Jump to ContentJump to Main Navigation
Targeted KillingsLaw and Morality in an Asymmetrical World$

Claire Finkelstein, Jens David Ohlin, and Andrew Altman

Print publication date: 2012

Print ISBN-13: 9780199646470

Published to Oxford Scholarship Online: September 2012

DOI: 10.1093/acprof:oso/9780199646470.001.0001

Show Summary Details
Page of

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (www.oxfordscholarship.com). (c) Copyright Oxford University Press, 2018. All Rights Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a monograph in OSO for personal use (for details see http://www.oxfordscholarship.com/page/privacy-policy). Subscriber: null; date: 22 June 2018



Targeted Killings

Fernando R. Tesón

Oxford University Press

Abstract and Keywords

On May 2, 2011, a special unit of the U.S. Navy killed Osama bin Laden in Pakistan. Many rejoiced at this development and felt that justice had been served. However, emotion is no substitute for dispassionate moral analysis. This chapter examines the morality of targeted killings in general. It addresses the killing of bin Laden, but the discussion goes beyond that: it probes the morality of all targeted killings by liberal governments. It applies to targeted killings by the United States as well as other liberal regimes, and it explores the justification of the practice in wartime and in peacetime. Given that the United States and Israel have announced that they will continue to kill named targets, and given that not all contemplated targets are as villainous or dangerous as bin Laden, a moral evaluation of the practice is especially required.

Keywords:   targeted killing, Osama bin Laden, morality, liberal governments, United States, Israel

I. Introduction

On May 2, 2011, a special unit of the U.S. Navy killed Osama bin Laden in Pakistan. Many (this writer included) rejoiced at this development and felt that justice had been served.2 However, emotion is no substitute for dispassionate moral analysis. In this chapter I examine the morality of targeted killings in general. I address the killing of bin Laden, but my discussion goes beyond that: it probes the morality of all targeted killings by liberal governments. It applies to targeted killings by the United States as well as other liberal regimes, and it explores the justification of the practice in wartime and in peacetime. Given that the United States and Israel have announced that they will continue to kill named targets, and given that not all contemplated targets are as villainous or dangerous as bin Laden, a moral evaluation of the practice is especially required.

The deliberate killing of another human being is presumptively a deeply immoral act. Targeted killings are deliberate killings, so any discussion must start with a strong moral presumption against those acts. However, the prohibition has some exceptions: killing in war, self-defense, and law enforcement of various kinds.3 This chapter examines whether targeted killing (which is a species of assassination) can be one of those exceptions. (p.404)

First, some definitions are in order (these are conventional stipulations for expository purposes; no substantive issue turns on them). I define targeted killing as the extrajudicial intentional killing by the state of an identified person for a public purpose. This chapter will examine only those targeted killings conducted by a liberal democracy. The definition calls for some clarification. The word “extrajudicial” excludes from the definition all instances where someone is killed in execution of a lawful sentence (whether this is morally justified or not I will not address). The word “intentional” means that the assassin directly intends to kill the victim. It excludes from the definition all killings that are incidental to combat in war or revolution (serious as those are). Thus, it excludes not only unforeseen deaths, but also foreseen yet unintended deaths.4 In a targeted killing the victim is precisely identified: the lethal action is directed at him. The requirement that the victim be identified is necessary to distinguish targeted killing from the anonymous intentional killing of enemy combatants in war. Killing an enemy soldier on the battlefield is not a targeted killing in our sense. Finally, I use the expression “public purpose” loosely, to exclude private purposes such as revenge or personal gain. A public purpose is still normatively neutral: it may or may not be morally justified.

In this chapter I examine the various forms of targeted killing. I proceed on the premise that the moral rules that govern killing in peacetime are very strict but that they are relaxed during wartime. Section II addresses peacetime situations; that is, the strict moral rules governing a liberal government's resort to targeted killing outside the war context. I then turn in section III to targeted killing in conventional war, where the central question becomes whether the more permissive moral rules governing lethal force in war relax the strict rules that govern targeted killing in peacetime. Section IV begins the examination of the targeted killing of terrorists.5 I focus on the question of whether such killing must be analyzed under the stricter peacetime framework, or under the more permissive wartime framework, or under a third framework that borrows from the other two. The conclusions of this section are presumptive. I try to show that there are moral arguments in favor of these admittedly repulsive acts. Section V presents several objections to targeted killings and examines whether the force of those objections should make us revise our presumptive conclusions. At the very least, these objections show that targeted killings are troubling, even accepting the force of arguments in their favor. (p.405)

II. Targeted killing in peacetime

I start with the general concept of targeted killing, the extrajudicial intentional killing by a liberal government of a person for a public purpose. It will be convenient to start with targeted killing in peacetime, because clarifying that concept will help us analyze targeted killing in wartime and targeted killing of terrorists. During peacetime, the state can use lethal force only in very limited circumstances, mostly in self-defense or to protect persons from deadly threats. Beyond that, a suspected criminal is entitled to due process and may not be killed except in execution of a lawful sentence pronounced by a court of law after a finding of guilt (and this assuming, controversially, that the death penalty is justified). The liberal state's moral obligation not to kill a person without due process includes foreigners, even if they are not entirely protected by domestic constitutional guarantees. Foreigners have a right to life, which is universal. However, I would like to make a prima facie case for the occasional permissibility of targeted killing in peacetime, leaving for later whether such presumptive case may be defeated by the general objections against targeted killing. Because the moral presumption entailed by the right to life is strong, a targeted killing in peacetime can only be justified, if at all, under very stringent conditions. I provisionally suggest the following four conditions.6 The conditions are individually necessary and jointly sufficient for the legitimacy of targeted killing in peacetime, again, disregarding for now the general objections discussed at the end of this chapter.

  1. (a) The targeted killing will save many lives, including many innocent lives.

  2. (b) The public purpose of the targeted killing is just.

  3. (c) The target of the killing is morally culpable, a villain.

  4. (d) There are no non-lethal alternatives available, such as diplomatic threats or capture.

(a) Saving lives

Targeted killing in peacetime can only start to make moral sense if it is likely is to spare the lives of a significant number of innocent persons.7 In general, this happens when the targeted killing avoids war or a similar catastrophe such as genocide. Several examples come readily to mind; the most obvious is the morality of killing Hitler before the Second World War: arguably, that act would have spared the world terrible ordeals. Notice that killing Hitler in early 1939 would have been a targeted killing in peacetime. If the contemplated target, vile as he may be, is not (p.406) threatening innocent lives, then he may not be permissibly killed; the default prohibition against murder resurfaces. Let's set up two imaginary examples, both of which involve targeted killing in the form of assassinating a political leader.

Genocide in Rhodelia: Rhodelia is ruled by Caligula, a vicious dictator who is perpetrating genocide8 against his own population. His neighbor, Freeland, is a liberal democracy with the military capability to stop the atrocities. The government of Freeland can do one of three things. It can do nothing; it can invade Rhodelia and fight a predictably successful war of humanitarian intervention; or it can kill Caligula and thus end the genocide. Let us assume that doing nothing is morally problematic. Invading Rhodelia to stop the atrocities, while predictably successful, will result in significant collateral deaths of civilians,9 deaths of combatants on both sides, and physical destruction. However, sending a special operations team to kill Caligula will end his crimes and restore peace without any of these consequences. What should the government of Freeland do?

Planned aggression in the Chosen Kingdom: King Vlad, a charismatic absolutist monarch with delusions of grandeur, rules over The Chosen Kingdom, a militarily powerful nation. Against his advisers’ best judgment, Vlad is planning a massive invasion of his neighbors, all liberal democracies, who are dreading the impending catastrophe. The government of Sunland, the most powerful of these democracies, is considering action. It can do three things: wait for the aggression and then react defensively; invade preemptively; or send a sniper to kill Vlad and predictably avoid the war. Again, the impending war is likely to have terrible costs in blood and treasure, whether started by the Kingdom or by Sunland's preemptive strike. What should the government of Sunland do?

These examples show why a blanket prohibition of targeted killing in peacetime is, on closer inspection, too quick. Moral considerations may favor targeted killing over war. Targeted killing, a prima facie immoral act, may appear as preferable because it will avoid genocide or war while placing the cost on a culpable person. Many people die in war. Those who bear arms to resist unjust attacks against themselves or others put their lives at risk for a just cause. Every one of those deaths is murder because inflicted by an unjust warrior.10 Importantly, war also brings about the incidental deaths of civilians. As is well known, this is a highly problematic aspect of war. One important pacifist objection is that any war, no matter how “clean,” will bring about the deaths of civilians. These persons have not given up their right to life, so starting a war that predictably will kill them is morally problematic, even if the country that initiates it has a just cause. Maybe this worry can be addressed by a properly (p.407) formulated version of the doctrine of double effect, but the worry persists nonetheless, because even if one reluctantly thinks that bringing about those collateral deaths is permissible under the right circumstances, surely achieving the same results with no deaths is morally preferable. In the Rhodelia example, killing Caligula will save Caligula's present victims, plus the innocent lives (both just combatants and civilians on both sides) that would be lost should Freeland decide to start a war. It will also avoid the terrible physical destruction that war typically causes. In the Chosen Kingdom example, killing Vlad would have similar effects: it would avoid the terrible losses that the impending aggression would cause. The difference between these two peacetime cases is that Caligula is not threatening his neighbors yet killing his own citizens, while Vlad is not killing its own citizens yet threatening its neighbors.

(b) Just cause

However, saving lives, even in large numbers, does not suffice to justify targeted killing in peacetime. The government that orders the killing must itself not be on the wrong side of a conflict. The public purpose that defines targeted killing must be a normatively compelling just purpose in the sense of the just war tradition.11 In the hypothetical examples discussed above, killing Caligula is justified because Caligula is committing a crime against humanity; and killing Vlad is justified because he is about to unleash a war of aggression, another major crime. In these cases, this condition is identical to the first: the just cause is preventing the humanitarian catastrophe. However, the two conditions must be carefully distinguished. Suppose that country A has decided to unlawfully attack country B. The leader of country B will predictably resist. On learning this, the government of A, the aggressor, decides to kill the leader of B, on the grounds that doing so will avoid the impending war and save many lives. This targeted killing is murder, notwithstanding the fact that it will save many lives. The reason is that A, as the aggressor, lacks a just cause. To take a real-life example, in 1939 Hitler knew, or should have known, that Winston Churchill would lead Britain to war if Germany invaded Poland. Suppose Hitler would have ordered the killing of Churchill on the grounds that it would have forced Britain to compromise and thus avoid the impending war. This would have been murder because Germany was the aggressor. So while saving lives may be a just cause for a targeted killing, not all targeted killings that save lives have a just cause. Determining what causes are just exceeds the confines of this chapter. Suffice it to say here that a just cause for war is only the defense of persons and liberal institutions against unjustified threats or attacks against them. National glory, economic gain, strategic advantage, redress of non-lethal wrongs, or territorial expansion never constitute just cause.12 (p.408)

(c) Moral culpability

The object of targeted killing must be culpable of having created the threat to, or destruction of, human lives in the first place. In most cases he will be a villain.13 In peacetime, a liberal government intentionally targets someone to avoid a war (either a defensive war or a humanitarian intervention). By doing so, the liberal government prevents the deaths of many combatants (its own and the enemy's) and civilians. As we saw, just combatants and civilians are not culpable. Even if one thinks that not all enemy combatants are free of blame for fighting an unjust war, surely they are less culpable than their leader who sends them to fight. Targeting the villain instead of targeting those who fight at his behest places the cost on the morally culpable person. This is ostensibly preferable to an alternative that, we know, will bring about the deaths of many persons who had nothing to do with the critical situation that forced the liberal government to act. Targeted killing, then, performs a double task: it reduces the number of victims to one, and it inflicts lethal force on the blameworthy individual.

The target must be the person responsible for creating the unjust threat or inflicting unjust violence. The targeted killing cannot be justified by the result alone. In the examples presented, imagine that the liberal government knows that by killing the children of Caligula or Vlad (or a few randomly chosen civilians, or equally non-culpable targets) it will avert the impending catastrophe. Such action is morally impermissible by straight application of general deontological principles that forbid using innocent persons to achieve a morally justified end.14 The liberal government is not allowed to target someone affiliated with the enemy if that person is not sufficiently culpable (but it may target a sufficiently culpable henchman). Notice that this condition is distinct from the just cause condition, although they may sometimes overlap. The government may have a just cause but unjustifiably kill an innocent person to pursue that cause. In the examples above, Freeland's government has a just cause (stopping genocide), yet it is not allowed to target an innocent person in the pursuit of that cause, even if that killing would stop the genocide.15 Likewise, Sunland's government has a just cause (preventing aggression) but may not target, say, the Queen instead of the King.

The requirement that the target be culpable adds a retributivist ingredient to the justification of targeted killings.16 The just warrior is moved by the imperative of saving many lives in the context of a just fight. Any alternative he chooses will impose costs on someone. Given this, it is morally preferable to impose the cost on the person culpable for the lethal threat than on non-culpable persons. Yet moral guilt is a (p.409) necessary condition for justification, not a sufficient one. The liberal government does not tell the victim: “We kill you because you deserve it.” Rather, it tells him: “We must stop these deaths. We have several ways to do this, but the least costly way in a moral sense is to kill you, who are responsible for this predicament. Any alternative action will bring about the deaths of many innocent persons.” Requiring culpability avoids killing innocent persons and, in a sense, “punishing” them for the misdeeds of another. The paradox is that assassination, a morally repulsive notion, is more in accordance with the liberal traditions of the criminal law because it requires a sort of mens rea on the part of the target of lethal force. This seems preferable to the rather illiberal practice of killing anonymous soldiers who, for all we know, have, if at all, diminished responsibility for the impending or ongoing disaster.

(d) Lack of non-lethal alternatives

These three conditions (saving innocent lives, having a just cause, and targeting a villain) are still insufficient to justify targeted killing in peacetime. In addition, the liberal government must lack non-lethal alternatives to resolve the crisis. In the Rhodelia example, suppose that the government of Freeland can credibly threaten Caligula with major destruction if he does not relent. This threat is preferable to killing Caligula, as are all other diplomatic maneuvers to get Caligula to desist.17 The same reasoning is available in the Chosen Kingdom hypothetical: if Sunland has non-lethal alternatives to prevent the impending war, then it must use them.

Now why is this so? If part of the reason that makes targeted killing palatable is that the victim is culpable, then the government who avoids the killing through these diplomatic moves is getting the villain off the hook. Killing him would not only give him his due, but would also prevent him from ruling despotically and making the lives of everyone around him miserable, not to mention the probability that he will pose a similar threat in the future. In this case, forgoing the targeted killing may be worse in the long run. This happened when the Coalition decided not to kill Saddam Hussein in the 1991 Gulf War. Yet these alternatives to killing are preferable for two reasons. First, although the target is culpable, the killing deprives the target of the due process that would have ensued after capture. Most non-lethal alternatives at least preserve that possibility; and capture ensures it. To be sure, as we shall see in the discussion of targeted killing of terrorists, process is not due in all cases; in particular, it is not due when capture is impossible. But when capture is possible, it always precludes the permissibility of killing. Second, the prohibition of intentional killing is partly grounded in agent-relative reasons. (I return to this topic below.) By choosing alternatives to the killing, the governments of Freeland and Sunland avoid being killers. They avoid the state of affairs where they deliberately kill someone in cold blood. (p.410) Punishing Caligula or Vlad (the retributivist impulse) is not a reason compelling enough to outweigh the immorality of their killing those villains, of getting their hands dirty with the deliberate extrajudicial killing of another human being.

We can imagine, however, that some would prefer the targeted killing over these alternatives. A strong retributivist would insist that the villain get his due. On the opposite end, a strong consequentialist would insist that the government weigh costs and benefits and proceed with the killing if, all things considered, it would cause more good than the alternatives. These positions are dubious, however. In response to the retributivist, giving people their due may perhaps be a necessary condition for the legitimacy of criminal punishment (outside of war), but it cannot alone justify extrajudicial killing. In most cases, a targeted killing based only on the culpability of the target, where capture and prosecution are possible, amounts to vigilantism or revenge. The answer to the consequentialist is well known. The morality of action cannot be determined only by their net benefit. There are immoral ways to produce beneficial consequences, and deliberate killing is surely a likely candidate. This does not mean that targeted killing is never permissible, but it does mean that the threshold for permissibility is high. If the liberal government can avert the humanitarian catastrophe that alone would justify the targeted killing in the first place by resorting to non-lethal alternatives, then it must do so, even if it means the survival of the villain. A non-lethal alternative to killing the villain is, of course, capturing him and bringing him to justice. This alternative is not simply preferable to killing: killing is prohibited when low-cost capture is possible (I return to the issue of capture in the discussion of targeted killing of terrorists, below).

III. Targeted killing in conventional war

As is widely recognized, these strict conditions for the legitimacy of targeted killing are significantly relaxed in wartime. Consider a conventional war scenario. Two armies are facing each other. One of them is the aggressor, so it lacks a just cause; the other is resisting aggression, so it has a just cause. Let us stipulate (although this is contested)18 that in a conventional war only the combatants on the side that has a just cause have a moral license to kill enemy combatants. Further, all the deaths brought about by just combatants are legitimate only if they comply with the morality of war, including an appropriate version of the doctrine of double effect as reflected largely in the Geneva Conventions, in particular the principle that obligates belligerents to discriminate between combatants and civilians.19 (p.411)

Can the government on the right side of a war target a named enemy for killing? Consider first a targeted killing in the battlefield. Notwithstanding the claims of some authors,20 identifying the victim in advance does not alter the general permission to kill in combat.21 A commander says to a soldier: “There behind the machine-gun is Colonel Sanders, the enemy's battalion commander. With his great skill he is decimating our troops and making our lives miserable, so make sure you take him out.” Here the victim is identified by name and individually targeted, yet I take it most people would regard this killing as justified under the laws of war. The commander permissibly targets Sanders because Sanders, in the pursuit of an unjust cause, is taking an active part in combat. He is an unjust enemy combatant and as such may be permissibly killed, named or unnamed.

So let us consider targeted killing of an enemy combatant who is not on the battlefield. The guiding principle is that a targeted killing in war is justified only if it is sufficiently close, in a moral sense, to standard killing in war. If the killing is instead morally removed from that category, then it will be evaluated under the more restrictive standard of killing in peacetime. Stating this principle, however, is easier than applying it, so some elaboration is in order.

In war, the moral framework for targeted killing is more permissive. The first condition for peacetime, that the targeted killing must save many innocent lives, must be relaxed. In a just war, a necessary condition for the permissibility of a targeted killing is that the soldier should reasonably believe that it will increase the chances of victory. For in wartime it is permitted to kill enemy combatants for just that reason, assuming a just cause. Some may object that the rationale for killing enemy soldiers is self-defense, not a mere increase in the chances of victory. Even if the general rationale for killing in war is self-defense, this does not mean that the just warrior must feel threatened every time he faces an enemy combatant. I take it that those writers who claim self-defense as the proper rationale for war are thinking about the overall justification of the war, and not as a reason that applies in every individual case of combat. In other words, the aggressor has forced us to fight in self-defense. That is the reason why we, the just army, are fighting. We are defending ourselves from the aggressor. But that does not mean that in every case where I face the enemy soldier, an agent of the aggressor, I must feel threatened in order for my firing to be justified. The right way to look at this is to say that a soldier on the right side of a just war may permissibly kill an enemy soldier if that will increase the chances of victory (a fortiori he may kill the enemy soldier if his life is threatened). (p.412)

The second condition applicable to targeted killing in peacetime, that the victim must be personally culpable, disappears in wartime, when it comes to the killing of a combatant who is on the battlefield and fighting for an unjust cause. This combatant is a piece in the war machine of a government or group that is fighting for an unjust cause, so he is liable to lethal attack by the enemy even if he is not personally culpable.22 The rationale is the same for not requiring culpability as a condition for killing (without targeting) the combatants fighting on the unjust side. The soldier on the right side of a just war is facing an unjustified threat posed by the enemy. Whether the enemy soldiers are individually culpable or not, they are armed agents of the unjust enemy bent on destroying us (now or later). If this is correct, then it is correct also for the targeted killing of an unjust enemy combatant, as in the Colonel Sanders example above.

However, the concept of combatant is vague. Is it confined to combatants in the battlefield, or does it also it also include (i) combatants not in the battlefield; or (ii) non-military leaders of the enemy? Consider first enemy soldiers who are not in the battlefield. Again, two situations are possible. Colonel Sanders, the enemy's battalion commander, is not currently in the battlefield but is resting in his military headquarters. Some may object to this killing on the grounds that Sanders is not posing any threat. However, I think most people would say this killing is justified under the laws of war. The reason is, I believe, that the Colonel is simply restoring his strength to go back to the battlefield and continue his aggression (remember he is fighting for an unjust cause). Sanders and his subordinates are engaged in a continued unlawful fight, so the fact that he is resting now is irrelevant to the general justification for killing unjust enemies in war.23 We can say that Colonel Sanders is in combat, although not in the battlefield. Again, Lieber's worry is not applicable here. Colonel Sanders is not killed because he is an outlaw, but because he is a piece in the unjust lethal machinery that the unjust enemy has mounted against us.

The second situation occurs when the enemy combatant is geographically removed from combat altogether—not just from the battlefield, but from any military installation such as a barrack (but not wounded, as different principles apply). He is in a private setting. Imagine Colonel Sanders is on vacation. Is the liberal government justified in killing him, for example, by a sniper gunshot? Daniel Statman has argued that there is no moral distinction between killing an enemy soldier while he is in his military headquarters and when he is on vacation. In neither case is he participating in active combat, so to make the legitimacy of the killing depend on location is arbitrary. As Statman points out, this is even (p.413) clearer with respect to high-ranking officers, as they rarely pose an immediate threat.24

Yet I believe there is a moral difference between killing Colonel Sanders when he is in a combat role and killing him while on vacation. The distinction comports with a central purpose of the laws of war: to minimize deaths by confining as much as possible the destruction to the theater of operations. If the war conventions are in place, and those conventions prohibit killing enemies while on vacation, then killing him is morally impermissible on those grounds alone. Moreover, there are reasons to prohibit targeted killing of enemy combatants removed from the theater of operations, even conceiving the theater of operations quite broadly. This is a close call, but I think, contrary to Statman's claim, that the geographical location matters, for a couple of reasons. First, there is a general (not just conventional) obligation to minimize deaths in war where possible. Second, recall that in war the individual culpability of the victim (beyond the fact that he fights for an unjust cause) is not required for the permission to kill. If Colonel Sanders is commanding his troops or even resting in the barracks for next day's battle, he is participating in the war machine that threatens us. If he is on vacation he has removed himself from the war machine. Given that he may not be culpable, the default prohibition should re-emerge and assassinating him should be prohibited. These two reasons, plus the two general objections I will examine below, tip the balance against targeted killing in this case. The (admittedly tenuous) difference between killing Sanders while he is in the barracks and killing him while he is on vacation is that in the first case he is in combat, broadly conceived. In contrast, while on vacation Sanders has provisionally shed the role of combatant; he is truly acting in his civilian capacity. Given the longstanding conventional prohibition against this kind of targeted killing, and given the strong presumptive reasons against these kinds of acts, I think the scale should be tipped against permissibility (but the killing may be still justified under the stricter standard of targeted killing in peacetime examined above).

We turn now to the targeted killing in wartime of someone who is not formally a member of the enemy's army displayed on the battlefield. Here the situation changes and, again, there are two cases. The first is the case of the political leader of the enemy. Call it the Qaddafi case. As I write these lines, a civil war has just ended in Libya with Qaddafi's death.25 Were the rebels, aided by NATO, justified in targeting Qaddafi? If we view the Libyan conflict as one where NATO assists justified revolutionaries against a tyrant, then Qaddafi was a legitimate target in the war to liberate Libya.26 The reason is that Qaddafi was the commander-in-chief (p.414) of the unjust enemy forces. He was a combatant, even though for a while he sat in his comfortable Tripoli palace while his loyalists troops fought the rebels. The misgivings in Qaddafi's case arise, not because he was not a justified target in the civil war, but because apparently he was shot after he had surrendered.27 But aside from that difficulty, the assassination of Qaddafi is not morally different from the assassination of Hitler during the Second World War, and not very different from killing any enemy combatant in war. The leaders are part—indeed, the essential part—of the enemy's military structure. A combatant is a person who is part of the logical chain of agency leading to the unjust threat.28 The political leader is a crucial link in that chain. Moreover, the leader is culpable because he fully endorses the unjust cause and also sends others to die for that unjust cause. Most likely, he invented the unjust cause. His high position in the chain of agency in the pursuit of a criminal enterprise (since an unjust war is a criminal enterprise) is decisive. the killing of the political leader of an unjust enemy is, in a sense, morally overdetermined: The leader is morally culpable for unleashing and endorsing an unjust war, and he is an enemy combatant in the sense of jus in bello. He is a culpable enemy combatant. By the same token, it is not enough to claim, as a justification for a targeted killing, that the killing will avert the threat or win the war. For example, targeting Qaddafi's s relatives was impermissible, even if that act would have eliminated the threat or helped to achieve victory (because Qaddafi would collapse, or whatever) because those persons are not culpable.29 The just warrior cannot target just anyone whose death he reasonably thinks will help him win. Here as elsewhere, the doctrine of double effect applies, so one can think of cases where the incidental deaths of civilians could be justified en route to killing the villain.

The second case involves a high-ranked government official who is not the commander-in-chief and is not otherwise part of the enemy's military structure. Here, as in the previous case, the culpability condition applies. Suppose that a cabinet minister of the Third Reich is an important piece in the war effort. Killing him, the Allies think, will considerably improve chances of victory. Call this the Albert Speer case. Albert Speer, a civilian architect, was the Minister of Armaments during (p.415) the Third Reich.30 Would the Allies have been justified in killing him? I would think the answer is yes, provided that the stipulated conditions apply (killing him would increase the chances of victory; Speer is culpable; there are no non-lethal alternatives). However, the more removed the official is from the war effort, the less acceptable the killing will be. Speer is a relatively easy case, but what about, say, the German Undersecretary of Public Parks at the time? The weaker the connection with the war, the less probable will be that the killing will comply with the required conditions, and the closer the killing will be to terror killing; that is, killing someone solely to demoralize the enemy. This is why the culpability condition must hold in the case of a targeted killing of someone who is not part of the enemy's armed forces. If the victim is sufficiently removed from the war effort, then targeting him comes dangerously close to using an innocent person to achieve legitimate war objectives. This case differs importantly from targeting a combatant, because in the latter case the victim has voluntarily identified himself, by wearing the real or symbolic uniform, as a permitted target. The person who does not belong to the military, does not participate in combat, and is sufficiently removed from the war effort is morally immune to targeted killing, even if the killing will increase the chances of victory.

IV. Targeted killing of terrorists

(a) Two views

Most of the literature on targeted killings has centered on whether the practice is acceptable to combat terrorism.31 There are essentially two camps. Those in the first camp (the law enforcement view) claim that confronting terrorism is no different from confronting crime generally. The liberal state has at its disposal enough tools to respond to this kind of crime.32 Thus, the government can arrest a terrorist when it has probable cause and prosecute him in court where it must convince a jury of his guilt beyond a reasonable doubt. If convicted, he can be sentenced to the penalties stipulated by the criminal law. True, terrorists pose special, large threats. The liberal state can counter them, however, in a number of ways: it can (p.416) increase the penalties for terrorist crimes or invest in improved techniques of crime detection, prevention, and apprehension. Above all, by using these standard tools to deal with terrorists the liberal state preserves the integrity of the liberal system, and especially of its constitutional guarantees. As the Supreme Court of the United States said in the celebrated Milligan case:

The Constitution of the United States is a law for rulers and people, equally in war and in peace, and covers with the shield of its protection all classes of men, at all times and under all circumstances. No doctrine involving more pernicious consequences was ever invented by the wit of man than that any of its provisions can be suspended during any of the great exigencies of government. Such a doctrine leads directly to anarchy or despotism, but the theory of necessity on which it is based is false, for the government, within the Constitution, has all the powers granted to it which are necessary to preserve its existence.33

Those in the law-enforcement camp, then, claim to take the liberal constitution more seriously than their opponents, and they reject the idea that confronting terrorism requires departing from the Constitution.

Those in the second camp (the just-war view) claim that the nature of the terrorist threat is such that the only rational and effective way of confronting it is to use the tools of war making.34 Terrorist crimes are not common crimes, for two reasons. First, the harm threatened by terrorism is so grave that the ordinary tools of the criminal law are insufficient. Terrorists will target innocent persons in large numbers using weapons with indiscriminate destructive power. Second, the terrorist threat is particularly ubiquitous. Terrorist networks usually operate in foreign nations whose governments sometimes protect and even encourage them. And whether they operate here or abroad, terrorists live normal lives as civilians, thus significantly decreasing the chances of capture. All of these facts became painfully evident on September 11, 2001, and they are equally evident in the frequent terrorist attacks that Israel experiences. For these reasons, the liberal state must defend itself by declaring war, as it were, on terrorism. The main consequence of doing this is that the liberal state has a broader moral permission to use lethal force against terrorists than the law-enforcement view would countenance. On the just-war view, the terrorist is an enemy combatant; in fact, he is an unlawful enemy combatant, because he is at war with the liberal state without overtly wearing the enemy uniform. The law-enforcement view unduly ignores the fact that, by his own admission, the terrorist is at war, yet he has chosen not to identify himself as a combatant, thus preventing the laws of war to operate normally.35 The terrorist has an unfair advantage because whereas a conventional combatant can either be killed or captured, a terrorist (under the law-enforcement view) can only be (p.417) captured and brought to justice. He is just another criminal suspect. Worse, the terrorist claims combatant license to kill others (he regards himself at war with us) yet claims civilian immunity when the army is looking for him. This seems unacceptable. The just-war account solves this problem by treating targeted killing of a terrorist as the functional equivalent of killing in combat. The just-war view also claims to take the liberal constitution seriously. Doing so requires interpreting the liberal principles in a way that allows citizens to defend those very same principles.

The consequences of this debate for targeted killing are straightforward. If the law-enforcement camp is right, the liberal state is absolutely prohibited from intentionally killing a terrorist without affording him due process of law, except in the narrow set of situations in which the police may kill a terrorist under the same rules that apply to the use of lethal force in peacetime: they may kill the terrorist who fires at them or imminently threatens others, and so on. The government may try to arrest the terrorist or obtain his extradition, but it may not dispatch a sniper team to kill him. If the just-war camp is right, the liberal state may permissibly kill known terrorists on sight, just as it can kill enemy soldiers in conventional war. As in any war, in the “war on terror” the liberal state may not violate the jus in bello rules, but it does not have an obligation to afford the terrorist due process.

(b) What is a terrorist?

Before taking sides in this debate, we need to explore the concept of terrorism. Defining terrorism has proven especially daunting. The main reason is that it is impossible to provide a definition that is not condemnatory. The word “terrorism” has strong negative connotations. No one says “I'm a proud terrorist.” The definitions offered by writers and legal documents differ, but they converge on two factors: the method of violence chosen by the terrorist, and his purposes.36 Whatever else he does, the terrorist targets innocent persons in order to further, actually or symbolically, a political cause.

I will accept the pejorative connotation of the concept and provide a definition that reflects that condemnation. A terrorist, I stipulate, is someone who (1) does not identify himself as a combatant, (2) uses immoral violent means (the deliberate killing of civilians), (3) in the pursuit of an unjust political cause. Someone who uses violence (p.418) in the traditional war context is a belligerent, not a terrorist. A belligerent who violates jus in bello, for example by deliberately targeting civilians, is a war criminal, but not a terrorist. If he pursues an unjust cause while abiding by the laws of war he is an unjust belligerent, but neither a war criminal nor a terrorist. Someone outside the traditional war context who pursues a just cause in violation of jus in bello, for example by targeting civilians, may be considered a freedom fighter who is also a criminal,37 but not a terrorist. Finally, someone outside the war context who pursues an unjust cause with moral means is using illegitimate violence but is not a terrorist. I believe that this definition captures the two elements that explain why terrorism is so wrong. Terrorists are wrong on two counts. They are wrong because they use immoral means (they target civilians) and because they attempt to further an unjust political cause. In the case of Al Qaeda and similar religious extremists, the unjust cause is the desire to impose divine governance in accordance with the Koran.38 Their immoral means consist in targeting the innocent in pursuit of that cause. Symbolic objectives are also encompassed by the unjust cause. When the violent immoral act conveys the message that the West is corrupt and should submit to divine governance, it communicates an unjust cause, even if the act makes no causal contribution to that objective.

The attackers of September 11, 2001, then, were doubly wrong. They were wrong because they targeted and killed civilians. This is enough, of course, to condemn them. However, they were also wrong because they pursued an unjustified political objective. If those groups decided to impose divine governance by war but respected jus in bello, for example by targeting only military objectives, they would be unjust enemies deserving of defeat, but not terrorists. Conversely, if a group pursues a just cause by immoral means they are criminals, but not terrorists under my definition. For consider: a commando attack by the French Maquis during the German occupation of France that blows up a school killing innocent children is, under my definition, a crime. But under my definition the Maquis were not terrorists, because their cause was just. Of course, this is a purely verbal question, but I want to reserve the term “terrorist” for persons who use violence meeting these two conditions. It clarifies issues better and separates terrorists from freedom fighters who commit crimes; that is, those who fight for a just cause with immoral means.

There is another reason that supports my suggestion that a terrorist should be defined by his advancement of an unjust cause, in addition to his targeting of civilians. A typical terrorist is a principled evildoer.39 A theory of evil must distinguish (p.419) between opportunistic evil and principled evil. Most criminals are opportunistic; they act in self-interest. Their goal is to gain something for themselves: wealth, power, or whatever. The ordinary murderer for monetary gain is in this category. But other criminals are principled. They do evil, not out of selfish motives, but because they act out of evil maxims. Terrorists such as Osama bin Laden belong in this category. These persons are typically fanatic and immune to corruption or other temptations. Here, being principled is a vice, not a virtue, because the value of fidelity to principle is entirely parasitic on the value of the principle. Because the terrorist has an unjust cause, his tenacious efforts to impose divine governance (to advance his unjust cause) count, morally, against him. The terrorist's fearlessness (shown by his willingness to die for his cause) and perseverance make him particularly objectionable, fearsome, and difficult to confront. Al Qaeda's proud admission of their crimes and their firm attachment to the principles in the name of which they commit them make them morally worse than if they acted for personal gain. Some may think the opposite, that the fact that they do these things sincerely in the name of Islam makes them less open to criticism. Perhaps, the argument goes, that is part of a culture that we Westerners do not fully understand. I suggest exactly the opposite: there are stronger reasons to fight principled criminals than opportunistic criminals, both because their maxim is evil and because they cannot be bribed or persuaded. The only way to stop principled evildoing is by the sword.

This intentional attachment to an evil maxim is a crucial feature of terrorists. By focusing almost exclusively on the terrorist's desire to terrorize, the mainstream literature has missed an important feature of this very dangerous political phenomenon.40 It matters why the terrorist kills innocents. The account offered in this chapter assumes that the terrorist's reason for his killings is objectionable. If instead someone kills innocents in the pursuit of a justified aim, he should still be held accountable as a criminal, but an analysis other than the one suggested here would be required, one that would have to account for the killer's just cause. Defining the terrorist as a principled evildoer allows us to make room for his particular form of moral culpability.

(c) Peacetime and wartime settings

This analysis, however, still leaves open which violence the liberal state is entitled to use against terrorists. We must examine the morality of targeted killing of terrorists in two quite different settings. The first case is killing a terrorist in a setting that is sufficiently close to a war theater. An example would be killing a terrorist in a war zone in Afghanistan. The second and more problematic case is killing a terrorist in a setting that is sufficiently close to a peacetime setting. An example would be the United States’ government targeting a known terrorist leader who lives in (p.420) the United States or in a foreign country not at war—say, the United Kingdom or Saudi Arabia. I suggest that the liberal state's killing of a terrorist in a wartime setting is subject to the permissions and restrictions of killing in war, whereas the liberal state's killing of a terrorist in a peacetime setting is permissible only when necessary to avert a terrorist threat, and never if the terrorist can be captured at acceptable cost.

My proposal requires distinguishing between a wartime setting and a peacetime setting. I said something about this in my discussion of the concept of combatant in a conventional war (Section III). Here I must specify the distinction further because it is pivotal for the legitimacy of targeted killing of terrorists. To distinguish the two settings I use a geographical-institutional criterion. A wartime setting is a territory where the ordinary tools of crime control cannot operate because (1) rival political groups stake opposing claims to political supremacy, or (2) there are no claims to political supremacy whatsoever. A wartime setting is metaphorically a no-man's land. In contrast, a peacetime setting is a territory where, while there may be violence, there is an actual sovereign with undisputed claim to political supremacy who, for that reason, can use the standard tools of crime control. Thus Afghanistan and parts of Somalia are wartime settings; Paris, the United States, and Saudi Arabia are peacetime settings. To be sure, there are difficult cases (some of which I discuss in the next section), but this is the general line that must be drawn. If law and order reign, then the tools of crime control will presumptively apply; if instead law and order have broken down because of war, revolution, or anarchy, then the tools of war will presumptively apply.41

(d) Targeted killing of terrorists in a theater of war (counterinsurgency)

The analysis of targeted killing in conventional war (Section III above) applies equally to the targeted killing of a terrorist in a war theater, such as Afghanistan. There, the Western coalition is facing Taliban insurgents who use terrorist methods and have encouraged and protected Al Qaeda terrorists. This is a wartime setting as defined in the previous section, and therefore the coalition may kill insurgents because they are morally equivalent to enemy combatants. The Taliban members are terrorists to the extent that they meet the three conditions I stipulated: they do not identify themselves as combatants, they pursue an unjust cause, and they target civilians in the pursuit of that cause. They are close enough to enemy (p.421) combatants and thus the liberal state may kill them in combat, just as it may kill enemy combatants—subject, of course, to jus in bello restrictions.42 The just war view is right, I think, to deny civilian status to terrorists in a wartime setting. The restrictions on targeted killing in combat discussed in Section III apply, however. Just as the government may not kill Colonel Sanders while he is on vacation, it cannot kill a Taliban member sufficiently removed from the theater of operations. Also, compiling a list of terrorists in a wartime setting is not objectionable (just as it is not objectionable in regular war; see section III). The list is the substitute for the uniform and puts the terrorist on notice that he is a fair target.43

The difficulty with this approach is that the line between a peacetime setting and a wartime setting is often difficult to draw. I said that the distinction rests on the ability of the regular tools of crime control to operate; thus, Paris is a peacetime setting, whereas Afghanistan is a wartime setting. Other cases are not clear, however. I believe it was permissible for the Colombian government to treat terrorist organizations (or drug cartels) as enemies and target their members individually—as they targeted Pablo Escobar for assassination.44 The Colombian government could reasonably treat the territory occupied by drug cartels and extremist groups as a wartime setting. But many settings where terrorists operate defy classification. Terrorists typically perpetrate sporadic attacks on civilians without mounting a full-blown insurrection. Can the government target them for killing? Or what about a territory that is formally at peace (in the sense that no military combat is taking place there) but where the local government harbors terrorists?45 Again, all one can say is this: if the magnitude of the threat is sufficiently generalized to resemble an insurrection where the only plausible response is to use the armed forces, then the war paradigm will apply and targeted killings are permitted under the conditions I suggested above. If, however, the threat can be contained by law enforcement officials, then targeted killings are prohibited. I will say at once that I do not regard the present threat in the United States or in any European country as justifying treating their territories as wartime settings and thus enabling Western governments, without more, to conduct targeted killings there. (In the next section I specify when governments may target terrorists in a peacetime setting.) But I do regard the situation in Afghanistan as justifying the killing of insurrectionists, by application of the normal rules that apply to insurrections. (The right way to look at (p.422) the situation in Afghanistan is that the western Coalition assists the local government in suppressing an unjust insurrection).46

(e) Targeted killing of terrorists in a peacetime setting

Defenders of the just-war view allow the targeted killing of terrorists in peacetime settings, such as the territories of the United States or Europe. Their argument is as follows. Terrorists do not wear uniforms and therefore they do not identify themselves as combatants. They live amongst the general population where they hide while they plan their next attack, in the assurance that the liberal state will not fire indiscriminately. Targeted killing is necessary because, as Daniel Statman says, “tanks, jets, and submarines are helpful when confronting other tanks, jets, and submarines, not hijackers carrying knives or terrorists wearing explosive belts.”47 On this view, it is not the political purpose of the terrorists that brings the conflict closer to the war paradigm: it is the nature of the threat. If this is correct, should the Mafia or a drug cartel pose a threat of similar magnitude, the government could resort, perhaps, to targeted killing. If a society becomes utterly paralyzed by drug wars, frequent street gun battles and so forth, perhaps the government can permissibly kill named criminals, even though they are not strictly terrorists but persons who kill for personal gain.48 On the just-war view, the government can identify the possible targets. If the government knows who the terrorists are, it can announce publicly that these persons are enemy terrorists at war with the state, and therefore vulnerable to targeted killing. This can be done in two ways. The government may publish a list of known terrorists, and thus put them on notice that they can be lawfully targeted. The list performs a role analogous to the uniform in conventional war, and the publication of the list amounts to a declaration of war against terrorists.49 The second method is to identify the terrorists by their membership in a terrorist organization. The government can simply announce that it is at war with Al Qaeda and that all present or future members are henceforth vulnerable to killing. This announcement would also operate as a declaration of war against Al Qaeda. For the just-war camp, either of these public acts (a list of individual names or the criterion of membership) solves the problem of identification. The liberal state is at war with the terrorists and knows who they are. Just as the liberal state can kill (p.423) unjust enemies in a conventional war, so it can kill terrorists (who, by definition, are unjust enemies) in this war.

This argument must be rejected. The state cannot unilaterally repeal the strict moral and legal restrictions on state killing by just declaring war on terrorists. On the facts, the choice of the war paradigm may or may not be justified, but certainly it cannot be established by a mere statement from the government. Otherwise, the government could declare war on the Mafia, compile a list of Mafiosi, and announce that from now on the government will shoot them on sight. Sexual predators are especially repulsive. Why not allow the government to declare war on known sexual predators, compile a list, and start killing them? The prohibition of intentionally killing persons cannot magically disappear by the government's unilaterally declaring war on the persons it intends to kill. The criterion of membership does not fare any better, as it is overinclusive. Persons may be loosely affiliated with criminal organizations, yet in peacetime liberal principles require more than that to even prosecute them, let alone kill them. While in a liberal democracy we tend to trust our government, the experience with declared “wars” on terrorists and “subversives” in other societies should give us pause before abandoning the strict prohibition on deliberate killing.50 If the named terrorist is an outlaw criminal, then it is hard to explain why the state should not treat him just like it treats any other outlaw criminal.51 The state must in principle try to arrest and prosecute him.

Yet, while the just-war view fails to justify a broad permission to kill terrorists in peacetime, it has enough weight to prompt reconsideration of the law-enforcement paradigm. The weakness of the latter is its failure to recognize the special threat that terrorists pose, especially those whose centers of command are overseas. In particular, the dual nature of terrorists as individuals who claim license to kill indiscriminately but seek simultaneously the protection afforded by civilian status, strongly suggests that the solution to this problem must be sui generis.

(f) A sui generis standard for terrorism

In line with the conditions I set forth for targeted killing in peacetime, I propose a third solution for confronting terrorism: the liberal state may kill a targeted terrorist in peacetime only when necessary to prevent the deaths of a substantial number of innocent personstypically, when necessary to foil a deadly plot or a broader plan to conduct terrorist activities. I hasten to add that the other three conditions for targeted killing in peacetime must hold as well: the contemplated target must be culpable, the liberal state must have a just cause, and capture should be impossible or prohibitive. I insist on (p.424) this point because the existence of the threat is a necessary but not a sufficient reason to justify a targeted killing in peacetime. In addition, the liberal state must have a just cause (which, by the above definition of terrorism, it has) and the contemplated target must be morally responsible, both in the sense of endorsing the unjust cause and being causally responsible for posing the threat. A typical terrorist will often meet these conditions, but the threat alone does not suffice to justify a targeted killing. The liberal state is not justified to kill someone anytime it is necessary to avert a threat. Thus, killing innocent persons is always banned, even if necessary to suppress a threat.

A targeted killing of a terrorist (as defined above) is justified when necessary to avert a specific terrorist attack that is reasonably believed to be likely to cause the deaths of many innocents or to foil a broader plan to conduct such terrorist attacks. Notice that this condition is analogous to the first condition I specified for targeted killing in peacetime. The condition is somewhat relaxed here, however, because the number of deaths averted by killing the terrorist need not be as great as in the other peacetime cases. Recall that the normal justification for targeting a person in peacetime is to avert either a genocide or an impending war of aggression. The targeted killing of a terrorist need not avert as many likely deaths as genocide or aggressive war threaten to cause. This standard is more permissive than the normal standard for targeted killings in peacetime, yet not as permissive as the standard for state killings in a just war, where any combatant on the unjust side is permissibly targeted, regardless of how many persons he is likely to kill. This is why the treatment of terrorism is sui generis: it does not fall neatly into the law-enforcement camp or the just-war camp. Because the terrorist threat is ubiquitous, the threatened harm is great, and the terrorist is committed as a matter of principle to perpetrating the harm, the law-enforcement standard does not suffice. However, this does not mean that the liberal state should adopt the just-war standard, which would allow killing the terrorist on sight. The reasons supporting restraint in peacetime are powerful enough, I believe, to ban the killing of a terrorist merely because he is a known terrorist, or he has committed crimes in the past, or he is a member of a terrorist organization. The liberal state must protect its citizens from terrorist threat, but must also observe the restraints on the use of deadly force in peacetime, even when the temptation to take out a known enemy is great. The standard I suggest, then, mediates between the law-enforcement and just-war positions by trying to address the main concerns of each. It rejects the law-enforcement view that terrorists are just like any other criminals, by allowing the targeted killing of a terrorist who threatens to kill innocents. It likewise rejects the just-war view that known terrorists are enemy combatants who can be killed on sight regardless of the threat they actually pose.

As I indicated, the liberal state fulfills the just cause requirement by definition of what it is to be a terrorist: we saw that a terrorist pursues an unjust cause. In contrast, someone who targets innocents in pursuit of a just cause, for example because the government is a human rights violator and the perpetrator seeks restoration of (p.425) human rights, is not a terrorist. This chapter does not address that difficult situation; for our purposes, terrorists always have an unjust cause. If on reflection, citizens of the liberal state become convinced that the perpetrators of attacks against civilians have a just cause, a different analysis would be required. The requirement that the target must be culpable, or morally responsible, is also satisfied. The terrorist is culpable in a double sense: he endorses the unjust cause in the name of which he kills, and he contributes causally to the threat that the liberal state attempts to suppress. Therefore, the condition of culpability (or moral responsibility) that I required for killing in peacetime fully applies in the case of the terrorist. Here too, the condition prohibits the liberal state to kill innocent persons (such as the terrorist's family) even if the state reasonably thinks that would avert the terrorist threat.

The final condition also applies to the killing of a terrorist in peacetime: the government may not kill the terrorist without giving him the chance to surrender (subject to certain exceptions discussed below).52 Often, if the special operations team can reach their victim they can apprehend him too. Now, either the terrorist surrenders or he resists arrest. If he surrenders, then it is morally impermissible to kill him; his captors should bring him to justice. If, on the other hand, the terrorist resists arrest, then the state agents can permissibly kill him under the rules applicable to resistance to justified arrests. Killing a person, no matter how culpable, without giving him a chance to surrender is morally impermissible.

A critic may challenge this requirement. He may argue that the onus of surrender should be on the terrorist. After all, he could have decided to surrender to the authorities at any time.53 Because he has failed to surrender, the terrorist is now a fair target, and the liberal state does not have an obligation to give him the opportunity he himself disdained. This criticism, however, is a simple return to the war paradigm, where the just warrior can kill the enemy on sight without giving him the chance to surrender (although, of course, if the enemy does surrender he is hors de combat and cannot be killed). Here, however, we are in a peacetime setting and thus outside the war paradigm, and so the governing rule is the standard rule that operates in that setting. When the police encircle a common criminal, they do not kill him on sight. This is true even if the criminal, like the terrorist, failed to surrender voluntarily to the authorities. In a peacetime setting, state action against a terrorist is subject to a similar constraint. When the agents of the state encircle the terrorist they must order him to surrender before shooting. The fact that the criminal is a terrorist and not a common criminal does not make a moral difference here. (p.426)

However, there are complications. Capture may be physically impossible or morally prohibitive. Suppose the target is unreachable. The only way to kill him is shooting from afar or using drones or similar devices. In that case the condition is satisfied because the government cannot physically capture the terrorist, and killing the terrorist is permissible. (Recall that the main necessity condition applies.) Suppose, however, that the team can capture the terrorist but at the cost of the lives of five of its members. Killing him from afar would avoid these casualties. It is problematic to sacrifice five innocent persons (by our earlier definition of innocence, these are just warriors) just to spare the life of the terrorist. Perhaps one can say that the particular job description of the capturing team contemplates that they will risk their lives for these purposes: they would have given their lives to apprehend the terrorist. One raison d'etre for these special teams is precisely that they will risk their lives to protect innocents. Like police officers, SWAT teams are supposed to put their lives on the line to protect the public, so perhaps it is appropriate to pay that price for capturing the terrorist in these kinds of cases.

Now let us suppose instead that the terrorist can be captured but at the cost of the lives of five civilians. Here the job description rationale does not apply: these are innocent persons, and killing the terrorist instead of attempting capture would save them. This puzzle raises the question of collateral deaths in peacetime. What is the right way to think about these issues? Are deaths of bystanders during a police raid (regretfully) justified? If the answer is no, then the same is true in the case of capturing the terrorist, and an attempt to capture the terrorist might well be prohibited if it is likely to cause these collateral deaths. In those cases, killing the terrorist is morally preferable to attempting capture: better for him to die than the civilians.

But what if the liberal state has located a terrorist who is armed and willing to resist in a place where, the state reasonably knows, any attempt to kill him (he has sworn not to be taken alive) will bring about deaths of civilians? Here (unlike the previous case where the issue was capture and not killing) bringing about the deaths of civilians may be justified under some version of the doctrine of double effect. I cannot develop this complex matter here, but here is, in concise form, the view of double effect I discuss elsewhere. The deaths of civilians brought about by the state en route to killing a terrorist will be justified only if the threat to be averted is serious enough relative to the seriousness of bringing about those collateral deaths (and additional conditions are satisfied).54 By the same token it is prohibited, in a peacetime setting, to bomb the terrorist's home without regard for his family or other innocent persons. The principles that apply to this situation are complex and problematic, to be sure, but not more so than those that apply to killing in conventional war.

The upshot, then, is this. The insight that terrorism calls for a sui generis response is correct. The law-enforcement view is wrong to claim that targeted killing of (p.427) terrorists in peacetime is never justified. The just war view is wrong to claim that targeting known terrorists in peacetime is justified as the functional equivalent of killing in war. The moral framework for killing terrorists is close to the one for targeted killing in peacetime, but not identical, because the threat in the case of the terrorist need not be as massive as the threat that would justify killing a political leader in peacetime. The permission to kill in conventional war, then, is inapplicable to terrorists outside a war zone. Because the license to kill is morally problematic even in conventional war, it should be strictly confined to that setting. Any extension to peacetime must be in principle avoided. The moral prohibition of deliberate killing is too strong to set aside every time the criminal threat increases. The liberal government should not be in the business of killing people except in extreme situations such as conventional war. However, because the terrorist scourge is sui generis, the government may kill a terrorist if necessary to prevent a terrorist crime, as explained. Targeted killing of a terrorist would then be, mutatis mutandi, the functional equivalent of killing Hitler in 1939, rather than the functional equivalent of killing an enemy combatant in conventional war.55

Critics may variously object. Some will say that my standard is too permissive because it does not require imminence of the threat. For example, Phillip Heymann and Juliette Kayemm suggest that killing a terrorist is justified only when “it is necessary to prevent a greater, imminent harm, or in defense against a reasonably imminent threat to the lives of the targets of the planned terrorist attack.”56 The idea is to restrict as much as possible this troubling practice. Targeted killing would not be available against terrorists who are planning a deadly attack that will take place some time in the more distant future, even if the killing is the only way to foil it. I am unpersuaded, however, because the requirement of imminence is clearly under inclusive, especially in this context.57 Terrorists are ubiquitous and often engage in long-term planning. For that reason, the liberal state cannot easily forgo chances to foil their plot in the hope that it will be able to act effectively later when the threat is imminent. More generally, the requirement of imminence is in tension with the rationale for targeted killing, which lies in the legitimate interest of the liberal state to effectively protect its citizens from terrorist threats. Should imminence be required, then in some cases the state will forgo its only chance to protect civilians from deadly attack. I cannot see what can be gained by adding imminence as a requirement when the state reasonably knows that acting now may be its only chance to avert a terrorist strike. (p.428) As several criminal law scholars have argued in the context of individual self-defense, the reason to require imminence is to ensure that the defensive action is necessary.58 Conceptually, imminence is subservient to necessity.59

The debate about imminence in other contexts is only marginally relevant to targeted killing. First, in international law many observers require imminence for lawful self-defense because (among other things) the harm done by a defensive war is catastrophic.60 That worry does not apply to the much less harmful targeted killing of terrorists. Second, criminal law scholars who require imminence in individual self-defense are dealing with a different setting: in a targeted killing the “defender” is the liberal state, and the “attacker” is a terrorist. Importantly, in individual self-defense scholars require imminence because usually the victim has available other forms of redress. She has a background obligation not to respond with deadly force and summon the police instead. In contrast, in a targeted killing there is no one to summon. The action by the liberal state is its only chance to save innocents. And finally, in individual self-defense the attacker poses a threat to the victim, but not a larger threat to innocent third parties. The terrorist threat is typically more serious.

In short, in peacetime the liberal state may kill a terrorist only when the killing is necessary to avert a terrorist plot and capture is not physically or morally possible. Even though the government operates under a more permissive standard for using force when it confronts a terrorist than when it confronts a common criminal, the standard is far from licensing the government to treat the terrorist as a sub-human creature that can be hunted down and killed on sight.

V. The killing of Osama bin Laden

On May 1, 2011, an elite team of United States Navy SEALs killed Osama bin Laden, leader of Al Qaeda and responsible for the attacks of September 11, 2001, at a compound in Abbottabad, near Islamabad, Pakistan. According to early reports, bin Laden was shot in the head “while resisting.”61 In a speech minutes after the (p.429) operation, President Obama (who had made the capture of bin Laden a priority) said that bin Laden had been killed “after a firefight.”62 Shortly thereafter, however, the official account changed. Apparently, bin Laden was unarmed when confronted by the Navy SEALs, although it is hard to know if the fatal shot was made in cold blood or upon some semblance of resistance.63 John Brennan, the White House counter terrorism adviser, said that the United States would have captured bin Laden alive, “should it have had that opportunity.”64

How does this targeted killing, the most important of all in the “war on terror,” fare under the analysis presented here? The factual uncertainties are such that I cannot presume to issue a verdict here. The rationale that most people, including President Obama, endorse, that killing bin Laden was a lawful execution, an act of justice, is unavailable for the reasons I explained above. A person is not morally vulnerable to being killed on sight only because he committed heinous crimes in the past. Liberal principles do not condone extrajudicial execution of even the worst criminals.

The first step, then, is to determine whether Abbottabad, Pakistan, was at the time a wartime setting or a peacetime setting. If the former, then the United States had a right to kill bin Laden, an enemy combatant. If the latter, the United States could only kill him if it was necessary to avert a terrorist threat, and this only if bin Laden was given the chance to surrender (if surrender was physically possible at a relatively low cost). This incident shows how hard it can be to determine the nature of the setting. Paris is clearly a peacetime setting; Afghanistan is clearly a wartime setting. But a suburb of Islamabad, where the terrorist is hiding, according to all indications, under the protection of segments of the Pakistani army, is in between the two. So if the notion of wartime setting extends to this kind of theater, where the local government is harboring terrorists, then the killing was justified as akin to killing in combat. This is a close call, but on reflection I think the concept of wartime setting (p.430) should be interpreted narrowly. Otherwise, the state would have a license to kill the terrorist anywhere whenever the liberal government thinks there is some local support for him. This is extending the notion of wartime setting too far.

Assuming, then, that the correct standard is the one that applies in peacetime, the next step is to determine if killing bin Laden was necessary to avert a terrorist threat. Although there is room for some doubt here, it was reasonable for the United States to believe that killing bin Laden would foil ongoing terrorist plots. After all, bin Laden's mission in life was to plot killings of civilians to further his political goals. Based on what he had done and what he had announced he would do, the United States could reasonably believe that killing bin Laden would avert terrorist threats, and so the killing of bin Laden, given the information we have, satisfied the necessity condition.

The real doubt arises with the question of capture. As I indicated, when capture is possible at acceptable cost killing the terrorist is morally prohibited. The crux of the issue is whether the person who killed bin Laden could reasonably have thought that bin Laden was resisting. Unfortunately, I have no way of knowing exactly what happened. As we saw, some credible reports say that bin Laden was unarmed; other equally credible reports say that he had weapons at hand. So I will put the conclusion in conditional form. If the officer who killed bin Laden reasonably thought that bin Laden was about to fire or was searching for a weapon, then the killing was permissible. But if the officer shot bin Laden while he was unarmed and not resisting, then the killing was impermissible: the officer had an obligation to capture him alive. The fact that the President's instructions were secret prevents us from establishing one crucial fact: did he order the commandos to execute bin Laden, or did he order them to capture him alive if possible? Perhaps some day we will know. The standard I propose in this chapter is incompatible with an instruction to execute a named person.

VI. General objections to targeted killings

I now turn to two objections to targeted killings. These objections may make us revise the conclusions reached above.

(a) The epistemic objection

In every instance where targeted killing is presumptively permitted, the liberal government must assess whether or not the permissibility conditions exist. In the case of Rhodelia above, the government must be sure that a genocide is afoot, that the target is culpable, and, most important, that the killing will save many innocent lives. This is seldom certain. Perhaps there is no genocide but a revolution with casualties on both sides. Perhaps the leader whose death the government is planning has nothing to do with those events. And perhaps killing him will make things worse. Governments do not have a particularly good record of making (p.431) assessments of this kind. The same epistemic difficulties arise in the case of targeted killing in wartime. Here the government must ascertain even murkier facts. Is the enemy's government official in question involved in the war? Is Colonel Sanders really on vacation? Will the targeted killing really increase our chances of victory? And finally, the difficulties in ascertaining the permissibility conditions before killing a terrorist are daunting. In addition to getting the right person (identity mistakes are not uncommon), the liberal government must determine, first, if the theater is a wartime or a peacetime setting. Second, the government must determine that killing this person is necessary to avert a terrorist crime. And third, it must make sure that the operation will have acceptable collateral costs. These are unusually high epistemic barriers. While we should allow for the fact that sometimes the evidence on the ground will be clear to anyone, these barriers should give us pause before enacting a legal permission of targeted killing.

(b) The objection from virtue

Why are targeted killings morally repulsive even when they give villains their due and lead to highly beneficial consequences? One reason is the heightened intentional focus that characterizes targeted killing. The law assigns varying degrees of blameworthiness for outcomes. The criminal law teaches us that killing someone in self-defense is (perhaps) not blameworthy at all; killing someone as a result of negligence is somewhat blameworthy; killing someone in a rage of passion is blameworthy; and killing someone for monetary gain is very blameworthy.65 When Colonel Sanders is commanding his troops, his unjust threat to us is imminent and proximate. Our soldiers kill him, even naming him, knowing who he is, in a situation that is quite close (though perhaps not identical) to individual self-defense.66 The more removed he is from that situation of direct threat, the less defensible the targeted killing will be, because killing him requires more planning. In the law of homicide, the more premeditated the killing the more blameworthy it will be. However, premeditation is an aggravating circumstance when the killing is wrongful in the first place. If the killing is otherwise justified, can premeditation make it wrongful? Perhaps not, but targeted killings that are justified on the merits can still be morally troubling, even if that worry does not suffice to make it wrongful.67

This troubling aspect can be illuminated by reference to the idea of moral philosophers that each person has agent-relative reasons to refrain from killing, and not simply impersonal or agent-neutral reasons. Consider a prima facie justified case of targeted killing: killing the political leader of a nation that has perpetrated (p.432) aggression against us, where the killing will predictably end the war. All of these good consequences are agent-neutral reasons to kill; that is, impersonal considerations to kill the aggressor. Yet targeted killing involves detailed planning, a sure hand, cunning behavior, and nerves of steel. A morally sensitive person has reasons not to be so cold-blooded as to be able to perform such a killing or to undertake the necessary preparatory acts for it. He agrees that it would be a good thing, due to the consequences, should the villain be killed, but does not want to create the state of affairs where he kills. (This is not to say that in moral deliberation the agent-relative reasons will always prevail over the agent-neutral reasons, the good consequences.) There is an important difference between the sentences “it is a good thing that bin Laden died” and “it is a good thing that I killed bin Laden.”

Can agent-relative reasons apply to the government? Possibly yes. The idea is that liberal governments should attempt to behave in accordance with values and virtues for which they stand.68 This includes rejecting self-help, revenge, and random violence in favor of lawful coercion, coercion under the rule of law. This excludes assassination of any sort. The prohibition on assassination is an expression of the values embedded in the liberal social contract. Targeted killing would perhaps be understandable in the state of nature, but not in civil society, where due process and the rule of law reign supreme. Call this the political virtue argument. I think this argument, while not conclusive, has some weight. In considering the morality of targeted killing we must weigh not only the goodness of the villain's death, but the badness of our government's killing. These considerations may collapse in the face of supreme emergency, but they certainly carry substantial weight.

(c) Are the objections conclusive?

These objections carry considerable weight. The fact that governments will often err about the existence of the permissibility conditions, and the fact that the modi operandi of targeted killings are (arguably) troubling, point, perhaps, to one conclusion: targeted killing in peacetime should be, in principle, legally prohibited.69 Not every morally permissible practice ought to be legally permitted. The law has its own logic and creates its own specific incentives. Given the proven tendency of governments to err—the many instances of government failure—it seems salutary (p.433) in a liberal democracy to prohibit the government from killing persons outside of war. Given how important it is for the state to stop terrorism, however, I think the highest authority in the land should have the power to waive the prohibition in cases where killing the terrorist is indeed necessary to avert a deadly terrorist attack, as I have specified in this chapter. The government should fully explain to the citizenry his reasons for waiving the prohibition. The secrecy that surrounds these operations in current liberal democracies does not help to ascertain their justification. It is important for the citizenry to understand the reasons why their government has resorted to an act as serious as a targeted killing—why the default rule against killing has been waived in a particular case. The justification for secrecy is the desire not to help the enemy; perhaps, then, the government should explain its reasons publicly after the fact. But secrecy should not serve as a way to hide from the state's own citizens the fact that these killings were, after all, summary executions.

VII. Concluding thought

A legitimate function of the liberal state is to protect persons from one another and from foreign enemies. In extreme cases, the state protects citizens against foreign enemies by waging war. But custom and morality have confined war, and the license to kill that it entails, to cases where the liberal state faces an organized enemy. The license to kill in war can plausibly be extended to situations where the liberal state faces commandos or terrorists in the battlefield. But the licence does not extend beyond those cases, so the state may not declare war against individuals. If the state's institutions function normally, it is bound by the strictures of the rule of law. One of the central precepts of the rule of law is the prohibition against killing anyone without due process, should that process be available. In times of peace due process is available, so the default rule is that the state may not conduct extrajudicial killings. This prohibition extends to foreigners, even if they have committed crimes. Nonetheless, I have argued that the terrorist threat justifies a departure from the prohibition when killing a terrorist is necessary to avert a crime that is likely to kill many innocents, even if the crime is not imminent. The permission to kill a terrorist in peacetime is an exception to a fundamental prohibition of state violence and must be interpreted strictly. In particular, the state must give the terrorist the chance to surrender if that option is available at an acceptable cost. These constraints reflect the fact that in a liberal democracy the morality of state coercion is not determined solely by the blameworthiness or dangerousness of bad persons, but by the values, goals, and purposes of the liberal state itself. The terrorist is a public danger and a moral monster, but those facts do not exhaust the relevant reasons for justifying state deadly violence. What we are, and what we may become, also matters. Perhaps that is what matters most.


(2) A sentiment echoed by the President of the U.S. See the President's speech in “Osama Bin Laden Dead,” available at 〈http://www.whitehouse.gov/blog/2011/05/02/osama-bin-laden-dead〉 accessed November 3, 2011.

(3) This chapter is about the morality of targeted killing, not about its lawfulness under international or domestic law.

(4) The distinction between intended outcomes and unintended yet foreseen outcomes has a crucial role in the morality of war, and is captured by the famous doctrine of double effect. See, inter alia, R.G. Frey, “The Doctrine of Double Effect” in R.G. Frey and C. H. Wellman (eds), A Companion to Applied Ethics (Blackwell, 2003); Joseph Boyle Jr, “Toward Understanding the Principle of Double Effect” in P.A. Woodward (ed.), The Doctrine of Double Effect: Philosophers Debate a Controversial Moral Principle (Notre Dame University Press, 2001) 12; and M. Walzer, Just and Unjust Wars: A Moral Argument with Historical Illustrations, 4th edn (Basic Books, 2006) 128.

(5) I avoid the term “asymmetrical war” because it prejudges the issue by deciding that the conflict with terrorists is sufficiently close to conventional war.

(6) In addition to the condition that the author must be a liberal government, which I do not discuss.

(7) See Andrew Altman and Christopher Heath Wellman, A Liberal Theory of International Justice (Oxford University Press, 2009) 116.

(8) I use the word “genocide” loosely, to denote mass murder.

(9) I use the term “civilians” to denote innocent noncombatants.

(10) As Jeff McMahan has convincingly argued. See J. McMahan, Killing in War (Oxford University Press, 2009) esp. chs 1 and 2. See the discussion below. I use the word “murder” as meaning prohibited killing.

(11) The literature on the just war tradition is extensive. See generally A. Bellamy, Just Wars: From Cicero to Iraq (Polity Press, 2006).

(12) This account of just cause is proposed by L. Lomasky and F.R. Tesón, Justice at a Distance (unpublished manuscript).

(13) I say “in most cases” because one can think of the rare case of someone who created the threat of war in a non-culpable manner.

(14) This is an extraordinarily difficult problem which I cannot examine here. See, inter alia, J. Thomson, “The Trolley Problem,” 94 Yale Law Journal (1985) 1395.

(15) I leave aside even more extreme circumstances where this prohibition may collapse.

(16) See S. David, “Israel's Policy of Targeted Killing,” 17 Ethics & International Affairs, (2003) 111.

(17) Interestingly, Freeland's government's threat must be credible, and this depends on Caligula's not realizing that Freeland's government is threatening him to avoid killing him! Caligula, in other words, must believe that Freeland's government is as callous as he is.

(18) The position in the text largely follows McMahan, supra n. 10, but see n. 22 below. The classic locus for the view that McMahan opposes, the “moral equality of combatants,” is Walzer, supra n. 4, 34–41. See also L. May, “Killing Naked Soldiers: Collective Identification in War,” 19 Ethics and International Affairs (2005) 39.

(19) See Protocol Additional to the Geneva Conventions of August 12, 1949, Relating to the Protection of Victims of International Armed Conflicts (Protocol I) June 8, 1977, arts 51–54. I assume without argument that, with respect of jus in bello, the laws of war generally track the morality of war, with the already noted exception that unjust warriors lack permission to kill.

(20) See the Francis Lieber Code [1863], available at 〈http://www.civilwarhome.com/liebercode.htm〉 accessed November 3, 2011.

(21) See also C. Finklestein, “Targeted Killing as Preemptive Action,” in this Volume, ch. 6.

(22) I tend to depart from McMahan's views on individual culpability of soldiers fighting an unjust war. Average foot soldiers fighting an unjust war should not be held individually culpable, even though, as McMahan says, each one of their killings is an unjustified killing. See Lomasky and Tesón, supra n. 12.

(23) See J. McMahan, “War as Self-Defense,” 18 Ethics & International Affairs (2004) 75.

(24) See D. Statman, “Targeted Killing,” 4 Theoretical Inquiries in Law: Targeted Killing (2004) 179, 196.

(25) See K. Fahim, A. Shadid and R. Gladstone, “Violent End to an Era as Qaddafi Dies in Libya,” New York Times, October 20, 2011, available at 〈http://www.nytimes.com/2011/10/21/world/africa/qaddafi-is-killed-as-libyan-forces-take-surt.html?pagewanted=all〉 accessed November 3, 2011.

(26) An analogous case is the assassination of Reinhard “The Hangman” Heydrich, ordered by the Czech government in exile in 1942. See R.C. Jaggers, “The Assassination of Reinhard Heydrich,” CIA's declassified document, available at 〈https://www.cia.gov/library/center-for-the-study-of-intelligence/kent-csi/vol4no1/html/v04i1a01p_0001.htm〉 accessed November 3, 2011.

(27) See D. Bentley, “Call for Inquiry into Gaddafy Death,” The Independent, London, October 20, 2011, available at 〈http://www.independent.co.uk/news/world/africa/call-for-inquiry-into-gaddafi-death-2373545.html〉 accessed November 3, 2011. If that was indeed the case, then killing him was impermissible.

(28) See J.G. Murphy, “The Killing of the Innocent” in Retribution, Justice, and Therapy (Reidel 1979) 6–9.

(29) Thus, the strike by NATO forces on April 30, 2011 that killed three of Qaddafi's grandchildren is highly problematic. See MSNBC News, “Gadhafi's Youngest Son, Grandkids Killed in NATO Attack,” available at 〈http://www.msnbc.msn.com/id/42829913/ns/world_news-mideast_n_africa/t/gadhafis-youngest-son-grandkids-killed-nato-attack/〉 accessed November 3, 2011.

(30) For Speer's own version of his role, see A. Speer, Inside the Third Reich (New York: Simon & Schuster, 1997). For a challenge to Speer's account, see G. Sereny, Albert Speer: His Battle with Truth (Vintage Books, 1996).

(31) See, in addition to the works already cited, M. L. Gross, Moral Dilemmas in Modern War, (Cambridge University Press, 2010) 100; T. Meisels, “Combatants—Lawful and Unlawful,” 26 Law & Philosophy (2007) 31; A. Plaw, Targeting Terrorists: A Licence to Kill? (Ashgate, 2008); S. Miller, Terrorism and Counter-Terrorism: Ethics and Liberal Democracy (Oxford: Blackwell, 2009) 139.

(32) See, inter alia, D. Luban, “Eight Fallacies about Liberty and Security” in R. Ashby (ed.), Human Rights in the “War on Terror” (Cambridge University Press, 2005) 242. The position in international law is analyzed in N. Melzer, Targeted Killing in International Law (Oxford University Press, 2009). See also Jens Ohlin, “Targeting Co-belligerents,” in this Volume, ch. 2.

(33) Ex Milligan, 71 U.S. 2 (1866).

(34) See the works by, Gross, Meisels, and Miller, supra, n. 31.

(35) See Gross, 104–9; and Miller, Terrorism and Counter-Terrorism, 139–45, supra n. 31.

(36) Bruce Hoffman's definition is typical: “Terrorism is . . . political violence in an asymmetrical conflict that is designed to induce terror and psychic fear . . . through the violent victimization and destruction of noncombatant targets.” B. Hoffman, Inside Terrorism, 2nd edn (Columbia University Press, 2006) 34. See also the complex definition offered by S. Miller, Terrorism and Counterterrorism: Ethics and Liberal Democracy (Oxford: Blackwell, 2009) 53–4. Legal definitions of terrorism focus on the method of terrorism, not its purposes. For a sample, see S. Dycus et al., National Security Law (Aspen Publishing, 2007) 465–9.

(37) I do not want to say here “war criminal” because I do not want to prejudge the issue whether anti-terrorist action is war.

(38) See J. Kelsay, Arguing the Just War in Islam (Harvard University Press, 2007) chs 3 and 4.

(39) I introduced the distinction in my essay “Enabling Monsters: A Reply to Professor Miller,” 25 Ethics & International Affairs (2011) 165. There I compared different kinds of despots; here I compare different kinds of criminals.

(40) See the various definitions of terrorism in n. 36 above.

(41) On September 16, 2011, John Brennan, the President's senior advisor on counterterrorism and homeland security, delivered a speech at the Harvard Law School clarifying the current Administration's security policy, including targeted killing. There he expressly rejected the geographical distinction in the text, and reaffirmed the long-standing view that the U.S. is at war with Al Qaeda. This means that, subject to a number of considerations such as “respect for a state's sovereignty and the laws of war,” the U.S. is entitled to kill members of Al Qaeda regardless of location. The speech can be found at 〈http://www.lawfareblog.com/2011/09/john-brennans-remarks-at-hls-brookings-conference/〉 accessed November, 3, 2011(hereinafter Brennan Speech).

(42) The laws of war distinguish between lawful and unlawful combatants, see Article 4, Third Geneva Convention, Relative to the Treatment of Prisoners of War. This distinction has been much debated in the literature; however, the argument in the text is agnostic about the moral justification of those distinctions. The distinction does not bear on the morality of targeted killing, except for the fact that the failure of the terrorist to openly identify himself as combatant contributes to the magnitude of the terrorist threat.

(43) See Gross, supra n. 31, 108–9.

(44) See M. Bowden, Killing Pablo: The Hunt for the World's Greatest Outlaw (London: Atlantic Books, 2001).

(45) For Pakistan and bin Laden, see below, Section V.

(46) I examine some aspects of the morality of the war in Afghanistan in Tesón, supra n. 39.

(47) Statman, supra n. 24, 179.

(48) For the current situation in Mexico, see “Mexico Under Siege: The Drug War at Our Doorstep,” LA Times, April 4, 2011, available at 〈http://projects.latimes.com/mexico-drug-war/#/its-a-war〉 accessed November 3, 2011.

(49) Curiously, this method is consistent with Lieber's aversion to assassination, supra n. 20. Lieber, let us recall, abhorred declaring a regular enemy soldier an outlaw and targeting him for assassination. The terrorist, however, is not a lawful enemy combatant but an unlawful enemy combatant, because his chosen method, targeting civilians, is immoral. Therefore, it is perfectly appropriate for the state to declare him an outlaw.

(50) A poignant example is Argentina's “dirty war” in the 1970s, where the regime targeted and killed thousands of persons that the government called terrorists and “subversives.” See the chilling account in Comisión nacional sobre la desaparición de las personas, Nunca Más (Buenos Aires, 1985).

(51) See Gross, supra n. 31, 107.

(52) The policy of the current U.S. Administration seems close to the view in the text, but it is not quite the same: “Whenever it is possible to capture a suspected terrorist, it is the unqualified preference of the Administration to take custody of that individual so we can obtain [vital] information,” Brennan speech, supra n. 41. While this view is better than the view that terrorists can be freely killed without trying to capture them, it is still wanting, because it makes capture a preferred alternative, whereas I suggest capture is morally required unless impossible or morally prohibitive.

(53) I owe the point to Andy Altman.

(54) I discuss the doctrine of double effect in war in Lomasky and Tesón, supra n. 12.

(55) The argument here is strictly confined to targeted killing. It does not address other areas of disagreement between the two camps, such as whether terrorists should be tried by regular courts or military commissions.

(56) P.B. Heymann and J.N. Kayyem, Protecting Liberty in an Age of Terror (MIT Press, 2006) 66 (my emphasis).

(57) See R. Christopher, “Imminence in Justified Targeted Killing,” in this Volume, ch. 9. Christopher makes the point that imminence may be useful as an evidentiary guideline, but it must not be confused with the underlying moral principle which, more plausibly, requires necessity.

(58) In the words of Paul Robinson: “If a threatened harm is such that it cannot be avoided if the intended victim waits until the last moment, the principle of self-defense must permit him to act earlier—as early as is required to defend himself effectively.” P. Robinson, Criminal Law Defenses (West, 1984) § 131(c)(1), 78.

(59) The current U.S. Administration likewise rejects a strict concept of imminence. See Brennan Speech, supra n. 41.

(60) A summary of the debate can be found in T. Rheinold, “State Weakness, Irregular Warfare, and The Right to Self-Defense Post-9/11,” 105 American Journal of International Law (2011) 244.

(61) See BBC News, May 2, 2011, available at 〈http://www.bbc.co.uk/news/world-us-canada-13256676〉 accessed November 3, 2011; CBS News, May 2, 2001 at 〈http://www.cbsnews.com/stories/2011/05/01/national/main20058777.shtml〉 accessed November 3, 2011; “Bin Laden is Dead, Obama Says,” New York Times, May 2, 2011, available at 〈http://www.nytimes.com/2011/05/02/world/asia/osama-bin-laden-is-killed.html?pagewanted=all〉 accessed November 3, 2011.

(62) See “Full text of Obama's Speech on Bin Laden's Death,” available at 〈http://www.cbsnews.com/8301-503544_162-20058783-503544.html?tag=contentMain;contentBody〉 accessed November 3, 2011.

(63) Jim Miklaszewski, “Bin Laden ‘Firefight’: Only One Man Was Armed,” NBC News, May 5, 2011, available at 〈http://www.msnbc.msn.com/id/42906279/ns/world_news-death_of_bin_laden/〉. Here is a “senior official's” description of the killing: “Then, without hesitation, the same commando turned his gun on Bin Laden, standing in what appeared to be pajamas, and fired two quick shots, one to the chest and one to the head. Although there were weapons in that bedroom, Bin Laden was also unarmed when he was shot.” One the most detailed accounts of the incident also reports that bin Laden was unarmed. See N. Schmidle, “Getting bin Laden,” The New Yorker, August 8, 2011, available at 〈http://www.newyorker.com/reporting/2011/08/08/110808fa_fact_schmidle?currentPage=all〉. The New York Times gives a different account: “When the commandos reached the top floor, they entered a room and saw Osama bin Laden with an AK-47 and a Makarov pistol in arm's reach. They shot and killed him, as well as wounding a woman with him.” New York Times Topics, “The Death of Osama Bin Laden,” available at 〈http://topics.nytimes.com/top/reference/timestopics/people/b/osama_bin_laden/index.html〉.

(64) See 〈http://www.cbsnews.com/8301-503544_162-20058913-503544.html〉 accessed November 3, 2011.

(65) See Model Penal Code, sections 210.1–210.4.

(66) For a criticism of this assimilation, see David Rodin, War and Self-Defense (Oxford University Press, 2003) esp. ch. 5.

(67) Perhaps this premeditated planning is what makes the death penalty objectionable. On the other hand, one can say that the state would be blameworthy if it did not carefully plan executions.

(68) I suggested this argument in my article “International Abductions, Low-Intensity Conflicts, and State Sovereignty: A Moral Inquiry,” 31 Columbia Journal of Transnational Law (1994) 584–5. Altman and Wellman, in their excellent book on international justice, take me to task for wrongly conceiving assassination of a tyrant as punishment without due process, as opposed to suppression of a threat. While I was not altogether clear, my emphasis was not so much on the villain's due process, but on the government as assassin. I wrote: “agents of a liberal democracy must conduct themselves in a way that honors the civic virtues for which they stand.” Ibid.

(69) Thus I concur with Jeremy Waldron's conclusion that targeted killings should not be legally allowed. Jeremy Waldron, “Justifying Targeted Killing With a Neutral Principle?”, in this Volume, at ch. 4.