least in one sense, insensitive to numbers. Provided that each Taliban is liable to be killed, or has no right not to be killed, narrow proportionality is unaffected by the number killed.

But there is another way in which, in practice, narrow proportionality tends to be affected by numbers. The degree to which a person is liable to defensive action, and thus how much harm it can be proportionate to inflict on him, depends on various factors, such as the magnitude of the threat he poses, the degree to which he is morally responsible for that threat, whether the threat can be eliminated or diminished (and if so, by how much) by harming him, and so on. But it is often the case that how much an individual unjust combatant contributes to the threat his side poses depends on how many such combatants there are. Often, the more there are, the less each one individually contributes. And the less an individual contributes to a threat, the less good it will do to eliminate his contribution. And the less good it will do to eliminate his contribution, the less harm he is liable to suffer as a means of eliminating that contribution. If, for example, eliminating his individual contribution to a threat will prevent only a relatively small expected harm, he cannot be liable to be killed as a means of eliminating his contribution. Killing him would therefore be disproportionate in the narrow sense. This provides the explanation of why it would be wrong, in Miller's example, to kill all the members of the Pakistani army as a means of eliminating the threat of an unjust annexation of Kashmir by Pakistan.

It is not obvious, however, that these considerations are relevant to the killing of Taliban fighters in Afghanistan. While the responsibility of unjust combatants is often diminished by the fact that they fight under duress and under the influence of carefully crafted government propaganda, these excuses are not available to members of the Taliban. If an individual Taliban fighter makes a significant contribution to the aim of imposing an unrepresentative and highly repressive government on the people of Afghanistan, against the will of the great majority, and threatens the lives of those who oppose this imposition as well as the lives of innocent bystanders, he thereby makes himself liable to be killed as a means of eliminating the threats he poses or to which he contributes. To the extent that this is true of each Taliban fighter, the war in Afghanistan is not disproportionate in the narrow sense.

Final Reflections

I also agree with Miller, perhaps even more than he agrees with himself, that the aim of preserving American global dominance is a wholly inadequate reason for continuing the war. He correctly observes that this aim is not a just cause for war: no one in Afghanistan has made himself liable to be killed in the service of maintaining the United States' dominant position in the world. Miller also interprets me correctly when he notes that I accept that a war could in principle be morally justified on grounds of necessity, or lesser evil, even when those who are warred against had not made themselves liable to be attacked. But while this form of justification for harming innocent people does have significant application in the in bello context—it is, for example, the only justification for harming innocent people as a side effect of military action—its application in ad bellum contexts is highly restricted precisely because the wide proportionality constraint is so demanding, particularly when what must be justified is the intentional killing of people who are not liable to be killed. The expected benefits (usually in the form of the prevention of wrongful harms) of a war justified on grounds of necessity must substantially outweigh the harms the war can be expected to cause. And, as Miller shows with a few examples selected from many he could have cited, the advertised benefits of American dominance for people other than Americans often turn out to involve an awful lot of suffering and death.

Just as Middle Eastern dictators will not be able to retain power indefinitely by slaughtering those they wish to rule, so the United States will not be able to maintain its power in the world by means of war. There are better ways of trying to maintain the strength of the U.S. economy than squandering our wealth on protracted wars in the Middle East. And in most cases (Libya being an apparent exception), there are better ways of supporting the spread of democracy and respect for human rights than the use of military force. Rather than sacrificing the lives of our young and devoting our ingenuity and wealth to destroying things, we would do better, in Afghanistan and elsewhere, to use our resources to provide medical aid (for example, in reducing under-five mortality), education, and other forms of assistance necessary for a society in which people flourish and terrorism does not. Realists will find this comically naïve, but I really believe that generosity and respect for ways of life we do not share—dispositions that are despised by Tea Partiers and other opponents of foreign aid—are often far more effective, even as tools of foreign and economic policy, than war.