This paper explores the relationship between the morality of war and the law of war. The focus is on permissible conduct in war, specifically the principle of noncombatant immunity, which confines belligerents to targeting only military objectives (see article 48 of the first Additional Protocol (API) to the Geneva Conventions, Roberts and Guelff 2000: 447), and the legal equality of combatants, which grants soldiers the same permissions and holds them to the same prohibitions irrespective of the justice or injustice of their cause (preamble to API: 422–23).
Call soldiers whose side satisfied jus ad bellum—the principles governing justified resort to war—j-combatants, and those whose side did not satisfy those principles ucombatants (for unjustified). The dominant figure in late twentieth-century just war theory, Michael Walzer, argued that all combatants enjoy equal permissions to target one another, grounded in the threat that they pose to each other’s lives (Walzer 2006: 42–45). In virtue of that threat, u-combatants and j-combatants alike lose their rights against lethal attack, so are not wronged when their adversaries kill them. By contrast, noncombatants, unthreatening by definition, retain those same rights, so are not permissible targets. Noncombatant immunity and combatant equality, then, are grounded in a single argument: one may permissibly target only those who have lost their rights against lethal attack; all combatants have lost that right; all noncombatants retain it.
Walzer’s position has developed into a conventional orthodoxy, bolstered by its consonance with international law. In recent years, however, many philosophers have become increasingly dissatisfied with his arguments. In particular, these revisionist critics have discredited his account of how one loses the right to life, showing its inconsistency with other plausible beliefs about permissible harming (see especially Coady 2008; Fabre 2010; McMahan 1994, 2004; McPherson 2004; Rodin 2002). Against Walzer, they argue that the morality of harming is almost always asymmetrical—a person who defends himself against unjustified attack does not become liable to be harmed by his attacker, simply by now posing a threat. To become liable to be killed, the threat one poses must be unjustified. Moreover, posing an unjustified threat is neither necessary nor sufficient for liability. A politician who sends a u-combatant to fight an unjustified war might be liable, despite not posing any threats himself; a child soldier, out of his mind on drugs and with a gun to his head, might not be liable, despite posing a threat. Walzer’s critics have concluded that what matters for liability is (1) responsibility for (2) contributing to threats of (3) unjustified harm (Coady 2008; Fabre 2010; McMahan 2004; Rodin 2008).
Though their criticisms of Walzer are shared, revisionists diverge in how they combine these three elements. How much and what kind of responsibility? What degree of contribution? Must the threatened harm be pro tanto or all things considered unjustified? Their views’ practical implications depend on the answers to these questions, which determine where the liability bar is set; and revisionist though they are in theory, most of these critics endorse quite conservative practical conclusions. Though they reject the moral equality of combatants, they endorse conventional views, such as noncombatant immunity, and the rejection of pacifism (that is, they reject the view that warfare can never be justified). Insofar as they endorse these commonsense conclusions, however, they each face the same problem. If a significant number of u-noncombatants and u-combatants, in any given conflict, are responsible to just the same degree, for just the same quantum of contribution to the same unjustified threats, we cannot retain both noncombatant immunity and the rejection of pacifism.
Distinguish between micro-threats to specific lives and the macro-threat posed by a state. Many u-combatants neither pose micro-threats themselves, nor contribute to those posed by their comrades, while their contribution to the macro-threat posed by their state is negligible (Lazar 2010). By definition, u-noncombatants do not pose microthreats, but many of them contribute to micro-threats, for example by producing the goods (military and nonmilitary) that sustain their armed forces’ ability to fight (Fabre 2010). This also contributes to the macro-threat, which they also further through their taxes, popular support and, in a democracy, their vote. Moreover, many u-combatants—both those who contribute to micro- and macro-threats and those who do not—are guiltless for their actions, fighting either from duress or under a reasonable belief that their cause is justified. Meanwhile, many u-noncombatants will make their contributions without duress and without the nonculpable ignorance excuse. They will be culpable.
The liability theorists face a dilemma. If they endorse a low threshold of responsibility for liability—say, minimal responsibility for some small contribution to micro-or macro-threats—they will render too many noncombatants permissible targets (and their criterion of liability may also be independently implausible: Lazar 2009). If they endorse a high threshold—requiring a significant contribution to micro- and macrothreats, a significant degree of responsibility or both—they will struggle to justify the killing inevitable in justified wars, because too many u-combatants will not be liable to be killed, and j-combatants cannot discriminate between liable and nonliable u-combatants. This is the “responsibility dilemma” for the liability view (Lazar 2010). The first horn leads to permitting the justified side to target too many noncombatants; the second leads to contingent pacifism—the view that although wars can hypothetically be permissible, in all feasible cases we should not fight, for fear of violating our enemies’ rights. If the liability theorists aspire to less controversial practical conclusions than these, their view needs additional support.
Some find this support in the distinction between the morality and law of war. They argue that the contingent pacifist and noncombatant immunity objections might be good reasons not to implement their view in the laws of war, but do not undermine their account of war’s underlying morality (Fabre 2010: 39; Frowe 2011; Hurka 2005; McMahan 1994, 2004, 2008, 2009, 2010). The following sections first set out the most fully developed version of this argument, then criticize it, before asking just what the relationship between war’s law and its morality should be.
The appeal to law is quite simple. People often act wrongfully. Sometimes they choose to do so; other times they do so by mistake or accident, for example because they lack important information. How should this predictable wrongdoing impact our morality, and our laws? One response is that the predictable failure to act rightly should not alter what the right is, but might be relevant to choosing our laws. In war, two causes of predictable noncompliance are particularly troubling. First, combatants’ lack of important information: whether their side satisfied jus ad bellum, whether this operation proportionately conduces to that just cause and whether their targets are liable are all very complex and urgent questions whose answers depend on often ambiguous or unavailable information. Second, voluntary noncompliance: of course, in one sense this is endemic to warfare, since without some voluntary wrongdoing, it is unlikely there would be any conflict in the first place. But especially salient here is the tendency of u-combatants to arrogate to themselves any permissions that are made available to j-combatants. Even if the injustice of their cause should be clear to them, they are likely to affirm themselves to be j-combatants, and so entitled to the relevant schedule of permissions.
Revisionists deny that this predictable noncompliance is relevant to the morality of war. The epistemic shortfall might make acting morally difficult, but that is to be expected: doing the right thing is often hard (McMahan 2005: 770). If we ought to X if we had full information, then we ought to X even when our information is incomplete (though we might be excused for failing to do so). Likewise, that others will abuse our principles is no argument against them, qua principles. However, these are both appropriate worries when devising the law. If people will routinely disregard a law, or if it makes unachievable demands, then the law will be regarded as irrelevant, unfair and unrealistic, and will lose its capacity to guide action. If the laws of war have any value or purpose (McMahan, for example, argues that they can be used to minimize wrongful suffering—2008: 28) then we must guard against this outcome.
Morality and law, the argument goes, should therefore come apart (McMahan 2008: 32; McMahan 2010: 506). Though combatant equality and noncombatant immunity lack substantial foundations in moral principle, they might nonetheless be justified as laws. If the laws of war enforced combatant asymmetry they would be unworkable, because uncertainty over who is liable to be killed would render them impossible to enforce (Fabre 2010: 57; McMahan 2008: 32). Moreover, any permissions granted to the justified side would be abused by the unjustified side as well. Similarly, noncombatant immunity must be retained in law because extending the permission to kill noncombatants to jcombatants would mean its abuse both by them and by u-combatants.
This approach enables the following response to the responsibility dilemma. Perhaps it does identify salient pragmatic concerns, and the liability view would be difficult to implement. But these pragmatic concerns are irrelevant to the moral principles that govern warfare. That a principle is hard to follow, or abused by some, does not make it false. The contingent pacifist objection worries that the liability view will deny states and people the moral option of fighting justified wars. But in practice, soldiers and states need not worry about killing only the liable, since they can instead adhere to the laws of war, which permit them to kill enemy combatants and noncombatants who are directly participating in hostilities. The noncombatant immunity objection worries that the liability view renders too many noncombatants permissible targets—but the laws of war will retain noncombatant immunity, so this worry is misplaced.
Although the appeal to law raises interesting questions, it does not resolve the problems with the revisionist position. We can begin with two general worries before turning to specifics.
How should soldiers respond when legal and moral injunctions diverge? Though the appeal to law is often made, this vital topic has scarcely been discussed. If legal reasons always trump moral reasons, then the revisionist morality of war would be redundant in practice. If moral reasons always trump legal reasons, then the appeal to law would be equally pointless. For the appeal to law to work, moral reasons must sometimes trump legal ones, and vice versa. Most will agree that moral reasons can override legal ones; the reverse ordering is much harder to explain. First, because it requires an account of the duty to obey the law of armed conflict (LOAC), and it is notoriously difficult to ground a duty to obey any law—especially one that explicitly diverges from our other moral reasons. Second, even if we can produce some such account, if we ought to obey the law when it clashes with morality, then the law in this case describes what we ought to do, and the “morality of war” can be no more than a subset of the relevant moral reasons.
The second general worry is that the appeal to law cannot resolve the contingent pacifist and noncombatant immunity objections, it can only deflect them. To resolve the contingent pacifist objection, it must show that soldiers (at least j-combatants) may obey their legal permission to kill enemy combatants, even when morally prohibited from doing so because their adversary is not liable. But if our most fundamental moral prohibitions—against killing the nonliable—can be overridden by a mere legal permission, then the morality of war truly is irrelevant, and we should simply focus on the laws of war. Where noncombatant immunity is concerned, even if the law does reduce the targeting of noncombatants, that does not answer the objection that with a low liability threshold too many noncombatants will be liable. The appeal to law would then look like the utilitarian’s response to the slavery objection—enslaving some will never in fact maximize utility. Even if this were true, it would not adequately resolve the objection, which is that people’s freedom should not be vulnerable in this way to calculations of overall utility. Similarly, even if, given these laws, noncombatants will not often be targeted in practice, we can object that they should not, in principle, be morally vulnerable to attack.
Turning to specifics: McMahan (2008: 37–38) helpfully distinguishes between moral and legal permissions, prohibitions and positive requirements, then argues that when morality requires what law permits or prohibits, and when morality prohibits what the law permits, soldiers should obey their moral reasons. But when the law prohibits what morality permits, combatants should adhere to the law. He says little about clashes between legal requirements and moral permissions and prohibitions, since he thinks the law rarely requires positive action. This is an important oversight, since the additional protocols do require, for example, combatants to observe a duty of constant care toward civilians (article 57—see Roberts and Guelff 2000: 452–53). Moreover, soldiers are required by their own military law (often backed up by an oath of allegiance) to obey lawful orders. However, let us set that aside to concentrate on the orderings McMahan does defend.
For the appeal to law to have any purchase against the noncombatant immunity objection, we must first show that j-combatants are morally permitted, not required, to kill liable u-noncombatants, and then that legal prohibitions trump moral permissions. Neither is easy to do. Combatants generally have strong positive moral reasons for action—they have natural duties to protect their comrades and their country, and contractual duties grounded in their roles and the oaths they took. If they can save their comrades or advance their just cause by killing liable u-noncombatants, then one could readily argue that they are required to do so. Since moral requirements trump legal prohibitions, the appeal to law would not resolve the noncombatant immunity objection.
Even if j-combatants are merely permitted, not required, to kill liable u-noncombatants, we can justify denying them this option only if we have a strong argument for a duty to obey the law. Again, despite the frequency with which the appeal to law is made, discussions of this crucial point are scarce and brief. McMahan adumbrates two relevant arguments. The first sees the fact that X is against the law as a reason not to consider other reasons for X-ing. He argues that “combatants should be reluctant to give their individual judgment priority over the law, for the law has been designed in part precisely to obviate the need for resort to individual moral judgment in conditions that are highly unconducive to rational reflection” (McMahan 2008: 41). This presupposes a particular conception of law’s authority, according to which law gives us “protected reasons,” which preempt the need to appeal to our own judgment (Raz 1979). The second sees the fact that X is against the law as providing an additional reason not to X—it argues for a pro tanto duty to obey the law, grounded in the fact that disobedience will lead to further breaches by others (McMahan 2008: 38).
The first argument fails: the laws of armed conflict do not provide protected reasons. If they did, then adherence would be required without exception, since if other moral reasons could justify disobedience to the law, then we would have to consult those reasons in any situation to determine whether it is exceptional. The law would not, therefore, obviate the appeal to our own judgment. Since McMahan thinks (plausibly) that some moral reasons do override legal reasons, the protected-reasons logic is unavailable.
Could revisionists make the narrower argument that the law governing noncombatant immunity, at least, grounds protected reasons? They might argue that few u-noncombatants will in fact be liable to be killed, while even j-combatants with the best intentions will likely inflate the permissions available to them, and so end up mistakenly killing nonliable u-noncombatants. Adherence to the laws of war will then better enable them to comply with their reasons not to kill nonliable noncombatants, since their own judgment will be excessively permissive.
There are at least two problems with this argument. First, if j-combatants are indeed sometimes morally required to kill the liable u-noncombatants, then they ought to disobey the law. They cannot therefore defer to the law’s authority: they must assess each case on its merits, against the full range of moral reasons. Second, the argument presupposes a high liability threshold, given the premise that few u-noncombatants will be liable. With a low liability threshold, j-combatants could plausibly argue that restricting their attacks to adult u-noncombatants is discriminating enough. Raising the liability threshold to ensure few noncombatants cross it must also mean rendering more u-combatants non-liable (those who make no significant contribution to micro- and macro-threats, and on some accounts those who are only minimally responsible for their contributions). This puts us squarely back at the contingent pacifist objection.
The second argument mooted by McMahan is that combatants have a duty to obey LOAC, since even morally permissible breaches will encourage others to impermissibly breach it. Evidently this applies only if and when disobedience will actually lead to further disobedience by others. Breaking the law in secret, or when one’s adversary lacks the capacity to retaliate, would still be permissible. But even when the empirical speculation holds, we still need more argument, which would have to be developed along these lines: if some soldiers kill liable noncombatants, other soldiers will likely end up killing nonliable noncombatants. The original soldiers are partly responsible for this outcome, and so are morally required to forego killing liable noncombatants even when it is otherwise permissible.
This argument’s weakness is the idea that combatants are responsible for their adversaries’ unjustified actions and retaliations, and should forego options that might save their lives and those of their comrades, as well as advance their just cause, to avoid bearing that responsibility. Elsewhere in most versions of the revisionist view (Fabre is an exception), our responsibility for our own wrongdoing is thought far greater than for the wrongful acts of others that we fail to prevent. And yet here we expect j-combatants to sacrifice their lives, and the opportunity to contribute to a just cause, because of speculative claims about how their conduct might connect with the voluntary wrongful actions of other combatants in the future. On the one hand this seems to demand too much of j-combatants. Mala prohibita