I. INTRODUCTION
It is customary, in a piece for a special issue on a philosopher's works, to be critical (if constructively so) of the latter, to show where he or she ‘has got it wrong’. I shall do no such thing here, if only because (for what it is worth) I am broadly sympathetic to his stance.Footnote 1 Instead, I bring to bear his account of soldiers’ right to kill in war on an entirely neglected but nevertheless interesting issue, to wit, what the military call ‘blue-on-blue’ killings or, as I refer to such acts here, internecine war killings. Such killings or threats thereof have always occurred, of which the following are classic examples: the First World War officer who incompetently and recklessly orders his men to march into pointless butchery and who is revengefully killed by them in the heat of combat; the GI who ‘frags’ his commanding officer in the middle of the night or under cover of battle, hoping that the latter will be replaced by a more competent officer;Footnote 2 the soldier who is ordered by his officer, at gunpoint, to go into action or to kill innocent civilians; the soldier who collapses under the strain of combat and starts shooting wildly and indiscriminately at his comrades;Footnote 3 the air-force pilot who, in a case of so-called friendly-fire, fires missiles at tanks which he mistakenly believes to be the enemy's.Footnote 4
In this article, I assess whether McMahan's account of the right to kill in war licenses internecine war killings. I restrict my inquiry to cases where commanding officer O attacks his soldier S as a means to secure S's compliance with an order and where S kills him in self-defence.Footnote 5 In those cases, in contrast with internecine killings caused by friendly fire or committed by soldiers running amok, the fact that attacker and victim stand in a special relationship of authority and obedience must be taken into account when assessing the moral status of both the attack to which O subjects S and S's self-defensive response. I argue that, at the bar of McMahan's account of the right to kill in self-defence, O lacks a justification for attacking S as a means to enforcing his order, and S thus sometimes (but not always) has the right to kill O should he so act. I proceed as follows: I set out the central features of McMahan's account in section II. I examine internecine killings in cases where O's order is unjust in section III, and turn to the case of just orders in section IV.
Two final remarks before I start. First, as far as I am aware, McMahan has not written on internecine war killings. As a result, the conclusions I reach here as to the implications of his account for the issue at hand are inevitably speculative – so much so that it might perhaps be more appropriate to speak of a ‘McMahanian’ account of internecine war killings than of McMahan's putative take on those kinds of act. That being said, it seems to me that there is more scope in our field than is usually acknowledged for precisely that kind of speculation, which consists in applying controversial theories (as his surely is) to neglected issues as opposed to embarking on the familiar task of demolishing them.
Second, the essay's subject matter might seem unduly removed from the challenges which our soldiers face (by whom I mean American, British, French, German soldiers, etc.), belonging as they do to armies which regard themselves as subject to the rule of law and which do not currently use lethal force as a means to secure compliance with an order.Footnote 6 It is all too easy to forget, however, that a great many armies function rather differently and that their soldiers are often compelled under threat of serious harm to commit atrocities. In a case recently brought to the International Criminal Tribunal for the Former Yugoslavia (ICTY), Serb soldier Dragan Erdemovic was charged and convicted for participating in the 1995 killing of hundreds of unarmed male civilian Bosnian Muslims in Pilica. In mitigation, Erdemovic argued that he would have been shot by his commanding officer had he refused to obey that order. Further afield from Europe, it is a well-documented fact that many child soldiers are forced at gunpoint to commit atrocities against civilians and to engage in senseless combat.Footnote 7 To those soldiers, the issue of internecine killing is all too pressing, and it is to their predicament, thus, that the article addresses itself.
II. McMAHAN'S ACCOUNT
In mainstream just war theory, soldiers are permitted to kill enemy soldiers once the war has started, on the grounds that the latter subject them to a lethal harm. On that view, the moral status of the war – as just or unjust ad bellum – is irrelevant to the conferral on soldiers of the permission to kill the enemy: thus, although the German army wrongfully (I assume) invaded Poland on 3 September 1939, German and Polish soldiers were equally permitted to kill one another from that point onwards (until the end of that phase of the war). In a number of his writings, and most notably in his recent Killing in War, McMahan has revived a rather different account of soldiers’ rights and liabilities in war – to be found in pre-modern, classical just war theory. On that account, a soldier from belligerent community A is not permitted to kill a soldier from belligerent community B unless the latter either unjustifiably subjects him to a wrongful and serious harm to his life and limbs, or unjustifiably contributes to B's wrongful collective attack on A's territorial integrity or national sovereignty and is morally responsible for so doing. Suppose, then, that soldierB takes part in a wrongful aggression against community A and uses lethal force against soldierA as the latter seeks to resist the aggression. Suppose further that soldierB is morally responsible for his action. In this case, soldierB acts wrongfully, in just the same way as a robber acts wrongfully when shooting an armed bank guard in the course of the robbery. By that token, German soldiers were not, but Polish soldiers were, permitted to kill their enemy. According to McMahan, the basis of a soldier's liability to attack, and of his enemy's right to kill him, thus lies in his moral responsibility for the wrongful harm to which he subjects them.
I cannot, within the scope of this article, rehearse current debates between McMahan and his critics. Instead, I shall comment on those features of his account (as I understand it) which are relevant to internecine killings. First, a point of terminology. McMahan often describes attackers as subjecting their victims to lethal threats of harm. However, the word ‘threat’ is ambiguous in a way that is particularly problematic in the context of this article. On the one hand, a threat is (in the words of the OED), a ‘painful pressure, oppression, compulsion; vexation, torment; affliction, distress, misery; danger, peril’. On the other hand, a threat is also a risk, or (again in the words of the OED) ‘a declaration of hostile determination or of loss, pain, punishment, or damage to be inflicted in retribution for or conditionally upon some course; a menace’. Often, of course, attackers inflict on their victims a risk of misery or danger (in other words, subject them to a second-sense threat of a first-sense one), and one might think that this is what O does – that by telling or indicating to S that he will kill him unless S complies with the order, he is merely threatening to attack him. I believe, however, that by placing a gun to S's head, O just is attacking him – particularly if his threat is not empty (as I shall postulate throughout the article). For this reason, I shall use ‘attack’, rather than ‘threat’, as my main ‘descriptor’; and I shall use the word ‘threat’ only in the second, non-obsolete sense.Footnote 8
Second, an agent is liable to be killed only if he is morally responsible for contributing to a wrongful (severe) harm at time t1 on some other party – a harm, in other words, which the latter has done nothing to warrant. To say that the agent is liable is to say that he has lost his right not to be killed – more precisely, that neither his victim (V) nor her potential rescuers (R) wrong him by using lethal force at time t2, and that he is under a duty not to use lethal force against them at time t3 in defence of his own life. That last clause is important, for in saying that the agent may not seek to defend his own life, we accept that V and R not only may, but indeed have the right (as correlating into precisely that particular duty on his part) to kill him. By implication, subjecting someone to a wrongful and serious attack is a necessary condition for being liable to defensive force either at the hands of one's victim or at the hands of her rescuers. Note, however, that it is not necessary, for an attacker to be liable to defensive force, that he should threaten someone's life. An attack which results in grievous harm short of death, such as a severely disabling bodily or psychological impairment, can (sometimes) warrant a lethal response. To put the point differently, it is not necessary for someone to have the right to kill an attacker that the latter violates someone's right not to be killed: attacks on other fundamental rights, e.g. for example to bodily integrity, may provide both victims and rescuers with a just cause for killing him.Footnote 9
Third, an attacker can be liable to defensive force even if the attack has not already started. As I have already indicated, it strikes me as entirely appropriate to describe O (in the order-at-gunpoint cases) as engaging in an ongoing attack against S. Some might dispute that claim and counter that his attack is merely imminent. Be that as it may, McMahan in fact believes that, in principle at least, one does have the right to resort to lethal defensive force in order to block an imminent attack.Footnote 10 The point is particularly important in the present context, as it disposes of the claim that, insofar as O ‘merely’ puts a gun to S's head, and does not fire at him, the latter may not kill him as a matter of right. It could very well be, of course, that McMahan's account precludes permitting S to kill O – but if that is so, this cannot be because O's attack is imminent rather than ongoing.
Fourth, being subject to a wrongful attack is not sufficient for conferring on agents the permission and the right to kill attackers in self- or other-defence. It must also be the case that killing the attacker contributes to stopping him (the instrumentality condition), that it is a necessary means to that end (the necessity condition), and that it is a proportionate response (the proportionality requirement). What counts as ‘proportionate response’ depends, not merely on the magnitude of the good which it will bring relative to the harms thereby inflicted on the attacker or on third parties such as innocent bystanders, but also on the attacker's degree of causal and moral responsibility for his victim's predicament. (See Appendix for elaboration on this fourth point.)
Fifth, those conditions do not suffice to support the claim that an agent has the right (as opposed merely to a permission) to kill an attacker in self- or other-defence. For there to be such a right, the attacker must act in an unjustifiable way. Thus, a strategic bomber who fights a just war might well have a justification for dropping a bomb, at t1, on a munitions factory whose output is crucial to the enemy's fighting abilities even though the bomb will kill scores of civilians as a side-effect of the factory's destruction. But even if he does have such a justification (a claim which must in any event be substantiated), insofar as the civilians have not done anything to warrant the loss of their right not to be killed, they are not liable to being attacked, and thus are being wrongfully endangered. Accordingly, they are at liberty to defend themselves against him by shooting him down at t2. But insofar as the bomber may be deemed to have a justification for acting as he does, he in turn may defend himself against them at t3. In short, the bomber and the civilians are permitted to kill each other in self-defence, but lack the right (at least against each other) to do so precisely because the attack which they are mounting, though it will wrong or infringe the rights of certain people, is nevertheless justified. By contrast, a terror bomber who intentionally targets densely populated areas with a view to killing and terrorizing the civilian population not only subjects them to a wrongful attack, but lacks a justification for so doing. The civilians have the right to kill him in self-defence, which implies that they do not wrong him by so acting, whilst he himself is under a duty not to retaliate against them, in just the same way as he is under a duty not to attack them deliberately in the first instance.Footnote 11
Sixth, McMahan's account serves to justify defensive killing: it does not license punitive killings as carried out through the institution of the death penalty. This is because on his account, an agent has the right to kill some other party only if the latter unjustifiably infringes his (or someone else's) right not to suffer serious injury to life and limbs or contributes to an unjust attack on some collective good, and if the act of killing, in fulfilment of the instrumentality condition, serves to neutralize the attack. Insofar as a punitive killing is an act of retribution for past, rather than ongoing, wrongdoing, and insofar as it does not stop the wrongdoing to which it is a response, it is not permissible (at least on his liability account). I shall return to this issue in section IV.
Seventh, it has become standard to say that, for McMahan (and those who adopt a similar line) whether soldiers have the right to kill depends on the ad bellum moral status of their war. That, in fact, is a considerable oversimplification: combatants who fight an ad bellum unjust war sometimes kill permissibly, and combatants who fight an ad bellum just war sometimes kill impermissibly. Accordingly, whether combatants have the right to kill depends on the moral status of the particular phase or specific aim of the overall war in which they are involved.Footnote 12 This renders the account extraordinarily complex. For my purposes here, however, the oversimplification will broadly do, though I shall revisit the matter in section IV. But it does raise a terminological issue, namely, what is meant by the labels ‘just/unjust combatants’. Throughout this article, by an unjust combatant, and following McMahan's usage, I shall mean a combatant who is fighting a war which is unjust ad bellum for lacking a just cause, and who is morally responsible for so doing. By just combatant, conversely, I shall mean a combatant who is fighting a war which has a just cause (though it may be overall unjust on the grounds, for example, that it is disproportionate).
Finally, it is important always to bear in mind that McMahan's account of the right to kill in war is located in what he calls the ‘deep morality of war’ – a morality, in other words, which is largely untouched by practical concerns such as feasible implementation, and which is thus distinct from morally directed laws of war. Many have queried the usefulness of that distinction.Footnote 13 Defending or rejecting the very idea of a deep morality of war is not my concern here. Accordingly, the thesis that, on McMahan's account, S sometimes has the right to kill O in self-defence must be read as a part of the deep morality of war, insofar as the latter applies not merely to killing enemies but also to internal discipline within armies.
III. INTERNECINE KILLINGS AND UNJUST ORDERS
To recapitulate, a soldier A is permitted to kill another soldier B if the latter infringes his right not to incur grievous harm to his life and limbs (subject to requirements of instrumentality, necessity and proportionality). He has the right to do so only if B unjustifiably infringes – or, as I shall say henceforth, violates – his right not to incur such harm. Suppose, then, that O issues a grievously unjust order to S, that S refuses to obey the order, and that O then reiterates the order, but this time at gunpoint, so that S has no option of defence against O. Suppose further that S has every reason to believe that O will make good on his threat – for example because he has already seen O kill another soldier under similar circumstances. S, thus, is faced with the following three courses of action:
(1) Obey the unjust order.
(2) Disobey the order and let O kill him.
(3) Disobey the order and kill O.
On McMahan's account of the right to kill in war, I shall now argue, S has the right to kill O in self-defence in such cases. (I take no stand in this article on cases where S has the option of defending himself from O after he refuses to comply with the latter's order, and/or where the wrongdoing which he is ordered to carry out is not grievous.)
Officers often give wrongful orders to their soldiers. One might think that, in such cases, soldiers clearly do have the right to kill them in self-defence if the latter seek to enforce their orders at gunpoint. For even if using lethal force as an enforcement tool is in principle permissible (a rather strong assumption to make in any case), it is hard to see how it could ever be permissible if that act is not mandatory or, indeed, in some cases, is a grievous wrongdoing. Imagine the following two cases: O orders S, at gun point, to kill innocent civilians on the grounds that they belong to the ‘wrong’ ethnic group; O orders S, at gunpoint, to go into action into enemy fire with the aim of capturing a small hill of no tactical or strategic importance. In both cases, it is obviously and grievously wrongful of O to give that particular order: wrongful to the civilians in the first case, and to S in the second case. Moreover, in the first case, S would act grievously wrongly if he obeyed: option (1) is not open to him. In the latter case, that option is open to him, since he would not act wrongly (I submit) by obeying, but he is certainly not under a duty to choose that option, in virtue of a more general permission (which McMahan would clearly grant himFootnote 14) not to sacrifice one's life and limbs needlessly. By that token, S is not under a duty to go for (2), in either case. He thus may opt for (3) and kill O. More importantly still, as we saw in section 2, it is a necessary condition for someone to be liable to an attack that he should contribute to the imposition of a wrongful harm on some third party. Given that in either case S's refusal does not constitute such an act, O is subjecting him to a lethal attack which he has done nothing to warrant – to which, in other words, he is not liable. In addition, O lacks a justification for doing so, since there cannot be any justification for deliberately killing innocent civilians, or for subjecting one's subordinates to a wholly unnecessary risk of dying. Insofar as O therefore unjustifiably attacks S, S clearly has the right to kill him in self-defence.
In yet another kind of case, however, the issue deserves closer scrutiny. Suppose that O orders S at gun point to kill just enemy combatants. As I noted in section II, S lacks the right to kill those soldiers and, indeed, is not morally permitted to do so, unless, as we also saw, those ad-bellum just enemy combatants are committing in-bello unjust acts of killing. (Recall, however, that I set aside those complications in this article.) O's order is wrongful. But before concluding that S has the right to kill him in that case too, we must attend to the fact that, in that case, unlike the previous two, S's refusal to obey would put someone else's life at risk – namely the lives of his fellow comrades, and/or the lives of those, back on the home front, on behalf of whom the war is fought and who may well die if the enemy's progress is not checked. Might it not be the case that S, in refusing to obey, unjustifiably contributes to the imposition of a wrongful lethal harm on those third parties? And if that is so, might it not be the case that O's lethal attack is, in fact, one to which S is liable?Footnote 15
Recall that on McMahan's view, for S to be liable to being attacked by O, it must be the case not merely that S's refusal to act as per O's order itself constitutes a contribution to the imposition of a wrongful lethal harm either to O or to some other party: as per the instrumentality requirement for permissible defensive killing, it must also be the case that O's use of lethal force against S will block the latter's wrongdoing. I shall address the instrumentality requirement in section IV. In the remainder of this section, I focus on the first condition. As I suggested in the previous paragraph, one might think that the condition does obtain, on two distinct kinds of ground:
(a) S's refusal to obey wrongfully endangers the lives of his fellow, ex hypothesi, unjust, combatants, because it constitutes a violation of their right to life; or
(b) S's non-compliance wrongfully endangers the success of the mission or the war, thereby imposing a risk of lethal harm on those on whose behalf the war (though unjust) is fought.
The view just described contains an important insight, to wit, that the fact that one is engaged in a wrongdoing is not a sufficient basis for desisting here and now: extricating oneself from a course of action, to use Tony Coady's apt label, is often costly on third parties, and those costs must be taken into account when assessing the moral status of one's decision to withdraw even if that course of action was wrong to begin with. Thus, as Coady suggests, it might be wrong for John to have an extra-marital affair with Mary, but if the latter is prone to suicidal depression and would suffer a serious breakdown were John to break up with her prior to, for example, her seeing a psychiatrist, it is not clear at all that John ought to leave her now (in fulfilment of his prior obligation to his wife), as opposed to waiting until such time as she has started receiving professional help. Analogously, though it is morally wrong for a soldier to take part in an unjust war (according to McMahan), that alone does not support the claim that the soldier, once he has started fighting, must stop immediately.Footnote 16
Still, there are wrongdoings of such magnitude that one should desist immediately from committing them. Let us begin with claim (a). In what sense, if any at all, can S's refusal to obey O be seen as a violation of his comrades’ right to life? An agent's right against third parties not to incur a particular harm can be read either as a negative right that they should not inflict such harm on him, or as a positive right that they help him thwart the harm, or a combination of both. In the case at hand, S's refusal is not an act of killing directed at his comrades and thus cannot be read as violating their negative right not to be killed. Rather, it constitutes a withholding of assistance as the latter are subjected to an attack that is ex hypothesi just on the part of just enemy combatants. Accordingly, if it violates a right of theirs, it violates their positive right to receive life-saving assistance. But it may also perhaps be read as the violation of a negative right not to suffer other kinds of harms, such as the harms attendant on disappointed expectations or the harm of having to readjust their conduct and being subject to additional risks.Footnote 17
Recall the three courses of action between which S must choose:
(1) Obey the order by killing just enemy combatants.
(2) Disobey the order and let O kill him.
(3) Disobey the order and kill O.
Once again, I assume (in keeping with McMahan's account) that S is not under a duty to opt for (2). The question thus is which of options (1) and (3) he may, or must, choose. In the case at hand, enemy combatants are just: ex hypothesi they have a right not to be killed in that war. To claim that S must opt for (1) is to claim that the rights which his fellow unjust comrades have against him outweigh just combatants’ right not to be killed. But that seems wrong. For by complying with O's order, S, in the conflict which pits his comrades against enemy combatants, would help the former kill the latter. On McMahan's account, however, it is not permissible to kill someone in defence of her wrongful attacker: to claim otherwise would be tantamount to endorsing murder. S thus breaches no positive right to life-saving assistance of his comrades by refusing to obey O's order. Nor does he breach their negative right not to have their expectations of him disappointed: for whilst they may well have such a right against him when their expectations are justified, surely they lack a right to have their unjustified expectations fulfilled. Insofar as an expectation that one will continue to receive help and in so doing be exposed to lesser risks as part of an unjust enterprise is not a justified expectation, S does not violate any corresponding right of his comrades.
At this juncture, one might insist that war killings differ in one crucial respect from private killings, namely that the former, unlike the latter, are carried out by individuals who occupy the very specific role of a combatant, with its attendant privileges and duties. Combatants, one might argue, have a role-based obligation to look after one another which, in situations of forced choice between lives, outweighs obligations which they might have to spare just enemy soldiers. On that view, S has an obligation to his fellow soldiers to help them thwart the enemy's attack, even if he thereby contributes to the continuation of an unjust war or phase of war. However, those putative objections strike me as implausible, at least if one grants McMahan the claim that a judgement about the moral status of the war fought by O and S (and of their responsibility for the war's constitutive acts of killing) can be reached. Consider the following domestic example. Police officers have a number of role-based obligations, vis-à-vis members of the public of course, but also vis-à-vis one another – such as an obligation to come to the help of fellow officers, particularly from their own unit, e.g. if the latter find themselves under threat from gang members. Suppose now that a group of corrupt officers, led by Detective Chief Inspector Bloggs, are locked in a shootout with the anti-corruption squad, which has been tasked to arrest them. Bloggs orders his men to keep shooting, but one of them, Smith, decides to stop fighting. Surely we would not say (either at the bar of McMahan's account or, indeed, of any plausible account of the right to kill) that Smith has a prior obligation to his corrupt unit members that outweighs whatever obligation he has not to kill individuals who are not ex hypothesi subjecting him to a wrongful attack. If so, S's general obligation not to kill just enemy combatants trumps (prima facie) special obligations he may have to his comrades not to impose harm on them or to provide them with life-saving assistance. And insofar, thus, as S's refusal to obey O's order does not violate his comrades’ right to life, O is subjecting him to a wrongful attack – and unjustifiably so, since their war is unjust. This confers on S the permission, indeed the right, to opt for (3) – namely to kill O in self-defence.
Let us now consider the case where, as per claim (b) above, S's refusal to obey O's order exposes innocent civilians to a high risk, or even certainty, of dying at the hands of just combatants. Recall the case of the strategic bomber SB who, in prosecution of a just war, bombs a weapons factory which is located in a densely populated area, and, in so doing, unintentionally but foreseeably kills scores of civilians. Suppose that the latter are members of the political community on behalf of whom O and S fight. O orders S, at gunpoint, to shoot SB down. If S refuses to do so, fellow compatriots will die. If he complies, they will survive, at the cost of the life of a just combatant. In the previous case, I denied that S was violating his comrades’ right to assistance, on the grounds that he could have no obligation to help them, unjustified attackers as they were, to the detriment of their innocent victims. In this case, however, the civilians do have a justification for killing SB. The question, then, is whether S is violating their right to assistance by refusing to shoot him down in compliance with O's order. To claim that he does is to imply that he owes them a duty of assistance which he does not owe to SB. In fact, and more strongly still, it is to imply that he is under a duty to kill SB in their defence rather than to allow them to die (since killing SB is the only way in which he can, in this instance, promote their right to life). The difficulty with that claim is that, ex hypothesi, SB has not lost his right not to be killed. But if we say, instead, that S must disobey O and let SB drop his bomb, at the cost of the lives of fellow compatriots, we must confront the discomforting thought that they too still have a claim to live.
Two different sets of considerations, both of which McMahan endorses, pull in opposite directions. On the one hand, the doctrine of acts and omissions, stipulating as it does that killing someone is worse, generally, than letting them die, gives us a prima facie reason for affirming that S may not kill SB but, rather, ought to let the civilians die. On the other hand, S is under a contractual duty to protect the civilians on whose behalf his army is fighting, and that gives us a prima facie reason to affirm that given the choice between killing SB or letting the civilians die, neither of whom have lost their right not to be killed, he ought to do the former.Footnote 18 In fact, it is tempting to think, in addition, that insofar as those civilians are under attack partly because S shares responsibility for the fact that his community is fighting a war against SB's community, he owes it to them as a matter of restitutive justice to help them thwart the attack. His duty to them, thus, is contractual, restitutive, or both. And if those points are correct, then we might well be drawn to the thought that he is unjustifiably contributing to a wrongful attack on their lives by refusing to comply with O's order.
I believe that McMahan's account pulls towards the view that S ought not to kill SB, from which it follows that his refusal to obey O's order does not constitute a wrongful contribution to the death of the civilians. For if S kills SB who, remember, is fighting a just war, he is contributing to the continuation of a war for which, on his side, there is no justification. But if he desists, though he allows some killings to take place, those killings are ex hypothesi justified precisely by the fact that SB's war is just. Given the choice between contributing to an unjust war by killing just combatants, and omitting to block the permissible killing of those one has been hired to protect, I believe that McMahan would prescribe the latter. On his view, participating in an unjust war is wrong, and killing in prosecution of that war is wrong. Although that view is compatible with the thought that ad-bellum unjust soldiers sometimes kill justly (when, for example, they protect their compatriots from unjustified killings as carried out by ad-bellum just combatants), it does imply that there is a prima facie case for ad-bellum unjust combatants to withdraw from the war, at the cost of the lives of their compatriots.Footnote 19
To recapitulate, then, pending further argument to the effect that S is unjustifiably contributing to the imposition of a wrongful lethal harm on third parties by refusing to obey, O is subjecting him to a wrongful lethal attack by ordering him, at gun point, to act in the aforementioned ways. From that it follows (subject to the requirements of proportionality, necessity and instrumentality) that S may kill him in self-defence. Moreover, it should be clear that O lacks a justification for so acting: the fact that, by obeying O, S would kill innocent civilians or enemy soldiers who ex hypothesi have not lost their right not to be killed deprives the former of any sound justification for enforcing that order at gunpoint. S, therefore, not merely is permitted, but also has the right, to kill him in self-defence.
IV. INTERNECINE KILLINGS AND JUST ORDERS
Suppose now that O orders S, at gunpoint, to kill unjust combatants, without thereby exposing him to disproportionate and unnecessary risks to life and limb. Here again, S, who to repeat does not have the option of defending his life once he disobeys, must choose between the following three courses of action:
(1) Obey the just order by killing unjust enemy combatants.
(2) Disobey the order and let O kill him.
(3) Disobey the order and kill O.
Ex hypothesi, O's order is just. The question, thus, is whether O violates S's right not to be attacked by enforcing the order at gunpoint and, if so, whether the latter may, indeed has the right to, kill him in self-defence. One might think that he may not do so, on the ground that he has after all a morally acceptable alternative to killing O, since he can kill enemy combatants who ex hypothesi are liable to be killed, in compliance with an order which he is under a prima facie duty to obey. As I shall now argue, however, O violates S's right not to be attacked; and although S does, in some cases, lack the right to kill O in self-defence, the fact that O attacks him unjustifiably nevertheless sometimes provides him with a justification for killing him, as a means to save his life.
Let us begin, then, by assessing the moral status of O's resort to a lethal attack as a means of enforcing his just order. In the light of our discussion in section III, one might think that there are two related reasons for affirming that O does not, in this case, wrong S by attacking: unless S kills those unjust combatants, his comrades will die wrongfully at the hands of the enemy, and/or the just mission or the just war will be in jeopardy, as a result of which innocent people will die at the hands of unjust combatants.Footnote 20 On McMahan's account, however, the argument works only if S's refusal to comply constitutes a right-violation which is severe enough to warrant a lethal attack, and, in that case, if the attack not only stops S from acting wrongly (as per the instrumentality requirement) but is also a necessary means to that end (as per the necessity requirement).
Clearly, S's refusal to obey O exposes his fellow soldiers and citizens to danger. Moreover, ex hypothesi, O's order is just, in the twofold sense that the enemy are liable to be killed and that S is not at greater risk than he is under a duty to incur. Let us grant for the sake of argument that S is under a duty to protect his fellow soldiers and compatriots – not least because (or insofar as) he has voluntarily enlisted in the army – an organization whose task, which necessarily involves killing and risking one's life, is precisely to prosecute wars. To the extent that his refusal to fight would compromise the lives of his comrades, it does constitute a rights-violation, in sharp contrast with the case which we considered at the close of section III.
That said, the foregoing does not exonerate O from the charge of violating S's right not to be killed. For as we saw in section II, the claim that S violates the rights of a third party does not suffice to make him liable to attack: in addition, the rights-violations must be such as to warrant a lethal response. In the case at hand, although S is under a duty to kill unjust enemy combatants as per O's order, and although he violates other soldiers’ and civilians’ rights by refusing to obey, it is not clear that his wrongdoing warrants a lethal response. Two different considerations seem to support that view. First, take the case where S wrongfully exposes his comrades to death by refusing to fight. Suppose, a contrario, that wrongfully letting someone die does cause one to lose one's right not to be killed. That would imply, for example, that if A refuses to give B the loaf of bread which the latter, unlike A himself, desperately needs as a matter of survival, and if B can secure the bread by killing A, then A has lost his right not to be killed by B. On that view, wrongfully letting someone die and wrongfully killing them are met with the same radical, irreversible response (the wrongdoer's loss of life). The difficulty with this view, however, is that it seems to ride roughshod over the intuition (which, remember, McMahan accepts) that there generally is a meaningful moral difference between wrongfully killing someone and wrongfully letting him or her die. And if that is correct, then surely a juncture at which that difference must manifest itself is in the kind of acts by which agents may seek to stop or redress those different kinds of wrongdoing: for it would seem somewhat odd, on the one hand, to assert that (other things equal) killing is morally worse than letting die, and, on the other hand, that killing and letting die, though not morally on a par, may both warrant an imminent threat of lethal harm.
But perhaps this is wrong – perhaps S is a legitimate target for O's attack even though he merely lets his fellow comrades die. I confess to being unsure on that point. Even so, there is another reason for doubting that his wrongdoing warrants a lethal response. S's duty is a duty to provide assistance by way of personal services – in that instance, killing. To put it differently, it is a duty to perform the relevant services. In law, however, an employee who refuses to do the work which his contract binds him to do will not be made to do it; rather, he will have to pay a fine, be sacked – in the most serious cases, perhaps, go to prison. In other words, whilst penalties will be attached to non-compliance, there is no requirement of specific performance – partly on grounds of efficiency, but also out deep-seated and powerful objections to the institution of forced labour. And it seems to me that if S would not be forcibly made to fight were he to go AWOL (for example), he ought not to be subject to a lethal attack by O as a means to coerce him into fighting.
Note, moreover, that if O does actually start shooting at S, his act is wrongful not only for the reason just mentioned, but under the terms of the requirement of instrumentality. As we saw in section II, it is a necessary condition for an agent to be liable to defensive force that the use of such force should be a means to stop him. In the case at hand, the instrumentality condition would not be met, for the very simple reason that (bluntly put) killing someone is not the best way to get him to fulfil his duty. If that is correct, then insofar as O's lethal attack would not fulfil the instrumentality condition for permissible defensive force, it is wrongful. S, therefore, is not liable to it.
At this juncture, some might be tempted to press that even though O's subjecting S to a lethal attack will not stop S's wrongdoing, and thus constitutes an infringement of S's right to life, it may induce S's fellow soldiers to comply with O's (just) orders and help ensure the success of the mission. The thought is worth addressing, because it brings to light different construals of the instrumentality requirement. On a narrow construal, one has the right to kill a wrongdoer only if one thereby stops him from violating the right(s) of another party. Given that killing S cannot stop S himself from wrongfully violating the rights of his comrades, S is not liable to be killed by O, however wrongful his refusal to disobey. But even if that is correct (some might think), and even if, thus, O's attack on S is wrongful at the bar of narrow instrumentality, O might nevertheless be acting in compliance with a requirement of wide instrumentality, whereby one has the right to kill a wrongdoer only if killing him stops a wrongful attack simpliciter.Footnote 21 On the wide construal, O's attack on S is an exemplary attack (as opposed to a punitive one, which, as we saw in section II, McMahan's liability account disallows): for O kills S for insubordination, not as a means to stop the wrongdoing he (S) committed (or to punish him for disobeying), but as a means to induce S's fellow soldiers to comply with his just orders. The First World War is one theatre in which many such killings occurred. On the French side, the most often cited case is that of the general who reportedly ordered his gunmen to fire on French trenches to force soldiers to ‘go over the top’. On the British side, one Brigadier-General Crozier talked, in a provocatively and eponymously titled book published in 1937, of ‘the men he killed’ – not, as one may be forgiven for surmising, German soldiers, but British ones who refused to die in the carnage of mindless trench-warfare.Footnote 22
The question, then, is whether the right to kill is dependent on narrow or wide instrumentality. Now, to deem S liable to attack at the hands of O on the grounds that killing him would – through example – induce some of his comrades rightfully to kill unjust combatants in defence of one another, is to say that S is liable to being treated as a means only to his comrades’ ends. Put differently, the narrow instrumentality requirement yields the conclusion that, once other conditions are met (such as proportionality), a wrongdoer is liable to be killed in an eliminative fashion (to kill him thwarts the attack which he poses). By contrast, the wide instrumentality requirement implies that once those other conditions are met, a wrongdoer is liable to be killed in an opportunistic fashion, as a means only to someone else's ends. Crucially, McMahan need not deny that one sometimes has the right deliberately to kill wrongdoers opportunistically. Suppose, for example, that the anti-terrorism squad have managed to corner members of a terrorist cell in an enclosed space, moments after (at time t3) one of them has programmed a bomb (at time t2) to detonate five hours from thence in a busy shopping centre. Assume further that the bomb programmer cannot stop the timer and that only his cell leader (who is present at the scene) has the deactivation code. Killing the programmer will not stop the threat. But suppose that killing him would induce the cell leader to deactivate the bomb. It is entirely open to McMahan to advance the following view: to the extent that the programmer is morally responsible for a wrongful lethal attack on innocent people and lacks a justification for so acting, he is liable to eliminative defensive force as he is about (at time t1) to set the timer; if killing him at t1 is not possible, then killing him opportunistically, in an exemplary fashion, at t3 is the next best thing – so to speak. However, if I am right that S's wrongdoing is not of a kind as to warrant eliminative killing, then I believe that McMahan would not endorse O's opportunistic killing of S. For in his brief remarks on the distinction between eliminative and opportunistic killing, he seems implicitly to suggest that justifying the latter is harder than justifying the former. If that is so, it would be in tension with McMahan's account of the right to kill, and in particular of the right to kill deliberately, to claim that one has the right opportunistically (and deliberately so) to kill wrongdoers whose contribution to the imposition of a wrongful lethal harm is not such as to warrant eliminative defensive force in the first instance.Footnote 23
Finally, it is worth bearing in mind that in so acting, O is subjecting S's comrades to an attack which is relevantly similar to that which he poses to S himself, and which is therefore wrongful. One might think that the cases are relevantly different, on the grounds that O is not, after all, attacking S's comrades: rather, he is threatening to do so by showing, in killing S for insubordination, that he ‘means business’. Clearly, however, those men now have every reason to believe that he will make good on his threat to them if they disobey. Accordingly, he is subjecting them to a lethal attack which is an inappropriate response to their wrongdoing by omission. If they comply, in fear for their lives, O will have satisfied the wide instrumentality requirement for permissible killing, at the cost of the requirement that he not treat S as a means only. If they do not comply, and if O makes good on his threat, killing them all, he will have failed under the terms of both that requirement and the requirement of instrumentality. On either count, they too are not liable to being attacked by O.
Does O nevertheless have a justification of a lesser-evil kind for attacking S, and, thereby, his comrades? On McMahan's account as articulated in Killing in War, it is not clear to me that he may. To be sure, in subjecting S to a wrongful lethal attack, thereby inducing S's comrades to obey his just orders, O serves the just end of contributing to a just war or, more locally as it were, of ensuring that his (just) soldiers are protected by one another from unjust combatants. In so acting, however, he would violate the Kantian requirement that we should not treat one another – including wrongdoers – as means only to our ends. Contrast O with the strategic bomber, who has a justification for (unintentionally) killing innocent civilians precisely insofar as his dropping the bomb will serve a (just) end. SB does not fall foul of that requirement, since he does not intend the civilians to die. O clearly does, since he opportunistically attacks S (as a means to secure compliance from S's comrades). It is true that McMahan seems to think that lesser evil justifications may sometimes permit the deliberate killing of those who are not liable to be killed. But as far as I can tell, he does not provide a full argument for that view.Footnote 24 Insofar as the Kantian requirement plays an important part in his account of the right to kill, it seems at least plausible to hold that the requirement should apply not merely to the determination of a wrongdoer's liability to harm, but also to the determination of justifications for the deliberate imposition on agents of harms to which they are not liable.
There is one further consideration in support of the claim that O acts unjustifiably: although the exemplary killing of wrongdoers might have some success at first, it is unlikely that it would be effective in the long term, as it would in all likelihood adversely affect morale in O's unit and undermine the latter's cohesiveness, and in turn jeopardize its ability to fight well and advance their just cause. Accounts of troops’ rebellion in the First World and Vietnam against harsh measures meted out to them by superior officers bear those points out. Thus, even if O manages to get his men to comply at first, thus ensuring the success of the mission, it is likely that over time his conception of order-enforcement might jeopardize the success of the larger phase of the war of which the mission is a part – which in turn suggests that his attack on S would not satisfy the wide instrumentality requirement, and would thus be unjustified.Footnote 25
To recapitulate, since O unjustifiably uses lethal force to enforce his order to S, he is violating the latter's right not be killed: the moral status of the war or phase of war in which S and O are engaged is thus irrelevant to the determination of O's enforcement strategy as a rights-violation. Insofar as O violates S's right not to be subject to a lethal attack, S is not under a duty to let O kill him. Given that, by implication, S may choose to save his life, the question is whether he may do so by killing O or, on the contrary, must do so by obeying O's just order.
At the outset of section III, I noted, drawing together the various strands of McMahan's account of the right to kill in self-defence, that a soldier has the right to kill another soldier in self-defence only if the latter unjustifiably imposes a (serious) wrongful harm on him (or unjustifiably contributes to such imposition). But moral responsibility for the unjustified imposition of a wrongful harm is not a sufficient condition for liability to lethal defensive force: as we saw in section II, it must also be the case that the victim of the harm lacks a morally acceptable alternative. When O's order is unjust, consisting as it may do in killing innocent civilians or just enemy combatants, S lacks such an alternative. In this case, however, S is morally permitted to kill unjust combatants. More strongly still, S is under a prima facie duty to obey his commanding officer's just orders. It would seem clearly to follow, then, that although O does violate S's right not to be subject to a lethal attack, and although S is not under a duty to let O kill him, he is under a duty to kill unjust combatants rather than O (should he choose to save his own life).
Yet, the issue is not as straightforward as it might seem. For once O has issued his threat, let alone if he starts acting on it, he too is a wrongful attacker vis-à-vis S. To claim that S must kill the enemy rather than O is to imply that even though both the enemy and O are wrongful attackers, the fact that the former issued a just order suffices to tip the balance in his favour, even though it is not enough to provide him with a justification for attacking S in the first instance. Now, when other things are equal, then it does indeed seem that the justness of the order protects O from being killed by S. When they are not equal, however, it might not do so. Suppose, for example, that O starts shooting at S and his comrades to induce them to go out into open terrain and start the mission which, it is hoped, will result in the (justified) death of a number of unjust combatants. Suppose further that S has a greater chance of surviving if he kills O than if he complies. Granted, ex hypothesi, the latter risk is one which he may be expected to incur: otherwise, to repeat, O's order would be unjust. Still, given that O violates S's right not to be killed, it is not clear that, on McMahan's account, S is under an obligation to target the enemy in order to save his own life precisely because he would incur a greater risk of failure than if he were to kill O. In fact, it seems open to McMahan to say, in that case, not merely that S is permitted to kill O in self-defence but, in addition, that he has the right to do so. The point holds even more strongly, I believe, in the following kind of cases. Suppose that O orders S at gunpoint to bomb a munitions factory which is located in a densely populated area and that S refuses. Or suppose that O orders S at gunpoint to target enemy combatants who have taken up position in inhabited civilian dwellings. Insofar as O's order is assumed to be just, S is morally permitted to obey it even though he would thereby foreseeably though unintentionally cause the deaths of many innocent enemy civilians. In that sense, he may save his life from O by obeying the latter's order. But insofar as O violates his right not to be killed, it would seem that S, at the very least, is permitted instead to kill O – to put it differently, that he is not under a duty to protect himself from a wrongful attacker by acting in such a way as to kill many innocent persons. In fact, I would suggest that it is entirely compatible with McMahan's account to hold that S has the right so to act. There might well be other cases where the fact that O issues a just order, and thus provides S with a morally acceptable alternative to block the attack against him, is not enough to deny S the right to kill him in self-defence. Space does not permit me to pursue this line of inquiry here. Suffice it to say that, if the conclusion reached here is correct, the moral status of O's order as just or unjust does not make all the difference between cases where S has, and cases where he lacks, the right to kill O in self-defence.
V. CONCLUSION
Internecine killings, I noted at the outset, are more common than one might think, and yet they are overlooked by just war theory, concerned as it is with offering an account of the conditions under which combatants may kill the enemy. In this article, I focused on the internecine killing of commanding officers by their own men as committed by the latter on the grounds that the officer subjected the soldier to a lethal attack. I argued that soldiers sometimes have the right so to act, even if their officers’ orders are just orders. McMahan's central insight is that (as classical strands within the just war tradition aver) the moral status of the war ad bellum decisively bears on the conferral of the right to kill enemy soldiers. As we saw, however, such consideration is not always decisive with respect to the conferral on S of a right to kill O in self-defence: whether or not O's order is wrongful, and whether or not S is under a duty to obey it, O is not justified in enforcing it by applying lethal force, and the fact that he is not justified in so acting does give S a right to kill him, even in (some) cases where S is under a duty to do that which he is ordered to do. In that respect, the article did not merely scrutinize a kind of war killing which the relevant literature consistently overlooks: it also offered a modest contribution to McMahan's account. Of course, some might think that his account is all the worse for yielding the conclusions which, I argued, it delivers with respect to the killing of officers by their soldiers. Whether they would be right is another matter.Footnote 26
APPENDIX
In section II, I note that on McMahan's account of the right to kill in war, being subject to a wrongful attack is not sufficient for conferring on agents the permission and the right to kill attackers in self- or other-defence. It must also be the case that killing the attacker contributes to stopping him (the instrumentality condition), that it is a necessary means to that end (the necessity condition) and that it is a proportionate response (the proportionality requirement). Three additional points are worth making here. First, the necessity requirement is notoriously troublesome. Suppose that you are (wrongfully) attacked by two individuals A and B, that killing either one of them will suffice to thwart the threat and that you are in a position to choose (from your good vantage point) which one to kill. In that case, killing A is not necessary to your survival; nor is killing B. Strictly interpreted, thus, the necessity requirement does not permit you to kill either, which is absurd. Here, however, I have in mind the thought that if the act of killing itself is necessary to your survival (as opposed to, for example, breaking your attacker's leg), then you may commit it (provided other conditions obtain).
Second, I should say here that McMahan does not distinguish between the requirements of instrumentality and necessity: rather, instrumentality, on his view, seems to be part of necessity. As he puts it,
the assignment of liability is governed by a requirement of necessity. If harming a person is unnecessary for the achievement of a relevant type of goal, that person cannot be liable to be harmed. The infliction of a deserved harm is, by contrast, not governed by a requirement of necessity, since the value of a person's getting what he deserves is not instrumental and hence is not necessary for anything beyond itself. (Killing in War, p. 9)
Yet there is a difference between the claim that killing an attacker is unnecessary in the sense that some means other than killing can do the job, and the claim that killing an attacker is non-instrumental in the sense that it will not serve the purpose of stopping the attack. I do not think that drawing that particular distinction constitutes a gross distortion of McMahan's overall account.
Third, McMahan's complex account of proportionality differs from conventional accounts in that, unlike the latter, it includes considerations of agents’ responsibility for wrongful harms in the calculus. Mainstream theory only looks at the relationship between the harms which the defensive killing averts and the loss incurred by wrongdoers and others as a result of the killing.
