The Argument about Humanitarian Intervention

There is nothing new about human disasters caused by human beings. We have always been, if not our own, certainly each other’s worst enemies. From the Assyrians in ancient Israel and the Romans in Carthage to the Belgians in the Congo and the Turks in Armenia, history is a bloody and barbaric tale. Still, in this regard, the twentieth century was an age of innovation, first-and most important-in the way disasters were planned and organized and then, more recently, in the way they were publicized. I want to begin with the second of these innovations-the product of an extraordinary speedup in both travel and communication. It may be possible to kill people on a very large scale more efficiently than ever before, but it is much harder to kill them in secret. In the contemporary world there is very little that happens far away, out of sight, or behind the scenes; the camera crews arrive faster than rigor mortis. We are instant spectators of every atrocity; we sit in our living rooms and see the murdered children, the desperate refugees. Perhaps horrific crimes are still committed in dark places, but not many; contemporary horrors are well-lit. And so a question is posed that has never been posed before-at least never with such immediacy, never so inescapably: What is our responsibility? What should we do?

In the old days, “humanitarian intervention” was a lawyer’s doctrine, a way of justifying a very limited set of exceptions to the principles of national sovereignty and territorial integrity. It is a good doctrine, because exceptions are always necessary, principles are never absolute. But we need to rethink it today, as the exceptions become less and less exceptional. The “acts that shock the conscience of humankind”-and, according to the nineteenth-century law books, justify humanitarian intervention-are probably no more frequent these days than they were in the past, but they are more shocking, because we are more intimately engaged by them and with them. Cases multiply in the world and in the media: Somalia, Bosnia, Rwanda, East Timor, Liberia, Sierra Leone, and Kosovo in only the past decade. The last of these has dominated recent political debates, but it isn’t the most illuminating case. I want to step back a bit, reach for a wider range of examples, and try to answer four questions about humanitarian intervention: First, what are its occasions? Second, who are its preferred agents? Third, how should the agents act to meet the occasions? And fourth, when is it time to end the intervention?

Occasions
The occasions have to be extreme if they are to justify, perhaps even require, the use of force across an international boundary. Every violation of human rights isn’t a justification. The common brutalities of authoritarian politics, the daily oppressiveness of traditional social practices-these are not occasions for intervention; they have to be dealt with locally, by the people who know the politics, who enact or resist the practices. The fact that these people can’t easily or quickly reduce the incidence of brutality and oppression isn’t a sufficient reason for foreigners to invade their country. Foreign politicians and soldiers are too likely to misread the situation, or to underestimate the force required to change it, or to stimulate a “patriotic” reaction in defense of the brutal politics and the oppressive practices. Social change is best achieved from within.

I want to insist on this point; I don’t mean to describe a continuum that begins with common nastiness and ends with genocide, but rather a radical break, a chasm, with nastiness on one side and genocide on the other. We should not allow ourselves to approach genocide by degrees. Still, on this side of the chasm, we can mark out a continuum of brutality and oppression, and somewhere along this continuum an international response (short of military force) is necessary. Diplomatic pressure and economic sanctions, for example, are useful means of engagement with tyrannical regimes. The sanctions might be imposed by some free-form coalition of interested states. Or perhaps we should work toward a more established regional or global authority that could regulate the imposition, carefully matching the severity of the sanctions to the severity of the oppression. But these are still external acts; they are efforts to prompt but not to preempt an internal response. They still assume the value, and hold open the possibility, of domestic politics. The interested states or the regional or global authorities bring pressure to bear, so to speak, at the border; and then they wait for something to happen on the other side.

But when what is going on is the “ethnic cleansing” of a province or country or the systematic massacre of a religious or national community, it doesn’t seem possible to wait for a local response. Now we are on the other side of the chasm. The stakes are too high, the suffering already too great. Perhaps there is no capacity to respond among the people directly at risk and no will to respond among their fellow citizens. The victims are weak and vulnerable; their enemies are cruel; their neighbors indifferent. The rest of us watch and are shocked. This is the occasion for intervention.
We will need to argue, of course, about each case, but the list I’ve already provided seems a fairly obvious one. These days the intervening army will claim to be enforcing human rights, and that was a plausible and fully comprehensible claim in each of the cases on my list (or would have been, since interventions weren’t attempted in all of them). We are best served, I think, by a stark and minimalist version of human rights here: it is life and liberty that are at stake. With regard to these two, the language of rights is readily available and sufficiently understood across the globe. Still, we could as easily say that what is being enforced, and what should be enforced, is simple decency.

In practice, even with a minimalist understanding of human rights, even with a commitment to nothing more than decency, there are more occasions for intervention than there are actual interventions. When the oppressors are too powerful, they are rarely challenged, however shocking the oppression. This obvious truth about international society is often used as an argument against the interventions that do take place. It is hypocritical, critics say to the “humanitarian” politicians or soldiers, to intervene in this case when you didn’t intervene in that one-as if, having declined to challenge China in Tibet, say, the United Nations should have stayed out of East Timor for the sake of moral consistency. But consistency isn’t an issue here. We can’t meet all our occasions; we rightly calculate the risks in each one. We need to ask what the costs of intervention will be for the people being rescued, for the rescuers, and for everyone else. And then, we can only do what we can do.

The standard cases have a standard form: a government, an army, a police force, tyrannically controlled, attacks its own people or some subset of its own people, a vulnerable minority, say, territorially based or dispersed throughout the country. (We might think of these attacks as examples of state terrorism and then consider forceful humanitarian responses, such as the NATO campaign in Kosovo, as instances of the “war against terrorism,” avant la lettre. But I won’t pursue this line of argument here.) The attack takes place within the country’s borders; it doesn’t require any boundary crossings; it is an exercise of sovereign power. There is no aggression, no invading army to resist and beat back. Instead, the rescuing forces are the invaders; they are the ones who, in the strict sense of international law, begin the war. But they come into a situation where the moral stakes are clear: the oppressors or, better, the state agents of oppression are readily identifiable; their victims are plain to see.

Even in the list with which I started, however, there are some nonstandard cases-Sierra Leone is the clearest example-where the state apparatus isn’t the villain, where what we might think of as the administration of brutality is decentralized, anarchic, almost random. It isn’t the power of the oppressors that interventionists have to worry about, but the amorphousness of the oppression. I won’t have much to say about cases like this. Intervention is clearly justifiable but, right now at least, it‘s radically unclear how it should be undertaken. Perhaps there is not much to do beyond what the Nigerians did in Sierra Leone: they reduced the number of killings, the scope of the barbarism.

Agents
“We can only do what we can do.” Who is this “we”? The Kosovo debate focused on the United States, NATO, and the UN as agents of military intervention. These are indeed three political collectives capable of agency, but by no means the only three. The United States and NATO generate suspicion among the sorts of people who are called “idealists” because of their readiness to act unilaterally and their presumed imperial ambitions; the UN generates skepticism among the sorts of people who are called “realists” because of its political weakness and military ineffectiveness. The arguments here are overdetermined; I am not going to join them. We are more likely to understand the problem of agency if we start with other agents. The most successful interventions in the last thirty years have been acts of war by neighboring states: Vietnam in Cambodia, India in East Pakistan (now Bangladesh), Tanzania in Uganda. These are useful examples for testing our ideas about intervention because they don’t involve extraneous issues such as the new (or old) world order; they don’t require us to consult Lenin’s, or anyone else’s, theory of imperialism. In each of these cases, there were horrifying acts that should have been stopped and agents who succeeded, more or less, in stopping them. So let’s use these cases to address the two questions most commonly posed by critics of the Kosovo war: Does it matter that the agents acted alone? Does it matter that their motives were not wholly (or even chiefly) altruistic?

In the history of humanitarian intervention, unilateralism is far more common than its opposite. One reason for this is obvious: the great reluctance of most states to cede the direction of their armed forces to an organization they don’t control. But unilateralism may also follow from the need for an immediate response to “acts that shock.” Imagine a case where the shock doesn’t have anything to do with human evildoing: a fire in a neighbor’s house in a new town where there is no fire department. It wouldn’t make much sense to call a meeting of the block association, while the house is burning, and vote on whether or not to help (and it would make even less sense to give a veto on helping to the three richest families on the block). I don’t think that the case would be all that different if, instead of a fire, there was a brutal husband, no police department, and screams for help in the night. Here too, the block association is of little use; neighborly unilateralism seems entirely justified. In cases like these, anyone who can help should help. And that sounds like a plausible maxim for humanitarian intervention also: who can, should.

But now let’s imagine a block association or an international organization that planned in advance for the fire, or the scream in the night, or the mass murder. Then there would be particular people or specially recruited military forces delegated to act in a crisis, and the definition of “crisis” could be determined-as best it can be-in advance, in exactly the kind of meeting that seems so implausible, so morally inappropriate, at the moment when immediate action is necessary. The person who rushes into a neighbor’s house in my domestic example and the political or military commanders of the invading forces in the international cases would still have to act on their own understanding of the events unfolding in front of them and on their own interpretation of the responsibility they have been given. But now they act under specified constraints, and they can call on the help of those in whose name they are acting. This is the form that multilateral intervention is most likely to take, if the UN, say, were ever to authorize it in advance of a particular crisis. It seems preferable to the different unilateral alternatives, because it involves some kind of prior warning, an agreed-upon description of the occasions for intervention, and the prospect of overwhelming force.

But is it preferable in fact, right now, given the UN as it actually is? What makes police forces effective in domestic society, when they are effective, is their commitment to the entire body of citizens from which they are drawn and the (relative) trust of the citizens in that commitment. But the UN’s General Assembly and Security Council, so far, give very little evidence of being so committed, and there can’t be many people in the world today who would willingly entrust their lives to UN police. So if, in any of my examples, the UN’s authorized agents or their domestic equivalents decide not to intervene, and the fire is still burning, the screams can still be heard, the murders go on-then unilateralist rights and obligations are instantly restored. Collective decisions to act may well exclude unilateral action, but collective decisions not to act don’t have the same effect. In this sense, unilateralism is the dominant response when the common conscience is shocked. If there is no collective response, anyone can respond. If no one is acting, act.

In the Cambodia, East Pakistan, and Uganda cases, there were no prior arrangements and no authorized agents. Had the UN’s Security Council or General Assembly been called into session, it would almost certainly have decided against intervention, probably by majority vote, in any case because of great-power opposition. So, anyone acting to shut down the Khmer Rouge killing fields or to stem the tide of Bengalese refugees or to stop Idi Amin’s butchery would have to act unilaterally. Everything depended on the political decision of a single state.

Do these singular agents have a right to act or do they have an obligation? I have been using both words, but they don’t always go together: there can be rights where there are no obligations. In “good Samaritan” cases in domestic society, we commonly say that passersby are bound to respond (to the injured stranger by the side of the road, to the cry of a child drowning in the lake); they are not, however, bound to risk their lives. If the risks are clear, they have a right to respond; responding is certainly a good thing and possibly the right thing to do; still, they are not morally bound to do it. But military interventions across international boundaries always impose risks on the intervening forces. So perhaps there is no obligation here either; perhaps there is a right to intervene but also a right to refuse the risks, to maintain a kind of neutrality-even between murderers and their victims. Or perhaps humanitarian intervention is an example of what philosophers call an “imperfect” duty: someone should stop the awfulness, but it isn’t possible to give that someone a proper name, to point a finger, say, at a particular country. The problem of imperfect duty yields best to multilateral solutions; we simply assign responsibility in advance through some commonly accepted decision procedure.

But perhaps, again, these descriptions are too weak: I am inclined to say that intervention is more than a right and more than an imperfect duty. After all, the survival of the intervening state is not at risk. And then why shouldn’t the obligation simply fall on the most capable state, the nearest or the strongest, as in the maxim I have already suggested: Who can, should? Nonintervention in the face of mass murder or ethnic cleansing is not the same as neutrality in time of war. The moral urgencies are different; we are usually unsure of the consequences of a war, but we know very well the consequences of a massacre. Still, if we follow the logic of the argument so far, it will be necessary to recruit volunteers for humanitarian interventions; the “who” who can and should is only the state, not any particular man or woman; for individuals the duty remains imperfect. Deciding whether to volunteer, they may choose to apply the same test to themselves-who can, should-but the choice is theirs.

The dominance that I have ascribed to unilateralism might be questioned-commonly is questioned-because of a fear of the motives of single states acting alone. Won’t they act in their own interests rather than in the interests of humanity? Yes, they probably will or, better, they will act in their own interests as well as in the interests of humanity; I don’t think that it is particularly insightful, merely cynical, to suggest that those larger interests have no hold at all (surely the balance of interest and morality among interventionists is no different than it is among noninterventionists). In any case, how would humanity be better served by multilateral decision-making? Wouldn’t each state involved in the decision process also act in its own interests? And then the outcome would be determined by bargaining among the interested parties-and humanity, obviously, would not be one of the parties. We might hope that particular interests would cancel each other out, leaving some kind of general interest (this is in fact Rousseau’s account, or one of his accounts, of how citizens arrive at a “general will”). But it is equally possible that the bargain will reflect only a mix of particular interests, which may or may not be better for humanity than the interests of a single party. Anyway, political motivations are always mixed, whether the actors are one or many. A pure moral will doesn’t exist in political life, and it shouldn’t be necessary to pretend to that kind of purity. The leaders of states have a right, indeed, they have an obligation, to consider the interests of their own people, even when they are acting to help other people. We should assume, then, that the Indians acted in their national interest when they assisted the secession of East Pakistan, and that Tanzania acted in its own interests when it moved troops into Idi Amin’s Uganda. But these interventions also served humanitarian purposes, and presumably were intended to do that too. The victims of massacre or “ethnic cleansing” disasters are very lucky if a neighboring state, or a coalition of states, has more than one reason to rescue them. It would be foolish to declare the multiplicity morally disabling. If the intervention is expanded beyond its necessary bounds because of some “ulterior” motive, then it should be criticized; within those bounds, mixed motives are a practical advantage.

Means
When the agents act, how should they act? Humanitarian intervention involves the use of force, and it is crucial to its success that it be pursued forcefully; the aim is the defeat of the people, whoever they are, who are carrying out the massacres or the ethnic cleansing. If what is going on is awful enough to justify going in, then it is awful enough to justify the pursuit of military victory. But this simple proposition hasn’t found ready acceptance in international society. Most clearly in the Bosnian case, repeated efforts were made to deal with the disaster without fighting against its perpetrators. Force was taken, indeed, to be a “last” resort, but in an ongoing political conflict “lastness” never arrives; there is always something to be done before doing whatever it is that comes last. So military observers were sent into Bosnia to report on what was happening; and then UN forces brought humanitarian relief to the victims, and then they provided some degree of military protection for relief workers, and then they sought (unsuccessfully) to create a few “safe zones” for the Bosnians. But if soldiers do nothing more than these things, they are hardly an impediment to further killing; they may even be said to provide a kind of background support for it. They guard roads, defend doctors and nurses, deliver medical supplies and food to a growing number of victims and refugees-and the number keeps growing. Sometimes it is helpful to interpose soldiers as “peacekeepers” between the killers and their victims. But though that may work for a time, it doesn’t reduce the power of the killers, and so it is a formula for trouble later on. Peacekeeping is an honorable activity, but not if there is no peace. Sometimes, unhappily, it is better to make war.

In Cambodia, East Pakistan, and Uganda, the interventions were carried out on the ground; this was old-fashioned war-making. The Kosovo war provides an alternative model: a war fought from the air, with technologies designed to reduce (almost to zero!) the risk of casualties to the intervening army. I won’t stop here to consider at any length the reasons for the alternative model, which have to do with the increasing inability of modern democracies to use the armies they recruit in ways that put soldiers at risk. There are no “lower orders,” no invisible, expendable citizens in democratic states today. And in the absence of a clear threat to the community itself, there is little willingness even among political elites to sacrifice for the sake of global law and order or, more particularly, for the sake of Rwandans or Kosovars. But the inability and the unwillingness, whatever their sources, make for moral problems. A war fought entirely from the air, and from far away, probably can’t be won without attacking civilian targets. These can be bridges and television stations, electric generators and water purification plants, rather than residential areas, but the attacks will endanger the lives of innocent men, women, and children nonetheless. The aim is to bring pressure to bear on a government acting barbarically toward a minority of its citizens by threatening to harm, or actually harming, the majority to which, presumably, the government is still committed. Obviously this isn’t a strategy that would have worked against the Khmer Rouge in Cambodia, but it’s probably not legitimate even where it might work-so long as there is the possibility of a more precise intervention against the forces actually engaged in the barbarous acts. The same rules apply here as in war generally: noncombatants are immune from direct attack and have to be protected as far as possible from “collateral damage”; soldiers have to accept risks to themselves in order to avoid imposing risks on the civilian population.

Any country considering military intervention would obviously embrace technologies that were said to be risk-free for its own soldiers, and the embrace would be entirely justified so long as the same technologies were also risk-free for civilians on the other side. This is precisely the claim made on behalf of “smart bombs”: they can be delivered from great distances (safely), and they never miss. But the claim is, for the moment at least, greatly exaggerated. There is no technological fix currently available, and therefore no way of avoiding this simple truth: from the standpoint of justice, you cannot invade a foreign country, with all the consequences that has for other people, while insisting that your own soldiers can never be put at risk. Once the intervention has begun, it may become morally, even if it is not yet militarily, necessary to fight on the ground-in order to win more quickly and save many lives, for example, or to stop some particularly barbarous response to the intervention.

That’s the moral argument against no-risk interventions, but there is also a prudential argument. Interventions will rarely be successful unless there is a visible willingness to fight and to take casualties. In the Kosovo case, if a NATO army had been in sight, so to speak, before the bombing of Serbia began, it is unlikely that the bombing would have been necessary; nor would there ever have been the tide of desperate and embittered refugees. Postwar Kosovo would look very different; the tasks of policing and reconstruction would be easier than they have been; the odds on success much better.

Endings
Imagine the intervening army fully engaged. How should it understand the victory that it is aiming at? When is it time to go home? Should the army aim only at stopping the killings, or at destroying the military or paramilitary forces carrying them out, or at replacing the regime that employs these forces, or at punishing the leaders of the regime? Is intervention only a war or also an occupation? These are hard questions, and I want to begin my own response by acknowledging that I have answered them differently at different times.

The answer that best fits the original legal doctrine of humanitarian intervention, and that I defended in Just and Unjust Wars (1977), is that the aim of the intervening army is simply to stop the killing. Its leaders prove that their motives are primarily humanitarian, that they are not driven by imperial ambition, by moving in as quickly as possible to defeat the killers and rescue their victims and then by leaving as quickly as possible. Sorting things out afterward, dealing with the consequences of the awfulness, deciding what to do with its agents-that is not properly the work of foreigners. The people who have always lived there, wherever “there” is, have to be given a chance to reconstruct their common life. The crisis that they have just been through should not become an occasion for foreign domination. The principles of political sovereignty and territorial integrity require the “in and quickly out” rule.

But there are three sorts of occasions when this rule seems impossible to apply. The first is perhaps best exemplified by the Cambodian killing fields, which were so extensive as to leave, at the end, no institutional base, and perhaps no human base, for reconstruction. I don’t say this to justify the Vietnamese establishment of a satellite regime, but rather to explain the need, years later, for the UN’s effort to create, from the outside, a locally legitimate political system. The UN couldn’t or wouldn’t stop the killing when it was actually taking place, but had it done so, the “in and quickly out” test would not have provided a plausible measure of its success; it would have had to deal, somehow, with the aftermath of the killing.

The second occasion is exemplified by all those countries-Uganda, Rwanda, Kosovo, and others-where the extent and depth of the ethnic divisions make it likely that the killings will resume as soon as the intervening forces withdraw. If the original killers don’t return to their work, then the revenge of their victims will prove equally deadly. Now “in and quickly out” is a kind of bad faith, a choice of legal virtue at the expense of political and moral effectiveness. If one accepts the risks of intervention in countries like these, one had better accept also the risks of occupation.

The third occasion is the one I called nonstandard earlier on: where the state has simply disintegrated. It’s not that its army or police have been defeated; they simply don’t exist. The country is in the hand of paramilitary forces and warlords-gangs, really-who have been, let’s say, temporarily subdued. What is necessary now is to create a state, and the creation will have to be virtually ex nihilo. And that is not work for the short term.

In 1995, in an article called “The Politics of Rescue,” published in these pages, I argued that leftist critics of protectorates and trusteeships needed to rethink their position, for arrangements of this sort might sometimes be the best outcome of a humanitarian intervention. The historical record makes it clear enough that protectors and trustees, under the old League of Nations, for example, again and again failed to fulfill their obligations; nor have these arrangements been as temporary as they were supposed to be. Still, their purpose can sometimes be a legitimate one: to open a span of time and to authorize a kind of political work between the “in” and the “out” of a humanitarian intervention. This purpose doesn’t cancel the requirement that the intervening forces get out. We need to think about better ways of making sure that the purpose is actually realized and the requirement finally met. Perhaps this is a place where multilateralism can play a more central role than it does, or has done, in the original interventions. For multilateral occupations are unlikely to serve the interests of any single state and so are unlikely to be sustained any longer than necessary. The greater danger is that they won’t be sustained long enough: each participating state will look for an excuse to pull its own forces out. An independent UN force, not bound or hindered by the political decisions of individual states, might be the most reliable protector and trustee-if we could be sure that it would protect the right people, in a timely way. Whenever that assurance doesn’t exist, unilateralism returns, again, as a justifiable option.

Either way, we still need an equivalent of the “in and out” rule, a way of recognizing when these longstanding interventions reach their endpoint. The appropriate rule is best expressed by a phrase that I have already used: “local legitimacy.” The intervening forces should aim at finding or establishing a form of authority that fits or at least accommodates the local political culture, and a set of authorities, independent of themselves, who are capable of governing the country and who command sufficient popular support so that their government won’t be massively coercive. Once such authorities are in place, the intervening forces should withdraw: “in and finally out.”
But this formula may be as quixotic as “in and quickly out.” Perhaps foreign forces can’t do the work that I’ve just described; they will only be dragged deeper and deeper into a conflict they will never be able to control, gradually becoming indistinguishable from the other parties. That prospect is surely a great disincentive to intervention; it will often override not only the benign intentions but even the imperial ambitions of potential interveners. In fact, most of the countries whose inhabitants (or some of them) desperately need to be rescued offer precious little political or economic reward to the states that attempt the rescue. One almost wishes that the impure motivations of such states had more plausible objects, the pursuit of which might hold them to their task. At the same time, however, it’s important to insist that the task is limited: once the massacres and ethnic cleansing are really over and the people in command are committed to avoiding their return, the intervention is finished. The new regime doesn’t have to be democratic or liberal or pluralist or (even) capitalist. It doesn’t have to be anything, except non-murderous. When intervention is understood in this minimalist fashion, it may be a little easier to see it through.
As in the argument about occasions, minimalism in endings suggests that we should be careful in our use of human rights language. For if we pursue the legal logic of rights (at least as that logic is understood in the United States), it will be very difficult for the intervening forces to get out before they have brought the people who organized the massacres or the ethnic cleansing to trial and established a new regime committed to enforcing the full set of human rights. If those goals are actually within reach, then, of course, it is right to reach for them. But intervention is a political and military process, not a legal one, and it is subject to the compromises and tactical shifts that politics and war require. So we will often need to accept more minimal goals, in order to minimize the use of force and the time span over which it is used. I want to stress, however, that we need, and haven’t yet come close to, a clear understanding of what “minimum” really means. The intervening forces have to be prepared to use the weapons they carry, and they have to be prepared to stay what may be a long course. The international community needs to find ways of supporting these forces-and also, since what they are doing is dangerous and won’t always be done well, of supervising, regulating, and criticizing them.

Conclusion
I have tried to answer possible objections to my argument as I went along, but there are a couple of common criticisms of the contemporary practice of humanitarian intervention that I want to single out and address more explicitly, even at the cost of repeating myself. A few repetitions, on key points, will make my conclusion. I am going to take Edward Luttwak’s critical review of Michael Ignatieff’s Virtual War* as a useful summary of the arguments to which I need to respond, since it is short, sharp, cogent, and typical. Ignatieff offers a stronger human rights justification of humanitarian warfare than I have provided, though he would certainly agree that not every rights violation “shocks the conscience of humankind” and justifies military intervention. In any case, Luttwak’s objections apply (or fail to apply) across the board-that is, to the arguments I’ve made here as well as to Ignatieff’s book.

First objection: the “prescription that X should fight Y whenever Y egregiously violates X’s moral and juridical norms would legitimize eternal war.” This claim seems somewhat inconsistent with Luttwak’s further claim (see below) that the necessity of fighting not only forever but everywhere follows from the fact that there are so many violations of commonly recognized norms. But leave that aside for now. If we intervene only in extremity, only in order to stop mass murder and mass deportation, the idea that we are defending X’s norms and not Y’s is simply wrong. Possessive nouns don’t modify morality in such cases, and there isn’t a series of different moralities-the proof of this is the standard and singular lie told by all the killers and “cleansers”: they deny what they are doing; they don’t try to justify it by reference to a set of private norms.

Second objection: “Even without civil wars, massacres, or mutilations, the perfectly normal, everyday, functioning of armies, police forces, and bureaucracies entails constant extortion, frequent robbery and rape, and pervasive oppression”-all of which, Luttwak claims, is ignored by the humanitarian interveners. So it is, and should be, or else we would indeed be fighting all the time and everywhere. But note that Luttwak assumes now that the wrongness of the extortion, robbery, rape, and oppression is not a matter of X’s or Y’s private norms but can be recognized by anyone. Maybe he goes too far here, because bureaucratic extortion, at least, has different meaning and valence in different times and places. But the main actions on his list are indeed awful, and commonly known to be awful; they just aren’t awful enough to justify a military invasion. I don’t think the point is all that difficult, even if we disagree about exactly where the line should be drawn. Pol Pot’s killing fields had to be shut down-and by a foreign army if necessary. The prisons of all the more ordinary dictators in the modern world should also be shut down-emptied and closed. But that is properly the work of their own subjects.

Third objection: “What does it mean,” Luttwak asks, “for the morality of a supposedly moral rule when it is applied arbitrarily, against some but not others?” The answer to this question depends on what the word “arbitrarily” means here. Consider a domestic example. The police can’t stop every speeding car. If they go after only the ones they think they will be able to catch without endangering themselves or anyone else, their arrests will be “determined by choice or discretion,” which is one of the meanings of “arbitrary,” but surely that determination doesn’t undermine the justice of enforcing the speeding laws. On the other hand, if they only go after cars that have bumper stickers they don’t like, if they treat traffic control as nothing more than an opportunity to harass political “enemies,” then their actions “arise from will or caprice,” another definition of “arbitrary,” and are indeed unjust. It’s the first kind of “arbitrariness” that ought to qualify humanitarian interventions (and often does). They are indeed discretionary, and we have to hope that prudential calculations shape the decision to intervene or not. Hence, as I have already acknowledged, there won’t be an actual intervention every time the justifying conditions for it exist. But, to answer Luttwak’s question, that acknowledgment doesn’t do anything to the morality of the justifying rule. It’s not immoral to act, or decline to act, for prudential reasons.

These three objections relate to the occasions for intervention, and rightly so. If no coherent account of the occasions is possible, then it isn’t necessary to answer the other questions that I have addressed. My own answers to those other questions can certainly be contested. But the main point that I want to make is that the questions themselves cannot be avoided. Since there are in fact legitimate occasions for humanitarian intervention, since we know, roughly, what ought to be done, we have to argue about how to do it; we have to argue about agents, means, and endings. There are a lot of people around today who want to avoid these arguments and postpone indefinitely the kinds of action they might require. These people have all sorts of reasons, but none of them, it seems to me, are good or moral reasons.

Michael Walzer is co-editor of Dissent. A slightly different version of this article was given as the Theodore Mitan Lecture at Macalester College, St. Paul, Minnesota.

*“No Score War,” Times Literary Supplement, (July 14, 2000), p. 11.