Cosmopolitan War

Cécile Fabre

https://doi.org/10.1093/acprof:oso/9780199567164.001.0001
Published online: 24 January 2013 Published in print: 27 September 2012 Online ISBN:
9780191746055 Print ISBN: 9780199567164

Search in this book

CHAPTER

2 Collective self-defence 

Downloaded from https://academic.oup.com/book/33082/chapter/281977852 by Library user on 09 May 2025

Cécile Fabre

https://doi.org/10.1093/acprof:oso/9780199567164.003.0003 Pages 51–96

Published: September 2012

Abstract
Chapter 2 offers an account of wars of collective self-defence against an unwarranted military aggression
—the paradigmatic case of a just war, and yet the kind of war, precisely, which might seem to pose
insuperable problems for cosmopolitanism, since it pits sovereign political communities against one
another. The chapter delineates a normative account of defensive killings in private contexts and applies it
to the case of collective self-defence. It argues that whether combatants have the right to kill enemy
combatants (largely) depends on the moral status of the cause for which they ght. The defence of the
territorial integrity and political self-determination of one's political community against unwarranted
aggression is a just cause because of the importance to individuals of those collective goods. The chapter
then examines tensions between, on the one hand, the cosmopolitan claim that membership in a given
political community is irrelevant to the conferral and infringement of rights, and on the other hand, the
collateral killing of enemy non-combatants. In the course of doing so it highlights some important
revisions to the principle of proportionality.

Keywords: collective self-defence, defensive killing, collateral killing, just cause, proportionality
Subject: Social and Political Philosophy, Practical Ethics, Moral Philosophy
Collection: Oxford Scholarship Online
 2.1 Introduction

On 24 June 1812 Napoleon ordered his forces to cross the Niemen river into what was then Russian territory. On
4 August 1914 the German Army invaded Belgium and quickly made its way to France which, though not
neutral, nevertheless had not declared war upon either Germany or her allies. On 22 September 1980 the Iraqi
government ordered its army to carry out an air and land invasion into Iran's territory. On 2 August 1990 it gave
a similar order, this time against Kuwait. The resulting wars were of the paradigmatic kind: namely, wars of
collective self-defence by a state against a military aggression carried out by another state. Yet states, I argued
in Chapter 1, are legitimate only if and to the extent that their institutions and of cials, through the laws which
they vote and enforce and the executive decisions which they make on the basis of those laws, do not violate
the fundamental rights of both the state's members and outsiders. Moreover, the rights, powers, immunities,
and liberties of state of cials ow from and supervene on the rights, powers, immunities, and liberties of

Downloaded from https://academic.oup.com/book/33082/chapter/281977852 by Library user on 09 May 2025

citizens. It follows that a state, qua state, does not have the right to defend itself, in the sense that it does not,
qua state, have rights in the rst instance; it also follows that individuals qua members of a sovereign political
community do not have the unconditional collective right to defend the latter's territorial integrity and
institutional structures. Rather, they have it only if the military aggression to which they are subject is
unwarranted.
The view that unwarranted aggression provides its victims with a just cause for war has always been central to
the tradition. Indeed, the so-called fathers of codi ed international law (Vitoria, Grotius, and Vattel) are
notorious for their highly moralized interpretation of the jus gentium. Yet international law from the eighteenth
century to the early decades of the twentieth century conferred on states the unlimited rights both to protect
their sovereignty and to resort to war as an instrument of foreign policy: neither the invasion of Russia by
Napoleon nor the invasion of France and Belgium by Germany were deemed unlawful at the time. Clearly,
p. 52 however, a world in which states and their leaders are entitled to go to war at will is not one in which state
sovereignty can ourish. Nor is it one which is particularly conducive to human ourishing. It took two world
wars and the concomitant death of dozens of millions of people for the international community to revise the
law. Key legal sources, in that regard, are the Charter of the United Nations (1945) and the Rome Statute
instituting the International Criminal Court (1998). The Charter prohibits UN members from using or
threatening to use armed force against the territorial integrity or collective self-determination of another state
(Article 2(4)), but confers on them the right to defend themselves against a military attack (Article 51). The
Rome Statute, for its part, explicitly lists aggression as one of four crimes over which the ICC claims jurisdiction
(Article 5 (1, d).) What counts as aggression is by no means uncontroversial in the law—as evidenced by the
facts that the UN and the ICC did not provide their own de nition of it until, respectively, 1974 (by way of
General Assembly resolution 3314) and 2010 (by way of a resolution passed by the ICC Assembly of state
parties.) By and large, international law de nes aggression as the use of force by a state against another state,
which includes, inter alia, land invasion, air bombardment, and blockades. Unlawful aggression consists in the
use of such force other than as (a) a response to provocation which itself imminently threatens territorial
integrity and collective self-determination, (b) in self-defence or other defence, or (c) as a result of UN
authorization.1
The international law of war has thus rediscovered its roots in the just-war tradition. And in that tradition,
Russia, Belgium, France, Iran, and Kuwait, faced as they were with a large-scale armed attack onto their
territorial integrity and political sovereignty, had a just cause for using military force to repel their aggressor.
Recently, however, some contemporary cosmopolitan theorists have argued that in the face of changes in the
nature of warfare, statist accounts of war as an instance of national self-defence ought to be abandoned. On
their views, wars should instead be regarded as a police operation undertaken by international peacekeeping
forces with a view to enforcing the fundamental human rights of individuals. I reject the enforcement model of
war elsewhere, on the grounds that it cannot, on its own, fully account for the fact that the right to wage and
p. 53 ght a war includes, inter alia, a right to kill and maim other persons in defence of one's and compatriots’
rights. The enforcement model, I argue there, does not provide a justi cation for that particular right.2
Moreover, the intuition that a territorially bounded and sovereign community has the right to defend itself by
force against an unwarranted aggression is too deeply rooted in common-sense morality to warrant
abandoning without attempting to defend it against its critics. In this chapter my primary aim is thus to offer a
normative account of those wars to which the right to kill is central and which is compatible with the tenets of
cosmopolitan justice as set out in Chapter 1. The task is not as straightforward as it may seem. For if
membership in such a community is largely irrelevant to individuals’ basic entitlements, it is not clear, as rst
sight, how the protection of territorial integrity and collective self-determination can warrant killing human
beings, particularly (as will more often than not happen in such a war) civilians who are not themselves directly
responsible for the wrongful aggression carried out by their military forces.
The chapter proceeds as follows. In Section 2.2 I sketch a normative account of defensive killings in private,

Downloaded from https://academic.oup.com/book/33082/chapter/281977852 by Library user on 09 May 2025

domestic contexts. I bring that account to bear on the case of collective self-defence in Section 2.3. Having thus
shown that members of a wrongfully aggressed country have the right to authorize their leaders to wage war
on their behalf in defence of their rights to territorial integrity and collective self-determination, I examine, in
Section 2.4, some tensions between, on the one hand, the cosmopolitan claim that membership in a given
political community is irrelevant to the conferral and infringement of rights, and on the other hand, the view
that the collateral killing of enemy non-combatants is (at least sometimes) permitted.
Let me make one further comment before I begin. In Chapter 1 I argued that inequalities in prospects for a
minimally decent life are traceable to residence in, rather than nationality or citizenship of, a given sovereign
political community. Clearly, long-term residents in a country who do not have citizenship rights often have a
vested interest in that country's not being wrongfully invaded. However, the rights which an aggression
primarily harms, to wit, interests in controlling both access to the country's territory (which includes its
airspace and territorial waters) and its institutions, are held by citizens, and it is on that special relationship
such as obtains between fellow citizens or compatriots on which I shall focus in this chapter.

p. 54
2.2 Defensive rights

As I noted in Section 1.2.2, a theory of justice must provide an account of what agents may and have the right to
do in defence of their rights. The right to wage a war of collective self-defence, I shall argue below, consists in a
right to maim or kill enemy combatants in defence of one's individual and jointly held rights. But here a
dif culty arises, pertaining to what some have called the dual morality of war.3 War, typically, is waged by
groups against other groups in the name of communal and often national values; however, those who do the
actual ghting and die from it are not groups as such but individuals acting together. The alleged dual morality
of war—as waged by groups ad bellum and fought by individuals in bello—is thought to pose a particular
problem for standard justi cations of the right to wage and kill in war. Those justi cations divide into so-
called analogical and reductive strategies. The analogical strategy claims that just as an individual has a right
to defend herself from a lethal threat by killing her attacker, states have a right to defend their territorial
integrity and sovereignty from other aggressive states. The eighteenth-century European jurists Vattel and
Wolff, who sought to apply to states the natural laws which regulate individuals’ conduct towards one another,
were foremost proponents of the analogical strategy—which underpins much of the international laws of war.4
The problem with the analogical strategy is twofold. First, it fails to account for the fact that war is about killing
and maiming, and that those acts are committed by individuals. Second, it should be clear from the accounts of
cosmopolitan justice and group rights which I deployed in Chapter 1 that a cosmopolitan theory of the just war
cannot subscribe to it, since it confers on groups, and in particular states, rights which are not reducible to the
rights of their individual members. From a cosmopolitan point of view, the right to wage war must be
 explainable in terms of, or by reference to, some rights of individuals. The reductive strategy seems more
p. 55 promising in that respect, for it regards individual rights as central to the right to wage war. That said, it has a
collectivist and an individualistic variants. In its collectivist variant it confers on the state the right to wage
war in defence of the fundamental rights of its citizens.5 As we saw in Chapter 1, however, the rights of state
of cials are reducible to rights held by individuals, either qua individuals or qua group members. Accordingly,
state of cials’ right to defend the rights of fellow citizens cannot be justi ed other than by justifying a citizen's
very same right.6 The right to wage war, therefore, is a right as held by individuals to kill in defence of their
rights—as per the individualistic variant of the strategy.
In what follows, I sketch out the account of defensive rights which underpins my account of the right to wage
war in general and a war of collective self-defence in particular. By defensive rights I mean both the right to kill
in self-defence and the right to kill in defence of others. The quali cation is crucial, because in a war of
collective self-defence, combatants do not kill merely in defence of their lives; they also kill in defence of their

Downloaded from https://academic.oup.com/book/33082/chapter/281977852 by Library user on 09 May 2025

comrades’ lives, as well as in defence of the lives and other fundamental rights of non-combatants back on the
home front. Accordingly, when thinking about killing in wars of collective self-defence, we should not focus
solely on one-to-one cases where a lone victim V is subject to a lethal attack at the hands of a lone attacker A.
We should also bear in mind cases where a number of victims are subject to a lethal attack at the hands of a
number of attackers, and where those victims, when killing those attackers, not merely each defend their own
lives but also defend the lives of fellow victims.
That said, justi cations for the right to kill in defence of others ow from arguments in favour of the right to
kill in self-defence, so we should start with the latter. Consider the following scenarii, involving a victim (V) and
p. 56 her attacker (A), both of which are drawn from Judith Thomson's seminal article on self-defence:7
Villainous truck driver. V is walking in the park minding her own business. A is a villainous truck-
driver who drives at speed towards V with the manifest intention of killing her—though she has not
provoked him in any way.
Psychotic gunman. V is walking in the park minding her own business. A is in the grip of a psychotic
breakdown and is convinced that he must kill V—though she has not provoked him in any way.
Those two cases differ from yet another Thomson case—that of the innocent passive threat (T):
Falling man. V is having a picnic on a beach. T, who is a very fat man, is pushed off the cliff above V by
some villain. T will crush V to death unless the latter pulverises him in mid-air, but will survive if he
lands on V.
Whereas the villainous truck-driver is morally responsible for his action, T is not, any more than is the
psychotic gunman. But whereas the latter deliberately (though not culpably) imposes a harm onto V, the threat
to which V is subject in Falling man is not the product of T's exercise of his agency: it comes about purely as a
result of the location and velocity of his body, over which, in this instance, he has absolutely no control. Finally,
the truck driver and the psychotic gunman have this in common that they both act in such a way as to subject
V to a threat of lethal harm, for which they lack a positive and objective moral justi cation.
One of the most important questions in the ethics of self-defence is whether one may kill innocent attackers
and passive threats in self-defence or other defence. According to Thomson, A (in both cases) and T violate V's
right not to be killed and thus forfeit their own right not to be killed, which is why V may kill them in self-
defence. Many have persuasively argued that it is not appropriate to say of the psychotic gunman, who is not in
any way morally responsible for his actions, that he violates A's rights—any more than it is appropriate to say of
a murderously hungry tiger which jumps at my throat for the kill that he violates my right not to be eaten alive.
p. 57 It seems even less appropriate to charge T with a rights violation.8 Nonetheless, I agree with Thomson's
 conclusion (though not with her argument for it) that V may kill all three in self-defence. For to claim, a
contrario, that the victim of a wrongful threat may not kill her morally innocent attacker or passive threat is to
say, by implication, that she is under a duty to allow herself to be killed even though (ex hypothesi) she has done
nothing to warrant the attack. However, as we saw in Chapter 1, agents generally have a personal prerogative to
confer greater weight on their own projects and goals than on other agents’ similar projects, which in turn sets
limits to the sacri ces which they can be expected to make for the sake of others. I submit that whilst the
prerogative does not permit agents to kill a bystander in their own defence (for the bystander himself bears no
causal responsibility for their predicament and thus cannot be used as a means only to their ends), it does
permit them to kill their attacker if the latter, whatever might be said about their lack of moral responsibility
for the situation of forced choice between lives, nevertheless subjects them to a wrongful lethal threat. In other
words, agents may not be expected to sacri ce their life for the sake of not bringing about the death of another
person when they have done nothing to warrant the threat which he poses to them.9

Downloaded from https://academic.oup.com/book/33082/chapter/281977852 by Library user on 09 May 2025

A standard objection to the personal prerogative justi cation for self-defence is that if it permits the killing of
attackers—particularly innocent attackers—then it also licences the deliberate killing of innocent bystanders
as a means to one's preservation. For if a victim is permitted to kill her attacker on the grounds that she may
give greater weight to the preservation of her own life than to his, then (it is argued) she may kill an innocent
p. 58 bystander intentionally, on those very same grounds. Either one accepts, wholly implausibly, that the
intentional killing of bystanders is permissible in one-to-one cases, in which case one can endorse partiality as
a justi cation for self-defence albeit at considerable moral costs; or one holds on to the much more plausible
prohibition on the killing of bystanders, in which case one has to look elsewhere for an account of V's right to
kill her attacker. In so far as we assume at the outset that one may not deliberately kill innocent bystanders in
self-defence, the objection, if sound, strikes at the heart of the partiality justi cation.10
The objection invites some quali cations to the partiality-based account, though do not fatally undermine it.
Having an interest in p, though necessary, is not a suf cient condition for an agent to have a liberty or a right in
respect of p. Accordingly, the partiality justi cation is not committed to regarding the fact that one's interest in
survival is under threat as a suf cient condition for being permitted, let alone having the right, to kill in self-
defence. In fact, it can allow that there are limits to the extent to which one may give priority to one's interests,
such as (in that instance) whether the target of one's self-defensive move causes one to be under threat.
Innocent bystanders, by de nition, are not, whereas attackers are. Partiality is also compatible with the entirely
plausible claim that a culpable attacker, such as our villainous truck-driver, is not permitted to kill his victim in
his own defence. Thus, the partiality argument for self-defence can and should stipulate that one is permitted
intentionally to kill in self-defence (on grounds of partiality) only if one is directly threatened by the target of
one's self-defensive actions.
Opponents of the partiality view will remain unconvinced. They will maintain that partiality cannot on its own
discriminate between cases where V has the right to kill and cases such as that of the bystander, where V must
desist. What justi es V's permission to kill, therefore, is not partiality, but whatever consideration we invoke to
distinguish A from B. Such considerations typically take the following form: by culpably attacking V, A has
forfeited his right not to be killed, and V is permitted to defend herself by killing him precisely because he no
longer has a right to be killed. The bystander has not attacked V and therefore has not forfeited his right not to
be killed by her. Consequently, V is not permitted (and lacks the right) to kill him. We thus have a criterion for
distinguishing A from B—a criterion which partiality is unable to provide.
p. 59 Now, I agree that partiality cannot, on its own, explain why V may kill A and not B—put differently, that the
judgement that V is directly threatened by A and not B is not one that is made on grounds of partiality.
However, that claim does not entail that partiality has no role to play in justifying self-defensive (and, as we
shall see, other-defensive) killings. In fact, unlike agent-neutral justi cations for self-defensive killings,
partiality makes sense of the intuition that V has a special reason for killing her attacker which others lack: he
is threatening her life, and no one else's, and accordingly she has a vested interest, which others (on the whole)
 lack, in thwarting his attack. That factual point about V's special location, as it were, in relation to her own
survival, does not merely capture the fact that V's has special prudential reasons to kill A: it has some
normative implications as well. In particular, her special stake in thwarting A's attack may sometimes give her
latitude which she would lack if she only had impartial, agent-neutral reasons for stopping him. Thus, she
would have greater latitude vis-à-vis innocent bystanders who might be killed as a result of the defensive steps
which she would take; she would also have greater latitude for deciding that A really does subject her to a lethal
threat.
I shall return to this issue below when discussing rescue killings. Meanwhile, the claim that V may kill A or T in
self-defence does not entail that she has the right to do so. For V to have that right means that the target of her
self-defensive move is under a duty not to defend himself. Now, given that the falling man exercises no agency
whatsoever, it would be unfair to withhold from him the permission to defend his own life from V. In this case,
although V may kill him, she lacks the right to do so. By contrast, if the villainous truck driver is unsuccessful

Downloaded from https://academic.oup.com/book/33082/chapter/281977852 by Library user on 09 May 2025

in his attempt on V's life at t1, and if she starts shooting at t2, he clearly may not defend his life from her at t3.
The reason why that is so is not merely that he triggers a forced choice between their lives: it is also that he
does so without (objective) justi cation whatsoever, which in turn supports the view that V not merely is
permitted but also has the right to kill him. More controversially, I also believe that a morally non-responsible
attacker such as the psychotic gunman loses his right not to be killed—and thus that V has the right to kill him
—precisely because he too acts without (objective) justi cation. For to claim that V in those cases lacks that
right is to claim that third parties are not under a duty to her either not to interfere with her self-defensive
steps by helping her attacker, or to help her by thwarting him. This strikes me as extraordinarily implausible. It
would mean, for example, that the police would be permitted (vis-à-vis her) either to prevent a severely abused
woman from defending herself from her psychotic husband, or to refuse to intervene as he is about to kill her
even if she would die as a result.
p. 60 I should add that whether victims have the right to kill morally innocent attackers is a matter of controversy.
Fortunately, not much in this book hinges on the soundness of that claim, since the overwhelming majority of
war cases do not involve psychotic attackers—with the admittedly problematic case of child soldiers, some (but
not all) of whom cannot in any way be regarded as morally responsible agents. The key point, rather, is that if
an agent lacks an objective justi cation for attacking V, the latter has the right to use lethal force in self-
defence.11 Sometimes, however, agents are justi ed in subjecting others to a threat of lethal harm which they
have not warranted—in other words, to what I shall call a wrongful threat. Whereas the (morally responsible)
attacker who acts without justi cation is standardly said to violate V's rights, the attacker who does have such a
justi cation is said to (justi ably) infringe it. Consider the following case, drawn (with slight modi cation) from
McMahan's recent book on the ethics of killing in war:12
Ambulance driver. An ambulance driver drives at speed to the hospital, though not recklessly so by
ambulance-driving standards, with a grievously injured patient who will die unless he receives
medical help within twenty minutes. On the way, some freakish event causes the driver to lose control
of the ambulance, which veers towards a pedestrian.
p. 61 Suppose that the only way the pedestrian can survive is by killing the driver. True the driver poses a wrongful
threat of harm to the pedestrian. But in so far as he is justi ed, indeed is under a duty, to drive at speed, to
claim that he violates the pedestrian's right not to be killed and thus that he may not defend himself from the
latter's self-defensive move seems grossly unfair indeed. In this case, both are permitted to defend themselves
but neither has the right (at least against the other) to do so.
To recapitulate, V is permitted to kill A in self-defence in so far as she is permitted more generally to attach
greater importance to her life than to A's, provided that the latter is causally responsible (or, as I shall
sometimes say, bears contributory responsibility) for the fact that she is subjected to a wrongful threat of lethal
harm. Describing morally innocent attackers as causally responsible for their victim's predicament might seem
 tendentious in so far as the use of the term ‘responsibility’, evoking as it does the notion of moral responsibility,
loads the philosophical dice, as it were, in favour of killing them. I myself do not hear or read ‘causal
responsibility’ in that way. Still, to avoid any and all possible ambiguity: by causal responsibility, I mean just
that—that the agent acts in such a way that another party is subject to a lethal harm, no less, and no more. And
it is that fact about him which turns him into a legitimate target, whilst it is another fact, about his victim,
which provides her with a special justi cation for killing him. To put the point differently, prior to his attack, A,
like any other similar agent, as a right in rem not to be killed—a right, in other words, which he holds against
everyone else. By attacking V, A loses his right that no one kills him. However, it does not follow from the claim
that A has lost his right that no one kill him, that he is a legitimate target for any individual against whom he
had that right. That claim only suggests that it is no longer the case that everyone is under a duty not to kill
him. Whether or not any of all those against whom he previously had that right is now at liberty to kill him is
precisely what needs to be shown. Partiality, as I have argued, offers a convincing reason as to why V may kill A:

Downloaded from https://academic.oup.com/book/33082/chapter/281977852 by Library user on 09 May 2025

if A has lost his right not to be killed against anyone, then surely he has lost it against his victim. V, thus, is
permitted to kill A in self-defence. Moreover, she has the right to do so if A acts without justi cation. In those
cases where V has the right to kill A, I shall say that A is liable to being killed by her. Note, moreover, that it is
not a necessary condition for V permissibly or rightfully to kill A in self-defence that the latter actually pose the
threat himself. If he contributes signi cantly to it, for example by instigating it, authorizing it, ordering
someone else to carry it out, or supplying the means by which it is carried out, and if V can save her life by
killing him, then she may, and sometimes has the right, to do so.
p. 62 So much, then, for self-defence. Let us now turn to the defence of others. A standard objection to victim-
centred accounts of the right to kill in self-defence is that they do not justify the right to kill in defence of
others. For although it is easy to see why a victim may give priority to her life over her attacker's, it is not so
easy to see why a rescuer may choose V's life over A's on victim-centred grounds, particularly if R is not V's
friend or relative. I disagree. Suppose that a potential rescuer, R, is on the scene. In the con ict which opposes V
and A, and on the assumption that A no longer has a right not to be killed, R's options are the following:
(1) Do nothing and remain neutral between A and V.
(2) Intervene on A's side and help him kill V.
(3) Intervene on V's side and help her kill A.
Clearly R may not opt for (2): for if V's interest in remaining alive is important enough to hold A under a duty
not to save his own life by killing her in self-defence, surely it is important enough to hold third parties under a
duty not to prevent her from picking the gun. Likewise, it would be wrong of them to thwart V by helping A
obtain the additional weapon he would need in order to overcome V's self-defensive action—let alone by killing
her (indeed, in most jurisdictions, they would be charged with the criminal offence of aiding and abetting in
the commission of a crime.13) Not only may R not side with A and against V: more strongly still, R has a right to
come to V's rescue and kill A, precisely because V has defensive rights at the bar of the prerogative. According to
the interest theory of rights, you recall (as per Section 1.2.2), to say that an agent has a right is to say that an
interest of his is important enough to hold some other person(s) under some duty. As applied to permissions, to
say that an agent is permitted to φ is to say that an interest of his is important enough to deem it (morally)
permissible for him to φ. Not only does the interest theory account for rights and liberties; it also accounts for
powers—the ability to change not merely one's own but also others’ moral or legal relationships by granting
them rights and liberties. In the case at hand, V's fundamental interest in surviving A's attack is not merely
protected by a right to kill A: it is also protected by a prima facie power to transfer that right to R. For if V's
interest in survival is important enough (as it surely is) to be protected by a permission and a right so to act,
then surely it is important enough on principle to be protected by a prima facie power to transfer that right to
p. 63 third parties. To claim otherwise is to impose an arbitrary restriction on V's ability to promote this
fundamental interest of hers. By analogy, to say, on the one hand, that V has a right to procure food for herself
 and thus that others are under a duty to let her do so, and on the other hand that she lacks the power to get
others to procure food on her behalf, is to fail to protect her interest in getting food.
None of this is to deny that R might also have agent-neutral reasons for killing A—as reside in the importance
of blocking wrongful attacks in general. Moreover, the claim that R might have agent-neutral reasons for
killing A does not imply that agent-relative justi cations for defensive killings are redundant to those reasons.
In making space for agents’ personal prerogative, one allows them to confer greater weight on their own goals
than on the goals of others, but one also recognizes that they have a standpoint vis-à-vis those goals that others
lack. The point is important in two respects. First, by that token, V may desist from killing A, because she may
confer lesser weight on her life than on his—which means that, in some cases at least, it is up to her and to no
one else to decide who will survive the attack. Put differently, and to bring to bear on this issue the distinction
between procedural and substantive justice which was drawn at the close of Section 1.2.2, it is up to her to
decide whether or not to exercise her right not to be killed. This will prove relevant in subsequent accounts of

Downloaded from https://academic.oup.com/book/33082/chapter/281977852 by Library user on 09 May 2025

humanitarian wars and civil wars. Humanitarian wars are fought on behalf and for the sake of individuals who
often do not take part in belligerents’ decision-making procedures; civil wars are often fought by actors who
have not been given an explicit mandate by their compatriots. In both cases, the issue of consent is crucial.
Second, V's special stake in A's demise may sometimes give her latitude which R lacks vis-à-vis innocent
bystanders who might be killed as a result of the defensive steps which either she, or R, will take. This is why I
claimed two paragraphs ago that V's interest in her own survival, if important enough to be protected by a
permission or a right, is on principle important enough to be protected by a prima facie (and not all things
considered) power. Suppose that V, in the course of ring at the villainous truck-driver, is at risk of
unintentionally though foreseeably killing a pedestrian who happens to pass by. She may well have a justi ed
personal prerogative so to act; but even if that is so, this is not enough to show that R himself may do so on her
behalf. To be sure, he may well have agent-neutral reasons to prevent the truck-driver from killing V. But those
reasons, stemming as they do from interests of his (and of everyone else's) in not living in a world in which
wrongful attackers are allowed a free rein, are not as powerful as agent-relative reasons (here, V's) which stem
p. 64 from one's interest in survival. Accordingly, before concluding that R may kill innocent bystanders (who
have not lost their right not to be killed) on V's behalf, we must thus establish that his (agent-neutral) interest
(which is lesser than an interest in survival) in blocking the villainous truck-driver is strong enough to justify
causing him to die. In this particular case it is not clear that R may so act—though he may (I submit) shoot at A
if in so doing he will merely in ict a slight injury to B's arm. In other cases, agent-neutral reasons would justify
R's act, as where (for example) the only way to rescue 500 hostages from two attackers is by storming the
theatre in which they are held and, in so doing, foreseeably but unintentionally kill a bystander who cannot be
evacuated in time. To put the point in a language familiar to just war theory, rescuers who do not have a special
relationship to the victim must confer greater weight than the latter on the interests of innocent bystanders
when assessing whether their defensive move is a proportionate response to the wrongdoing which they would
thereby block. The point will prove important for assessing the permissibility of in icting collateral damage in
wars of humanitarian intervention (Section 5.5.2) and scrutinizing the ethics of contracting private soldiers
into helping us ght our own wars of defence (Section 6.4).
Let us now turn to cases where A is a justi ed attacker (JA), and where V and A thus are permitted, though lack
the right, to kill each other in self-defence. In the face of a con ict of permissions between JA and V—between,
for example, the pedestrian and the ambulance driver—and in the light of the view that killing is generally
worse than letting die, it might be tempting to think that R not only may but in fact must remain neutral. But
suppose that the number of letting‐dies which would take place were R remain neutral far exceed the number
of killings which R would have to commit in order to prevent those deaths—to such a degree, in fact, that R is
permitted to intervene.14 That both JA and V have an equal permission to kill, as grounded in their interest in
survival, does not imply that they both have the power to transfer their permission to R. For there is an
p. 65 asymmetry between them, which resides in JA's having a positive moral justi cation for starting a chain of
events leading to a forced choice between their lives. Consider the following case, which takes us closer to lethal
con icts in war:
Bank robbery. A bank customer C is under a lethal threat from a gangster who is carrying out a bank
robbery and has already killed some other customers. C is armed, and pulls out his gun to defend
himself and the remaining customers. However, unless another customer, R, also armed, shoots at C,
the latter will inevitably (though unintentionally) kill someone else, V, in the process of killing the
gunman. Suppose that V would not die anyway at the hands of the gangster.
There is an asymmetry between C and V in this case, for although both are at risk of dying, C also has a positive
moral justi cation for subjecting V to a lethal threat, in that he is stopping a gangster from robbing a bank and
more importantly from killing more people. One might think that V is under a duty to let C kill her, precisely
because C has a justi cation for acting as he does. However, to hold that V is under a duty to make that sacri ce

Downloaded from https://academic.oup.com/book/33082/chapter/281977852 by Library user on 09 May 2025

would y in the face of the personal prerogative. It seems, thus, that V has not lost her right not to be killed and
thus may defend herself from C on grounds of partiality. On the assumption that R may step in, he ought to side
with C rather than V, for if he kills C he allows the continuation of a seriously wrongful process in which the
fundamental rights (not to be killed or seriously injured) of a number of individuals are violated. Agent-neutral
considerations in this case outweigh V's interest in conferring on R the permission to kill C. Put differently, it is
(sometimes) permissible deliberately to kill morally and causally innocent agents whose lives are under threat
at the hands of justi ed attackers if they turn themselves into (justi ed) attackers and, in so doing, threaten to
undermine the successful completion of the justi ed attackers’ rightful course of action.
I shall return to these dif cult issues in Section 2.4 when examining the issues of collateral damage and
patriotic partiality in wars of collective self-defence. In the next section I apply the foregoing account of
defensive rights to the right (deliberately) to kill enemy combatants in such wars.
 2.3 Wars of collective self-defence

2.3.1 The right to wage a war of collective self-defence: a first cut

We saw in Section 1.4 that political self-determination is a communal good which has worth to individuals in
so far as it helps ful l the demands of suf cientist cosmopolitanism and offers citizens the opportunity to
p. 66 exercise their legitimate personal prerogative to associate with one another. Given that it cannot be
exercised other than over a bounded territory, large-scale attacks on that territory itself undermine it and
thereby the rights which it serves. When they are backed by lethal force, as they typically are, those attacks
sometimes give rise to a just cause for war—and more speci cally provide the victims of aggression with a
justi cation for killing their aggressors in self-defence. Let us suppose that the government of country A
orders its army to invade country V in a bid for territorial expansion, with strict instructions to its combatants

Downloaded from https://academic.oup.com/book/33082/chapter/281977852 by Library user on 09 May 2025

to kill anyone from V's community who attempts to resist the invasion. Let us also take it as xed throughout
that A's invasion of V is wrongful. V's government sends its own army to repel the invasion by force. According
to the reductive strategy, V's right to wage that particular war against A is best understood as a right, held by
V's citizens, to resort to force in defence of their homeland—a right which they have the Hohfeldian power to
transfer to their army and its individual members. CombatantsV thus have a right to use force, on behalf of their
own citizens, against combatantsA. In the course of defending their individual and jointly held rights,
combatantsV are subject to lethal attacks from combatantsA. So are non-combatantsV, in fact, if combatantsA
do not restrict their military operations to the battle eld but also seek to destroy V's infrastructure and, in so
doing, in ict lethal collateral damage on them. The right of combatantsV to wage war, thus, is not simply a right
to defend their compatriots’ and their (jointly held) rights to, for example, territorial integrity and political
sovereignty: it is a right to defend their and others’ life from lethal threats which combatantsA pose to them in
the course of prosecuting their war. Moreover, in some cases, violations of those rights by way of a military
aggression lead to further violations of the rights not to be killed or maimed of V's members, not merely in the
sense that the latter's refusal to surrender causes combatantsA to attack them, but in the sense that A's regime,
were it to win the war, might carry out genocidal policies against V's population, engage in the systematic
torture of opponents, and so on. Similarly, were A to win, its war of aggression might lead to systematic and
severe poverty within V's borders as a result of A's grossly incompetent mismanagement of V's resources; it
might also lead to the establishment of a dictatorship. In other words, violations of the rights to territorial
integrity and political sovereignty warrant resistance by lethal force not just because they are themselves
backed by lethal force, but also in so far as a defeat would result in violations of the requirements of justice
within V.15 By analogy, we have the right to use lethal force against an intruder who threatens to kill us
p. 67 unless we let him stay in our house, as well as against an intruder who, were we to lay down our arms as he
crosses the threshold, would still kill us, maim us, cause us to live a less than minimally decent life, and so on. It
is in that sense that the right to wage war is simply the right to use lethal force against those whose deadly
threat is one which they pose (as combatants), or contribute to posing (as those combatants’ leaders) in
furtherance of, or as a prelude to, the violation of our fundamental rights.
At this juncture, the following objection might be raised. Imagine that on 1 July 1914 two Frenchmen, André and
Bernard, are taking a stroll in the French countryside and are unjusti ably attacked by a German individual,
Carl. On my account of defensive rights, André and Bernard each have the right to kill Carl in self-defence as
well as the Hohfeldian power to transfer that right to potential rescuers and thus to each other: put differently,
André has the right to kill Carl in defence of Bernard, and vice versa. As should be clear, the reason why that is so
has nothing to do with the fact that André and Bernard are compatriots and everything to do with the fact that
Carl is unjusti ably subjecting them to a wrongful threat of lethal harm. And indeed, to claim, a contrario, that
André and Bernard have the right to kill Carl in defence of each other just because they are compatriots (whilst
 Carl is a German) would violate the central tenets of cosmopolitan justice: as we saw in Sections 1.2.1 and 1.3.1,
the principle of fundamental equality forbids agents deliberately to harm one another on the basis of morally
arbitrary factors, of which political membership per se is one. So far, so good. But now, fast-forward to 6
September 1914: France and Germany are at war, and André and Bernard are conscripted into the French army
and ght alongside each other at the Battle of the Marne.16 By happenstance, Carl, who survived their defensive
move two months earlier by eeing, is ghting them as part of the German army. In this case, it does seem as if
André has the right to kill Carl in defence of Bernard and vice versa precisely because André and Bernard are
compatriots whose rights (both individual and jointly held) are under threat at the hands of Carl and other
German soldiers. And if that is the case, then (the objection might continue) my account of André’s and
Bernard's right to kill Carl in defence of each other on 1 July 1914 does not support their right to kill him in
defence of each other on 6 September 1914.
The dif culty is more apparent than real. For suppose that on 1 July 1914 Carl throws an incendiary bomb on a

Downloaded from https://academic.oup.com/book/33082/chapter/281977852 by Library user on 09 May 2025

p. 68
house which André and Bernard jointly own and currently occupy—with a view to forcing them to relinquish
their rights to him. André and Bernard have the right to kill him in defence of their own lives, in defence of
each other's life, and in defence of their jointly held right of ownership over the house. In this example the fact
that they are compatriots is irrelevant, because the right at issue is not one which they have in virtue of that
particular relationship. By contrast, the jointly held rights which are under threat in a war of collective self-
defence are rights which compatriots together have with respect to the territorial integrity and political
sovereignty of the national or political community to which they belong. Notwithstanding this difference
between this case and the scenario involving joint rights over a house, the point just made holds. When
violating those rights by using lethal force, Carl (and his fellow German soldiers) provide André and Bernard
with a justi cation for killing him in defence of their own life as well as in defence of each other's life; he also
provides them with a justi cation for killing him in defence of their rights to territorial integrity and collective
self-determination—which rights they happen to hold jointly as members of the same political community.
Two points of clari cation are required before we proceed further. First, it is worth stressing, once more, that I
am not endorsing the claim that individuals may kill in defence of their compatriots qua compatriots. Second,
my point is not that we may give priority to those with whom we have a special relationship in general only if
that special relationship generates jointly held rights, and if, therefore, one's own interest is at stake. (For if that
were so, then we would not be permitted either to give priority to our children over other children, or to do for
our children certain things which we would be permitted to do for ourselves—which in some cases at least is
absurd.) My focus here is on the relationship of co-citizenship. When that relationship is under threat it is
because the jointly held rights which are constitutive of it are also under threat, and it is that, in turn, which
sometimes confers on actors in that relationship the right to exercise lethal force in its defence.
Note that the argument presupposes that agents may kill wrongdoers in defence of jointly held rights, which
are of lesser importance than the right not to be killed and whose violations themselves need not lead to the
loss of that particular right. On some views this undermines the reductive strategy, for in private cases (they
will press) one may kill only if one is subject to a lethal threat.17 However, in some jurisdictions the law does
p. 69 permit individuals to kill in defence of lesser rights than the right to life—such as, for example, the rights
not to be raped, kidnapped, and enslaved. I believe that the law has got it right. Killing someone without
justi cation is an egregious wrongdoing because it treats the victim as less than a person. In killing V (non-
accidentally) the attacker shows, in effect, that she counts for nothing, that she is not in any way worthy of his
concern and respect, and thus that she can appropriately be treated as a means only to his ends, and not as an
end in herself. So do rapists, kidnappers, and slave-owners: for rape, kidnapping, and enslavement also treat
their victims as less than persons. If V may kill her lethal attacker on the grounds (in part) that she cannot be
held under a moral duty to let him treat her as nothing better than an object, then she may kill him when he
treats her in dehumanizing ways which come close to in icting on her the loss of her life.
 More controversially, contrary to what the law stipulates in most jurisdiction, killing is not a disproportionate
response to the threatened loss of one's property—at least in those cases where threats to one's property rights
are backed by force or severely jeopardise one's prospects for a minimally decent life. Suppose that your house
is wrongfully invaded by a group of individuals who intend to stay there permanently and who use coercive
force against you if you dissent with whatever decision they make with respect to your house. Or suppose that
you are (non-lethally) coerced by an attacker into signing over all your possessions, with no possibility for you
to get compensation or help from anyone ex post. And suppose, nally, that in all those cases you have a choice
between fully complying with wrongdoers’ demands or killing them. I cannot think of any plausible
justi cation for the view that you lack the right to kill your aggressors. To reiterate, there are limits to the
sacri ces which we may be expected to incur for the sake of others. In particular, we are not under a duty to
lead a less than minimally decent life for their sake. Whilst this does not permit us deliberately to kill innocent
bystanders for the sake of living such a life, it does permit us (subject to considerations of proportionality,

Downloaded from https://academic.oup.com/book/33082/chapter/281977852 by Library user on 09 May 2025

chance of success, and so on) deliberately to kill those who wrongfully placed us in a position where we have to
choose between agreeing to live such a life, or being coerced into living it if we refuse to accede to their
(wrongful) demands.18
Clearly, this is what a war of aggression is like. When country A's army invades country V, and when A's
p. 70 government insists that V comply with its demands as a condition for withdrawing, A is typically subjecting
V to a multi-pronged attack which is very similar to that described in the previous paragraph: it is, typically, an
attack on the rights of V's members to determine their own future and decide who will live on, and dispose of,
their territory. Moreover, it is an attack that is more often than not accompanied by threats to some of the most
fundamental rights of V's individual members, such as the right not to be killed, raped, beaten up, maimed, and
so on.
Here is another way to put the point. A just cause for war is standardly seen as a wrongdoing of such magnitude
as to warrant a retaliatory lethal attack by its victims, understood collectively as a group of individuals, on the
wrongdoer, also understood in those terms. Yet we ought to distinguish between a narrow just cause and a wide
just cause. A narrow just cause is a justi cation for killing wrongdoers, whilst a wide just cause is a justi cation
for using non-lethal force even though one plausibly foresees that wrongdoers will counter this defensive move
by lethal means—the use of which is not warranted and provides victims with a narrow just cause for killing.
When we speak of a just cause for war, therefore, we might speak either of a narrow just cause or of a wide just
cause. Now, not all rights are so important that their violation on its own provides their holders with a narrow
just cause for subjecting wrongdoers to a lethal attack. Suppose that A's forces bloodlessly invade a tiny empty
island over which V claims sovereignty, but which is located many thousands miles away from V's mainland.
The invasion alone does not provide V with a narrow just cause for sending its soldiers to kill combatantsA.
This is because the violation of this particular right of V's does not in this case undermine the prospects of V's
members for a minimally decent life. However, A's act of aggression does provide V with a wide just cause for
using non-lethal force against combatants A. If combatantsA then pose a lethal threat to combatantsV, the latter
have a narrow just cause for killing or seriously maiming them. By contrast, A's large-scale traditionally
military invasion of V's mainland would, on its own, constitute a narrow just cause for combatantsV to kill
combatantsA, to the extent that those combatants could take it as read from that particular act that the invaders
will kill them unless they surrender, or that they have very strong reasons to believe that A's regime, if
unopposed and thus victorious, would commit further violations of their fundamental human rights.
An important question at this juncture is that of the grounds upon which—if any at all—illegitimate states can
have the right to wage a war of collective self-defence. As we saw in Section 1.4, a state is legitimate—that is, its
of cials collectively have a moral right to govern—provided, and to the extent that, it respects the fundamental
rights of both citizens and outsiders; moreover, the right to govern is a duciary right which the state's agents
p. 71 hold as trustees acting on behalf of citizens, who are under an obligation to obey those agents’ directives
 only if their fundamental rights are thereby better protected by those directives than if they disobeyed. By that
token, the right to wage war as held by the state's agents is one which they have as trustees acting on behalf of
those individuals. If the latter do better in the face of severe rights violations, such as territorial aggression by
another state, by obeying the agents’ directives during the war than they would do without it, those agents
collectively have the right to wage war on their behalf, even though the fact that they are themselves guilty of
severe rights violations provides citizens with a justi cation for overthrowing them once the war is over. The
example of the USSR during the Second World War illustrates the point. That the Soviet regime was illegitimate
is absolutely beyond question. But it does not allow us to conclude that Germany did not violate the rights to
sovereignty, territorial integrity, and life, of Soviet citizens, and that Stalin and his henchmen therefore did not
have the right (acting on behalf of those citizens) to take the steps necessary to repel the invasion. The latter
clearly did have that right, but, to repeat, only as the agents of Soviet citizens, who had morally-directed
pragmatic reasons to defer to their orders. Had Soviet citizens decided to overthrow Stalin and his regime in

Downloaded from https://academic.oup.com/book/33082/chapter/281977852 by Library user on 09 May 2025

the aftermath of the war, they would not have wronged them in any way at all.19

2.3.2 The moral status of combatants

Taken together, my account of defensive rights and the account of the right to wage and kill in wars of
collective self-defence which it underpins imply that whether combatants may—indeed, have the right to—kill
enemy combatants (in large part) depends on the moral status of their war ad bellum in general and whether it
has a just cause in particular. In private killings, V has the right to kill A in self-defence on the grounds that the
latter is subjecting her to a wrongful threat, but (as we saw) A clearly may not retaliate in his own defence. If A's
lethal threat is unjusti ed at time t1, such that V may take defensive steps against him at t2, then surely he may
p. 72 not continue to threaten V at t3, even if the latter, at that point, is threatening his life. To claim otherwise
would be to say, wildly implausibly, that an armed bank-robber may kill the cashier who, furnished with good
reasons to believe that the robber would start shooting, pulls out his gun and mounts a self-defensive and
other-defensive attack. By parity of reasoning, if an act of aggression as carried out by A's army against the
territorial integrity and sovereignty of community V is morally wrong, then its constitutive elements are
morally wrong as well, and combatantsA are therefore committing a wrongdoing. If it is morally wrong for A's
army to cross over the border into V's territory, it is morally wrong for combatantsA to do precisely that.
Crucially, if it is morally wrong for combatantsA to threaten, at t1, the lives of those of combatantsV who stand in
their way, then it is wrong for those individual combatants to kill combatantsV in self-defence at t3. On that
view, German soldiers who killed Belgian soldiers in pursuit of their army's wrongful invasion in the summer
of 1914 were not permitted so to act, even when they were attempting to defend their own lives from the
Belgians’ own self-defensive steps.
This view, variants of which have been revived recently by, inter alia, Tony Coady, David Rodin, and Jeff
McMahan, is not altogether new: thus, writing in the sixteenth century, Francisco de Vitoria avers that
combatants ghting on the objectively unjust side are acting impermissibly.20 But to many, this is unacceptable.
As Walzer puts it, once the war has started, combatants lose their right not to be killed by the enemy simply in
virtue of posing a lethal threat to them, and thus are legitimate targets (subjects to considerations of
proportionality and necessity), irrespective of the justness, or lack thereof, of the side for which they are
ghting. The point is not merely a point of law—for the laws of war do treat combatants on a par—but also a
claim of morality. The modern just-war tradition and the body of laws which it underpins do not draw any
distinction between German and Belgian combatants in 1914.21
p. 73 The dispute between proponents of that older view (which under their penmanship is known as the ‘neo-
classical account’) and proponents of the orthodox account of war killings as articulated by Walzer is at the
frontier of contemporary war ethics, and exploring its participants’ various moves and counter-moves in depth
is beyond the scope of this chapter. Instead, I shall brie y tackle three objections which are standardly levelled
 against the neo-classical account: the duress objection, the epistemic objection, and the lack-of-control
objection. Strictly speaking, those objections target an interpretation of that account which I do not endorse—
to wit, that one may, indeed has the right, to kill an attacker only if the latter is morally responsible for a
wrongful lethal harm. However, it is worth discussing them here, just in case moral responsibility is necessary
for liability—contrary to what I argued in Section 2.2.
Brie y put, the objections claim, respectively, that combatants act under duress, that they cannot be expected
to know whether they are ghting for a just cause, and that they lack control over their country's foreign
policy; they then insist that as a result, combatants are not to be held morally responsible for the killings which
they commit during that war, and conclude that they have therefore not lost their right not to be killed, which
implies that they may kill enemy combatants in self-defence even if their cause is unjust.
I do not think that either duress, or epistemic dif culties, or lack of control, furnish combatants who take part

Downloaded from https://academic.oup.com/book/33082/chapter/281977852 by Library user on 09 May 2025

in an unjust aggression on another country's territory and sovereignty with the kind of excuses which would
enable them permissibly to kill enemy combatants in self-defence.22 Consider duress. That it is felt by many
combatants is beyond question—though far less so in those countries which no longer resort to conscription. It
is worth noting, however, that private actors who kill in the course of, for example, drug-related gang warfare,
are also subjected to considerable duress. Yet one would not say that a gang hit-man may kill his victim, should
he fail at rst try and should she be armed, in his own defence: duress, in those cases, is not fully exculpatory,
and so it is hard to see how, exactly, war differs (with respect to duress) from private killings.
The epistemic objection for its part need not claim that soldiers are acting under duress. Rather, it asserts that
they act under conditions of severe epistemic uncertainty. For in most cases, the objection presses, determining
p. 74 whether the war is just ad bellum is so complex as to be beyond the cognitive reach of ordinary soldiers.
Consequently, even if soldiers are objectively in the wrong when killing in the prosecution of an unjust war,
they have very strong subjective justi cations for so doing—such as to warrant regarding them as morally on a
par with their enemy on the battle eld. By contrast, private actors can be expected to know whether or not they
are killing rightfully.23
However, the objection assumes—somewhat disrespectfully and without much empirical basis—that
combatants are incapable of determining whether the war is just. Relatedly, a decision to join the army and
thereby go to war can be, and in many cases is, made in the light of information about the facts of the case.
Finally, proponents of the objection are more than willing to hold soldiers morally responsible for the acts of
killing which they commit in violation of jus in bello, such as the deliberate killing of non-combatants, even
though the conditions under which they have to ght (‘the fog of war’) might make it extremely dif cult for
them to judge whether or not the orders they are given are just. If epistemic uncertainty in the fog of war does
not constitute a justi cation for a wrongful killing, it is not clear why it would constitute a justi cation for
taking part in an unjust war when the decision so to act can be made in the safety of the barracks.24 I do not
mean to deny that there are good reasons in practice for treating combatants as if they were morally on a par,
which would mean, in particular, not prosecuting for murder those who ght an unjust war. Such reasons
include, for example, the fact that combatants who fear prosecution may be less likely to obey just orders, or
more likely to ght without restraint if they sense that they will, in any event, be punished. But that is not the
same as to say that combatants are morally on a par, and in so far as my concern, in this book, is with the moral
p. 75 permissions and moral rights which agents do have in war rather than morally directed laws of war, the
epistemic objection leaves the moral responsibility-variant account of the right to kill untouched.
Finally, the lack-of-control objection insists that most combatants have no in uence on the overall course of
action taken by their army or, for that matter, their regime; as a result, to withhold from them the permission
to defend their lives on the grounds that they are participants in their army's collective wrongdoing is to
subsume them under their collective identity. The objection is worth scrutinizing, because in so far as it is a
central premise of cosmopolitanism that agents are not reducible to their membership in a group, the
 objection, if correct, is seriously damaging to any cosmopolitan theory of the just war. Again, however, gang
members too often have little control over the decisions made by gang leaders. If lack of control in such cases is
not thought to provide gang members with a subjective justi cation for killing wrongfully, then it is not clear
why it should furnish combatants with a subjective justi cation for taking part, and killing, in an unjust war. In
any event, although it is true that rank-and- le soldiers have close to little in uence on their political leaders’
decision wrongfully to invade a neighbouring country, acting as such they do have control (for the most) over
their own decision to stay in the army, to board the aircraft which will take them to the border on the eve of the
invasion, to drive their tanks over the border, and, most importantly, to re at enemy soldiers who refuse to let
them pass through. They cannot invoke lack of control as an excuse, let alone a justi catory excuse, for
committing those lesser wrongdoings, and particularly for backing those wrongdoings with the threat of
(lethal) force. Moreover, drawing on the account of individual responsibility for collective wrongdoings which
was set out in Section 1.2.3, to the extent that they have a choice as to whether or not to participate in this

Downloaded from https://academic.oup.com/book/33082/chapter/281977852 by Library user on 09 May 2025

particular wrongful collective venture (their army's aggression against V), and to the extent that they intend to
do that which they know is part of that collective venture, they can be held responsible for the wrongdoing
which consists, not just in setting foot into V's territory without V's consent, but, in doing so collectively, in
invading V. The claim that soldiers are not morally on a par, thus described and defended, is entirely of a piece
with the individualistic principles of cosmopolitan justice.
To recapitulate, I have argued that even if, pace my argument in Section 2.2, attackers are liable to being killed
only if they are morally responsible for the wrongful threat which they pose to their victims, whether or not
soldiers have the right to kill enemy soldiers largely depends on the moral status of their war. That said, the
foregoing discussion raises the further and important issue of the liabilities of agents other than combatantsA
p. 76 —such as A's non-combatting of cials who play a signi cant role in the decision to wage the war of
aggression and in its subsequent prosecution, or A's citizens who elected those of cials in the rst place. It
might seem tempting to suppose that only combatants are legitimate targets, on the grounds that they alone
pose a threat to combatantsV. On that view, however, A's leaders, and perhaps some of its citizens, are not liable
—which, in some cases at least, seems counter-intuitive. To give but one example, it is not plausible on the one
hand to claim that generals who give unjust orders to their soldiers are liable to being killed whilst those
generals’ civilian prime minister or head of state, who took the decision to wage the war in the rst instance, is
not liable. It thus seems more plausible to say that causal responsibility for a wrongful threat posed by
unjusti able action is the basis for liability. The dif culty here is that more often than not, individual
wrongdoers’ contributions to the collective harm which they in ict on V's members are very marginal and not
signi cant enough on their own to warrant the judgement that they have lost their right not to be killed. To
give but one example, we might be reluctant to claim that citizens who elect a warmongering leader are
legitimate targets, on the grounds that the act of casting a vote, when taken on its own, cannot possibly justify
killing the voter.
The reason why casting a vote is not a basis for liability in this particular case is that the voter makes an
exceedingly small contribution to the resultant war. Causal signi cance often greatly matters: the secretaries
who were sharpening the pencils used by the German General Staff during World War II were not liable; the
generals themselves clearly were, and not merely in virtue of being soldiers. However, one should not infer that
making a signi cant contribution to a wrongful threat of lethal harm is a necessary condition for liability. For
there are cases where an agent is liable to being killed even when his contribution to such a harm is very small.
Suppose that 20,000 individuals are sending 10V of electricity into the body of an innocent victim, as a result of
which she dies. Or suppose that a mob of thousands of people are each throwing a small stone at a woman on
the grounds that she committed adultery, as a result of which she too dies. Imagine that a rescuer could save
the victim's life, in both cases, by killing one, and simply one, wrongdoer (for example, because the others
would ee, fearing for their lives.) I believe that each and every one of those torturers or stone-casters has lost
their right not to be killed in this case—in other words, is liable to being killed—even though their
 contribution to V's fate, each taken on their own, is very small. An important difference between those two
cases on the one hand and the aforementioned voting example on the other hand is that torturers and stone-
p. 77 casters would be liable to less-than-lethal force if they acted alone, since they would be in icting a wrongful
(though lesser) harm onto their victim; voters, by contrast, would not be liable for a single voter cannot on its
own in ict a wrongful harm to third parties (at least in cases where outcomes are not settled by margins of one
vote only). I believe that this difference matters to the determination of agents’ liability to being killed when
their individual contributions to collective wrongdoings are marginal. If an agent's contribution is such that he
has lost his right not to have force used against him, then he is liable to being killed for his participation
(subject to other conditions such as necessity.) On the other hand, if his individual contribution is such that one
would violate his right by resorting to force as a means to block the harm to which the individual contributes,
then that agent is not liable to being killed for his participation in a collective (wrongful) lethal harm.25 That
said, the fact that agents who are not liable to being killed in the sense just de ned but nevertheless

Downloaded from https://academic.oup.com/book/33082/chapter/281977852 by Library user on 09 May 2025

unjusti ably contribute to the imposition of a wrongdoing provides their victims, or agents acting on her
behalf, with a justi cation for deliberately imposing a lesser harm (than the loss of their life) on them
collectively. It may also provide victims and their rescuers with a justi cation for acting in such a way as to kill
them unintentionally, though foreseeably. In many cases, of course, we will not be in a position to ascertain
whether civilians who belong to the enemy do in fact contribute to the latter's wrongdoing—notably in those
cases where the wrongdoing occurs as a result of collective actions or omissions involving thousands of
individual agents. Under conditions of epistemic uncertainty of that kind, we ought to err on the side of caution
and proceed on the assumption that those agents are innocent. The rationale for adopting this precautionary
principle lies in the familiar two-pronged thought that the deliberate imposition of harm, rather than
immunity from it, stands in need of justi cation, and that agents generally ought to be regarded as (causally)
p. 78 innocent of wrongdoing until proven (causally) responsible for it. I shall revisit those questions in Section
3.4 in the course of my normative account of subsistence wars.26 Meanwhile, let me present the conclusion
reached so far in the following way. As we saw in the Introduction, the principle of discrimination is central to
both the morality and the laws of war. In its orthodox mainstream interpretations, it holds that combatants are
legitimate targets, irrespective of their cause, just because they pose a threat to one another; by contrast, non-
combatants are not legitimate targets, precisely because they do not take part in hostilities. On my account, by
contrast, not all combatants are legitimate targets, whereas some non-combatants are.
An important quali cation is in order. Although my account of the right to kill in a war of collective self-
defence implies that the ad bellum justness, or lack thereof, of the war is a decisive consideration for evaluating
combatants’ in bello rights and liabilities vis-à-vis one another, it does not entail that combatants who take part
in an unjust war may never kill enemy combatants.27 Thus, even if combatantsV are at liberty to kill
combatantsA, they are not permitted to use whatever means they wish. For example, they may not use weapons
which are designed not merely to kill but to kill atrociously painfully. If they do, combatantsA are permitted to
retaliate against combatantsV’ self-defensive steps. Moreover, to the extent that combatants who are unjust (for
lacking a just cause) at time t1 may come to acquire a just cause for continuing with the war at time t2, they may
and should, in those phases, be regarded as just combatants.
Nor does the account entail that combatants who take part in a just war always have the right to kill
combatants from the ad bellum unjust side. Suppose that country A wrongfully invades V. V, I assume, has a just
cause for waging a war of self-defence against A but goes into war for additional unjust reasons—such as the
seizure of A's diamond mines, over which it has no legitimate claim. Some of combatantsV will be deployed to
repel the invasion and in so doing will act justly, whilst others will be sent deep into A's territory to take control
p. 79 of the mines and in so doing will act unjustly.28 By implication, those of combatantsA who oppose V's self-
defensive steps act unjustly; but those who kill V's soldiers in protection of the diamond eld act justly. Or
suppose that although A went to war against V at t1 without a just cause, it subsequently acquired a just cause
 for continuing with the war at t5. Correspondingly, combatantsA who, at t1, posed a wrongful threat to their
enemy, are acting permissibly at t5. And so on. A full account of combatants’ moral status vis-à-vis one another
—which it is not my aim to provide here—would offer more ne-grained principles for discerning, on the basis
of the justness of the war ad bellum, when combatants are liable to being killed and are permitted to kill, and
when they are not. I shall return to this dif cult issue in Section 7.5, where I shall examine the moral status of
ad bellum just combatants who use unjust tactics on the grounds that they cannot win their war in any other
way. In the remainder of this chapter, however, I assume that combatantsA are violating the rights of V's
members, be they combatantsV or citizensV.
To recapitulate, combatantsV have the right to kill combatantsA in one another's defence and in defence of their
compatriots in so far as combatantsA violate their jointly held rights to territorial integrity and political
sovereignty via the use of lethal force. The point might seem entirely straightforward once one rejects the

Downloaded from https://academic.oup.com/book/33082/chapter/281977852 by Library user on 09 May 2025

thesis that combatants are not on a par once the war has started. But one of its very controversial implications
is worth drawing out. Ex hypothesi, combatantsA are unjust attackers. Suppose that they are sent on a mission
to kill some combatantsV. Whilst it might be tempting to describe war killings in collective terms—‘platoon A
killed soldiers of platoon V’—they are, in fact, carried out by individuals against other individuals. In the lethal
con ict which pits those two sets of individual enemy combatants, individual combatantsA will often kill, not
so much, or not merely, in their own defence, but also in defence of their comrades. However, in so far an
individual combatantA lacks the right to kill an individual combatantV in his own defence, he lacks the power to
authorize his comrades to kill the latter—and thus his comrades may not kill combatantV in his defence. More
strongly still, if combatantsA who are not themselves directly threatened by combatantsV decide to intervene in
the set of lethal interaction which pits their comrades with combatantsV, then they must intervene on the side
of combatantsV (or, indeed, of non-combatantsV who are unjusti ably threatened by their comrades). In other
words, they must kill fellow members of A rather than members of V. Moreover, if there is such a thing as a duty
to kill unjusti ed attackers in defence of their victims (and I shall argue in Section 5.3 that there is), then
combatantsA not only are permitted, but are under a duty, to kill other combatantsA in defence of combatantsV,
non-combatantsV, or any other innocent third party who might be embroiled in that con ict.
p. 80 That conclusion will seem unacceptable to many: for how could it be that combatantsA may, let alone must,
turn against their own side? And yet I believe that there is no place for patriotic partiality under those
circumstances. The point is entirely uncontroversial in private cases. Suppose that, as André and Bernard are
taking a stroll in the countryside, Bernard (this time) unjusti ably attacks Carl, who attempts to defend
himself. One does not need to be a cosmopolitan to af rm that the fact that André and Bernard are both French
whereas Carl is German does not provide André with a justi cation for killing Carl in defence of Bernard: one
just may not come to the help of an unjust attacker by killing his victim. The same applies to the case of war. As
we saw in Section 1.3, individuals can act in such way as to deliberately confer greater weight on the interests of
fellow members in two cases: once the demands of suf ciency have been met, and if not giving greater weight
to patriotic partiality (than to the demands of non-community-members) would threaten the very survival of
the community as a politically sovereign body. But although cosmopolitan justice does allow individuals to
confer greater weight on their political community's projects and goals in such cases, it forbids them to do so in
furtherance of grievous wrongdoings such as unjusti able and wrongful killings. Accordingly, for combatantsA
to kill combatantsV in defence of other combatantsA simply on the grounds of common membership in A
would constitute such participation, and is therefore wrongful.
I suspect that at this juncture some would be tempted to insist that war sharply differs from the private case I
have just described, in that soldiers who belong to the same army—an institution whose task it is to defend
 communal interest, and which trains its members to perform that task by creating and fostering very strong
ties of loyalty between them—are permitted, or at the very least have the right, to defend one another at the
expense of the enemy, even if the war is unjust ad bellum. More strongly still, some might object, at the very
least they must not side with the enemy.
On that view, what does the work is a combination of patriotic partiality and the contractual and associational
duties and rights which soldiers are putatively said to have towards one another. Again, however, it is not clear
at all whether killing enemy combatants, who ex hypothesi are just combatants, falls within the remit of
legitimate patriotic-cum-role-based partiality. Partiality so construed would not, I think, forbid combatants A
from killing fellow combatantsA if the latter are unjusti ably subjecting innocent non-combatants to a
wrongful lethal attack. Take the case of the My Lai massacre. On 16 March 1968, American Warrant Of ce Hugh
Thompson—a helicopter pilot during the Vietnam War— ew over My Lai and noticed injured and dead bodies

Downloaded from https://academic.oup.com/book/33082/chapter/281977852 by Library user on 09 May 2025

p. 81 of Vietnamese civilians. Upon realizing that the latter were being randomly killed by American soldiers (led
by Lt Calley), he landed his helicopter between victims and attackers, and as he and his crew started lifting
some of the injured villagers to safety, (allegedly) instructed his men to re at the Americans if the latter kept
shooting at the Vietnamese.29 At the bar of cosmopolitan justice there is no doubt that Thompson acted
justi ably—that the fact that he shared US citizenship with Calley and his men could not possibly constitute a
consideration in favour of letting them commit such a grievously wrongful killing, let alone in favour of
helping them in that task by taking part in the killing themselves. Why, then, should shared citizenship
(combined with shared membership in the army) be a consideration in favour of helping one's fellow soldiers
kill morally justi ed enemy combatants?
Because, some will undoubtedly argue, combatants are armed, whereas the Vietnamese villagers were not: it is
one thing to kill (or threaten to kill) combatants of one's own army when the latter are wrongfully killing the
unarmed; it is another to do so when they are themselves under threat. But clearly this putative reply will not
do. For even though combatantsA might have a greater excuse for killing combatantsV than for killing unarmed
non-combatants, ex hypothesi, combatantsA are no more permitted, indeed no more have the right,
(deliberately) to kill combatantsV in self-defence than they are permitted, let alone have the right, (deliberately)
to kill innocent civilians at rst strike. And so if patriotic partiality cannot forbid combatantsA to kill their
comrades in defence of those civilians, it cannot forbid them to kill them in defence of combatantsV. If
cosmopolitan justice permits, indeed mandates, combatantsA to kill fellow combatantsA in defence of unarmed
innocent civilians, then it also permits, indeed mandates, them to do so in defence of armed justi ed
combatantsV. Pending further considerations to the contrary, in other words, prima facie they are under a duty
to defect to the ad bellum just side, however controversial the point might be.
p. 82
2.4 Patriotic partiality and collateral damage

Let us take stock. I noted at the outset that a normative account of wars of collective self-defence must speak to
war's constitutive acts of killing. The right to wage and ght in such a war, in other words, is inter alia a right to
kill. The reductive strategy, which regards the right to kill in wars as a particular example of defensive rights in
general, sought to provide such an account against the background of a cosmopolitan theory of justice which
on the one hand proscribes agents from in icting harm upon others solely on the basis of political
membership, and on the other hand permits them to kill aggressors in defence of their and their compatriots’
jointly held rights to territorial integrity and collective-self-determination. Combatants do not merely kill
other combatants, however: they also (usually) kill non-combatants, whose plight is central, as a matter of fact,
to both the law and the morality of war. In Section 2.3.2 I noted that participants in an unjust war of aggression
contribute to the wrongdoings in icted on the aggressed parties to different degrees and in different ways, and

Downloaded from https://academic.oup.com/book/33082/chapter/281977852 by Library user on 09 May 2025

that only those agents whose contribution would render them liable to the use of force simpliciter are liable to
being killed. Others, whose contribution does not meet that condition, might lose their right not to be subject
to lesser forms of harm or, indeed, might be liable to having their interest in survival somewhat discounted
when the aggressed party's military decide which tactical objectives to pursue and thus which civilians will be
killed foreseeably though not unintentionally. As we also saw earlier, however, an agent may sometimes have a
justi cation for subjecting innocent third parties to a wrongful threat of lethal harm. Often, the agent foresees
the harm but does not intend to cause it. In such cases, I argued, even if the agent may kill those individuals,
they in turn may kill him in self-defence—precisely because they have not acted in such a way as to warrant
the threat which he poses to them. Consider, paradigmatically, tactical bombersV who target A's munitions
factories in the knowledge (though not with the intention) that the bombs will kill scores of non-combatantsA
who live in the vicinity. It is a familiar claim, in just war theory, that the justness of their cause furnishes
tactical bombersV with a justi cation for posing a wrongful threat to non-combatantsA. Less familiar, but
nevertheless plausible, is the view that although they may kill those non-combatants, the latter in turn may
retaliate in their own defence—precisely because they have not acted in such a way as to warrant the threat
which tactical bombersV pose to them.
p. 83 But now some dif culty arises. For cosmopolitan justice, political membership is not, per se, an appropriate
basis for the conferral or denial of fundamental human rights. By that token, it is not an appropriate basis for
the justi ed infringement of such rights. In the case at hand, however, combatantsV bomb A's munitions
factories and foreseeably though unintentionally kill some non-combatantsA in order to weaken A's military
effort as a means better to defend their and their compatriots’ individual and jointly held rights. Whether one
can justify this course of action on cosmopolitan grounds remains to be seen. For it is one thing to confer
greater priority on one's interests and on the interests of one's compatriots than on the interests of
combatantsA who, ex hypothesi, act wrongfully; it is quite another to kill non-combatantsA, albeit
unintentionally, in self-defence and in defence of the rights of one's compatriots, given that those non-
combatants have not lost their own right not to be killed and that their membership in A per se and
combatantsV’ membership in V per se should have no bearing on the latter's justi cation for so acting. To be
clear: the dif culty is not that of justifying collateral damage itself, for on agent-neutral grounds there might
well be powerful justi cations for permitting combatants who ght in defence, not so much of their individual
and jointly held rights, but rather as a means to enforce norms against aggression, unintentionally (though
foreseeably) to kill innocent bystanders. The dif culty is that of justifying collateral damage from the
standpoint of the account of defensive rights which I offered in Sections 2.2 and 2.3.2, to which partiality—
here, collectively understood—is central.
 Those points raise the following three questions: (1) On what grounds, if any, may combatantsV give priority to
their own compatriots by unintentionally but foreseeably killing innocent non-combatantsA? (2) If non-
combatantsA in turn may kill combatantsV in defence of their own lives, do they have the (Hohfeldian) power to
transfer their permission to combatantsA? (3) If so, does this not undermine the point made in Section 2.3.2
that combatantsA, who prosecute an ad bellum unjust war, may not kill combatantsV, who wage an ad bellum
just war?
Before addressing those questions I need to take one important issue out of the way. My focus in this section is
on collateral damage, by which I mean in this instance killing by, for example, dropping bombs on munitions
factories which are located in densely populated areas. There are other kinds of harm which combatants and
their leaders typically in ict, or contribute to in icting, on non-combatants, such a shortern life-expectancy,
hunger, and general poverty. If it is permissible to in ict lethal collateral damage, then a fortiori and other

Downloaded from https://academic.oup.com/book/33082/chapter/281977852 by Library user on 09 May 2025

things roughly equal it is permissible to in ict collateral harm that is lesser than (immediate) death. In both
p. 84 kinds of cases, however, it will be appropriate to say that combatantsV justi ably infringe non-combatants’
right not to be harmed. As I noted in Section 1.2.1, agents are under a prima facie duty not to harm innocent
bystanders in the course of defending their or other people's fundamental rights, though are sometimes
justi ed in not ful lling their duty. But as I also averred then, agents are not under a prima facie duty to divest
themselves of the resources they need for a minimally decent life, thereby sacri cing their own prospects, in
order to save the needy. By deciding to wage, and thus nance, a war of collective self-defence against A,
citizensV might become unable to continue to help needy individuals in other communities without
jeopardizing their own prospects for a minimally decent life. There is thus a sense in which their decision to go
to war will result in severe harm (which they foresee but do not necessarily intend) for those individuals. In this
case however, citizensV cannot be described as infringing a right of those individuals, since there is no such
right in the rst instance under those circumstances. To put the point differently, they need not take those
harms into account when assessing whether their decision to go to war is a proportionate response to A's
wrongdoing.
In other cases, they must take those harms into account. Suppose that V's leaders and citizens face the
following choice: they can either wage their war of defence with sophisticated weapons thanks to which only
15,000 non-combatantsA will die in the bombings but 300,000 needy individuals from neutral community N
will be left without life-saving resources; or they can wage a cheaper war, in which case 30,000 non-
combatantsA will die but only 150,000 needy individualsN will starve to death (since V will be able to divert
extra resources towards them.) V would win their war under both courses of action—that is to say, they would
regain the resources and freedoms which are necessary for a minimally decent life. With that stipulation in
place, and given that (as we saw in Section 1.3) citizensV are under a duty of justice to help those in need under
those circumstances, it is appropriate to describe them as justi ably infringing the rights of non-combatantsA
not to be killed and the rights of individualsN to receive assistance. If the numbers are roughly equal, then V
ought to wage the more expensive war, on the grounds that it is morally worse to kill the innocent than to let
them die. But when numbers are not roughly equal, and when V themselves would not be worse off under one
course of action than the other, then it is not so clear that V ought to weigh the lives they destroy more heavily
than they weigh the lives they allow to be lost. This is one respect in which the cosmopolitan theory of justice I
defended in Chapter 1 has important implications for the ad bellum principle of proportionality.
In the remainder of this section, however, I focus on combatantsV’ act of foreseeably though unintentionally
p. 85 killing non-combatants in the course of defending theirs and their compatriots’ rights. Let us begin with
the rst of our three questions: namely, whether the act is permissible. Two paragraphs ago I outlined a
scenario where combatantsV must decide whether to bomb a munitions factory in the knowledge that, in so
 doing, they will kill some non-combatantsA. I shall argue presently that patriotic partiality does have a role to
play in such cases. Beforehand, however, let us examine a very different scenario, where it does not have place.
Suppose that, other things being equal, combatantsV must choose between bombing a munitions factory
which is controlled by A but is located on V's territory (factoryV), as a result of which they will kill some non-
combatantsV, and bombing a munitions factory which is controlled by A and is located on A's territory
(factoryA), as a result of which they will kill some non-combatantsA. May they opt for the latter on the grounds
that they are generally permitted to give priority to their compatriots over the lives of non-combatantsA,
however innocent the latter are of the wrongdoing committed by their leaders and combatantsA against V's
members? In private, domestic cases, patriotic partiality clearly has no place. Suppose that André, in the course
of defending his life, has a choice between killing Carl and foreseeably killing Bernard, and killing Carl and
foreseeably killing Werner, a German bystander who is wholly innocent of Carl's wrongdoing. According to

Downloaded from https://academic.oup.com/book/33082/chapter/281977852 by Library user on 09 May 2025

cosmopolitan justice he may not confer greater weight on Bernard's life than on Werner's simply on the basis
that the former is, whilst the latter is not, a compatriot.30 Can we infer, then, that combatantsV may not opt to
bomb the factory located on A on grounds of patriotic partiality?
We can infer that they may not: whatever justi cation they may invoke in support of this strategy cannot
appeal to patriotic partiality, at least not if one's account of war is to remain resolutely cosmopolitan. That said,
there are reasons in support of this particular act of bombing which a cosmopolitan can endorse.31 By waging
war against V, one might hold, A's regime wrongfully impose harm on non-combatantsV which they do not
impose on their own compatriots, and combatantsV may thus redistribute harm away from the latter to the
former. Put differently, it would be doubly unfair for non-combatantsV to suffer harm at the hands of
combatantsV given that they already suffer harm—and unjustly at that—at the hands of combatants A.. If non-
p. 86 combatantsV are exposed to a given risk of harm prior to their regime's defensive steps, and if combatantsV
have a choice between lowering this risk by bombing factoryA, thereby increasing the risk of harm to non-
combatantsA, and increasing the risk to their compatriots by bombing factoryV, then they may do the former
(at least if the total number of non-combatants harmed under both options remains the same.) But note that
the harm-distribution argument has nothing to do with patriotic partiality. Suppose that A locates its factories
in an area which is not inhabited by A's members, but which is near its borders with neighbouring neutral
country N. At the bar of just harm-distribution, V would not have a lesser reason to target those factories, thus
unintentionally killing non-combatantsN, than it would have to target factories located in the heart of A's urban
network. In a similar vein, suppose that some of the factories controlled by A though located on V's territory are
surrounded by refugees camps. Those camps are populated by non-V members whose predicament is caused
by A's decision to wage a wrongful war against V. Given that those individuals are subject to wrongful harm at
the hands of A, other things equal V has no lesser reason to redistribute harm away from them and onto non-
combatantsA than it has to redistribute harm away from non-combatantsV. To repeat, patriotic partiality plays
no role, it seems, in justifying the in iction of collateral damage by combatantsV on non-combatantsA.
However, it is central to yet another, fairly standard argument to the effect that combatantsV may so act, to wit,
that V's military and civilian of cials have a special responsibility to protect their compatriots which provides
them with a justi cation for instructing combatantsV so to act.32 Once again, this simply cannot be right, for
two related reasons. First, the standard argument thus implies that stateless individuals, be they combatants or
non-combatants, do not enjoy the protection awarded to state members. On cosmopolitan grounds this is
unacceptable: if membership in a given political community rather than another such community is not a
legitimate basis for conferring or justi ably infringing basic rights, neither is political membership rather than
no membership at all.
 Second, as we saw in Section 1.4, the rights, permissions, and duties of state of cials supervene on the rights,
p. 87 permissions, and duties of citizens, such that they have the right to govern only if and to the extent that they
do not violate the fundamental rights of both the state's members and outsiders. Moreover, it is precisely
because those of cials’ directives provide agents with better conditions for the protection of their moral rights
and enable them better to comply with their moral duties than they would have absent the state, that those
directives are binding. We thus cannot infer from the claim that V's of cials have special responsibilities to
their compatriots the thought that they may authorize combatantsV unintentionally to kill non-combatantsA
in preference to unintentionally killing non-combatantsV: we must rst discern whether V's members
themselves are permitted so to act.
In waging or authorizing a war of collective self-defence against A, V's members act in defence of their
individually held rights not to die, be tortured, raped, or maimed at the hands of combatantsA, as well as in

Downloaded from https://academic.oup.com/book/33082/chapter/281977852 by Library user on 09 May 2025

defence of their jointly held rights to political self-determination and territorial integrity. One might think that
the former rights do not permit them to give priority to one another over non-combatantsA (as per the
domestic case), whilst the latter rights do, on the grounds that the loss of non-combatantsV’ lives threatens
those rights themselves: as we saw in Section 1.4, self-determination is a communal good, in the sense that it
can be enjoyed only if a suf ciently high number of community members value it—which in turn requires that
a suf ciently high number of community members actually survive. In the case at hand, however, we assume
that combatantsV’ choice, whether to (unintentionally) kill their compatriots or non-combatantsA, would not
impair their chances of success overall. With that assumption in hand, the fact that the protection of jointly
held rights is at issue, far from providing combatantsV with a justi cation for targeting factoryA, would in fact
dictate in favour of factoryV: for account must also be taken of the fact that non-combatantsV are ex ante
bene ciaries of V's war, whereas non-combatantsA are not. The point is not unique to war. Suppose once again
that Carl attacks André and Bernard on July 1914, and in so doing violates not merely their right not to be killed
but also their jointly held rights over their house. Suppose further that André is in a position to kill Carl, and
can do so either by shooting from such an angle as to unintentionally subject Werner to a given risk of death or
by shooting from such an angle as to subject Bernard to such a risk. Suppose, nally, that Bernard stands a
higher chance of surviving the attack and thus enjoying the aforementioned rights if André takes action than if
he desists. Given that Bernard stands to bene t from André’s defensive move and that Werner not only does
not stand to bene t from it but in addition is innocent of Carl's wrongdoing, André must opt for subjecting
p. 88 Bernard to the risk. Likewise, I submit, with the choice faced by combatantsV in the case at hand—though
with two important quali cations. First, the argument from bene tting does not support the bombing of
factoryV if the bombing is more likely than not to kill the non-combatantsV in the vicinity and if it is more
likely than not that those non-combatants would not have died had V surrendered to A. Second, non-
combatantsA themselves sometimes stand to bene t from V's victory. (It seems clearly plausible that Germans
who survived the war and found themselves in what was then to become West Germany did far better in the
long term out of the Allies’ victory in World War II than they would have done had the Nazis not been defeated.)
This may well provide combatantsV with a justi cation for imposing collateral damage on them, though one
which would not in any way appeal to patriotic partiality.33
So far, we have examined a case where combatantsV must decide between unintentionally killing non-
combatantsA and unintentionally killing non-combatantsV. There, I claimed, patriotic partiality has no place.
But suppose now, in the standard case of tactical bombing, that they must decide whether to bomb a munitions
factory which is located on A's territory, as a result of which some non-combatantsA will die. In this case, there
is no alternative in the form of unintentionally killing non-combatantsV. V's leaders and combatants thus face
the following dilemma: either they protect their fellow compatriots from the harms attendant on losing the war
 by making strategic choices which will result in the (unintended) deaths of innocent non-combatantsA, or they
desist from making such choices at all, in which case those non-combatantsA will survive but their own
compatriots will suffer those harms at the hands of A's leaders and combatants. Bluntly put, they can either kill
non-combatantsA or let their compatriots die. Now, as we saw in the Introduction, other things equal, there is a
stronger presumption against killing than there is against letting die. In the case at hand, the innocent
individuals whose lives are at stake differ in that some have a special relationship with V's leaders and
combatants (of co-membership in a sovereign political community), whereas others do not. The question, then,
is whether this relationship makes things not equal, such that we have a case where letting one's compatriots
die is morally worse than killing non-combatantsA—indeed (more strongly put still) is impermissible whilst
killing non-combatantsA is permissible.
In private, domestic cases, that special relationship does not tip the balance in favour of killing foreigners qua

Downloaded from https://academic.oup.com/book/33082/chapter/281977852 by Library user on 09 May 2025

p. 89
foreigners and against letting one's compatriots qua compatriots die—any more than it tips the balance in
favour of killing a foreigner and against killing a compatriot. Suppose that André realises that the only way he
can save Bernard's life from Carl's attack is by shooting at him from such an angle as to also kill Werner—
foreseeably though unintentionally. Now, it is not clear at all (at least to me) that André may so act. But if he may
so act, it cannot be in virtue of the fact that he and Bernard share the same nationality, whereas Werner is, in
that respect at least, an outsider. To repeat, Werner's life should have exactly the same weight in André’s eyes as
would the life of a French bystander. The challenge, for the cosmopolitan, is to nd support for the claim that
combatantsV are permitted unintentionally to kill non-combatantsA in the course of missions the end of which
is to ensure that their compatriots do not continue to suffer the grievous rights violations which gave V its just
cause.
It might be tempting at this juncture to accept that cosmopolitan justice cannot accede to the imposition of
collateral lethal damage by V's agents onto non-combatantsA as an alternative to letting their compatriots die.
But the costs of such a move would be enormous indeed, since we would have to rule out as unjust any war of
collective self-defence which involves the imposition of those harms—in other words, in an age where
belligerents have long-range air power and are faced with enemies who are more than willing deliberately to
locate their military installations in densely populated areas, more or less any such war. Unless it can nd a way
to justify the unequal weighing of non-combatants’ lives on both sides of the divide without giving undue
importance to political membership, cosmopolitan justice seems indeed committed to condemning as unjust
all wars of collective self-defence in the modern age. If that is so, there is a strong case for rejecting it as a
plausible moral theory, such is the intuitive pull of the thought that sovereign political communities do have
the (prima facie) right to defend themselves against an unwarranted aggression.
It seems to me that a plausible justi cation for unequal weighing lies in agents’ personal prerogative to confer
greater weight on their own rights than on the rights of strangers. Let us return to an earlier case, whereby Carl
throws an incendiary bomb on a house which André and Bernard jointly own and currently occupy—with a
view to forcing them to relinquish their rights over it to him. Assume further (as I did in Section 2.3.1) that the
loss of their house would severely compromise André’s and Bernard's prospects for a minimally decent life. By
killing Carl, André protects his own right not to be killed as well as Bernard's, and their jointly held right of
p. 90 ownership over the house. Suppose now that the only way in which André can do so is by killing Carl and at
the same time cause the death, foreseeably but unintentionally, of bystander Werner. André’s permission so to
act, if he does indeed have it in that particular case, is grounded in the importance to him of securing not only
his individual right not to be killed but also his right, which he holds jointly with Bernard, over the house. By
parity of reasoning, combatantsV, when bombing A's munitions factories and thereby unintentionally killing
non-combatantsA, defend their own lives as well as the lives of their compatriots, and their jointly held rights
to territorial integrity and political sovereignty—rights which warrant protecting by force precisely because (as
 we saw above) they are constitutive of a minimally decent life. CombatantsV do, and indeed may, assign greater
weight to their compatriots’ fundamental interests than on the lives of non-combatantsA in such cases, in
virtue not of common membership in V alone, but rather as a necessary means to defend the fundamental
rights which they too hold jointly with those other agents.
This defence of patriotic partiality presupposes that V's members, qua combatants and/or citizens, are not
under a duty to surrender to A, even though their refusal to surrender causes innocent non-combatantsA to die.
One might wonder whether V ought not, in fact, to surrender (both in the private and in the war cases), in the
light of those particular costs. For suppose that they could surrender immediately and, as a result, ‘only’ lose
control over their territory and institutions: in so doing, they would survive A's admittedly unlawful aggression
and would avoid killing innocent persons. If the choice, thus, is between, on the one hand, losing rights lesser
than the right not to be killed and avoiding innocent deaths, and, on the other hand, successfully defending

Downloaded from https://academic.oup.com/book/33082/chapter/281977852 by Library user on 09 May 2025

those rights by causing such deaths, it might seem that they should opt for the former course of action, not the
latter. Given what we know of the conduct of unwarrantedly bellicose regimes, I have serious doubts as to how
likely it is that a peaceful and immediate surrender would spare the lives of V's members. But let us suppose for
the sake of argument that this indeed is the choice with which V's members are faced. To say that they ought to
surrender is tantamount to claiming that individuals are under a duty to submit to a regime which, after all, is
willing ex hypothesi to invade them by force unwarrantedly, and which will in all likelihood unwarrantedly
impose its unjust rule by force. It is in effect tantamount to claiming that V's members are under a duty to
submit themselves to an oppressive dictatorship, though the latter would seriously impair their prospects for a
minimally decent life, for the sake of ensuring that innocent persons should not die, and however
unintentional their death. It may well be that V's citizens must surrender if they cannot win their war other
p. 91 than by (unintentionally) killing ten million non-combatantsA. But to say that they must do so even if they
would kill only a few thousand in the course of defending themselves is far too demanding indeed.
I have assumed so far that A's leaders and combatants violate rights to the freedoms and resources necessary to
V's prospects for a minimally decent life. But suppose that they violate rights which protect interests above and
beyond the decency threshold. In such cases, to hold V's leaders and combatants under a duty to surrender does
not strike me as too demanding. I argued in Section 2.3.1 that the protection of V's right to one of its small
empty island provides combatantsV with a justi cation for killing combatantsA only if the latter subject them
to a threat of grievous harm. Suppose that the island is far from V's territory, but very close to A's, and that A's
soldiers would employ lethal force if met with resistance. Suppose further that V's of cials come to the
conclusion that defensive force short of war will not work, and that they can persuade A's forces to leave the
island only by bombing their supply lines at their source—in so doing foreseeably though unintentionally
killing a number of non-combatantsA. Suppose nally that A's regime will not use a successful invasion of the
island as a platform from which to mount a large-scale invasion of V's mainland. It is hard to see, in that case,
how the loss per se of their rights over this small island, which would not cause V's members to have a less than
minimally decent life, could justify killing innocent individuals. More controversially still, the point holds even
if A invades a portion of V's inhabited mainland under the same circumstances—even if, in other words, A's
invasion, if unchecked, would lead to redrawing V's borders without at the same time jeopardizing the
prospects for a minimally decent life of V's members. In a nutshell, for the in iction of lethal collateral damage
to be permissible, it must occur via defensive steps which are necessary to defending rights the violation of
which would undermine rights-holders’ prospects for a minimally decent life. If that condition is not met, and
if the war cannot be fought and won other than by in icting such damage, then surrender is mandatory.34
Before concluding, it is worth highlighting some of the features of my argument in favour of unequal weighing.
First, it is not an argument in favour of conferring lesser weight on non-combatantsA qua non-combatantsA.
p. 92 Rather, it only supports the view that combatantsV are permitted to give priority to their own compatriots if
 that is the only way to win a war in which their own rights, qua citizens, are at stake. Crucially, the justi cation
for the permission also applies when the non-combatants who die as a result of combatantsV’ actions belong to
a neutral community. Suppose that André has a choice between collaterally killing Werner and collaterally
killing John—a British subject who happens to pass by. At the bar of cosmopolitan justice, he certainly may not
opt to kill Werner rather than John on the grounds that the former shares Carl's nationality. By parity of
reasoning, combatantsV may not invoke non-combatantsA’ shared nationality with wrongdoing combatantsA
as a justi cation for bombing the factories near which the latter live, as opposed to factories which are located
next to the border which A shares with neutral community N and the destruction of which would cause the
death of non-combatantsN.35 Second, it is not an argument in favour of conferring greater weight on the lives
and interests of combatantsV qua combatants over the lives of non-combatantsA as such: rather, it is an
argument in favour of conferring greater weight on one's lives and interests as well on the lives of one's

Downloaded from https://academic.oup.com/book/33082/chapter/281977852 by Library user on 09 May 2025

compatriots precisely to the extent that one's individual and jointly held fundamental rights are under attack.
At rst sight, the conclusion reached in the standard case of tactical bombing seems inconsistent with the
conclusion reached in the case where combatantsV had to choose between the unintentional killing of non-
combatantsV and the unintentional killing of non-combatantsA. For in the latter case, I argued, patriotic
partiality has no role to play. How can it be, then, that combatantsV are not allowed to kill non-combatantsA as a
means not to let their compatriots qua compatriots die in the rst scenario, but are allowed so to act in the
second scenario? This is because in the rst case, combatantsV themselves are in a position to in ict harm on
either one of those two groups as a by-product of the steps which they take in defence of their jointly held
rights, and must thus decide how to distribute that harm. In the second case, by contrast, their choice is not a
choice between killing some versus killing others; rather, it is a choice between killing some as a result of
which others to whom they stand in a special relationship and with whom they hold some fundamental rights
will survive, or not killing some as a result of which those others will die and the jointly held fundamental
p. 93 rights will be threatened by combatantsA. It is not incoherent to invoke patriotic partiality in support of
doing the latter but to deny its appeal for the former, at least on the assumption that, by shifting harm onto
non-combatantsV, combatantsV do stand a reasonable chance of restoring the rights which they are seeking to
defend. Suppose that factoryV and factoryA both manufacture nuclear weapons ten times as powerful as those
dropped on Hiroshima and Nagasaki; suppose further that A will subject V to a nuclear attack in the near future
unless they either surrender or bomb one of those factories. In opting for bombing either, however,
combatantsV would trigger a crippling, large-scale nuclear disaster in the relevant community. Assuming for
the sake of argument that a community may defend itself in such a way when faced with that particular threat,
patriotic partiality would permit them to bomb factoryA rather than factoryV.
Furthermore, it might also seem that the argument from bene t, which blocks patriotic partiality in the non-
standard case, and the argument from the personal prerogative, which makes room for it in the standard case,
pull in radically different directions in yet another, crucially important scenario. Suppose that if tactical
bombersV bomb factoriesA at very high altitude, they will incur a 2 per cent risk of being gunned down by A's
anti-aerial defence systems and unintentionally kill 500 non-combatants A in the vicinity. If they y lower they
will be able to identify their targets much more precisely, and thus will kill far fewer non-combatantsA, but will
run a 10 per cent risk of being gunned down. At the bar of the personal prerogative, given that their lives are at
stake, it would seem that they may opt for the former strategy. At the bar of the bene tting argument, given
that they would bene t from V's victory, it would seem that they ought to redirect the risks attendant on the
operation onto them and away from non-combatantsA. Remember, however, that harm-redirection is not
permitted by the bene t argument if there is a higher than 50 per cent probability that those victims would not
have died had V surrendered. Suppose, then, that bombersV stand a higher chance of surviving if the war is
 fought and fought by them under either course of action, than if they did not participate in it (either because V
surrenders, or because they are not called up to ght.) Given that they would kill fewer non-combatantsV by
ying low, this is what they should do. But if they stand a lower chance of surviving the war by ying low than
they would have if the war were not fought at all or, though fought, not by them, then the personal prerogative
permits them to y high (subject to the numbers of non-combatants killed.)
Finally, it is worth pointing out that my account of the conditions under which V's members are (sometimes)
p. 94 justi ed in giving priority to one another's fundamental rights over non-combatantsA’ rights does not
support the view that they are under a duty to one another, qua compatriots, so to act. For on my account,
unequal weighting is justi ed in virtue of V's citizens’ personal prerogative to further their interests in the
communal goods of territorial integrity and collective self-determination. Self-interest can at best support a
permission but not a duty to one another, so to act. Clearly, if V's citizens resort to war as a means of collective

Downloaded from https://academic.oup.com/book/33082/chapter/281977852 by Library user on 09 May 2025

self-defence (standardly via their regime), then those of their compatriots who are best equipped to do the
ghting (in other words, combatants) must carry out those missions. Theirs, however, is a contractual duty
which is constrained by what they as citizens are permitted to do. But it is not a duty which they have, qua
citizens and together with fellow citizens, to one another qua compatriots, to act in such a way as to
unintentionally kill innocent non-combatants who belong to the enemy or to neutral parties. Crucially, this
point is compatible with the view (which I shall defend at greater length in Section 7.3) that conscription for the
sake of national defence is legitimate, provided that it is understood and justi ed not as a mechanism to
enforce citizens’ duty to their compatriots to help protect the latter's rights, but rather as a mechanism to
enforce citizens’ duty not to free-ride on their compatriots’ willingness to ght for rights which they all deem
in their individual interest to preserve.36
To recapitulate, I have argued that combatantsV have a justi cation (as provided by their just cause) for
in icting collateral damage on innocent non-combatantsA. But as we saw in Section 2.2, in so far as the latter
have not (ex hypothesi) lost their right not to be killed, they would be permitted to kill their justi ed attackers
in their own defence should they have the wherewithal to do so. The question, at this juncture, is whether they
have the (Hohfeldian) power to transfer their permission to third parties—most obviously to combatantsA. To
claim that they do is tantamount to granting the latter the permission to kill combatantsV. In Section 2.3.2,
however, I argued that combatantsA may not kill combatantsV in their own defence. May they do so in defence
of their compatriots? I do not think so. For the latter in transferring their permission to combatantsA would in
effect enable those combatants to continue to participate in the grievous wrongdoing of aggression vis-à-vis V.
Admittedly, in defending themselves by killing combatantsV they too would enable their regime to continue so
p. 95 to act. But whilst they are not under a duty to allow themselves to die for the sake of stopping that
wrongdoing (of which, crucially, they are innocent), their interest in survival is not strong enough to be
protected by a power to allow those very agents who carry out a wrongful war of aggression against V and its
members to target justi ed combatantsV—any more than bystander V, caught in the cross re between a bank
robber and a policeman, has the power to permit the former to kill the latter in her own defence. Agent-neutral
considerations (in this case, bringing an end to A's wrongful war of aggression against V) outweigh non-
combatantsA’ agent-relative justi cation for conferring on combatants A the permission to kill combatantsV.
Far from killing combatantsV, combatantsA ought instead to surrender to them, as a means both to stop their
regime's wrongful aggression and to forestall justi ed attacks by combatantsV on their compatriots.
 2.5 Conclusion

When thinking about wars of collective self-defence, we tend to say, loosely, that ‘state A wrongfully attacks V
and state V justi ably defends itself against A.’ But this will not do, for two reasons. First, so stated, the
statement does not do justice to the complex set of actions and actors which characterize an act of military
aggression and the concomitant act of collective self-defence. Instead, we ought to mean the following: A's
individual agents together violate the rights of V's individual agents, on several levels and in different ways. A's
political and military leaders order A's individual combatants (sometimes with the authorization of A's citizens)
to act in violation of those rights—for example, by crossing the border into V's territory. CombatantsA in turn
acquiesce in those orders, and not only act in the indicated ways but in addition use lethal force against V's
attempts to block their attack. V, in that last clause, must be understood in several ways: as V's political and
military leaders who, (sometimes) on the authorization (implicit or explicit) of V's individual citizens, order

Downloaded from https://academic.oup.com/book/33082/chapter/281977852 by Library user on 09 May 2025

individual combatantsV to use lethal force to block combatantsA’ attack; as those citizens themselves (when
appropriate), and as the combatants tasked with repelling the attack. The resultant war between A and V
consists in a series of acts of killing on both sides—some justi ed, others not. To the extent that the initial
wrongdoings committed by A's individual leaders and combatants (acting together) consist in violating the
rights of V's members, the right as held by V's individual combatants to kill combatantsA is a right to use lethal
force in defence of those rights. To the extent that those wrongdoings are backed by lethal force, the right to go
p. 96 to war is a right that soldiers have, not simply to defend their fundamental basic rights, as individuals or
citizens, but also to defend their compatriots’ similar rights (in which case they do so by way of the latter's
exercise of a justi ed Hohfeldian power).
Second, the claim that a given rights violation is a just cause for war is too rough. That claim in fact comprises
two distinct (though related) sub-claims: the claim that a given right is important enough on principle to
warrant defending by force against those who violate it even if one in icts lethal collateral damage on innocent
non-combatants, and the claim that there are identi able agents who do violate that right in such a way that
they are liable to having force used against them. I have argued in this chapter that individual rights not to be
killed or grievously maimed, as well as jointly held rights to political self-determination and territorial
integrity, are important enough in that sense. In addition, we need to distinguish between a narrow just cause
and a wide just cause. The former is a wrongdoing which provides its victims with a justi cation for launching
a lethal attack against wrongdoers; the latter is a wrongdoing which provides its victims with a justi cation for
making a non-lethal forceful defensive move against wrongdoers in the knowledge (or at least on the
reasonable assumption) that the latter will retaliate with unwarranted lethal force and thus provide victims
with a narrow just cause for (this time) waging war. Paradigmatic cases of collective self-defence against
military aggression obscure the distinction, since they typically are responses to a wrongdoing which, carried
out as it is through the use of lethal force, gives rise to a narrow just cause. Yet the distinction is absolutely
essential, because unless we can identify, in any given case, wrongdoers who are liable to being killed, or
wrongdoers who are liable to being subject to non-lethal force (as the rst step towards a full-scale war), we
will not be able to claim that the war is justi ed. Subsistence wars, as we shall now see, represent a case in
point.

Notes
1 On the international law of aggression, see Y. Dinstein, War, Aggression and Self-Defence, 3rd edn. (Cambridge: Cambridge
University Press, 2001) 10.1017/CBO9781139164726 . For a philosophical treatment of that particular body of law, see L.
May, Aggression and Crimes Against Peace (Cambridge: Cambridge University Press, 2008) 10.1017/CBO9780511611551 .
2 For avowedly cosmopolitan defences of the model, see Atack, The Ethics of Peace and War; D. Held, ʻViolence, Law and
 Justice in a Global Ageʼ, no. (2001), http://www.theglobalsite.ac.uk/press/112held.htm ; M. Kaldor, Global Ccivil Ssociety:
An Answer to War (Cambridge: Polity Press, 2003), esp. ch. 5. I reject the model in C. Fabre, ʻCosmopolitanism and Wars of
Self-Defenceʼ, unpublished typescript.
3 See Atack, The Ethics of Peace and War; N. Zohar, ʻCollective War and Individualistic Ethics: Against the Conscription of
“Self-Defense”ʼ, Political Theory 21 (1993): 606–22 10.1177/0090591793021004003 .
4 E. d. Vattel, Le droit des gens, ou, Principes de la loi naturelle appliques á la conduite et aux a aires des nations et des
souverains (London: 1758); C. Wol , Jus gentium methodo scientifica pertractatum (Oxford: Clarendon Press, 1934). For an
analysis of the international law pertaining to aggression, see Dinstein, War, Aggression and Self-Defence; and for a recent
defence of the analogical strategy, see G. P. Fletcher and J. D. Ohlin, Defending Humanity (Oxford: Oxford University Press,
2008).
5 For the individualistic variant, see J. McMahan, ʻInnocence, Self-Defense and Killing in Warʼ, Journal of Political Philosophy
2 (1994): 193–221 10.1111/j.1467-9760.1994.tb00021.x ; ʻSelf-Defense and the Problem of the Innocent Attackerʼ, Ethics
104 (1994): 252–90; ʻThe Ethics of Killing in Warʼ, Ethics 114 (2004): 693–733; ʻJust Cause for Warʼ, Ethics & International

Downloaded from https://academic.oup.com/book/33082/chapter/281977852 by Library user on 09 May 2025

A airs 19 (2005): 1–21; ʻWar as Self-Defenseʼ, Ethics & International A airs 18 (2004): 75–80. For a recent book-length
restatement and refinement of the views he expressed in those articles, see Killing in War. Both Nagel and Luban also
insist that war pits individuals against one another (T. Nagel, ʻWar and Massacreʼ, Philosophy & Public A airs 1 (1972): 123–
44; and D. Luban, ʻJust War and Human Rightsʼ, Philosophy & Public A airs 9 (1980): 160–81.) Walzer himself seems to hold
that view, alongside the analogical view, so that his overall account of the justification for the right to wage war is
ambiguous. See M. Walzer, Just and Unjust Wars: A Moral Argument with Historical Illustrations, 4th edn. (New York: Basic
Books, 2006), 51. For the collectivist variant, see H. Grotius, The Rights of War and Peace. ed. R. Tuck. (Indianapolis, IN:
Liberty Fund, 2005); F. R. Tesón, Humanitarian Intervention: An Inquiry into Law and Morality, 2nd edn. (Irvington-on-
Hudson, NY: Transnational, 1997); F. de Vitoria, On the Law of War, in F. de Vitoria, Political Writings, ed. A. Pagden and J.
Lawrance (Cambridge: Cambridge University Press, 1991), and N. Zohar, ʻCollective War and Individualistic Ethics: Against
the Conscription of “Self-Defense”ʼ, Political Theory 21 (1993): 606–22 10.1177/0090591793021004003 .
6 From now on, therefore, when I refer to the reductive strategy I shall mean the strategy in its individualistic form (unless
stated otherwise).
7 J. J. Thomson, ʻSelf-Defenseʼ, Philosophy & Public A airs 20 (1991): 283–310. The literature of defensive killings is
enormous, and the most important contributions are referenced in the next few footnotes.
8 McMahan, ʻSelf-Defense and the Problem of the Innocent Attackerʼ, Ethics 104 (1994): 252–90 10.1086/293600 ; M.
Otsuka, ʻKilling the Innocent in Self-Defenseʼ, Philosophy & Public A airs 23 (1994): 74–94 10.1111/j.1088-
4963.1994.tb00005.x ; J. Quong, ʻKilling in Self-Defenceʼ, Ethics 119 (2009): 507–37 10.1086/597595 . In that article,
McMahan argues that whilst we have good reasons to reject the view that killing innocent attackers is impermissible, it is
nevertheless very di icult to construct a plausible argument in support of the converse claim that such killing is
permissible. Otsuka explicitly rejects that claim in one-to-one context. Quong accepts it, on grounds similar to those I
adduce below (pertaining to victimsʼ personal prerogative.) Incidentally, the claim that killing innocent attackers in self-
defence is permissible is compatible with the claim that one ought to exercise greater restraint when thwarting them than
would be warranted if one were faced with a culpable attacker—at least given equal chances for survival (precisely
because they are not responsible for forcing the choice between lives.) See McMahan, ʻSelf-Defense and the Problem of
the Innocent Attackerʼ, 264–5.
9 In Grotius's words, our permission to kill stems ʻfrom the Care of our own Preservation, which Nature recommends to
every one, and not from the Injustice or Crime of the Aggressor.ʼ For Grotius, moreover, the wrongfulness of the act, as
opposed to moral fault in the aggressor, does make a di erence to the permissibility of self-defensive killing: as Grotius
holds, an unjust attacker may not defend himself from his victim unless he has desisted and o ered her restitution even
though she still threatens his life. (Grotius, The Rights of War and Peace, Book II, Ch. 1, para. 3, and Book II, Ch. 1, para. 18.)
See also Quong, ʻKilling in Self-Defenceʼ, and F. M. Kamm, ʻGrouping and the Imposition of Lossʼ, Utilitas 10 (1998): 292–
319 10.1017/S0953820800006233 , at 310–16. In this piece, Kamm rejects Peter Unger's claim that we must treat
ourselves exactly as we may treat others—in other words, that there is no such thing as a personal prerogative. (See Unger,
Living High and Letting Die.) Note that I use the phrase ʻwrongful threatʼ to denote a threat which its target has not done
anything to warrant, such that compensation is owed to her (should she survive the attack). I do not mean to suggest that
an innocent attacker can be appropriately described as acting wrongly in such cases. In fact, the case of justified attackers
show that one can pose a wrongful threat and be justified in so doing.
10 See, for example, McMahan, ʻSelf-Defense and the Problem of the Innocent Attackerʼ, at 270 . The rejoinder that follows
appears in C. Fabre, ʻPermissible Rescue Killingsʼ, Proceedings of the Aristotelian Society 109 (2009): 149–64 10.1111/j.1467-
9264.2009.00262.x .
11 One further point on innocent attackers. Some attackers are what Catholic theologians call ʻinvincibly ignorantʼ, in that
 they cannot possibly know, or be expected to know, that they are inflicting harm on someone else (consider, for example,
the case of someone who dreams that he is about to drown, who gesticulates in an attempt—he thinks—to claw his way
back to the surface of the water by holding on to a branch, but who in fact is strangling his wife.) On my account, in so far
as an invincibly ignorant attacker lacks an objective justification for attacking V, the latter has the right to kill him. This too
is a very controversial claim, though the overwhelming majority of war cases do not involve invincibly ignorant
combatants. More generally, on the view I articulate here, whether or not A has an objective justification for attacking V is
decisive for determining whether V has the right to kill A. Whether A has a subjective justification (or an excuse) is another
matter. For a detailed scrutiny of the complex relationship between excuses and justifications in the context of war, see
McMahan, Killing in War, ch. 3. For an account of permissible self-defence which assigns a central role to subjective
justifications, see H. Frowe, ʻA Practical Account of Self-Defenceʼ, Law and Philosophy 29 (2010): 245–72 10.1007/s10982-
009-9062-1 . For normative accounts of the killing of child soldiers, see, for example, McMahan, Killing in War, pp. 198–
202, and M. Vaha, ʻChild Soldiers and Killing in Self-Defence: Challenging the “Moral View” on Killing in Warʼ, Journal of
Military Ethics 10 (2011): 36–52 10.1080/15027570.2011.561639 . For an empirical account, see P. W. Singer, Children at

Downloaded from https://academic.oup.com/book/33082/chapter/281977852 by Library user on 09 May 2025

War (New York: Pantheon Books, 2005).
12 See McMahan, Killing in War, p. 165, for the case of the ambulance-driver. McMahan agrees that the ambulance-driver has
a subjective justification for acting as he does, but denies that he is objectively justified in subjecting the pedestrian to a
risk of being killed. On his account, then, the pedestrian has the right to kill the driver—the latter, in other words, is under
a duty not to save his own life. I submit, on the contrary, that the ambulance-driver has an objective justification for so
acting. For penetrating criticisms of McMahan's position on this and, more generally, of his account of the right to kill, see
S. Lazar, ʻResponsibility, Risk, and Killing in Self-Defenseʼ, Ethics 119 (2009): 699–728 10.1086/605727 . More generally,
my account of defensive rights implies that it is sometimes permissible to kill innocent obstructors, pace (for example)
Zohar, ʻCollective War and Individualistic Ethicsʼ; N. Zohar, ʻInnocence and Complex Threats: Upholding the War Ethics and
the Condemnation of Terrorismʼ, Ethics 114 (2004): 734–51 10.1086/383444 . Note, finally, that on my view, culpability is
not relevant to liability (on which point McMahan clearly disagrees).
13 This is so even if R is related to A, and even if A is not morally responsible for his action. I do not believe, therfore, that a
mother is justified in siding with her psychotic son against the latter's victims (though she may be excused for so doing.)
14 For the view that R must remain neutral, as applied to conflicts between innocent passive threats and victims, where (it is
said) the former are permitted to kill the latter in self-defence and vice versa, see N. Davis, ʻAbortion and Self-Defenseʼ,
Philosophy & Public A airs 13 (1984): 175–297. To anticipate Chapter 5, unless we are really prepared to condemn more or
less all humanitarian interventions as wrong, we must accept that letting many die is sometimes worse than killing a few.
For if third parties intervene, they almost always inevitably kill innocent victims (though unintentionally so). If they do not
intervene, they will let (some of) those victims die at the hands of their regime. To say that intervening is permissible is
tantamount to claiming that, in those cases at least, killing is permissible though the alternative is ʻmerely letting dieʼ.
More strongly still, to say that intervening is mandatory is tantamount to claiming that letting die, in those cases, is worse
than killing.
15 In another paper I scrutinise cases of so-called bloodless invasions, where A violates V's sovereignty rights without killing
anyone within V. I argue that even in (some of) those cases, combatantsV have the right to use force in defence of those
rights. See Fabre, ʻCosmopolitanism and Wars of Self-Defenceʼ, unpublished typescript.
16 The Battle of the Marne, which lasted only a few days and saw dozens of thousands of casualties on all sides, was the first
major battle of the First World War. It more or less stopped the German advance into French territory, and marked the start
of trench warfare. See H. Strachan, The First World War: To Arms, vol. 1 (Oxford: Oxford University Press, 2001), 242–62.
17 Rodin, War and Self-Defense, 128 and 132–8; R. J. Norman, Ethics, Killing, and War (Cambridge: Cambridge University
Press, 1995), 128–31 10.1017/CBO9780511554568 .
18 The argument which I deployed one paragraph ago regarding killing as a means to block rape (roughly) follows F. Leverick,
Killing in Self-Defence (Oxford: Oxford University Press, 2006) 10.1093/acprof:oso/9780199283460.001.0001 , ch. 8. For
interesting parallels between the right to use force in defence of one's home and the right to wage a war of collective self-
defence, see Hurka, ʻProportionality in the Morality of Warʼ, at 55–66.
19 For the view that only legitimate regimes have the right to wage war, see Orend, The Morality of War, 35 . For the view that
illegitimate (on cosmopolitan grounds) regimes nevertheless may have a just cause for war, see Caney, Justice Beyond
Borders, 203–4. Caney argues that to deny illegitimate regimes the right to go to war—on the grounds that, illegitimate as
they are, they cannot have a just cause—is to overlook the fact that for them not to wage war would create greater
injustice. Though he does not explicitly distinguish (at this juncture of his discussion) between the rights of political
regimes and the (individual and collective) rights of the lattersʼ members, I essentially agree with him.
20 C. A. J. Coady, Morality and Political Violence (Cambridge: Cambridge University Press, 2008); Vitoria, On the Law of War;
McMahan, Killing in War; Rodin, War and Self-Defense.
21 Walzer, Just and Unjust Wars, ch. 9. For spirited and recent defences of the orthodox account and the moral symmetry
thesis, see Y. Benbaji, ʻThe Responsibility of Soldiers and the Ethics of Killing in Warʼ, Philosophical Quarterly 57 (2007):
558–72 10.1111/j.1467-9213.2007.497.x ; ʻA Defence of the Traditional War Conventionʼ, Ethics 118 (2008): 464–95; L. May,
War Crimes and Just War (Cambridge: Cambridge University Press, 2007) 10.1017/CBO9780511841002 ; P. Emerton and
T. Handfield, ʻOrder and A ray: Defensive Privileges in Warfareʼ, Philosophy & Public A airs 37 (2009). See also Rodin and
Shue (ed.), Just and Unjust Warriors, for a very good set of essays on the dispute. For criticisms of McMahan's position
specifically, see Lazar, ʻResponsibility, Risk, and Killing in Self-Defenseʼ; ʻThe Responsibility Dilemma for Killing in War: A
Review Essayʼ, Philosophy & Public A airs 38 (2010): 180–213 10.1111/j.1088-4963.2010.01182.x . It is worth stressing
here, once more, that there is one crucially important di erence between McMahan's account and my own account.
Whereas he locates the basis for liability in moral responsibility for a wrongful threat (for which there is no justification),
and thus holds that one does not have the right to kill morally innocent attackers, I locate it in contributory responsibility
alone.
22 For powerful rebuttals of the duress and epistemic objections, see McMahan, Killing in War, ch. 3. For further decisive

Downloaded from https://academic.oup.com/book/33082/chapter/281977852 by Library user on 09 May 2025

criticisms of the epistemic objection, see D. Rodin, ʻThe Moral Inequality of Soldiers: Why jus in bello Asymmetry is Half
Rightʼ, in D. Rodin and H. Shue (ed.), Just and Unjust Warriors (Oxford: Oxford University Press, 2008).
23 See, for example, Shue, ʻDo We Need a “Morality of War”?ʼ, 99 ; see also C. A. J. Coady, ʻThe Status of Combatantsʼ, in D.
Rodin and H. Shue (ed.), Just and Unjust Warriors (Oxford: Oxford University Press, 2008), 164 . A full treatment (either in
defence or in rebuttal) of the epistemic objection would require distinguishing soldiersʼ putative ignorance of the
empirical facts at hand (for example, did Saddam Hussein's regime have nuclear weapons, and did it conceal them from
UN inspectors?), and their putative ignorance of the moral facts at hand (for example, assuming that Saddam Hussein did
have and conceal such weapons, did this constitute a justification for invading Iraq?). It would also have to trace the
connection, or identify the lack thereof, between ignorance (and knowledge) of either kind and moral responsibility for
the ensuing war.
24 A point pithily made, for example, by L. McPherson, ʻInnocence and Responsibility in Warʼ, Canadian Journal of Philosophy
34 (2004): 485–506. There is a sense in which soldiers might be deemed closer to the facts as pertain to targeting decisions
made once the war has started, than they are with respect to the ad bellum decision to resort to war in the first instance.
Perhaps. But there is also a sense in which the circumstances under which they find themselves are much more conducive
to dispassionate reflexion on whatever facts they have ad bellum than in bello. And that, surely, is highly relevant—to a
degree that the epistemic objection overlooks. I am grateful to Jon Quong for pressing me on this.
25 I used to think that an agent's contribution to a grievous wrongdoing had to meet a certain threshold of causal
significance in order for the agent to be deemed liable to being killed. I develop that point in Fabre, ʻGuns, Food, and
Liability to Attackʼ, part V. As is clear here, I now think that I was wrong, and that causal significance is not a necessary
condition for liability (though it o en matters). I owe this change of mind to Jon Quong, to whom I grateful for saving me
from that mistake. The torturers example is loosely drawn from Parfit's well-known discussion of collective responsibility
in D. Parfit, Reasons and Persons (Oxford: Clarendon, 1984), 70. The example of the stone-casting is Quong's. The criterion
for distinguishing marginal contributions which render one liable from those which do not is mine. The criterion of causal
significance has a troublesome implication, which was brought to my attention by both Quong and Krister Bykvist: to wit,
that individuals acting together can avoid being liable to lethal force by distributing responsibilities in such a way that the
act of every single one of them has a negligible probability of harming third parties (as negligible as the probability of
one's vote being a casting vote, for example.) In the cases at hand, however, where at least some policy decisions have to
be made by small groups of people, it is not clear to me that there would be no agent whose contribution would warrant
attacking him.
26 The precautionary principle I set out here dovetails with the view (which was defended in Section 1.2.1) that pending
information to the contrary, a needy individual is not responsible for his predicament, and prima facie has a claim to our
assistance. Note that my discussion in this paragraph partly follows McMahan's account of the complex relationship
between just cause and proportionality in McMahan, Killing in War, 21–32. I say ʻpartlyʼ because McMahan rejects appeals
to thresholds in general (ibid.,), and because he believes that moral (as opposed to merely causal) responsibility for a
wrongful threat (by way of unjustified action) is the basis for liability.
27 Pace Tesón, ʻSelf-Defense in International Law and Rights of Personsʼ. I phrase the point by reference to combatants for
ease of exposition, but it applies to participants in war in general.
28 See McMahan, ʻThe Morality of War and the Law of Warʼ, 31, the argument of which I follow closely in the remainder of this
paragraph. McMahan's own example—that of an oil field—appears in another context. I use the case of a diamond mine to
forestall the objection that, as A's army relies on regular supplies of oil to fight its unjust war, V might be wholly justified in
seizing its fields.
29 I say ʻallegedlyʼ because accounts di er as to whether Thompson really issued that threat. See D. L. Anderson (ed.), Facing
My Lai (Lawrence, KS: University Press of Kansas, 1998), 27 , for Thompson's own recounting of the massacre at a 1994
 conference organised by Tulane University—at which he does not, in fact, unambiguously either deny or admit having
used those words. See also M. Bilton and K. Sim, Four Hours in My Lai (Harmondsworth: Penguin Books, 1992), esp. 136–
40. There is a di erence, of course, between one's fellow soldiers attacking unarmed civilians and their attacking just
combatants: namely, that at least as things stand, soldiers can reasonably be expected to know that the former is
grievously wrong (as well as the most serious o ence under the laws of war), whereas they usually will not know that they
may not kill enemy combatants in prosecution of an unjust cause (given the training they receive and the enduring appeal
of the orthodox account). But lack of knowledge at best provides them with an excuse, and not a justification, for so acting
—as does duress, in fact. My concern here is with the latter, not the former. I am grateful to Jonathan Quong for pressing
me on this point.
30 One need not be a cosmopolitan to endorse that view: David Miller, for example, explicitly claims that in Thomson-type
trolley cases one may not redirect the runaway trolley away from five workers stuck on the line towards a foreigner in
preference to turning it towards compatriots. See D. Miller, ʻReasonable Partiality towards Compatriotsʼ, Ethical Theory and
Moral Practice 8 (2005): 63–81 10.1007/s10677-005-3296-2 , 74–5.
31 I owe the essence of the argument that follows to Jonathan Quong.

Downloaded from https://academic.oup.com/book/33082/chapter/281977852 by Library user on 09 May 2025

32 See, for example, Hurka, ʻProportionality in the Morality of Warʼ, 60–1; F. M. Kamm, ʻFailures of Just War Theory: Terror,
Harm, and Justiceʼ, Ethics 114 (2004): 650–92, 674 10.1086/383441 . Hurka invokes the special responsibility argument
not merely in support of a nation's preference for the lives of its civilians as opposed to the lives of enemy civilians, but
also in support of its preferences for its own soldiers as opposed to the lives of enemy civilians. For apt criticisms of
Hurka's and Kamm's arguments, see D. Lefkowitz, ʻPartiality and Weighing Harm to Non-Combatantsʼ, Journal of Moral
Philosophy 6 (2009): 298–316 10.1163/174552409X433409 .
33 I thus agree with McMahan that one must take into account benefits accruing from the war when considering whether one
may subject non-combatants at risk of dying. Unlike McMahan, however, I believe that one must take into account not
merely ex ante but also ex post benefits. See J. McMahan, ʻThe Just Distribution of Harm Between Combatants and
Noncombatantsʼ, Philosophy & Public A airs 38 (2010): 342–79 10.1111/j.1088-4963.2010.01196.x , esp. 365. I shall say
more about harm-shi ing towards those who stand to benefit from a war in Section 5.5.2, where I address the issue of
collateral damage in wars of humanitarian intervention.
34 This is another way of making the familiar point that war must be a proportionate response to the wrongdoing inflicted.
But note that in the proportionality calculus, only harms to the innocent are included: combatants who carry out the
wrongdoing of aggression have lost their claim to have the loss of their lives included in the calculus.
35 Kamm also seems to think that members of a political community may confer greater weight on saving themselves as a
group, at the cost of the lives of non-combatants who belong to the enemy. On the issue of neutral non-combatants,
however, I disagree with her, for she believes that V ought to confer greater weight on the lives of non-combatantsN than
on the lives of non-combatantsA, on the grounds that the latter belong to the community with which V is at war. See
Kamm, ʻFailures of Just War Theory: Terror, Harm, and Justiceʼ, 674–5.
36 For if it were understood and justified in that way, conscription would be tantamount to institutionalizing a duty which
citizensV in fact do not have: namely, a duty to their compatriots to protect the latter's rights at the cost of distant and
innocent strangersʼ interest in survival. For the view that the relationship of co-citizenship generates associative duties in
war which trump general duties not to harm outsiders, see Lazar, War and Associative Duties.