1 Introduction

Humanitarian interventions over the last couple of decades have resulted in large numbers of civilian casualties as well as the destruction of countless homes, businesses, and vast infrastructure.Footnote 1 Interventions in Bosnia, Kosovo and Libya, for instance, caused the deaths of over a thousand civilians, whereas interveners suffered no casualties in these cases.Footnote 2 Consider NATO’s intervention in Kosovo in 1999. The prospective intervening countries were faced with a critical dilemma. On one hand, they felt significant pressure to intervene, after the US and other countries had failed to take decisive action to stop the genocide in Rwanda five years earlier. On the other, there was significant pressure not to expose their own combatants to risks of lethal harm, after the gruesome killing of 18 American soldiers during the intervention in Somalia in 1993.Footnote 3 The alliance’s response to this challenge was to rule out ground forces, and fly at altitudes out of reach of Serbian anti-aircraft weapons. However, this not only reduced the risk to combatants; it also reduced precision in targeting as pilots were unable to do their own reconnaissance, significantly increasing the risk of collateral harm to civilians. Whilst this strategy was successful in the sense that NATO forces suffered no casualties over the course of the war, this came at the cost of many innocent lives. An estimated 500 civilians were killed by a NATO intervention originally intended to protect innocent civilians.Footnote 4

This case is by no means exceptional. It is standard practice for militaries to seek to minimise risk to their own forces to the extent legally permissible, a policy usually referred to as ‘risk transfer’.Footnote 5 This paper challenges one line of argument which has been advanced by philosophers to justify transferring risks from intervening combatants to certain civilians in armed interventions. I call this argument the ‘Beneficiary Principle’. Against the Beneficiary Principle, which holds that certain civilians’ immunity to being harmed as a side effect may be diminished by their prospects of benefiting from an intervention, I defend the view that beneficiary status is morally void when the benefit consists in the mitigation or prevention of wrongful harms. To the extent that individuals stand to benefit by virtue of being victims of wrongful threats of harm, as is the case in just humanitarian interventions, I argue, their expected benefit does not plausibly justify their exposure to additional risks of harm. I maintain that the BP is inadequate as a distributive principle for determining the permissibility of risk imposition, unless we disambiguate it in a number of critical aspects.

I begin by outlining the Beneficiary Principle in some detail (Section 2). In Section 3, I show that it can neither be plausibly grounded in a liability-based justification for harming nor constitute a lesser-evil justification in any conventional sense. I then consider whether the BP may provide a distinct justification for imposing risks on some individuals rather than others. However, I argue that a conception of the BP as a self-standing justification faces two critical challenges. The first concerns the BP’s applicability to collectives (Section 4), the second questions the normative significance of benefiting in the context of humanitarian wars (Section 5).

1.1 Preliminaries

First, although not strictly the same, I use civilians and noncombatants synonymously here.Footnote 6Bystanders are those individuals who are not involved in the conflict. They neither contribute to nor are affected by it, and do not stand to benefit from the intervention. They might be residents in the target party or individuals in neighbouring states who are in no danger. Beneficiaries or ‘to-be-liberated civilians’ are individuals in whose defence the intervention is carried out.Footnote 7 They are the feared victims of atrocity crimes who are usually, but not necessarily, civilians. They are expected beneficiaries insofar as the intervention’s aim is to protect them from harm they would otherwise suffer from unjust threats. Although persons other than potential victims of atrocities may actually benefit from an intervention, I use the term to refer to those individuals in whose protection the intervention is conducted. By ‘beneficiaries’ (or ‘prospective’ beneficiaries to emphasise this fact), I therefore mean intended or expected beneficiaries, as opposed to actual beneficiaries, unless otherwise specified.

Second, my concern in this paper is with the unintended infliction of foreseeable harm on innocent (i.e., non-liable) individuals. All we need to accept for the purposes of this discussion is that it is possible to foresee - recognise as possible and likely under certain circumstances - an outcome without intending to bring it about. My concern is with the permissibility of ‘acting in ways which knowingly put the lives of non-combatants at risk.’Footnote 8

Third, my concern is with the principles that determine what constitutes a just distribution of risks of harm. My aim here is not to give an account of when, or why, intervening combatants ought to bear greater risks to protect civilians.Footnote 9 This question is orthogonal to the question of which principles we should use to determine this.

2 The Beneficiary Principle

What I call ‘the Beneficiary Principle’ (BP) is a mosaic of corresponding claims advanced by several theorists.Footnote 10 It broadly submits that, when a determinate amount of risk can be borne by either beneficiaries, bystanders, or intervening combatants, it should be imposed on beneficiaries because they stand to benefit. A number of theorists have thus argued that prospective beneficiaries of an intervention are, by virtue of being beneficiaries, less immune to being harmed as a side effect than those who do not stand to benefit. On this view, the constraint against putting the lives of beneficiaries at risk is weaker than that against putting the lives of bystanders at risk.Footnote 11 What this means in practice, as Jeff McMahan has suggested, is that ‘[i]n trade-offs between harms to just combatants and harms to noncombatant beneficiaries, the immunity of the noncombatants is diminished by their status as beneficiaries.’Footnote 12 It is therefore permissible for intervening combatants to expose noncombatants to certain risks which intervening combatants could bear themselves, so long as those noncombatants nonetheless stand to benefit from the acts carrying the risk. As McMahan puts it:

'[i]t may not be wrong for combatants to fight in ways that involve a lower risk to themselves but expose noncombatants to new risks, provided that the noncombatants are nevertheless expected beneficiaries of the defensive action – that is, provided that the action's reduction of the risks they face from the original threat exceeds the risks to which the action itself exposes them.'Footnote 13

Gerhard Øverland has defended a similar position, placing the emphasis more explicitly on beneficiaries’ interests. On Øverland’s view, individuals who would benefit from an intervention have reasons for accepting its concomitant risks because it is in their own interest to be protected from the unjust threats posed by their government:

'The civilian population of a nation may accordingly have good reason to accept a certain level of risk if it means getting rid of the government. These civilians may have a lot to win by a humanitarian intervention; it is after all essentially done in order to help protect them from a bad regime. In such cases, if it means putting to-be-liberated civilian lives at risk, it could be permissible because it is in their interest that we do so'.Footnote 14

Kai Draper has appealed to a similar notion, according to which

'intended beneficiaries, even the ones who will ultimately be killed, are actually better off (in terms of expectable benefit) at the outset of the war, because, for each individual, the small chance of being killed as a side effect of the liberation effort is more than compensated for by the high likelihood of reaping the benefits of liberation.'Footnote 15

I use ‘the’ Beneficiary Principle', or ‘BP’, to capture the central idea which these arguments share: that individuals’ status as prospective beneficiaries may diminish their immunity to collateral harm and thus weaken the prima facie constraint against harming the innocent, even if only as a side effect.

None of the prominent accounts categorically distinguish between distributing risks between beneficiaries and bystanders on one hand and distributing risks between beneficiaries and just intervening combatants on the other. I therefore take the relevant distinction to be between beneficiaries and non-beneficiaries. Whether these non-beneficiaries are bystanders or just intervening combatants is, for the limited purposes of my argument here, irrelevant.

I, moreover, make no assumption that it is impermissible in general to expose beneficiaries to grave risks of harm in the course of an intervention, or to afford them only sub-optimal protection. It is not my aim to challenge the view that some innocent individuals’ claim on protection may legitimately be weaker than that of others. What I challenge is the view that their claim on protection is contingent on whether or not they stand to benefit from being liberated.

3 Justifications

None of the leading accounts specify what it is about benefiting that undermines innocent individuals’ rights against being harmed. Advocates of the BP invariably argue that beneficiaries, rather than non-beneficiaries, should be exposed to risks of collateral harm, because it is they who stand to benefit. But, to provide a real justification for infringing innocent individuals’ rights not to be harmed, this claim must be supported by a substantive theory which explains why their standing to benefit constitutes a justification for exposing them to risks of harm. The first question to consider is therefore - what kind of justification for harming might the BP offer?

Since justifications for harming in war usually appeal to the target’s liability or the lesser evil, there are three possibilities: First, beneficiary status might ground a liability-based justification for harming civilians who stand to benefit. Second, harming beneficiaries might be a pro tanto lesser evil than harming non-beneficiaries. Or, third, the BP may provide an altogether new and distinct justification which differs from both liability-based and lesser-evil justifications. The paper discusses each possibility in turn.

3.1 Liability?

Individuals can be liable to both intended and unintended harms. If advocates of the BP are right to claim that beneficiaries would not be wronged by being made to bear certain risks of being harmed as a side effect, one possible explanation for this is that beneficiaries are liable to bear these risks.Footnote 16 However, beneficiaries have done nothing, by virtue of standing to benefit from being liberated, to forfeit their right not to be harmed. This is supported by accounts on which even innocent threats do not make themselves liable to defensive harm. Many theorists deny that it is permissible to kill innocent threats in self-defence, on the grounds that there is nothing that morally distinguishes them from their victims.Footnote 17 But if what makes it impermissible to harm innocent threats and bystanders is their lack of responsibility for threats posed, a similar argument must apply to beneficiaries, whose standing to benefit is but an external fact over which they have no control. If circumstances for which individuals are not morally responsible cannot generate the type of moral asymmetry between innocent victims and innocent threats, which would make innocent threats liable to defensive harm, then, a fortiori, they cannot generate the relevant moral asymmetry between innocent beneficiaries and non-beneficiaries either.Footnote 18 Since beneficiaries have done nothing to forfeit their right against being harmed, they cannot have made themselves liable to be harmed.

There is an alternative conception of liability, according to which liability merely tracks the absence, rather than the forfeiture, of a right.Footnote 19 On this view, beneficiaries might lack a right not to be harmed without having done anything to forfeit it. Such a view, however, would merely beg the question as to why beneficiaries should lack the right to be harmed in the first place.

3.2 A Lesser Evil?

Whilst individuals who are liable to harm have by definition lost their immunity against being harmed, lesser-evil justifications acknowledge that individuals’ rights against being harmed remain intact, but assume that they can permissibly be infringed in order to prevent a significantly greater evil. Instead of relying on internal facts about individuals, such as their responsibility for an unjust threat, lesser-evil justifications aggregate external facts, such as the number of people to be saved or the gravity of harm to be prevented, although they are also sensitive to deontic features of actions.Footnote 20 Could expected benefit be part of such wide proportionality calculations, such that imposing risks of collateral harm on beneficiaries is a discernibly lesser evil than imposing the same risks on non-beneficiaries?

Consider the following two cases:

  1. 1)

    A unjustly threatens to kill B. You can stop A and save B in one of two ways. Option 1 inflicts 10 units of harm on B as a side effect of your defensive action. Option 2 inflicts 1 unit of harm on an unaffected bystander.

  2. 2)

    A unjustly threatens to kill 5 victims. You can stop A and save the five in one of two ways. Option 1 risks moderately harming each of the five as a side effect. Option 2 risks moderately harming one bystander as a side effect. The harm each of the victims and the bystander would suffer is the same.

It is not obvious how the BP should interact with requirements to minimise harm in wide proportionality calculations.Footnote 21 If the constraint against harming bystanders is stronger than that against harming beneficiaries, you ought to choose Option 1 in both cases - and, indeed, this is what McMahan’s account indicates.Footnote 22 If we take the BP to outweigh other considerations, harming a beneficiary severely will be preferable to harming a non-beneficiary minimally, and harming a large number of beneficiaries will be preferable to harming a small number of non-beneficiaries. This might strike us as counterintuitive. How beneficiary status should be weighed against other considerations in lesser-evil calculations is thus open to question.

Let us, however, set this question aside, and hold the degree of risk, gravity of harm, and numbers of individuals constant. All else thus being equal, is there any reason to think that imposing risks of collateral harm on beneficiaries qua beneficiaries is a lesser evil than imposing the same risks on the same number of non-beneficiaries?

An argument in favour of such a view might go as follows. Risks of harm to beneficiaries are automatically offset by their benefiting from having their chances of liberation and survival increased. Harms to presumptive beneficiaries may accordingly be discounted insofar as they are offset by beneficiaries’ increased chances of liberation and survival. Bystanders, on the other hand, do not stand to benefit. They would merely be made worse off, as harms suffered by them would not be offset by any benefits. We might therefore think that it is better to harm someone as a side effect of an action which might benefit her than to harm someone as a side effect of an action which can only benefit others. Another way to describe this thought might appeal to the notion of compensation: One might think that beneficiaries are automatically compensated for being exposed to risks of harm by having their chances of liberation and survival increased, whereas bystanders, who do not stand to benefit, would merely be harmed without being compensated for this. This is what Draper seems to have in mind when he argues that ‘for each individual, the small chance of being killed as a side effect of the liberation effort is more than compensated for by the high likelihood of reaping the benefits of liberation’, and what McMahan suggests when he argues that, ‘[w]hen a burden is borne by someone who on balance benefits from it, it is compensated for in a way that it is not when it is borne by someone who derives no benefit.’Footnote 23

I suggest that this argument succeeds only in a limited set of circumstances. These are circumstances in which the risk justified by the BP is of a compensable harm. I rely on David Rodin’s distinction between compensable and noncompensable harms here. What is crucial is that individuals whose rights are infringed thereby acquire a claim to compensation. When harms inflicted are compensable, matters are relatively straightforward, as the infringed right is simply ‘transmuted into a right to compensation’.Footnote 24 However, when harms are noncompensable - such as in the case of death, grievous bodily, injury, rape, or torture - this precludes the possibility of such a transmutation, because, in these cases, ‘the harm inflicted departs the system of rights altogether’, as Rodin puts it. The right that was infringed consequently cannot turn into a claim for compensation; as far as the rights-holder is concerned, it has, rather, been obliterated. As Richard Norman plainly put it, ‘there is truth in the cliché of the finality of death: once you’re dead, you’re dead, and nothing further can compensate for the loss.’Footnote 25 Insofar as rights rights infringements generate claims to compensation, constraints against rights infringements must be conceivably weighty in cases in which it is foreseeable that compensation ex post will not be possible.

  • Consider Cliff 1:

Perpetrator has pushed Victim off a cliff in an attempt to kill him, and has run away. Victim is now dangling from the edge of the cliff. Rescuer can save Victim in one of two ways. Rescuer can pull Victim up by one of his arms, which would break Victim’s arm, or Rescuer can yank Victim over the edge without harming Victim, which would knock over Bystander and break her arm.

In this case, the BP’s intuitive appeal is obvious. For Victim, having his arm broken is better than plummeting to his death - a broken arm is a very minor cost to pay for having his life saved. Whilst Victim derives a significant overall benefit from the rescue despite the broken arm, Bystander would merely be harmed. It is therefore easy to see why breaking Victim's arm is preferable to breaking Bystander's arm. However, my contention is that this argument only works because the harm of a broken arm is offset by the benefit of surviving.

  • Consider Cliff 2:

Perpetrator has pushed Victim off a cliff in an attempt to kill him, and has run away. Victim is now dangling from the edge of the cliff. Rescuer can save Victim in one of two ways. The first option comes with a 50% risk of Rescuer losing hold of Victim and Victim consequently plummeting to his death. The second option comes with a 50% risk of dislodging a rock that will kill Bystander who is standing at the bottom of the cliff.

In this case, the BP’s appeal seems significantly weaker. It is not obvious why imposing a 50% risk of death on Victim would be preferable to imposing a 50% risk of death on Bystander, given that both are equally innocent. After all, since the harm Victim would suffer if the risk eventuated is not compensable, there is no benefit that could offset the harm of plummeting to his death. If the risk imposed on beneficiaries is one of noncompensable harms, the assumption that harms to beneficiaries are automatically offset is therefore false.

Since rights infringements give rise to claims of compensation, beneficiary status can only plausibly constitute an element in lesser-evil calculations when the risk justified by the BP is that of a compensable harm. This is because beneficiaries only derive a benefit that could offset the harm suffered in these cases. McMahan similarly observes that it is clearer that beneficiaries, rather than bystanders, should be made to bear the costs of their defence in cases in which beneficiaries actually derive a net benefit rather than being made worse off.Footnote 26 If my argument is right, this is an understatement. That beneficiaries should be made to bear these risks is clear only in cases in which they are actual net beneficiaries. With this modification, the BP holds.

The problem, of course, is that the majority of harms inflicted in armed conflict are of the noncompensable kind. We should therefore consider a second possibility. One might argue that imposing risks of harm on prospective beneficiaries is a lesser evil than imposing it on non-beneficiaries, insofar as individuals who stand to benefit from the risky action have reason to consent to bearing the risk, whereas non-beneficiaries, who would risk being harmed without standing to benefit, have no reason to do so. But such a consent-based view would rest on the premise that beneficiaries’ likelihood of benefiting is increased to the extent that they bear the concomitant risks of harm. That is, it would assume that the risk of harm and chance of benefiting are intertwined in such a way that beneficiaries’ bearing the risk of being harmed is a necessary condition for their chances of benefiting to be increased.Footnote 27 If this were the case, beneficiaries would indeed have strong reasons to consent to the risk. However, the risk and the benefit need not fall on the same individuals. For beneficiaries’ chance of surviving to be improved, it is only true that someone must bear the risk of being harmed as a side effect; it is not necessary that the risk be imposed on them. In fact, their chances of benefiting would be increased to a greater degree if the risk was borne by others. If what matters is what beneficiaries have reason to accept, then the fact that they have stronger reasons to accept an alternative in which others shoulder greater risks refutes the above assumption.Footnote 28 Since individuals need not suffer a new risk for the original risk to be reduced, their consent to the new risk cannot be presumed. To the extent that a lesser-evil version of the BP would rely on presumptive consent, then, it would not hold. Of course, it might still be fairer to impose the risk on beneficiaries for other reasons, but this would not be explained by beneficiaries’ presumed consent.

What the analysis has revealed so far is that neither liability nor lesser-evil justifications can convincingly explain what it is about standing to benefit that weakens the prima facie constraint against harming those innocent individuals who are prospective beneficiaries. This leaves us with one more possibility. Rather than falling into one of the standard categories of justification for defensive harming, the BP might provide distinct grounds for the permissibility of causing collateral harms, and for protecting some civilians less effectively - grounds which differ significantly from both liability-based and lesser-evil justifications. The remainder of this paper examines how plausible this is.

4 The Problem of Collective Risk

One challenge to the BP as a standalone justification results from what we might call the ‘problem of collective risk’. Although the BP’s intuitive appeal is obvious in cases in which an individual is exposed to some risk of harm in her own interest and to her own benefit, it does not apply as straightforwardly to cases in which a group is exposed to some risk of harm, which will benefit most of its members while making others worse off.

Consider first a single-person case: Aggressor culpably threatens to kill Victim. You can save Victim in one of two ways. You can either break Victim’s arm as a side effect of rescuing him, or you can break Bystander’s arm as a side effect of rescuing Victim. According to the BP, you ought to choose the option that will harm Victim as a side effect rather than Bystander, because Victim will still be overall better off with a broken arm, whereas harming Bystander as a side effect would only make Bystander worse off.Footnote 29

Let us grant that this is right. In this scenario, imposing the risk of being harmed as a side effect of rescuing Victim on Victim himself is fairer than imposing it on Bystander. Given that the harm of a broken arm is offset by the overall benefit, the BP offers a perfectly plausible explanation for why some inevitable amount of harm should be inflicted on Victim as the beneficiary of the defensive action, rather than on Bystander. What is crucial is that Victim will, as a matter of fact, still be better off after suffering the harm concomitant with the rescue than he would have been in the absence of an intervention.Footnote 30

The application of this principle is much less straightforward when we are dealing with multiple intended beneficiaries, some of whom will foreseeably be made worse off. In his discussion of the 1999 Kosovo intervention, McMahan suggests that, because ‘Albanian Kosovar civilians were the expected and, as a group, actual beneficiaries of the intervention’, ‘NATO combatants were therefore entitled to shift some of the costs of the intervention to those who would, unlike themselves, benefit from it’.Footnote 31 On this view, the permissibility of exposing a group of to-be-liberated civilians to risks of lethal harm is grounded in the expectation that the group will derive a net benefit from being rescued - even if this means that some individuals will actually be made worse off. But, of course, killing some members of a group as a side effect of rescuing other members of that group is not the same as breaking an individual’s arm as a side effect of saving his life. To ascribe beneficiary status to collectives in the way McMahan’s discussion suggests, we would have to presume some Walzerian conception of a ‘collective life of the community’, on which killing some individuals as a side effect of rescuing the rest of the group would, in all relevant respects, be like injuring an individual as a side effect of saving his life.Footnote 32 But we can safely assume that this is not what McMahan has in mind. He has convincingly - and rightly - argued against the collectivisation of moral status ‘in a way that is independent of what individuals are actually doing’.Footnote 33 Problematically for advocates of the BP, however, this is precisely what it does in stipulating that what justifies unintentionally harming certain individuals is their expected membership in the vaguely defined group of beneficiaries. Collectivist iterations of the BP thus run the risk of conflating benefiting from being rescued with group membership.

As McMahan has argued, [t]he idea that people can be liable to attack, or immune from attack, merely by virtue of their membership in a group . . . is both false and morally repugnant.’Footnote 34 I contend that the same reasons which ground McMahan’s objections to collectivism must bear on the BP’s applicability in collective cases. If individuals’ immunity to harm can only be determined by their own actions and decisions, and not by their involuntary membership in some morally arbitrary group, then belonging to the collective of expected beneficiaries should have no bearing on their claim to protection. This is also why beneficiaries do not, by standing to benefit, make themselves liable to harm, as I argued earlier. Their beneficiary status is entirely independent of their moral agency. As victims of unjust threats, they are beneficiaries through no choice or fault of their own. If we are to take seriously individual rights and the separateness of persons, then incidental membership in the collective of ‘expected beneficiaries’ does not plausibly diminish individuals’ immunity to collateral harm.

Here is another way of putting the issue: If what justifies risk imposition is the expectation that the individuals exposed to this risk will be better off as a result, then it is a problem that some individuals will actually be made worse off. We are thus dealing with a different type of risk in this case. Whilst the uncertainty in the single-person case concerns whether a particular individual will be harmed if she is exposed to a certain risk, the uncertainty in the collective case concerns which individuals, and how many, will be harmed or killed. In cases in which not all intended beneficiaries will actually benefit, and some will foreseeably be noncompensably harmed, the BP's intuitive appeal is severely weakened by the problem of collective risk..

4.1 The Ex Ante Contractualist Response

Proponents of the BP might respond to the problem of collective risk (PCR) as follows: What matters is not that each individual is actually made better off ex post, but merely that individuals’ chance of benefiting is improved ex ante.Footnote 35 Some of McMahan’s and Øverland’s formulations point in this direction.Footnote 36 On this view, the relevant ‘benefit’ consists in the mere ex ante reduction of individuals’ overall risk of being harmed, rather than an actual ex post benefit. What matters, accordingly, is the ability to justify the risky action to every potential beneficiary ex ante.

We might conceive of the BP as what Johann Frick calls an ex ante rule, ‘the adoption of which, at some time t1, is in everyone’s individual interest, but which licenses or requires some agent to act at a later time t2 in a way that benefits some but significantly burdens others.’Footnote 37 Accordingly, before the rescue operation (at t1), it is in all beneficiaries’ individual interest that it be carried out; however, it licenses combatants during the rescue operation (at t2) to adopt means, such as dropping bombs from high altitudes with limited precision, which will harm some individuals as a side effect. What distinguishes ex ante rules, and what distinguishes the ex ante justification of the BP from its ex post version, is, in Frick’s words, that ‘[t]he locus of social risk . . . is not in the effects of the agent’s actions at t2, which may be perfectly predictable. Rather, it is the result of not knowing, ex ante, whether one will be among those benefited or burdened by the application of the rule at t2.’Footnote 38 This second claim is crucial: ‘not knowing, ex ante, whether one will be among those benefited or burdened.’

This condition may hold the key to defusing the PCR. Since individuals do not know whether they themselves will actually benefit or be made worse off, they have reason to consent to incurring the risk concomitant with the rescue operation. According to such a defence of the BP, it is the uncertainty about which individuals will be made better or worse off which grounds the justification of additional risk imposition to all affected individuals, including ex post non-beneficiaries.

Proponents of the BP could thus claim that prospective beneficiaries face a gamble and have reason to accept a comparatively small risk of being killed as a side effect of a military intervention in order to have a larger chance of surviving the unjust threat posed by, for instance, a genocidal regime. A contractualist ex ante defence of the BP could thus appeal to the notion that it is in beneficiaries’ own interest that they be exposed to additional risks of being harmed as a side effect of a rescue operation which would decrease their overall risk of harm, and that it is therefore permissible to impose this risk on them.Footnote 39 Although some of those individuals will actually be severely harmed or killed as a side effect, this does not compromise the BP’s ex ante justifiability. So long as it is uncertain which individuals will be made worse rather than better off, no particular individual has a reason for complaint, and it will be in each beneficiary’s own interest to accept the risk.Footnote 40 On this view of the BP, the PCR poses no real challenge, as it is irrelevant whether and for whom the risk eventuates ex post. The ex ante likelihood of any person benefiting from the risky action is, on this view, more important than the foreseeable outcome that some individuals will be made worse off.

Although this seems a powerful defence, there are several reasons for why we should remain cautious about endorsing the BP on these grounds.

4.2 Two Worries

My first concern is that the BP as an ex ante rule is best understood as a theory of a fair distribution of risks amongst a fixed, specified group of individuals, and not as a justification for imposing risks on some individuals rather than others. The BP’s ex ante appeal derives from its potential to present a fair distribution of risk amongst a given group of innocent individuals, but this is not the same as providing a cogent justification for exposing certain individuals to greater risks than others. The fact that victims’ identities are unknown behind the contractualist veil of ignorance not only explains why no person has a reason to reject the risky action, but it also fulfils some distributive ideal by treating morally indistinguishable individuals as having morally equal statuses. If the choice is between, on the one hand, harming a particular person while not harming nine others, and, on the other hand, exposing all ten people to a 1/10 risk of harm, ex ante contractualism is right that the latter option is preferable, assuming that all ten are equally morally innocent. But this is fairer only because there is no particular person who is less immune to harm than any of the others. Exposing everyone to the same 1/10 risk of harm therefore brings about a more just distribution of risk on egalitarian grounds. Choosing the option in which we do not know who will be harmed is the right thing to do in these cases because this allows us not to violate any particular person’s right not to be harmed. But not knowing who will be harmed is both what makes it possible for there to be no one who could reject the principle on personal grounds, and what makes everyone equal in terms of facing the same degree of risk. In this light, the BP’s primary ex ante appeal seems to lie in a desirable side effect - that of justly distributing risks amongst innocent individuals - rather than in the purpose it claims to fulfil - that of justifying why certain innocent individuals, rather than others, can be made to bear risks in the first place. Rather than giving grounds for why beneficiary status may diminish expected beneficiaries’ immunity to collateral harm, it suggests that a fair distribution of risks is one that equalises chances of benefiting for all individuals. It therefore fails to show why beneficiary status should diminish individuals’ immunity to collateral harm.

The second worry is that ex post considerations must enter into deliberations about what is permissible in contexts of war. Taking seriously the requirement of proportionality, for instance, requires interveners to take into account impersonal considerations such as predictions that a certain number of non-liable individuals will be killed. However, a contractualist view of the BP would prohibit appeals to impersonal principles such as the expectation that there will be unspecified non-liable individuals who will be killed as a side effect.Footnote 41 Moreover, it is a standard requirement of the jus ad bellum that belligerents be committed to establishing a just peace once hostilities have ceased, which requires interveners to take into account the ex-post justifiability of their actions. There are, then, important reasons for why, even as non-consequentialists, we should be concerned with the outcomes of interveners’ actions. Ex-post justifiability must therefore at least inform ex ante justifications.

5 The Normative Insignificance of Benefiting

My most substantial concern about the BP is that it overstates the normative significance of beneficiary status. The BP is ultimately insensitive to how individuals come to be in a situation where they would benefit from being rescued. What the BP consequently neglects is that the benefit consists in the prevention of a wrong which beneficiaries have a right not suffer in the first place. After all, the beneficiaries with whom the BP is concerned are by definition victims of wrongful threats of harm, whose rescue constitutes the very purpose of a just intervention. I submit that this renders beneficiary status ineffectual in diminishing individuals’ immunity, insofar as the benefit consists in the prevention or mitigation of a harm against which beneficiaries have a right.

I argue that it would be implausible and incoherent to ground the permissibility of exposing beneficiaries to further risks of harm in their beneficiary status, since they are beneficiaries only on account of being victims of wrongdoing. The wrongness of the threat of which beneficiaries are victims must counterbalance the benefit they will derive from not suffering the wrongful harm. If the fact that they are beneficiaries matters, so must the fact that they are such only as victims of injustice. This is because the notion of benefiting is comparative: beneficiaries are better off as a result of benefiting than they would have been in the absence of the benefit. The problem is that the BP treats the counterfactual against which the benefit is measured, i.e., the alternative in which threatened individuals are not rescued, as a valid baseline for the comparison. However, this baseline is flawed, insofar as it describes a counterfactual in which individuals are severely wronged. To the extent that this baseline against which the benefit is compared constitutes a wrong, I contend that it is misguided to treat ‘benefit’ as morally relevant. What strikes me as misguided is the presumption that the benefit of not having one’s rights violated can be taken to diminish one’s immunity to harm. Yet, this is precisely what the BP does. By maintaining that individuals’ beneficiary status justifies imposing additional risk on them, the BP essentially presumes that it is the very fact that those individuals have already suffered unjust threats of harm which permits exposing them to further risks of being harmed.Footnote 42 However, it is exceedingly difficult to see how individuals’ rights against being harmed could effectively be vitiated by the very fact that they are already victims of injustice.

It seems more plausible to assume that, if individuals would not benefit but for the fact that they are victims of injustice, the fact that they stand to benefit from not suffering the injustice should play no role in determining the permissibility of exposing them to further risks of harm. In this case, after all, the BP cannot plausibly provide an independent justification for imposing risks of collateral harm on them qua beneficiaries. Standing to benefit from the mitigation or prevention of an injustice should therefore be extraneous to innocent individuals’ moral status, contrary to what the BP postulates. McMahan’s assumption that beneficiaries must ‘bear the costs of their own defence’ is consequently open to doubt in cases in which individuals’ rescue is necessitated by the fact that they are, through no fault of their own, victims of wrongdoing.Footnote 43 It is therefore not obvious that they should be under an enforceable obligation to bear the lethal risks of their own rescue.

This argument requires some qualification in light of my earlier point that the BP holds when the risks it justifies are of a harm that can be offset. Specifically, my account must explain how these two considerations - whether harms can be offset, and whether beneficiaries are victims of wrongdoing - cohere.

In cases in which the risky act provides an offsetting benefit to the harm it inflicts, it is inconsequential whether beneficiaries are victims of injustice. As I explained earlier, when individuals would still derive a net benefit from an albeit harmful rescue, this distinguishes them from bystanders in such a way as to justify imposing the risk on them rather than others. If the harm risked is thus offset by the benefit, interveners have a BP-based justification for imposing those risks.

But it is for this same reason that the argument fails in cases in which the risk justified by the BP is of a harm which cannot be offset. In these cases, when individuals would not stand to benefit but for the fact that they are victims of wrongdoing, this negates the weight we could otherwise ascribe to the benefit they would derive from being rescued. In this case, the BP is undermined by the fact that the beneficiaries are such only as victims of injustice, because it relies on the premise that they will be better off than they would have been in the absence of an intervention even if the risk eventuates. However, as I argued earlier, this comparison does not hold for victims of injustice who are exposed to risks of harm which will not be offset. Since imposing risks of harms on non-liable individuals constitutes a rights infringement which gives rise to claims to compensation, and noncompensable harms cannot be offset, the claim that beneficiaries were still better off despite being thus harmed is both false and invalid. It is false because individuals will be overall worse off, and it is invalid insofar as the baseline of the comparison describes their suffering wrongful harms against which they have a right. The BP thus fails on at least two counts in cases in which beneficiaries are victims of wrongful threats and the risky act provides no offsetting benefit, as is usually the case in humanitarian intervention.

Recall the earlier example of Cliff 2, in which Victim has been maliciously pushed off the edge of the cliff by Perpetrator, and you can either pull him up and impose a 50% risk on him of being killed, or you can pull him up and impose a 50% risk of being killed on Bystander. On the view I advanced, the fact that Victim stands to benefit does not set him apart form Bystander in a decisive way, because the baseline of the comparison, Victim being killed, is the result of a wrong he has a right not to suffer. An adequate response, in this case, would be to flip a coin, or to appeal to an alternative principle which does not rely on Victim’s beneficiary status.

Consider now cases in which harms cannot be offset and beneficiaries are victims of misfortune, rather than wrongdoing. If Victim had simply tripped and fallen over the edge of the cliff, would this change anything? On my view, it would. It is true that, in most cases, reasons we have for prioritising victims of wrongdoing over victims of misfortune are plausibly outweighed by considerations such as the severity of harm to be averted or the number of persons to be saved.Footnote 44 However, this is compatible with the view that the duty to aid victims fo injustice is weightier than the duty to aid victims of misfortune when all other things are equal.Footnote 45 As Tom Parr has argued, if there is a morally significant difference between harmless wrongdoing and other non-harmful events, this indicates a difference between harmful wrongs and harmful misfortune.Footnote 46 We might, moreover, think that there is a consequential difference in ‘deontic badness’ between being a victim of someone’s wrongdoing and being a victim of bad luck.Footnote 47 In addition, rights violations plausibly generate different rectificatory obligations from wrongless harms. To the extent that these considerations set victims of wrongdoing morally apart from victims of misfortune, my argument that the benefit of not suffering a harm ought to be discounted only applies to the former. In Cliff 2, accordingly, if Victim merely tripped and fell, it would be permissible for Rescuer to apply the BP and choose the course of action which imposes a 50% risk of killing Victim in the process, despite the fact that, were the risk to eventuate, there would be no overall benefit to offset it. This is because the baseline of the comparison, one in which Victim dies, is not the result of wrongdoing.

Bear in mind that nothing in the view I have defended suggests that beneficiaries have a greater claim to protection than non-beneficiaries.Footnote 48 All I have argued is that beneficiary status does not tip the balance between beneficiaries and other non-liable individuals when beneficiaries only benefit by virtue of being victims of injustice. The upshot of this is that, when all other things are equal, beneficiaries’ claims to protection are just as strong as those of bystanders. Beneficiaries would consequently have the same complaint as innocent bystanders if they were made to bear the risks concomitant with an action aimed at preventing or mitigating a wrong they have a right not to suffer.

6 Conclusion

The aim of this paper was to throw doubt on the claim that civilians’ beneficiary status plays a decisive role in grounding interveners’ permission to expose them to risks of collateral harm in armed humanitarian interventions. Once we look closer, I argued, this argument for the limited immunity of beneficiaries fails to convincingly show why innocent individuals’ immunity to harm should be diminished by virtue of their expected or intended benefit from an intervention. However, I suggested, once we disambiguate certain aspects, we can delineate the limited conditions in which beneficiary status does affect the permissibility of risk imposition, although this is unlikely to be the case in armed interventions, as many harms inflicted in their course are noncompensable.

None of what I have argued means that it is never permissible to impose risks of harm on intended beneficiaries of humanitarian intervention. What I have argued is that it is not their beneficiary status on which their immunity depends. Ensuing questions of when beneficiaries would be wronged by being exposed to risks which could be otherwise distributed, and when they might, as a consequence, be permitted to engage in defensive harm against interveners must await another occasion.