An open access publication of the American Academy of Arts & Sciences
Fall 2016

Just & Unjust Targeted Killing & Drone Warfare

Michael Laban Walzer
View PDF

Targeted killing in the “war on terror” and in war generally is subject to familiar and severe moral constraints. The constraints hold across the board; they don't change when drones are the weapon of choice. But the ease with which drones can be used, the relative absence of military risks and political costs, makes it especially tempting not only to use drones more and more, but also to relax the constraining rules under which they are used. It seems clear that the rules have, in fact, been relaxed in the course of the American experience with drone warfare – by presidential decision and without public debate. This essay is an argument for the opening up of the decision process to democratic scrutiny and in defense of the familiar constraints.

MICHAEL WALZER, a Fellow of the American Academy since 1971, is Professor Emeritus of Social Science at the Institute for Advanced Study. He was for many years the Coeditor of Dissent and a contributing editor to The New Republic. He is the author of Just and Unjust Wars (1977), Spheres of Justice (1983), Thick and Thin: Moral Argument at Home and Abroad(1994), and, most recently, The Paradox of Liberation: Secular Revolutions and Religious Counter-revolutions (2015).

It is always a hard question whether new technologies require the revision of old arguments. Targeted killing isn’t new, and I am going to repeat an old argument about it. But targeted killing with drones? Here the old argument, though it still makes sense, leaves me uneasy.

First things first. Untargeted killing, random killing, the bomb in the supermarket, café, or bus station: we call that terrorism, and its condemnation is critically important. No qualifications, no excuses: this is wrongfulness of the first order. We need to be firm in rejecting all apologetic efforts on behalf of terrorists. But someone who takes aim at a particular person, a political official, say, is engaged in a different activity. He may be a “just” assassin, as in Albert Camus’s play by that name, though I don’t think that the justice of the killing depends, as Camus argues, on the killer’s willingness to accept death himself – obviously, Camus hadn’t heard of suicide bombers.1  Justice in assassinations depends on the character of the targeted official, the character of the regime he or she serves, and the immediate political circumstances: what else is there to do? There are often many other things that should be done, or attempted, before sending in the assassins.

But even if assassination is a wrongful act, as it usually is in history if not in literature, the wrongfulness is of a second order. By aiming carefully at a person thought to be guilty of something, by choosing weapons that can be used with discrimination, the assassin indicates his rejection of indiscriminate killing and also his concern about collateral damage. Someone in his organization probably thought that it would be better to kill the official’s extended family or to put a bomb in the restaurant where he and “his kind” regularly dine. A “just” assassin refuses to do that – or, at least, he doesn’t do it. Hence his act, if it is wrong, is not as wrong as terrorism is.

We should set clear limits on political assassinations. In democracies, they can never be justified; the killing of an American president, even if you disagree radically with his tax policies, is easy to make out as a criminal act. It is only the blood of tyrants that waters the tree of liberty. And even with tyrants, a trial is preferable to an assassination whenever it is possible to bring down the tyrannical regime without killing its leader. Postwar and postrevolutionary trials of tyrannical leaders are often criticized as “show trials” or “victor’s justice.” The trials are certainly meant to “show” something: they serve as an argument against the old regime and in favor of the new one. But they can also be just if the deposed leaders are accorded the same rights to which ordinary criminals are entitled. We know what a fair trial looks like, and a fair trial ranks far above assassination as a practical and also a moral response to tyranny.2  The trial of surviving Nazi leaders at Nuremburg is an example of “victor’s justice” that gave the second word in that phrase its due.

The international laws of war bar the killing of political leaders on the grounds that they are the ones who will, in the end, negotiate the peace treaty, and that is a moral as well as a legal argument, given the value of peace. But some political leaders, with whom we cannot imagine negotiating, are legitimate targets: Hitler is the obvious example. Killing Hitler would have been “extrajudicial” but entirely justified. Tyrants do have to be targeted, however; blowing up the neighborhood where they live is not a moral option.

In wartime, military leaders are obviously legitimate targets. A sniper sent to a forward position to kill a visiting colonel or general is engaged in targeted killing, but no one will accuse him of acting extrajudicially and therefore wrongly. We distinguish the military leaders of the state from its political leaders. It is probably best to think of insurgent organizations in the same way and to make the same distinction. If they have separated their political cadre from their military cadre, as in the case of Sinn Fein and the Irish Republican Army, it is only members of the second group who should be targeted (but only if they can’t be arrested). We may eventually decide, with whatever reluctance, to negotiate with the first group. I don’t believe that the same distinction is morally required in the case of terrorist organizations, which the IRA was, though it may be prudent to make it, as the British did.

Individuals who plan, organize, recruit for, or participate in terrorist attacks are, all of them, legitimate targets. It would be better to capture them and bring them to trial, but that is often not a reasonable option: the risks are too high; innocent bystanders would be killed in the attempt; the planning would take time and the terrorist attacks are imminent or actual. In such cases, the idea that we are at war with terrorists makes sense. More often the “war” is police work, and targeted killing by the police is not permissible.

A van carrying Al Qaeda militants in Afghanistan in 2002 was obviously a legitimate target: Afghanistan was a zone of war where enemies could rightly be attacked. By contrast, in Philadelphia (the city of brotherly love), the targeted killing of the same militants in the same van would not be justified; they would have to be arrested, indicted, provided with defense attorneys, and brought to trial. Philadelphia is a zone of peace where we don’t encounter enemies but rather criminals, whom we treat differently even if they share the ideological commitments of our enemies. What then about Yemen, where we actually attacked a van carrying Al Qaeda militants in 2002 (and where we have killed many times since, not always hitting our target)?3 Yemen was not then a zone of war; nor was it a zone of peace where it was possible to capture criminals and bring them to trial. Much of the “war on terror” is being fought in places like Yemen, and the engagements don’t seem to fit the zone of war or the zone of peace model; they aren’t exactly armed conflict, and they aren’t exactly law enforcement. I think that we have to work by analogy here. If the case can be made that Yemen is closer to Afghanistan than it is to Philadelphia, the attack on the van can be justified. But perhaps we need tougher criteria in intermediate cases like this one than we do in an actual war zone.4

I want to insist again that in a zone of peace, targeted killing is absolutely ruled out. Governments that make “kill lists” of their domestic political opponents and send out death squads to murder them are murderous and tyrannical governments without legitimacy. Lists of enemies abroad have their own problems, which I will come back to: Who makes the list? What are the criteria for inclusion and exclusion? But domestic political opposition raises an entirely different set of problems: opponents have to be engaged democratically and nonviolently; they have to be given a chance to win.

Insurgents who are not terrorists, who fight like soldiers, should be treated like soldiers. If they are captured, they are entitled to “benevolent quarantine for the duration of the war.” They are not criminals, to be imprisoned and brought to trial. But the targeted killing of insurgents in wartime is no different from the targeted killing of terrorists: all killing is subject to the same moral constraints. It will have to meet strict standards of proportionality and, given that the target is usually a single person, it will be difficult to justify injury to innocent bystanders. Remember the standard rule: collateral damage – that is, the death or injury of noncombatants – must not be disproportionate to the value of the military target. Even a high-ranking insurgent or terrorist leader is not so important – especially since these leaders are quickly replaced – as to justify many civilian deaths. I can’t say how many or, better, how few might be justifiable in any particular case, but a historical example suggests the caution that is required. After the attack on the USS Cole in October 2000, the Clinton administration undertook a search for Osama Bin Laden, intending to kill him and any members of his organization who were with him. The CIA located his camp in Afghanistan, but its surveillance photos showed children’s swings in front of some of the huts. The camp was not attacked.5

So this is the rule: targeting has to be undertaken with great care; collecting information about the targeted individuals, their daily schedules, and their families and neighbors is critically important, and if the necessary work involves risks for agents in the field, if the surveillance photos aren’t clear enough, the risks must be accepted before the killing can be justified. If force is used on the ground rather than from the air, the commandos must get close enough to make sure they are killing the right people. These are standard requirements for the zone of war. In the intermediate zone, the proportionality standards might well be tougher, the required care would be more extensive, and greater risks would have to be accepted in the collection of information or in the actual attack.

(I won’t attempt a discussion here of the constitutional question: can an American citizen be targeted and killed in a foreign country by agents of the American government? I will just say, without argument, that if someone is involved in terrorist attacks on innocent men and women or in military attacks on American soldiers, it is hard to see why his nationality should protect him from counter-attack.)

Thus far, I have sketched the moral argument about targeted killing; it overlaps with the legal argument, but since I am not a lawyer, that isn’t my subject. Now, does it make any difference, morally, if the actual killing is the work of a drone, operated by a technician sitting in a room five thousand miles away? Surely the same criteria apply to the drone as to any more closely manned machine. Why should we think it different from the sniper’s rifle?6  Indeed, this is a weapon ideally suited to asymmetric warfare, where it isn’t easy to get snipers in range. Asymmetric wars are not fought along a “front”; the insurgents don’t wear uniforms, they hide among civilians, and they prepare for battle in cross-border sanctuaries. They are more vulnerable to drones than to any more conventional weapon – so why should we worry about using drones?

But there are reasons to worry: killing-by-drone poses no immediate risk to the persons running the drone and so it is technically easier than other forms of targeted killing; there is no need to maneuver armed men into difficult terrain or contested airspace. I should qualify that last statement: drones are actually flown from bases fairly near their targets and it requires some 170 people to maintain the drones and get them into the air; these people are vulnerable to counterattack (though I don’t know of any actual attacks).7  The drone operators, however, aren’t vulnerable, and so killing-by-drone is also politically easier than other forms of targeted killing: it invites us to imagine a war in which there won’t be any casualties (on our side), no veterans who spend years in VA hospitals, no funerals. The easiness of fighting with drones should make us uneasy. This is a dangerously tempting technology. Of course, intelligence gathering will still be risky, but the drones “see” so much more than any soldier or agent in the field that they make fieldwork seem less important. It isn’t less important, but appearances here have consequences. Drones combine the capacity for surveillance with the capacity for precise attack. The technology is so good that the temptation is very strong to relax the criteria in accordance with which it is used so that drones can be used more and more often.

This is what seems to have happened with the U.S. armed forces or, more accurately, with the CIA in Pakistan and Yemen (and possibly other places). We need to be careful here, for the exact character of our policy and its consequences on the ground are still disputed; much is not known or not known with any degree of certainty. Our government’s commitment to secrecy with regard to drone warfare makes moral analysis problematic, and is itself one of the problems.

The most important dispute is about who is actually being killed: how many insurgents/terrorists and how many civilians? U.S. government officials and their congressional supporters have alternately claimed that no civilians have been killed in drone attacks, that the number killed each year is in the single digits, and, in President Obama’s words, that “drones have not caused a huge number of civilian casualties.” Independent researchers, some critical of drone warfare and some supportive, have released estimates for drone attacks in Pakistan (up to early 2015) that range from 2,200 to 4,000 people killed overall, including 150 to 900 civilians. The differences are extreme: by some accounts, civilian casualties make up only 7 percent of the total; by others, 40 percent. In July 2016, the Office of the Director of National Intelligence issued the first official figures, covering all airstrikes outside conventional war zones (that is, in Pakistan, Libya, Somalia, and Yemen) in the Obama years. In 473 strikes, the report says, between 2,372 and 2,581 combatants and between 64 and 116 civilians were killed.8  Taking the high figure for civilians and the low figure for combatants, this suggests that only about 4 percent of all casualties were civilians, a little over half of the lowest estimate produced by the most sympathetic independent researchers. It also indicates that if no strike produced more than one civilian death, then 357 strikes produced none at all: a remarkable result, which has already drawn highly skeptical responses.9  I don’t know what President Obama’s “huge number” would be, but it seems fairly certain that more civilians have been killed than our government would like to admit.10

The nonlethal costs imposed on the civilian population are also disputed. These costs have been described and harshly condemned in the Stanford International Human Rights and Conflict Resolution Clinic and nyu Global Justice Clinic’s report Living Under Drones, which is based largely on interviews in Pakistan. The authors describe what life is like in a Pakistani village with a drone hovering overhead, sometimes for days at a time. The hovering allows for accurate targeting of a particular individual; it is possible to wait and watch for the best opportunity for an attack. At the same time, the report says, the buzz of the hovering drone terrifies the villagers, who live in constant fear of an explosion from the sky.11

The fear may indeed be constant; there is much evidence for that and for the buzzing drone. But the very effectiveness of drone attacks raises questions about these accounts of the fear they provoke. Attacking drones must hover at such high altitudes that they can’t be seen or heard. If they didn't do that, the intended targets, who presumably know they are targets, would simply stay out of sight.12  Reconnaissance drones may fly at lower altitudes. Or, maybe, drones sometimes hover visibly and audibly precisely in order to terrify the villagers, so that they expel Taliban militants hiding among them. We really don’t know. Like much else in the early history of drone warfare, reports from the field are not necessarily accurate, but they may be accurate in part. And it is undoubtedly true that the experience of drone attacks leaves fear behind; villagers who have watched one explosion from the sky may plausibly fear another. So the Stanford/NYU report may have something important to tell us. Drones don't win hearts and minds. There are political arguments, in addition to moral arguments, against their overuse.

But I want to focus here on one striking example of how the moral criteria were relaxed, or seem to have been relaxed, in order to justify the costs that drone warfare imposes on innocent people. According to a May 2012 article in The New York Times by Jo Becker and Scott Shane, President Obama adopted “a disputed method of counting civilian casualties” that made it (or makes it; I am not sure what tense to use here and in the following sentences) much easier to claim that the casualties caused by a drone attack were “not disproportionate” to the value of the target. In effect, it “counts all military-age males in a strike zone as combatants” (so it was actually a way of not counting civilian casualties). If the targeted insurgent or terrorist leader was surrounded by, or simply in the vicinity of, a group of men between the ages of fifteen and sixty (and even drone surveillance can’t be precise about that), an attack was permitted, and the dead or injured individuals were not counted as collateral damage subject to the proportionality rule, but rather as legitimate military targets. But this wasn’t targeted killing. The men who happened to be in the vicinity of the actual target were never themselves targeted; they were not the specific object of the attack; we had no knowledge of what they had done, or were doing, or were planning to do.

There are ancient precedents for this sort of thing. According to Thucydides (in his History of the Peloponnesian War), when the Athenians captured the rebellious city of Melos, they “slew all the men of military age.” And according to the biblical book of Deuteronomy, when the Israelites besieged a city and “God delivers it into your hands … you shall put all its males to the sword.” Since the Deuteronomist goes on to exclude children, the two policies, Greek and Israelite, are identical.13  The new American doctrine isn’t the same as these two. We are not aiming to kill all the men of military age, but we have made them all liable to be killed. We have turned them into combatants, without knowing anything more about them than their gender and (approximate) age. That wasn’t right in ancient Greece or Israel, and it isn’t right today.

This new doctrine of liability may well have been reconsidered and rescinded after the New York Times report. U.S. policy with regard to drone warfare is, as I have said, still largely secret. The Obama administration has issued new rules for drone attacks, which the president described in a speech at West Point in 2014: “In taking direct action, we must uphold standards that reflect our values. That means taking strikes only when we face a continuing, imminent threat, and only where … there is near certainty of no civilian casualties.”14  But this says nothing about how we define “civilians.” The report of the Director of National Intelligence cited above states that the U.S. government, using many different sources of information, including assets on the ground and “geospatial intelligence,” determines whether people killed in airstrikes had “undertaken certain acts that reliably connote meaningful integration into an enemy group.”15  Exactly how “reliable” is defined is not specified. Nor do we know what the president’s “near certainty” means. After a recent drone attack that killed an American hostage held by the Taliban, an unnamed “senior administration official” told The New York Times that “you wonder whether the intelligence community’s definition of ‘near certainty’ is the same as everybody else’s.”16

These definitional problems are evident again in the case of “signature attacks” directed not against a particular individual with whom we are well acquainted, whose hostility we are certain about, but rather at assemblies or encampments of suspicious individuals.17  In these attacks, there is no “near certainty” that innocent men and women won’t be killed (and probably also no “imminent threat”). It seems that signature attacks have been cut back radically since President Obama’s new rules were promulgated. But they probably have not been stopped entirely. Again, we don’t know as much as we should.

All this suggests how important it is to open up the process by which lists of targets are put together and decisions about drone attacks are made. It can’t be right that a couple of people appointed by the president, with the president looking on, make these decisions entirely on their own. Particular decisions about particular targets will always be made by a small group, but these decisions should be subject to periodic review by a larger number of people inside the government but independent of the president’s office (and of the CIA). The general criteria for selecting targets in the zone of war and in the intermediate zone should also be considered and debated not only inside but also outside the government, by the body of citizens, by all of us. I don’t object to these killings because they are “extrajudicial”; my argument here refers only to dangerous enemies who can’t be captured and brought to trial. But I do want their deaths to be the subject of ongoing political and moral arguments, and there should be known government officials accountable to the rest of us for attacks that go badly wrong.

There is another issue raised by drone technology, which, even if not now real, certainly warrants discussion. Drones not only make it possible for us to get at our enemies, they may also lead us to broaden the list of enemies, to include presumptively hostile individuals and militant organizations simply because we can get at them – even if they aren’t actually involved in attacks against us. One example of this is the attempt by American officials to expand the target list in Afghanistan to include drug dealers, on the grounds that the Taliban benefited from the drug trade. The attempt was apparently blocked by our NATO allies.18  Our engagement with Islamist groups in West Africa that right now have only local ambitions is another example: these don’t seem to be enemies in the same sense as the people we are attacking in Afghanistan and Pakistan. The engagement may be justifiable, but it definitely needs justification; it should be, though it hasn’t yet been, the subject of public debate.

Drone warfare, when it involves only the targeted killing of real enemies, can be justified under tough constraints. But these constraints are not easy to maintain and enforce in the circumstances of asymmetric war. Up until now, at least, high-tech armies like our own have not been able to win these wars. Drones give us what I suspect is only an illusion of victory, though it is a powerful illusion. They make it possible to attack enemies in places where the conventional use of ground troops, even commandos, is difficult, sometimes impossible; they provide a way of coping with a war that has no “front.” That’s why they have been called “the only game in town.” Whether victory is actually possible in this “game” is another question, and not for this essay. A long history of exaggerated expectations should make us skeptical about the possibility of winning wars or defeating insurgencies from the air. But all that is necessary for my qualified defense of drone warfare is that there are real advantages to be gained by the use of drones against our enemies. Those advantages cannot, however, justify any relaxation of the constraints that we ought (and that we claim) to uphold, however attractive it may be to play the “game” without limits. Adopting a broad definition of combatants or justifying signature attacks are options that should be rejected.

Indeed, we should think very carefully before revising and expanding the targeting rules. The moral and political value of drones lies in their precision, which means using them only against individuals or small groups of individuals whose critical importance we have established, about whom we have learned a great deal, and whom we can actually hit without killing innocent people nearby. Using them indiscriminately, as we have sometimes used artillery and bombers, isn’t morally right or politically wise.

This last point can be driven home very simply: imagine a world, in which we will soon be living, where everybody has drones.

I want to stop now and try to restate in my own words a few of the most common arguments against the qualified defense of drone warfare that I have just provided. And then I will defend my defense.19

Before I get to the strongest arguments, let me quickly dispose of one bad argument. The likelihood that many states and many insurgent and terrorist organizations will soon possess drones might be a reason for the United States to adopt a set of tough constraints on their use, but one could say exactly the opposite: that the United States should use drones right now without constraint in order to defeat their most dangerous future users. Why shouldn’t we go all-out while we have the technological advantage? Consider a useful though inexact analogy: in the late 1940s, some American “hawks” argued in favor of using nuclear weapons against the Soviet Union while we had them and they didn’t. That was a very bad argument, given the number of civilian deaths that would have resulted from a nuclear attack. A preventive war would have been a crime of vast proportions, and I think that American leaders understood this.

By contrast, even the fiercest, most unrestrained use of drones would not produce casualties on anything like that scale. The idea of preventive drone attacks is nonetheless a bad idea. For in fact we don’t know, and can’t know, who the most dangerous future users of drones will be. So whose future use should we prevent? The technology is already easy to acquire and will soon be easier; all sorts of people will have it and will use it; some of them will be enemies. There is no way to prevent all future use of drones. Nor would it be possible to defeat potential enemies by preventive drone attacks, for drones are most useful in asymmetric wars, where conventional victory is probably not available.

So the argument for an early and heavy use of drones fails. I now want to look at three arguments for the proposition that we shouldn’t use them at all; that we should seek to ban drone warfare. The first of these arguments begins from the plausible claim that there is no reason to think that future users of drone technology – states or nonstate actors; officials, insurgents, or terrorists – will use drones in accordance with the rules I have urged the United States to adopt. Surely other users will adopt rules of their own, suited to their interests and purposes, or they will ignore all the rules. And they will certainly use drones not only against specifically targeted individuals but against enemy targets more generally. Imagine Al Qaeda or isis with drones: why wouldn't their militants use them not only against a particularly fierce congressional supporter of the “war on terror,” on his way to lunch, say, but also against Congress as a whole? Or, claiming belligerent rights, against the Pentagon, or Fort Jackson, or any stateside military encampment? Or, more in line with their previous and current behavior, randomly, as a terrorist weapon, against the inhabitants of our cities? The CIA hasn’t always observed, and may not now be observing, the necessary moral rules. Why would anyone else observe them?

If this is right, then here is the real choice that we face. We can stop using drones entirely and work for an international treaty outlawing their use, manufacture, and sale, and prescribing strong sanctions against any state or organization that makes or uses them. Or we can acknowledge that they already are or will certainly become a common weapon, sometimes used legitimately, within proportionality limits, but also and probably more often used indiscriminately and illegitimately, beyond those limits. If those are the alternatives that we face, the choice should be easy: ban the drone while we can (if we can).

The second argument for a ban has less to do with our possible enemies than with our own political leaders. Drone attacks are a form of warfare uniquely available to the executive branch of the government. In addition to all the other ways in which they are “easy,” these attacks are easy for the president to undertake. He can strike in secret, without congressional approval, and without taking any political risks at home. He doesn’t require the participation or even the support of his fellow citizens. Conscientious objection to the use of drones would have little point, and political opposition would have none of its usual reasons, for the president would not be asking Americans to risk their lives; he would not be asking them to do anything. Democracy was born, we have been told, when Athens recruited its poorest citizens into the navy. Thus, Xenophon wrote: “It is the ordinary people who man the fleet … it is these people who make the city powerful, much more than the … noble and respectable citizens. This being so, it seems that all should share in public office.”20  And when it isn’t so, when the leaders of the state don’t need the citizens of the state, the democratic rights of “ordinary people” may be gravely weakened.

It can be argued, indeed, that this has already happened, that President Obama is waging war or, at any rate, killing putative enemies in military campaigns that have never been authorized by Congress, campaigns that American citizens know very little about.21  There seem to be few if any constitutional constraints on the use of drones to kill people. It is, of course, possible to imagine constitutional constraints, but surely it would be better to remove this tempting weapon from the president's arsenal. Once again: ban the drone.

The third argument for a ban is a radical extension of the second. Drones, in this view, are the harbingers of a new, totally mechanized and impersonal kind of warfare. They require us to think about what was once only science fiction: a war of machines to which human beings are entirely superfluous. Governments capable of waging wars of that sort will not only be free from any democratic control, they will also be free to deal brutally with any citizens who dissent from their policies (if there are any). Enemies of the state will be hunted by machines programmed to recognize or anticipate their enmity. This is the ultimate use of drone technology. Whoever controls the machines rules the country. Readers will recognize the apocalyptic overtones of this argument: drone warfare points us toward, or leads us inexorably to, a mechanized and soon-to-be totalitarian politics.22  And the best way, perhaps the only way, to hold off the radical mechanization of war is to act now, early in the process, and ban the drone.

I will respond to these three arguments in reverse order. There isn't much to say about the apocalypse; it remains at this point science fiction. One might think, though, that wars between machine armies might be much like wars between mercenary armies in the early modern period. With both sides eager to limit the damage to hugely expensive “soldiers,” they would fight very limited wars. The threat of domestic tyranny is more frightening, but we know the remedy for that: eternal vigilance, always the price of liberty. Ordinary democratic politics is inherently antiapocalyptic.

This is also the necessary response to the second argument, which claims that drone warfare expands executive power and undercuts constitutional government. What is at issue here is political will, not technological determinism. Congressional oversight depends on a will to oversee, and this has been largely missing in recent decades. New technologies always require new regulatory systems, and we have fallen far behind: we have not yet figured out how to regulate the use of guns. So, again, there has to be a will to design and enforce the necessary regulations. In any case, for the foreseeable future, it will not be possible actually to win wars without putting soldiers in the field, so the president will still have to seek the support of the people and their elected representatives. Wars will still be subject to democratic debate, so long as there are citizens ready with arguments.

But I have to admit that banning the drone is almost as tempting as using the drone. Wouldn't the future of democracy, indeed, the future of humankind, look a little brighter without this new technology? It is, however, very hard to see how a ban could possibly succeed. The ban on the use of poison gas, which has been relatively successful, makes for a helpful comparison. Poison gas is a weapon that not only fails to discriminate between soldiers and civilians, it also fails to discriminate between our soldiers and enemy soldiers. Using it over a city or town is terrorism on a large scale, which is already a war crime. Using it on the battlefield is militarily dangerous; the winds change, and the gas drifts back across the lines. It is necessary for all soldiers to wear masks, and even the most advanced models are clumsy and difficult to wear while fighting. Perhaps the strongest argument in favor of the ban is that our own soldiers find gas warfare horrifically frightening.

Drones are obviously very different. They are discriminating weapons that do not kill civilians, or not many civilians, so long as they are used with discrimination. They are relatively risk-free for our own soldiers, and the fear that they will be used by our enemies isn't especially frightening – except, perhaps, to political officials who imagine themselves as particular targets. For the rest of us, drones are probably less frightening than carpet bombing or any other sustained attack from the air. I am sure that drone strikes leave fear behind, but this will be a localized and temporary fear. The strong feelings that led to the ban on poison gas don't seem to exist in the case of drones.

In any case, it looks as if the proverbial cat is out of the bag, and we seem to be the ones who opened the bag. According to a report in The Los Angeles Times in February 2013, a private American company is selling predator drones to the United Arab Emirates, presumably with our government’s approval, though obviously without any public debate.23  Surveillance drones are already being used or considered for use by police forces in the United States and, I would guess, in other countries as well. Surveillance isn’t the same as killing, but the technology is close enough so that if nonlethal drones are widely used, lethal ones won’t be far behind. In the not-too-distant future, very small surveillance drones, with limited range, will be sold in hardware stores, so that you can watch your neighbors from the sky or follow your wandering children. The defense of privacy is not my subject here, so I won’t comment on the legitimacy of domestic surveillance. But given the likely proliferation of drones of all sorts, there doesn’t seem any prospect of an enforceable ban on their military use.24

So we have to come back to the establishment of constraints on how they are used: moral and legal rules of engagement for drone warfare. This is the only possible response to the first argument for banning the drone. Of course, the rules will be violated, but that’s true of all the rules of war; indeed, it’s true of all legal and moral rules, like the ones that govern democratic politics, say, or market behavior. Still, it is important to have rules, if only to make possible the public condemnation and perhaps also the legal indictment of those who violate them. Public opinion is an important force in international politics today, especially with regard to asymmetric warfare, so it is worthwhile to educate the public about the moral and immoral use of drones. One way to establish the moral rules is for the United States, the first country to use drones on a large scale, to proclaim and observe a code for this kind of warfare. And one way to press this project upon the U.S. government is for all of us who write and worry about these issues to be firm in our critique of signature strikes, of the immoral incorporation of military-age males into the class of combatants, and of the studied ambiguity of our leaders’ pronouncements about the rules they claim to be following.

Philosopher Allen Buchanan and political scientist Robert Keohane have argued for something more than this: the establishment of an international drone accountability regime.25  They recognize the visionary character of their proposal, but argue that it is possible to imagine a “step by step” process through which states establish their own regulatory codes and look for reciprocity from other states – and from nonstate organizations. In any case, their proposal has the great advantage of forcing us to think about what the right regulations are and about the degree of visibility that would be necessary to make them morally and politically enforceable (Buchanan and Keohane don’t envision a legally coercive regime).

The rules of the national and international codes shouldn’t be anything radically new. They should be reasonably detailed, emphatic, and widely publicized, and they should attempt a definition of something like Obama’s “near certainty.” The codes should deny that the use of drones is or can be entirely risk-free; it requires intelligence work that is always dangerous, and that should be defined as morally necessary.26  Drones should aim with great care at known individuals chosen in accordance with criteria that are publicly defended. The proportionality standards should be tough, and they too should be publicly defended.

The United States could adopt a code of this sort right now, and if we did, we would have a right to insist that other countries and insurgent organizations of all sorts follow suit. If some don’t, as some won’t, our criticism of their behavior will, at least, be honest and strong.


  • 1Albert Camus, The Just Assassins in Caligula and Three Other Plays, trans. Stuart Gilbert (New York: Vintage, 1958).
  • 2See my introductory essay in Regicide and Revolution: Speeches at the Trial of Louis XVI (Cambridge: Cambridge University Press, 1974), which defends the trial of Louis XVI.
  • 3On civilian casualties in Yemen, see Open Society Justice Initiative, Death by Drone: Civilian Harm Caused by U.S. Targeted Killing in Yemen (New York: Open Society Foundations, 2015).
  • 4Critics of drone warfare consider cross-border attacks in places like Pakistan and Yemen especially bad. See Jeanne Mirer, “U.S. Policy of Targeted Killing with Drones: Illegal at Any Speed,” in Drones and Targeted Killing: Legal, Moral, and Geopolitical Issues, ed. Marjorie Cohn (Northampton, Mass.: Olive Branch Press, 2015), 135–168. Not enough is said by critics like Mirer about the responsibility of the governments of the countries in which these attacks take place. See the Open Society booklet on drone strikes in Pakistan for a discussion that, I think, gets things right: Open Society Foundations, After the Dead are Counted: U.S. and Pakistani Responsibilities to Victims of Drone Strikes (New York: Open Society Foundations, 2014).
  • 5Steve Cole, Ghost Wars (New York: Penguin Books, 2004), 145–146.
  • 6I have been greatly helped in thinking about drone warfare by Hugh Gusterson, who made available to me an advance copy of his just-published book Drone (Cambridge, Mass.: The MIT Press, 2016). The following books have also been helpful: Scott Horton, Lords of Secrecy: The National Security Elite and America’s Stealth Warfare (New York: Nation Books, 2015); Benjamin Wittes and Gabriella Blum, The Future of Violence: Robots and Germs, Hackers and Drones (New York: Basic Books, 2015); Peter Singer, Wired for War: The Robotic Revolution and Conflict in the 21st Century (New York: Penguin, 2009); Cohn, ed., Drones and Targeted Killing; Sarah Knuckey, ed., Drones and Targeted Killings: Ethics, Law, and Politics (New York: IDebate Press, 2015); Adam Rothstein, Drone (New York: Bloomsbury, 2015); Gregoire Chamayou, A Theory of the Drone (New York: The New Press, 2015); and Medea Benjamin, Drone Warfare: Killing by Remote Control (London: Verso, 2013).
  • 7Gusterson, Drone, 21.
  • 8Office of the Director of National Intelligence, “Summary of Information Regarding U.S. Counterterrorism Strikes Outside Areas of Active Hostilities,”
  • 9Charlie Savage and Scott Shane, “U.S. Reveals Death Toll from Airstrikes Outside War Zones,” The New York Times, July 1, 2016,
  • 10Gusterson, Drone, 83–87. See also Alice K. Ross, “Documenting Civilian Casualties,” in Drones and Targeted Killing, ed. Cohn, 99–131. The reluctance to acknowledge civilian casualties obviously makes it difficult for the relatives of victims to claim the compensation that the U.S. government says it is ready to pay.
  • 11See Stanford International Human Rights and Conflict Resolution Clinic and NYU Global Justice Clinic, Living Under Drones: Death, Injury, and Trauma to Civilians from U.S. Drone Practices in Pakistan (2012), See also Rothstein, Drone, 131–132: “Pashtun tribespeople in Pakistan refer to the drones overhead as ‘wasps’ or ‘mosquitoes’ due to their sound.”
  • 12Even the most nuanced accounts are contradictory: Gusterson quotes reporters who liken the sound of drones to “lawnmowers in the sky,” but then describes a targeted killing taking place “without warning.” Gusterson, Drone, 40–43.
  • 13The Thucydides quote is from Richard Schlatter, ed., Hobbes’s Thucydides (New Brunswick, N.J.: Rutgers University Press, 1975). See also Deuteronomy 20:13–14.
  • 14The White House Office of the Press Secretary, “Remarks by the President at the United States Military Academy Commencement Ceremony,” May 28, 2014, The speech is also available on YouTube.
  • 15Office of the Director of National Intelligence, “Summary of Information Regarding U.S. Counterterrorism Strikes.”
  • 16Peter Baker and Julie Hirschfeld Davis, “Amid Errors, Obama Publicly Wrestles with Drones’ Limits,” The New York Times, April 24, 2015.
  • 17Signature attacks are defined in Cohn, ed., Drones and Targeted Killing, 13. The phrase was originally classified; see Gusterson, Drone, 93–96.
  • 18See Jane Mayer, “The Predator War,” in Drones and Targeted Killing, ed. Cohn, 70–71; and Der Spiegel, January 28, 2009.
  • 19For a no-holds-barred critique, to which I will respond only in part, see Benjamin, Drone Warfare.
  • 20Xenophon, Constitution of the Athenians, 1:1–2.
  • 21For a critique of executive war-making, see Peter W. Singer, “Do Drones Undermine Democracy?” in Drones and Targeted Killings, ed. Knuckey, 36–372. But Singer does not advocate a ban on the use of drones.
  • 22Gregoire Chamayou offers a full-scale account, with French philosophical élan, of our totalitarian future, already partly present, in his A Theory of the Drone.
  • 23There have also been more recent sales: see, for example, the Center for the Study of the Drone, “Weekly Roundup 11/9/2015,”, reporting on the forthcoming sale of Hellfire missiles and laser-guided bombs to the Italian Air Force to arm its fleet of Reaper drones.
  • 24For the full range of possibilities, see Wittes and Blum, The Future of Violence, ch. 1. See Dan Fesperman, Unmanned (New York: Vintage Crime, 2015) for a fictional account of the dangers of the commercialization and proliferation of drone technology.
  • 25Allen Buchanan and Robert O. Keohane, “Toward a Drone Accountability Regime,” Ethics and International Affairs 29 (1) (2015): 1–23.
  • 26A reader pointed out that in the case of U.S. drone attacks, the intelligence work is often done not by U.S. military personnel, but by native informants. Nonetheless, the risks they take are taken on our behalf; they are necessary participants in the attack, and its success is radically dependent on their reliability.