One of the most commonplace worries about humanitarian intervention relates to the perverse incentives that it might create, or the adverse reactions that it might provoke. For instance, it is sometimes said that by weakening the norm of sovereignty humanitarian intervention can encourage unscrupulous states to wage aggressive wars of self-interest using human rights as a pretense. It is feared, in other words, that humanitarian intervention—even when it has the purest motives—might ultimately do more harm than good by inciting unwanted reactions from other states or substate groups. I will refer to these kinds of knock-on effects as the mediated consequences of intervention.Footnote 1 They are brought about via the interceding agency of parties other than the intervener.
It is generally assumed that when judging the proportionality of a humanitarian intervention, these consequences must be factored into the equation. If an intervention is expected to provoke adverse reactions the accumulated costs of which will outweigh the benefits that the intervention will deliver, then the intervention is thought to be disproportional and, therefore, unjustified. I want to challenge this assumption. I begin by considering what the principle of proportionality can reasonably demand of rebels who are defending their own basic rights against an oppressive government. I argue that a rebellion in such circumstances cannot plausibly be rendered impermissible solely by the expectation of negative mediated consequences, even when those consequences outweigh the anticipated benefits of the rebellion.Footnote 2 This would seem to imply that rebels may discount mediated consequence from the proportionality calculus. But if this is so, do we have sound reasons for withholding the same prerogative from humanitarian interveners pursuing similar ends and using similar means? Can we justify asymmetric standards of proportionality?
If a humanitarian intervention is expected to produce less favorable consequences than a rebellion in the same circumstances—to accomplish less than rebellion would, to cause more death and destruction—then the intervention might obviously be ruled out by the principle of proportionality despite rebellion being a legitimate option. This much is uncontroversial. But if asymmetric standards of proportionality apply to these two classes of belligerents—if we can justify demanding something more of interveners—then an intervention might fail to satisfy the principle of proportionality despite its expected costs and consequences being identical to those of an insurrection that is rightly judged to satisfy that principle. Whether this curious position stands up to critical scrutiny is the question I take up here. I will explore three potential justifications for asymmetric standards of proportionality and show that none of them is entirely persuasive.
Before we begin, however, a clarification is in order. I will concentrate on cases where a decision to use force is expected to have adverse mediated consequences for people other than the intended beneficiaries of that decision. Imagine two cases of rebellion. In the first, the act of rebellion provokes a brutal reaction from the targeted government. The reprisals are directed at a segment of that government's own population—specifically the persecuted group of people whom the rebels represent. In the second case, a rebellion in one country provokes the dictator of another country to purge his population of citizens perceived as potentially rebellious out of fear of suffering a similar fate. In the first scenario, the victims of the mediated consequences actually stand to benefit from the success of the rebellion. They might even consent to it, despite the possibility of such reprisals, and be willing to endure the risk for the sake of a successful overthrow of their oppressors. In the latter scenario, by contrast, the bearers of the mediated harms—those purged by their fearful dictator—are not represented by the foreign rebels who incited the dictator's fear, and are unlikely to benefit from the success of that rebellion in the same way. This factor may be morally significant, but for the purposes of our discussion I want to neutralize it. Thus, I will focus only on cases of rebellion and humanitarian intervention that are expected to have unwanted mediated consequences for people other than their intended beneficiaries. As mentioned, many commonplace worries about humanitarian intervention revolve around precisely these kinds of consequences. A routine objection is that intervention begets more intervention, thus jeopardizing international peace and stability. Some argue that an act of unilateral intervention can set a precedent and/or erode the legal prohibition against such actions enshrined in treaty law and the UN Charter. This in turn makes other states more likely to wage war for political and economic gain using the pretense of “humanitarian intervention.” One state's decision to engage in unilateral action can therefore have reverberations that adversely affect people elsewhere, now and in the future.
Rebellion, too, can have far-reaching mediated consequences. Take the Arab Spring. Scenes of the successful overthrow of the Tunisian regime inspired the disaffected citizens of Egypt to rise up against their own government. The contagion—propelled by satellite television, social media, and other forms of modern communication—quickly spread to Libya, Yemen, and Bahrain. In Syria, President Bashar al-Assad (not unlike the fearful dictator described in the hypothetical case above) continues to brutally repress his citizens in a desperate attempt to avoid becoming the next domino to fall. Some have warned of the possibility that the violence and repression may spread beyond the Arab world—to sub-Saharan Africa, and even to China.Footnote 3
Asymmetric Standards of Proportionality
Unilateral humanitarian intervention is often said to weaken the norm of state sovereignty and to thereby encourage self-interested actors to “licence aggression with fine words” to carry out wars of political and economic aggrandizement masquerading as altruism.Footnote 4 Those countering such objections tend to attack the empirical premise that intervention breeds more intervention and more war. Some suggest this line of argument exaggerates the efficacy of international laws and norms on state behavior.Footnote 5 What is not denied is the proposition that foreseen knock-on effects or mediated consequences of this sort can act as a moral barrier to otherwise justified intervention. It is simply assumed that if and where an intervention can confidently be expected to trigger a series of actions by other parties whose accumulated costs will outweigh the benefits of that intervention, then it is morally unacceptable to proceed. The freedom of states to wage war in defense of human rights beyond their borders is thus constrained by how other players in the international arena can be expected to react.
And apparently it is not only the reactions of other states that must be taken into account but also the reactions of substate groups. The interventionist policies of states are sometimes said to give these groups a powerful incentive to act irresponsibly, essentially by extending them a kind of insurance coverage. Insurance tends to encourage careless, and in some cases fraudulent, activity. If my car is not insured I will take every precaution against theft and criminal damage—such as parking the car in a garage rather than on the street, activating the engine immobilizer, and so on. Once my car is insured against theft and vandalism, however, I am less likely to bother with these precautions. Furthermore, if my car is over-insured I have an incentive to deliberately set the car on fire in order to receive the insurance company's payout. At work here is the phenomenon of “moral hazard,” and some say that humanitarian intervention creates it.Footnote 6 If those considering a violent rebellion believe that their success is unlikely, they might decide not to rebel. But if they expect that their violence will provoke brutal retaliation, and if this retaliation is in turn expected to trigger a foreign intervention on their behalf, then rebellion might become a viable option.
Alan Kuperman is convinced that moral hazard was behind the Bosnian war of independence.Footnote 7 In June of 1991 Croatia seceded from Yugoslavia. Belgrade responded belligerently, and UN peacekeepers were ushered in to restore order, which effectively protected Croatia's newly established independence. This galvanized the Muslim Bosniaks—who were not, according to Kuperman, the victims of serious oppression or discrimination—into following suit. They assembled a militia and declared independence the following year, knowing full well that this would similarly provoke violent retaliation from Belgrade. The hope was that the Serb onslaught would offend the moral sensibilities of Western onlookers and bring in foreign forces that would ultimately secure Bosnian independence. In 2000, Bosnian foreign minister Haris Silajdžić admitted: “My main priority in the whole strategy was to get Western governments and especially the U.S. to get involved, because [the Serbs] had the whole army.”Footnote 8
Indeed, the international community eventually took the bait. The United States delivered weapons to the Bosniaks. Then NATO launched a bombing campaign against the Bosnian Serbs on the Bosniaks’ behalf. But the Bosnian Muslims were not successful in achieving a unitary Bosnian state. In 1995 the Dayton Accords were signed, and the Bosniaks agreed to a division of the country similar to that which they had earlier rejected in favor of a war that left 2 million displaced and 150,000 dead. Had it not been for the hope of intervention, engendered by the earlier action on behalf of the Croats, Kuperman suggests, this settlement would likely have been accepted to begin with.Footnote 9 The UN intervention in Croatia might therefore have benefited the Croats, but it created a perverse incentive for the Bosniaks and provoked reactions whose consequences were disastrous for many others.
Those commentators who accept that moral hazard influences the use of political violence unanimously believe that a state's policy on intervention should be responsive to and constrained by the expected effects of the phenomenon. Kuperman himself proposes that a policy of nonintervention be adopted toward persecuted peoples who provoke their own persecution by rebelling, as this will preclude moral hazard from taking effect. The feeling, then, is that we should not attempt to rescue foreign nationals from human rights abuses if this will provoke reactions whose overall costs will outweigh the benefits that the rescue will deliver, whether the interceding actions are those of other states, substate groups, or both. Humanitarian interveners are taken to be morally restricted by mediated consequences.
What I want to suggest here is that subjecting rebels to the same constraint has strongly counterintuitive implications. Imagine a group of citizens who are being severely persecuted and face mass murder or torture or enslavement at the hands of their own government. Suppose that they have just cause for rebellion and a strong prospect of success at an acceptable cost: the government is expected to fall quickly with minimal casualties on both sides and no major disruptions to ordinary civilian life. If one looks only at the direct and immediate consequences, the rebellion is clearly proportional. But suppose that the rebellion would incite an unjust and extremely costly civil war in a neighboring country, or that it would encourage a foreign dictator to start killing his own citizens, as in the earlier example. Finally, suppose further that these remote consequences are so bad that they would render the rebellion disproportional if taken into account.
My intuition in such a case is that citizens facing violations of their basic human rights still retain their right of self-defense, irrespective of how others will react to the exercise of that right. Morality cannot demand that they be led like lambs to the slaughter. Perhaps not everyone will share this intuition. One might insist that rebellion is indeed impermissible where such disproportional mediated harms are foreseen. I do not propose a counterargument to this line of thought. It is not the conclusion of my inquiry that rebels are not morally constrained by mediated consequences; it is the premise that I am starting with, and it rests purely on an intuition that I take to be fairly widely shared and forcefully felt. Intuitively, it seems that while a rebellion might contribute to instability and conflict elsewhere in the world, this cannot consign the victims of severe oppression to passivity. The rebel therefore must have the prerogative to discount mediated consequences when making judgments about proportionality.Footnote 10
At this point one might be tempted to interject that the oppressed subject is merely excused when he rebels despite the expectation of excessive mediated harms, but that he is not justified in doing so. In some circumstances we are prepared to concede that an agent is not blameworthy despite having performed a wrongful action—for example, when the misdeed is committed out of necessity or under duress or irresistible coercive pressure. This, one might say, explains our intuitions in relation to oppressed subjects who rebel knowing full well that disproportional mediated harms will follow. They do wrong, but for obvious reasons their wrongdoing is not culpable. The claim that the rebel is merely excused, however, still strikes me as unacceptable. One might have an excuse in addition to being justified, where this simply means that excusing conditions obtain—conditions that would serve to exculpate the agent in the absence of a justification.Footnote 11 The rebel clearly “has an excuse” in this sense. He acts under extreme coercive pressure. But to maintain that the rebel is excused entails that he is not justified, and this implies that he does not have the right to defend himself, that he is not entitled to act where there is an expectation of disproportional mediated harms.Footnote 12 If this is something we are not prepared to accept, then the rebel must be justified, not merely excused.
Why, then, are humanitarian interveners not justified in acting under the same circumstances—namely, where there is an expectation of mediated harms that would make the intervention disproportional if counted, but where the intervention is otherwise proportional? Do we have any good reason for holding humanitarian interveners to a different ad bellum standard than that to which we hold rebels? If so, then an intervention can be disproportional, and therefore illegitimate, despite accomplishing no less, and costing no more, than a rebellion that is rightly judged to be proportional. But if not, then we may need to reconsider whether a humanitarian intervention can be morally delegitimized by the expectation of disproportional mediated harms alone. In what follows, I will explore three possible justifications for applying asymmetric standards of proportionality to rebels and humanitarian interveners.
Common Exploitation
Legal scholars have long puzzled over cases in which an action brings about an injury, but only via some interceding event or act. In some circumstances the interceding event is judged to free the initiating agent from responsibility. In others, the harmful consequences are traced back through the interceding events or successive agencies such that the first link in the chain is held morally (sometimes even causally) responsible. Consider the following: Andy flicks a lit cigarette out of the window of his car as he speeds down the highway. The cigarette sets fire to a patch of grass on the side of the road. A gust of wind then carries the fire to a populated area nearby, destroying several properties. Had it not been for the gust of wind the fire would have flickered out harmlessly. Nevertheless, we insist that Andy caused the fire, and we hold him legally and morally responsible for the damage. Shortly after Andy arrives home he gets into an argument with his brother Barry over control of the television remote. Tempers flare and Andy shoves Barry onto the hardwood floor. At that very moment a tree crashes through the roof of the house and lands on Barry, killing him instantly.
Andy's action is a “but for” cause of his brother's crushing death (that is, but for the shove, the death would not have happened), but we do not hold Andy responsible for it in the same way we hold him responsible for the fire damage. In both cases Andy's action constitutes a sine qua non of a bad outcome. In both cases a natural event intercedes between the action and the outcome. Only in the latter case does this free Andy from responsibility. According to some legal theorists, the relevant difference is that gusts of wind are a “common recurrent feature of the environment.”Footnote 13 On the other hand, a tree suddenly crashing through the roof of a house and landing in the exact spot where its occupant is positioned is an “abnormal” event or a coincidence that is freakishly unlikely by ordinary standards.Footnote 14
A similar rule of thumb seems to guide legal decisions about whether an initiating agent is responsible for harms brought about via interceding human actions. As with natural events, their likelihood seems to be an important factor. Take the case of Stansbie v. Troman. The plaintiff's house was burgled after the defendant—a decorator who had been employed there—failed to lock up after leaving. The court held the decorator liable for the plaintiff's losses, despite the interceding agency of the burglar. H.L.A. Hart and A.M. Honore appeal to the notion of common exploitation in defending the judgment. Where an opportunity is “sufficiently commonly exploited” for wrongdoing, the provider of the opportunity cannot divest himself of responsibility for the mediated consequences that arise.Footnote 15
Compare this to the case of Lamb v. Camden London Borough Council. In 1973 local council workers for the London suburb of Hampstead fractured an underground water main while replacing a sewer. The escaping water weakened the foundation of the plaintiff's house, caused the walls to crack, and rendered the house unsafe. This forced the tenant to move out. The unoccupied house was invaded by squatters shortly thereafter. The squatters caused massive damage, pulling off the paneling for fuel and ripping out the central heating and other installations. The owners of the house sued the city council and its contractors not just for the cracked walls and weakened foundation but also for the damage caused by the squatters.
While there had been instances of squatting a mile or so away from the plaintiff's house, neighbors were reportedly “very shocked” to learn that it had occurred in this particular part of Hampstead, which was of a very different class and character than those parts of town that squatters had previously targeted. Thus, the court found that the invasion was not a probable or likely consequence of the damage caused by the fractured water main, and that the harm caused by the squatters was therefore “too remote” from the actions of the city council for it to be held liable.Footnote 16
I cannot do justice to all the details that contributed to these contrasting decisions, but the important insight for our purposes is just that, generally, the more “abnormal” an intervening event or act, the more likely the initiating agent will be freed from liability for the consequences.Footnote 17 A legal or moral code that holds people responsible for things that are not “reasonably foreseeable” is considered intolerably demanding, and clearly one cannot reasonably foresee abnormal events or freak occurrences. In upholding the decision of Lamb on appeal, Lord Justice Oliver explained that, given the unlikelihood of the squatter invasion, while it was foreseeable as a possibility, it was not reasonably foreseeable. On this standard, only a relatively probable reaction to one's conduct is reasonably foreseeable.Footnote 18 This is why interceding events or acts that are highly unlikely are judged to break the chain of causation and responsibility.
Can this justify asymmetric standards of proportionality for different classes of belligerents? Can we say that foreign intervention is, while rebellion is not, “sufficiently commonly exploited” for wrongdoing, such that potential interveners, but not potential rebels, should foresee and are responsible for the harmful consequences of such exploitation? Only if foreign intervention is much more likely than rebellion to provoke interceding actions with undesirable consequences. But this is an empirical premise whose verity is far from obvious. First, recent research by Alex Bellamy and Paul Williams shows there is no evidence that the norm of humanitarian intervention or “responsibility to protect” causes—via the dynamic of moral hazard—violence that would not have otherwise occurred.Footnote 19 Second, the pervasive worry that humanitarian intervention will lead to more wars of political and economic self-interest in disguise seems to be somewhat overblown. As James Pattison points out, states are already able to invoke national self-defense to justify going to war, and regularly do so where there is no real threat to their territorial integrity or political sovereignty.Footnote 20 A precedent of humanitarian intervention may perhaps alter the way that states publicly defend their actions, but it is doubtful that it would lead to wars that would not have occurred anyway and simply been justified by appeal to some other principle. Finally, the so-called revolutionary wave—where a violent uprising by oppressed people in one country provokes violent uprisings by people in others—is a well-documented historical phenomenon.Footnote 21
I do not mean to suggest that rebellion is more likely to produce harmful mediated consequences than humanitarian intervention. My claim is only the weaker one: that there is no reason to think, and no evidence to suggest, that the reverse is true—that intervention is much more likely than rebellion to trigger adverse reactions and produce unwanted mediated consequences. Thus the argument for asymmetric standards of proportionality cannot solidly rest on the empirical claim that abusive reactions to foreign intervention are more “normal” or common than harmful reactions to rebellion.
But perhaps we have reason to think that, objective probabilities aside, mediated harms resulting from intervention are nevertheless more foreseeable? The typical rebel group will have limited resources, limited numbers, limited engagement with the outside world, limited time for action, and thus a limited sense of the possible international repercussions of their actions. By contrast, states—especially rich ones—have vast amounts of resources and personnel devoted to international intelligence, research, and surveillance. They will therefore have a much stronger capacity to predict the indirect consequences of the various courses of action they might take. Before intervening in the internal affairs of another state, the political leaders of most advanced industrial countries will typically be able to obtain detailed assessments, based on up-to-the-minute intelligence and analysis, of instabilities and possible knock-on effects in neighboring countries. Thus, arguably, while states will have a reasonable capacity to predict the mediated consequences that their actions will bring about, rebels will not, especially if their repressive government strictly controls internal communications, Internet access, and so on.
Be this as it may, it seems to me that even where rebels can foresee, with reasonable certainty, that their actions will have disproportional mediated harms, it is still permissible for them to defend their basic rights. If this is correct, we cannot say that humanitarians are morally constrained by mediated consequences, while rebels are not, because states are generally in a better position to reasonably foresee these consequences. That would suggest that rebels are equally constrained by mediated consequences at least in those situations where they can anticipate them, which does not seem to be the case. Thus, these epistemic considerations cannot solve the puzzle before us. For a defense of asymmetric standards of proportionality we must look elsewhere.
Duty of Care
There is considerable support in international law for the notion that states owe a duty of care not only to their own citizens but also to people beyond their borders.Footnote 22 Some such duty was formalized by the International Commission on Intervention and State Sovereignty in its 2001 report the Responsibility to Protect. According to the report, governments are not merely permitted but obliged to ensure the continued enjoyment of human rights in countries other than their own. If states do owe such a duty of care to foreign nationals, this may explain why mediated consequences count against the proportionality of their military endeavors.
Again, case law is instructive here. In Chomentowski v. Red Garter Restaurant Pty. Ltd. the plaintiff was head waiter and manager at the defendant's restaurant. The plaintiff's duties included depositing each day's takings in the night safe of a bank half a mile away. One night while walking to the bank the plaintiff was seriously injured in a robbery. The owner was held liable. In his assessment of the decision, John Kidd identifies as particularly relevant the “breach of a duty of care arising from the relationship of employee and employer.”Footnote 23 The injuries may have been inflicted on the employee by the harmful intervention of an assailant, but since this harmful intervention was precisely what the employer—given his duty of care—had an obligation to guard against, he too was held liable. By the same logic, if sovereign states have a duty to protect foreign civilians against violations of their human rights, then states cannot avoid the attribution of responsibility where their decisions or actions predictably provoke rights violations in other countries.
This approach is, I think, promising, but as it stands it is far too simplistic. First, the duties of states are presumably derived from the duties of the people they govern. Any duty of care owed by a state must be owed ultimately by the collective of men and women in whose name, on whose behalf, and with whose resources it acts. The global duty of care is owed by individuals to one another regardless of nationality or political association, and by states only by extension. There is no reason, then, to think that rebels are entirely absolved from the international duty of care and that this is a duty by which states are uniquely bound.
Second, the argument in its current formulation lends itself to implications that are difficult to accept. It implies that there is something unique about states that justifies holding them to a higher standard of proportionality than other actors. But consider the following scenario: State A wages war to drive out an unjustly invading foreign force, state B. It is foreseen that this resort to war will lead neighboring state C to bolster its arsenal after witnessing the benefits of having a strong military, which is in turn expected to set off an arms race, and eventually war, between that state and another neighboring country, state D.Footnote 24 Suppose that the outcome of this series of events is death and carnage that far outweigh the benefits of the initial defensive war. If sovereign states owe a duty of care to foreign nationals, which compels them to fully count mediated harms to foreign civilians in the proportionality calculus, does it not follow that state A's defensive war must be deemed disproportional and, accordingly, immoral? Surely this cannot be right. State A cannot be expected to surrender its independence in order to keep its neighbors from quarrelling.Footnote 25
Thus, a more nuanced formulation of this argument is needed. While states may have a general duty of care toward foreigners, they also have special duties toward their own citizens, and this may justify states in deviating from the default standard of strictly equal concern for all affected parties when making judgments about proportionality. It may justify states in weighing the interests of their own citizens more heavily. If so, we might be able to resist the absurd implication just mentioned. Let us say state A does indeed have a duty of care toward the people of state D, who will be adversely affected by its actions via the interceding agency of state C. But state A also has special obligations toward its own people, and we could plausibly say that state A is therefore permitted to weigh the interests of its own citizens extra-heavily in the proportionality calculus. State A's defensive action, which fulfills that state's duty to protect the common life of its people, thus satisfies the proportionality condition, even though that same action would be disproportional if state A owed a duty of care toward all affected parties equally.
Rebels may also have some general international duty of care, but again we must acknowledge their special duties toward the people for whom they are fighting. In an important respect, then, oppressed citizens engaged in rebellion and sovereign states engaged in national self-defense are analogous: both can claim that special duties justify them in giving extra weight to certain people when making judgments about proportionality.
Arguably, humanitarian interveners cannot make the same claim. For the state engaged in humanitarian intervention there is no special duty in play, only the general duty of care owed to all people equally. If state A launches a humanitarian intervention in state B to protect state B's citizens from slaughter, knowing that this will provoke a slaughter in state C that will kill even more people, state A cannot cite any special duty toward the people of B that justifies giving their interests more weight than what is accorded to the people in state C. The intervention is disproportional, because the interests of B's citizens and C's citizens must be weighed equally. This might explain why, despite both humanitarians and rebels having an international duty of care, a rebellion that predictably causes a certain amount of mediated harm can be proportional even though a humanitarian intervention that causes the same amount of mediated harm is not. Humanitarian intervention is distinguished from rebellion (and from national self-defense) by the absence of a special duty that can alter how interests are weighed in the proportionality equation.
Even if this argument is persuasive, it rests on the premise that a state engaged in humanitarian intervention has equally stringent duties toward those it is rescuing and those who might be adversely affected by the mediated consequences of the rescue. But is this always the case? In some instances the intervening state may well owe a special duty toward its intended beneficiaries. If state A installed, armed, and bankrolled the very dictator who is now slaughtering the people of state B, for example, then state A has arguably acquired a special duty toward the people of B that it does not have toward others. Where the intended beneficiaries of an intervention are people to whom the intervener owes some such special duty, can the intervener not—by parity of reasoning—weigh their interests more heavily than the interests of those to whom only a general duty of care is owed?
To be clear, I am not challenging the cogency of this argument. Rather, I am highlighting its limited scope. Specifically, the argument's normative force is restricted to those cases where an intervention is intended to benefit people that have no special claims against the intervener. But this is not true of all cases of intervention, and perhaps not even of most cases. Thus, a duty of care argument cannot adequately explain why humanitarian interveners must always count mediated consequences in a way that rebels need not.
Appeal to Cost
Where there is an expectation of excessive mediated harms, potential interveners can forgo action without sustaining too high a cost. By contrast, if persecuted citizens do not resist, they will likely face continued oppression and human rights abuses. The third argument says that rebels can justify their actions by making an appeal to personal cost—an appeal that humanitarians typically cannot make.
This argument presents two problems. First, as a rule, while an appeal to personal cost can serve to justify an agent in failing to prevent harm, it can at most excuse an agent for actively causing harm (at least where the victim of the harm is not morally liable to it or is not herself contributing to an unjustified threat).Footnote 26 In philosophical parlance, there is a deontological “option” or “prerogative” to allow harm, but no option or prerogative to do it.Footnote 27 If saving my neighbor's drowning child would mean losing my own life, I am morally permitted to refuse the rescue. On the other hand, suppose there is some independent threat to my life that I can avoid only by executing my neighbor's child. The personal cost associated with keeping the child alive is held constant across both cases, but to kill the child in the latter case is clearly wrongful. When it comes to actively causing harm to the nonliable or nonthreatening, the appeal to cost seems to operate (if at all) at the level of culpability, but not at the level of justification.
In the cases of rebellion under examination, however, the victims of the mediated harms do not pose any threat and are not liable to attack. Take, for instance, the case where the oppressed citizens in one country decide to rebel, foreseeing that this will provoke the dictator of another country to begin killing his own citizens. Moreover, the rebels are not merely withholding assistance from the victims, they are causally contributing to their victimization. They knowingly provoke or set in motion decisions and actions by other parties that they know will result in nonliable, nonthreatening people being harmed. This would seem to rule out the possibility of the rebels availing themselves of a justification grounded in an appeal to personal cost.
But perhaps this particular piece of philosophical orthodoxy is mistaken, or at least admits exceptions. One might simply insist that while one cannot appeal to personal cost to justify intentionally killing the innocent, in certain unique kinds of cases—where the harm to the innocent is mediated by interceding agency, and is merely foreseen rather than intended—personal cost can function as a justifying, and not merely an excusing, condition. In such a case, the rebel would then have a justifying reason unavailable to the humanitarian intervener. This could provide some foundation for asymmetric standards of proportionality.
But I am still not convinced. Suppose that A threatens to kill B without justification. If B defends himself against A's aggression, A will kill dozens of other innocent people (presuming that B cannot kill or incapacitate A so as to prevent this). But if B does not resist, the other innocent people will be spared. B's self-defense thus elicits a violent reaction from A that results in many innocent deaths, such that the cost of B's self-defense is disproportional relative to the benefits of his actions—one innocent life is saved but many more innocent lives will be lost. Yet even if B knows exactly what is going to happen, we would surely not condemn him for defending himself. We blame A, the interceding agent, for the innocent deaths. Let us grant that considerations of personal cost are what justify B's actions in these circumstances.
Now imagine that B is physically incapacitated and cannot repel A's attack. To provide for his own security, B has employed a personal bodyguard, C. A threatens B's life without justification. If C steps in to protect his client, again it is foreseeable that A will kill dozens of other innocent people out of anger, which he would not do if left to kill B. C cannot appeal to personal cost to justify defending B, but does that matter? Is he not justified in repelling A's aggression regardless? After all, C is simply acting as B's agent. Where there is an author-agent relationship, the agent is only permitted to do things that the author himself would be morally permitted to do. Otherwise individuals could circumvent their moral obligations simply by acting through an agent rather than directly. But as long as a person is morally permitted to bring about certain ends through certain means on his own behalf, other things being equal, he may enlist the freely given aid of a third party, or empower an agent to bring about the same ends through the same means for him. James Rachels rightly identifies this as a basic principle of moral reasoning.Footnote 28 Thus, the fact that C cannot appeal to personal cost seems immaterial, as long as his author, B, would be able to justify defending himself through some such appeal.
Further, it does not seem necessary for C to be a contractually pre-appointed bodyguard when A attacks. C might simply be a Good Samaritan. Let us say B is pleading for C's assistance. C is able and willing to help. If B would have been justified in appointing an agent to defend him in advance of A's attack, I see no reason why he cannot authorize an agent to do so during the attack. The same principle applies: B has every right to defend himself in these circumstances, and so he should be able to solicit and accept the temporary security services of C. If C would be justified in defending B, despite the expectation of excessive mediated harms and the unavailability of a cost-based justification, why isn't a state permitted to act as the agent of an oppressed group in another country? Why can't humanitarian interveners be considered instruments of justice who act on behalf of oppressed citizens, at least where those citizens would welcome the intervention?
The answer might have something to do with the fact that an agent is never solely an agent. He is also an independent moral actor with his own rights and duties. There are cases where one's obligations as an independent actor could preclude one from functioning as another's agent. For instance: suppose that, instead of a privately appointed bodyguard, C is a policeman. A policeman has a duty to protect all law-abiding citizens equally. In other words, while the policeman has a duty to protect B, he has no moral justification for prioritizing the interests of B ahead of the interests of other innocent people. Thus, if the policeman foresees that his defense of B will provoke deadly (and unstoppable) violence against other bystanders, arguably it would be wrong of him to defend B. While B might appeal to cost to justify defending himself, C's preexisting fiduciary obligation to protect all citizens in his jurisdiction equally precludes him from acting purely as B's agent.
Similarly, one might argue that states have preexisting obligations that preclude them from entering into author-agent–type relationships with oppressed groups beyond their borders at the expense of the interests of other groups. Humanitarian intervention might be seen as the international analogue of police work. Policemen in the domestic setting must weigh the interests of all law-abiding citizens equally, and cannot validly enter into arrangements that commit them to do otherwise. Maybe the same applies to humanitarian interveners. But this analogy is flawed: police officers have voluntarily accepted a role that commits them to “protect and serve” all citizens without bias. We cannot say that every state in the international arena has implicitly or explicitly made a similar commitment. Thus, we need some other reason to think that states have a preexisting obligation that precludes them from acting as the agents of oppressed groups.
It might be helpful to resituate this issue within the broader literature on the ethics of self-defense. There are two approaches to justifying self-defensive violence: agent-neutral and agent-relative. On the agent-neutral approach, what matters is the justice or rights consistency of the objective state of affairs brought about by the defensive act. Take the fairness-based account, for example.Footnote 29 Where one person is to blame for the fact that somebody must sustain a cost, it is only fair that he be the one to sustain that cost. Thus, where an aggressor threatens a victim, the preservation of the victim at the expense of the aggressor is preferable from the point of view of justice. It shifts the losses back onto the agent whose culpable or negligent behavior has made it inevitable that someone will suffer them. On this account, the imperative—or at least the license—to promote justice or to prevent injustice is what yields the right to self-defense.
Notice that this justification makes no reference to the victim's unique point of view. This is important. It means that the very same justification available to the victim when defending himself is also available to the rescuer when defending the victim. Again, what matters is the justice or rights consistency of the objective state of affairs brought about. If the preservation of the victim at the expense of the aggressor is preferable from the point of view of justice, and this is what permits the victim to kill the aggressor, it equally justifies third parties in killing the aggressor. Therefore, unless there is some relevant difference between the costs and consequences of an act of self-defense and an act of other-defense in the same circumstances, an agent-neutral approach does not have the resources to differentiate the two morally. Suppose, for instance, that the victim is in a position where he can defend himself without causing harm to any bystander, but a third party cannot save the victim without inflicting lots of collateral damage. Here, an agent-neutral approach can easily explain why self-defense is permissible but other-defense is not: there is some morally relevant difference between the states of affairs that the two acts bring about. But where the consequences of the two acts are identical, an agent-neutral justification commits us to accepting that other-defense is permissible where self-defense is permissible. The implication for our present discussion is clear: if, according to agent-neutral considerations, rebels are justified in defending themselves despite the expectation of excessive mediated harms, humanitarian intervention should be permissible under the exact same circumstances, assuming all costs and consequences are identical.
But an agent-relative model of self-defense complicates the picture somewhat. This model presumes that an individual is entitled to show a certain degree of partiality toward herself and her own interests. To deny that self-defensive violence is permissible, the argument goes, is effectively to maintain that the would-be victim, in a forced-choice situation where she must choose between her life and that of her aggressor, is obliged to give priority to the latter—and surely this is too much to ask. There are limits to what morality can reasonably demand that one person sacrifice for the sake of others.Footnote 30 However, while I may be permitted to prioritize my own life over that of my attacker, where one stranger attacks another it seems that I have no agent-relative grounds for giving priority to the life of the victim. Thus, on an agent-relative account the self-defender has justifying reasons that are not available to the other-defender. At first blush, this may seem to explain why humanitarian intervention (an act of other-defense) is not justified where rebellion (an act of self-defense) is. But drawing that conclusion would be too hasty.
As Cecile Fabre points out, if a victim's personal interest in survival is important enough to establish her right to kill in self-defense, it is also important enough to establish a power to alter the rights and liberties of others.Footnote 31 The victim, whose right to self-defense derives from her right to personal partiality on an agent-relative account, may transfer her permission to a third party.Footnote 32 This is how an agent-relative account explains the right (and even the duty) to defend or rescue others in ordinary circumstances. Now perhaps a victim's right to self-defense cannot always be transferred in this way, but where we want to deny the transfer, we surely need a reason. This brings us back to where we started. Even if we adopt an agent-relative approach in order to justify holding rebels and humanitarian interveners to different standards of proportionality (which is essentially what the cost-based argument does), we need some reason to believe that the rebel's right to defend himself cannot be transferred to foreign parties that are willing to help. We need some reason to think that, where there is an expectation of excessive mediated harms, the transfer is blocked. To be sure, there may well be such reasons that I have failed to identify. Unfortunately, space does not permit a further and fuller exploration here. But until such reasons are provided, we should be wary of the assumption that humanitarian interveners are subject to more demanding standards of proportionality than those applicable to rebels.
Conclusion
I have not mounted a dispositive case against asymmetric standards of proportionality, but have only shown that none of the arguments for the asymmetry is entirely persuasive. What I have called the duty-of-care argument may have purchase in some cases, but not in all. Its scope is limited. The cost-based approach cannot succeed without a convincing explanation for why states are not permitted to function as the agents of oppressed groups beyond their borders. And the common exploitation argument seems to fail even if we concede its empirical premises. My conclusion is thus a contingent one: We currently have no good reason to accept that intervening states, always and without exception, are constrained by foreseeable mediated consequences in a way that rebels are not. This at least shifts the moral presumption and, with it, the burden of argument. Until a good reason is provided, concerns about intervention begetting more intervention or civil strife via moral hazard should not simply be presumed to function as a moral barrier to otherwise legitimate military action.
