The Moral Distinction Between Combatants and Noncombatants: Vulnerable and Defenceless

In Sparing Civilians, Seth Lazar claims that in war, with rare exceptions, killing noncombatants is worse than killing combatants. This paper raises some doubts about whether this is an important principle – at least, once we understand Lazar’s clarifications. It also suggests that however it is clarified, it seems false. And it suggests a related principle that more plausible. This related principle applies only to those with just aims, and it applies only to intentional killing rather than to all forms of killing.

I raise some doubts about whether Moral Distinction is an important principle -at least, once we understand Lazar's clarifications. I also doubt that it is true, even understood in this way. And I doubt that Lazar is committed to his clarifications of it. I suggest an alternative version of Moral Distinction that is not undermined by my arguments that I find more plausible. This version applies only to those with just aims, and it applies only to intentional killing rather than to all forms of killing. The alternative version is in some respects more revisionist than Moral Distinction -it conflicts more powerfully with the pre-reflective judgements that many people are inclined to make about the wrongness of killing in war.

I. MORAL DISTINCTION FOR THE JUST
Consider a simple war between a just side and an unjust side. Of course, simple wars like this are quite rare -in most wars, both sides act unjustly -but the aspect of WWII fought between the Allies and Nazi Germany is a good enough example.
There are good reasons to think that it is normally worse for those on the just side of the war to kill combatants than non-combatants. The just side can justify killing only if doing so is necessary to advance their just aims, and killing non-combatants often does not advance these aims. Killing non-combatants is typically wrong because it is pointless. Killing combatants is more often necessary to advance a worthwhile goal, so this objection to killing them does not apply as often.
As Lazar points out, it is not always instrumentally valuable to kill combatants, and it is sometimes instrumentally valuable to kill noncombatants. But even when killing non-combatants is instrumentally valuable, intentionally killing them often violates some other moral principle. Intentionally killing non-combatants can usefully demoralize the enemy, but doing this is wrong because they will harmfully be used to achieve the ends of the just side. 2 (I will use the label 'manipulative killing', where Lazar uses 'opportunistic killing'. There is an important distinction between manipulative and opportunistic killing that Lazar doesn't notice -the main objection, I think, is that non-combatants are used, and not that a new opportunity that they create is exploited). 3 Targeting combatants is also useful, but they need not be used. Killing them eliminates the threats they pose. Lazar joins several others in arguing that it is worse to kill manipulatively than it is to kill eliminatively, other things equal. If this is right, it is typically harder for those on the just side to justify targeting non-combatants than combatants.
These arguments powerfully support the view that it is typically wrong intentionally to kill non-combatants on the unjust side where intentionally killing combatants on the unjust side is permissible. At most it would be permissible intentionally to kill non-combatants in order to prevent them from contributing to the war effort -doing this would be eliminative. But, Lazar thinks, it is rarely permissible intentionally to kill them in this way. And even if it sometimes is, combatants will often lack sufficient information to justify it. If this is right, these cases do not undermine the idea that killing non-combatants is often wrong where killing combatants is often permissible.
These arguments do not support Moral Distinction. First, the argument offered only applies to intentional killing. Moral Distinction, though, is concerned with killing more generally. It suggests that it is worse to kill non-combatants than combatants as a side-effect. I am not confident that this is always true, or even typically true, especially if combatants on the unjust side are ignorant of the injustice, or have powerful excuses, subject to some further conditions. 4 More importantly for our purposes, these arguments support only the view that it is harder to justify targeting non-combatants than combatants. They do not show that it is worse to kill non-combatants than combatants when doing either thing is completely unjustified. But Moral Distinction is concerned with the gravity of a person's wrongdoing when she has acted wrongly, rather than with permissibility.
Suppose that one act is harder to justify than another: for example, that manipulative killing is harder to justify than eliminative killing. It does not follow that wrongful manipulative killing is worse than wrongful eliminative killing. The idea that it is more difficult to justify one act than another implies that one's good reason for acting must be weightier in the former case than in the latter, if it is to make that act permissible. It may not follow, and I will later suggest that it is not true, that manipulative killing for a very bad reason is morally worse than eliminative killing for the same very bad reason.
Thus, some of the main arguments that Lazar mounts in favour of moral distinction do not obviously apply to those with evil aims. But Moral Distinction claims that the combatant/non-combatant distinction applies to both the just and the unjust side of a war, including those with evil aims.
Here is a narrower principle that I find more plausible than Moral Distinction: Moral Distinction II: In war, with rare exceptions, intentionally killing non-combatants is harder to justify than intentionally killing combatants.
In what follows I assume that some of Lazar's arguments support Moral Distinction II (although we will also see why the exceptions are not so rare, and so perhaps Moral Distinction II also fails). I show that Moral Distinction is false, and that Lazar's own arguments do not support it. Lazar clarifies the principle as follows. We must hold constant the number of people affected, the degree of harm and the aim sought. 5 Combatants are members of the armed forces of a group at war, and non-members who directly participate in hostilities. Everyone else is a non-combatant. Thus, there are two independently sufficient conditions of being a combatant: that the person is a member of the armed forces, whether or not she is directly participating in hostilities, and that the person is directly participating in hostilities, whether or not she is a member of the armed forces. Here is what Lazar means by 'worse': X is worse than Y, for the purposes of Moral Distinction iff 'X is pro tanto more seriously factrelative wrongful than Y'. He also suggests that: 'X is pro tanto more seriously fact-relative wrongful than Y if the objective moral reasons against X are weightier than those against Y'. As I have already noted, we might doubt the relationship that Lazar thinks holds between the gravity of wrongdoing and reasons that apply to us, so I am happier with the first definition of 'worse' than the second.

III. THE AIMS AND CONSEQUENCES OF KILLING
Lazar's clarifications of Moral Distinction invite us to hold various things constant, and then focus on the combatant/non-combatant distinction. One of these things is the aim of the killing. But this is a bit puzzling. Combatants and non-combatants are normally killed with different aims, and many people think that this is central to our conviction that it is normally wrong, or more gravely wrong, to kill non-combatants.
Take the application of Moral Distinction to the just side of the war. If true, it is worse for a just combatant to kill an unjust noncombatant than an unjust combatant if we hold constant the aim of the killing. But this stipulation also deprives us of some central arguments that are thought to support the importance of the combatant/non-combatant distinction. The main reason to kill unjust combatants is that they pose unjust threats. As we have seen, this is a reason against killing non-combatants -they are not liable to be killed because they do not pose unjust threats to us. Moral Distinction, as clarified by Lazar, cannot be supported by offering such reasons, for this would aim to explain the difference between combatants and non-combatants in virtue of the different aims that we could have in killing these people, and we are supposed to hold our aims constant. This is not an argument against the truth of Moral Distinction, but it does suggest that Lazar problematically limits the scope of our investigation into the importance of the combatant/non-combatant distinction. And this seems troubling given his broader aims, which are to shore up support for a widely and viscerally affirmed view that VULNERABLE AND DEFENCELESS it is worse to kill civilians than soldiers, and the legal impermissibility of killing non-combatants. 6 As I understand the widely held and viscerally affirmed view, it is less bad to kill combatants precisely because this is done with the aim of preventing them from engaging in hostilities, and to avert threats that they pose, whereas the killing of non-combatants cannot be done with this aim, for they typically pose no threats. They are typically killed with retributive or terroristic aims, and killing for these reasons is normally thought especially bad. This also explains the prohibition on killing non-combatants in international law.
Perhaps Lazar might claim that Moral Distinction is nevertheless important, because non-combatants are normally killed with worse aims than combatants. Moral Distinction would provide additional support to the conventional view that it is worse to kill non-combatants than combatants because of the aim of doing it. But that depends on conventional wisdom being true, and Moral Distinction, clarified as Lazar suggests, does not support it.
I also have some doubts that Lazar is committed to this clarification. He considers 7 : Necessity: Unnecessary killing is worse than necessary killing; so killing civilians is worse than killing soldiers, because killing soldiers is necessary, killing civilians is not.
Lazar thinks that Necessity is one important foundation of Moral Distinction. 8 But Necessity offers no support for Moral Distinction as Lazar clarifies it. Necessity implicitly appeals to the different aims that we have, or can have, in killing combatants and non-combatantsthose that can be served by killing combatants but that cannot be served by killing non-combatants.
Furthermore, Necessity implicitly fails to hold the number of people affected constant. A pointless killing, Necessity suggests, is more gravely wrong because the killing does not achieve a good aim. But the relevant good aims concern goods for people. So Necessity implicitly relies on the idea that harming non-combatants fails to protect others from harm whereas harming combatants does. That is why harming combatants is normally necessary but harming non-6 SC, 2; 6-7. 7 SC 24. 8 SC 25. combatants is not -it is often necessary to harm combatants, but not non-combatants, to protect others because of the threats combatants pose. Necessity thus supports Moral Distinction only if we clarify the latter principle in a different way from Lazar -by dropping the constraint that we hold constant the aims of the person inflicting harm and number of people affected by it.
When we do hold constant our aims and the effects of our actions on others, Moral Distinction also seems much less attractive. Some non-combatants, as Lazar defines them, are extremely grave wrongdoers. During a war, there may be many such people. To take the most obvious example, many of the people who governed concentration camps and who directly perpetrated the Holocaust in Nazi Germany during WWII were non-combatants. They were not members of the armed forces, and they were not participating in hostilities.
We could consider whether it is more gravely wrong to kill a combatant with the aim of averting a threat to human life that he wrongly poses when compared with a person operating a concentration camp with the aim of averting a threat to human life that he wrongly poses. We would then hold constant our aims and the number of people affected in the killing. But Moral Distinction does not seem attractive in this case. If anything, it seems better to kill a Nazi who is participating in genocide than a Nazi combatant.
Perhaps Lazar might claim that Moral Distinction does not apply to this case, as the killing of these people would not have been 'in war' and Moral Distinction is explicitly about killing in war. I noted earlier that it is not clear what 'in war' means. But it is hard to see how killing these people would not have been in war, at least if we hold that retributive killing of non-combatants is done in war. And Lazar clearly means Moral Distinction to apply to the latter case.
Perhaps Lazar might think this a rare exception, and hence compatible with Moral Distinction. But it is not at all rare for people to be killed by non-combatants whilst a war is going on, but independently of hostilities. Here are two more examples of many: Stalin's government used law enforcement agencies to kill many civilians during WWII, as well as to deport many citizens to Germany for execution; some perpetrators of the Rwandan genocide, which took place during the Rwandan civil war, were combatants, but many were not -police officers and civilians were also involved. Generally, war creates chaos, and in that chaos a great deal of lethal violence occurs. Not all of these killings are acts of war. If right, this suggests that even Moral Distinction II runs into trouble, depending on some empirical questions, and how rarity is best understood.
Overall, I think it better to hold fewer things equal than Lazar suggests in attempting a defence of the broader conviction that it is worse to kill non-combatants than combatants, and I think so does Lazar. The main reason why the distinction is thought important has to do with the fact that most non-combatants pose no threats to innocent people, and so killing them cannot avert threats they pose.

IV. AGAINST MORAL DISTINCTION FOR THE UNJUST
But if we clarify Moral Distinction differently -allowing us to consider different aims and effects of killing combatants and non-combatants -we have other reasons to reject it. Arguments that favour the killing of unjust combatants over non-combatants also favour the killing of just non-combatants over combatants.
Almost all killing by the unjust side in a war is wrong. 9 There is little or nothing valuable that the unjust side can offer, at least in typical cases, that would do the necessary justificatory work. There are cases where the unjust side has good aims, but aims that are insufficiently good to justify going to war. I leave such cases aside, though they are quite common. Let us focus on purely unjust warsthose common wars mounted to secure evil aims. Our question is whether it is typically worse for a person participating in a purely unjust war against a just foe to kill a combatant or a non-combatant. I think that it is often worse to kill combatants in such cases.

A. Effective and Ineffective Killing
We saw above that it is often wrong for the just side to kill noncombatants because doing so would be pointless -it would fail to advance the aims of the just side. Now consider those with evil aims. It is better that nothing is done to advance these aims. One's evil cause cannot count against killing if killing will not advance that cause, and has no prospect of doing so. Hence, these aims do not count against the pointless killing of non-combatants. As it is often only pointless for unjust combatants to kill non-combatants, the evil aims of the unjust side do not enhance to the badness of such killings. In contrast, killing combatants on the just side of the war is both inherently bad, and bad because it is done with an evil aim that it has some prospect of advancing. In this respect, it is normally worse for those with evil aims to kill combatants than non-combatants.
Here is a closely related point. Killing combatants on the just side is bad not only because of the importance of the lives of those combatants, but also because it prevents them from achieving their just aims. These aims, we can suppose, are extremely valuable because they justify acts of war. Typically, they include saving other people from being killed in large numbers. Killing combatants on the just side frustrates these aims, and doing this is very seriously wrong independently of the harm inflicted on the person with the aims. When killing non-combatants is pointless, as it often is, the valuable aims of the just side are not frustrated, so killing them is not as bad.
To illustrate, compare Nazi soldiers killing Allied combatants and non-combatants. Killing Allied combatants prolonged the war, substantially increasing the overall death toll. And it sustained Nazi rule, providing a shield for the escalating efforts of the Nazis to massacre Jews and others. In contrast, killing Allied non-combatants probably did little to prolong the war, or to shield the operations of the increasingly lethal Nazi machinery. The implications of Moral Distinction in this case -that it was worse for Nazi combatants to kill Allied non-combatants than Allied combatants -seems hard to accept. But this is hardly a rare exception. For this reason, we should reject Moral Distinction, at least if we don't hold equal the number of people affected by the killing.
This, of course, does not count against Lazar's interpretation of Moral Distinction -the number of people affected is not held constant. But it does count against Lazar's more general ambition to offer support to the commonly held view that killing combatants is less bad than killing non-combatants, and the reflected view in international law. For these views rely on overall judgements about the wrongness of killing a person, including the effects of that killing. Furthermore, if we follow Lazar in thinking that Necessity provides an important foundation of Moral Distinction when applied to the just side of the war, it also provides an important foundation of its opposite when applied to the unjust side.
And there are related reasons to doubt Moral Distinction as applied to the unjust side of the war, even when numbers are held equal. Those killed on the just side of the war are killed in order to prevent them from fulfilling one of their vitally important goals, and one that they have made great sacrifices to try to secure. Killing them is worse not only because the aims are valuable, and should not be frustrated, in some impersonal sense. They have a very stringent right that they are not killed in order to prevent them from fulfilling these goals.
Those who fought on the just side in WWII had the centrally important aim of preventing Nazi domination in Europe. They were willing to risk their lives for this aim. They have a stringent right not to be killed to prevent them from achieving it. This is not only because of the importance of preventing Nazi domination in Europe, considered impersonally. It is also because they committed themselves to achieving it. Killing a person to prevent that person from securing an aim that is of central importance to him, and that is of very great value, wrongs that person more gravely than killing a person without that ambition.
Perhaps some might doubt that the gravity of a rights violation is contingent on whether it frustrates a person's important goals, or that it is done with the aim of frustrating these goals. But notice that it is highly plausible that we have a right that our attempts to achieve our most important valuable goals are not frustrated quite independently of any right we have not to be killed. Others should not interfere with my liberty to prevent me from doing what I am permitted to do. The gravity of the wrong of interfering with my liberty in this way surely depends on the importance of what I am trying to do -interfering with my liberty in a way that prevents me from securing my most trivial ends is much less bad than doing so in a way that prevents me from doing what I, for good reason, value the most.
And it is not at all unusual that the gravity of the wrong that is done to a person depends on the intentions of the wrongdoer -a person who intentionally frustrates my most important valuable goals wrongs me more gravely than a person who does so as a sideeffect. It should then be unsurprising that these additional wrongs can exacerbate the wrongness of killing. This reinforces the idea that killing combatants on the just side is often worse than killing noncombatants on the just side because the latter is only pointless.

V. MODES OF KILLING BY THE UNJUST
Perhaps other moral principles that Lazar considers apply equally to the just and the unjust side.

A. Using and Eliminating
Consider his argument that the different modes of killing combatants and non-combatants supports Moral Distinction: Mode: killing non-combatants is often more gravely wrong because they are killed manipulatively, in contrast with combatants who are killed eliminatively.
Let us accept that manipulative killing is harder to justify than eliminative killing. On this view, a greater good is required to justify manipulative killing than eliminative killing. Nothing follows about the relative badness of manipulative and eliminative killing when these killings are executed in the service of a bad end, and Lazar needs an argument that there is such a difference to support Moral Distinction rather than Moral Distinction II.
Here is such an argument. Even when one has a good aim, it is normally wrong, and morally very bad, to use a person to serve it in a way that harms that person. We can see this from the fact that it seems wrong lethally to use one person to save five from being killed. If so, surely it is even worse to instrumentalize a person in service of a bad end.
I agree that it is very badly wrong to use a person to serve a bad end. But it is also very bad to eliminate a threat that a person poses where one has a very bad end. It is often thought less bad to kill a person eliminatively because doing so is necessary to prevent a rights violation, or a harm, that the person killed will otherwise inflict or VULNERABLE AND DEFENCELESS cause. The person's involvement in the threat weakens their objection to being harmed: she has a reason to welcome others ensuring that she does not violate rights, or is otherwise involved in inflicting harm. 10 But when a person on the just side is killed to eliminate a threat that she justly poses, her complaint against being killed is strengthened rather than weakened. She will be prevented from executing her admirable plan to protect others, and this is something she has powerful reason to reject. Just as harming a person is less bad if it is done to prevent her from bringing about a bad aim, harming a person is worse if it is done to prevent her from bringing about a good aim.
Lazar might marshal one of his arguments against this: he claims that even if some combatants have just aims, they do not know enough to justify their acts from the evidence-relative point of view. They thus act recklessly. This, he might argue, diminishes the significance of the fact that a combatant is prevented from doing something good. 11 I have some doubts that just combatants are typically reckless. Those who fought the Nazis in WWII don't seem reckless to me. But even if just combatants do not know that their cause is just, the fact that it is just is still morally relevant. Suppose that a combatant fights justly, and in the belief that his cause is just, but he does not fully appreciate all of the facts or moral considerations that justify his act. He surely has an objection to being prevented from doing the just thing, even though he is not fully cognisant of the justifying evidence or facts.
This also helps to rebut Lazar's argument that just combatants partially waive their rights not to be killed. He argues that as just combatants fight to protect non-combatants at risk to themselves, they conditionally waive their rights not to be killed. They say (or would say): 'if you are going to wrongly kill someone -a noncombatant or me -kill me'. This more limited view of just-com- batant waiver is much better than the more standard idea that combatants unconditionally waive their rights against being killed. I doubt that many combatants do waive their rights, or would do so at the point of death, where waiving them would increase the death toll overall. 12 But even if just combatants would prefer to be killed than to have non-combatants killed, other things equal, other things are often not equal in just wars. So they have not actually, or hypothetically, waived their rights. If a just combatant is killed he is prevented from making future contributions to the war. His death thus decreases the likelihood of winning. And when just combatants are killed in large numbers, this may result in overall defeat.
It is hard to believe that just combatants would prefer to be killed when compared with non-combatants when the killing would prevent them from securing victory, offering the much greater protection to non-combatants that victory would secure. Allied combatants in WWII did not plausibly even conditionally waive their right against being killed -they wished to protect non-combatants by defeating the Nazis, and their deaths prevented them from achieving this aim. They thus had every reason to prefer the deaths of some non-combatants than their own deaths where there was still a prospect of victory.
For this reason they cannot be said to have waived their rights against being killed. Perhaps they would waive their rights were other things equal (though I doubt it), but the fact that they would do so cannot make killing them less bad where they haven't done so, or would not do so in the actual situation they are in, where other things are unequal.
It is difficult to draw a general conclusion from this discussion. But it is at least casts doubt on the idea that our judgements about how difficult it is to justify manipulative and eliminative harming in pursuit of a good aim have direct implications for the gravity of our wrongdoing when we do these things in pursuit of evil.

B. Side-Effect Killing
I noted above a doubt about the implications of Moral Distinction with respect to side-effect killing on the just side of the war. I am even less convinced about this implication of Moral Distinction when applied to the unjust side.
Suppose that the unjust side of a war wins in one of two ways. The first involves killing many just combatants as a side-effect. The second involves killing as many non-combatants as a side-effect. Moral Distinction implies that winning the war in the latter way is normally morally worse. I don't see why. In a moment I will cast doubt on one set of Lazar's arguments to support this view -that it is worse to kill non-combatants because they are vulnerable and defenceless.
Aside from those arguments, it might be thought equally bad, or even worse, to kill just combatants as a side-effect than non-combatants. Consider: Whom to Kill: Alf wrongly threatens to kill Bob. Charlie is trying to defend Bob against Harry, but Harry is much more powerful than Charlie. Dave stands by and does nothing. Alf kills Bob in one of two ways: in the first way he kills Charlie as a side-effect; in the second he kills Dave as a side-effect.
Obviously, this simplified case is in some respects it is unlike war. However, it does have some salient similarities to war. Charlie is like a combatant on the just side, who aims to protect non-combatants against wrongful attack. Dave is like a non-combatant who does not do so. Now consider three views: (1) it is worse for Alf to kill Bob and Dave than Bob and Charlie; (2) it is worse for Alf to kill Bob and Charlie than Bob and Dave; (3) it is equally bad for him to kill Bob and Charlie and Bob and Dave.
I have some sympathy for (2). Mike puts himself at serious risk for the sake of Jake. Alf's wrongful conduct gives rise to this admirable decision. This seems intuitive to me both in the case where Charlie and Dave have a duty to intervene, and in the case where doing so is supererogatory because of the risk they will face. This seems to make it worse for Alf to kill Charlie than to kill Dave. Many people think it especially bad to kill police officers when they are acting well for reasons like this, and I am tempted by that view. But even if this is not right, (1) is very hard to believe. Even those who are not persuaded that it is worse to kill police officers than ordinary civilians when the police officers are acting well do not think it better to do so. Why think differently about combatants?
Perhaps it might be argued that these stylized examples tell us little about war. Admittedly, we should not draw conclusions about war too quickly from simplified interpersonal cases like this. But the lesson does seem to extend to war. When combatants support a just cause, they admirably put themselves in danger to protect those people that are threatened by the unjust side. This makes it especially wrong for the unjust side to kill them.

VI. VULNERABILITY
Lazar does offer some arguments to support Moral Distinction that more naturally apply to the unjust side of a war. Even if his arguments are successful, I doubt they are sufficiently powerful to support Moral Distinction all things considered, given the arguments against it just offered. It is nevertheless interesting to see whether they offer some support for it.
This section focuses on Lazar's view that it is more gravely wrong to kill non-combatants because they are vulnerable. The next section focuses on his view that it is more gravely wrong to kill them because they are defenceless. 13

A. Lazar on Vulnerability
Lazar understands vulnerability probabilistically in two ways: (1) Simple vulnerability: X's vulnerability to a threat, T, is the overall expected harm that T will cause to X. (2) Conditional vulnerability: X's vulnerability to a threat, T, is the overall expected harm that T will cause to X if T occurs.
These ideas come apart because simple vulnerability makes X less vulnerable to T in virtue of a lower probability of T occurring whereas conditional vulnerability does not. Lazar claims that each kind of vulnerability matters morally. One question is whether these ideas well capture what vulnerability is, or whether they describe something else. I am not sure. At first sight, vulnerability seems dispositional, like fragility. Admittedly, vulnerability and fragility are different in this respect: the semantics of vulnerability invite us to consider what X is vulnerable to -some 13 Lazar's views are defended in SC ch.5. bad thing that might happen to X. For example, X may be vulnerable to the flu. This is not true of fragility. X is fragile if X is generally disposed to break; X is not fragile to any specific thing. Nevertheless, I have some sympathy for a dispositional view. As this debate is mainly terminological, though, let us leave it aside and focus on the notion that Lazar actually focuses on: the risk that a person will be harmed if a certain action is performed.

B. Intentions, Side-Effects and the Risk of Death
Lazar, I think, believes this: Risky Killing: If D wrongly kills V, the gravity of D's wrongdoing is proportionate to the magnitude of the risk that D's act kills V. This principle has the inviting implication that how grave it is to kill a person by reckless driving depends on how recklessly one drove, where the degree of recklessness is a function of the magnitude of risk that one imposed on V. This conclusion is sufficiently uncontroversial that any plausible view should yield it. However, while it is sometimes true that the degree of risk of harm that one imposes on others is morally significant, Risky Killing is too general.
As we have seen, the combatant/non-combatant distinction might be understood to apply generally to all killing in war, or more narrowly to intentional killing, or to intentional killing by the just side. I have argued for the narrowest of these readings above. Here I focus on the problematic implications of Risky Killing in the case of intentional killing. The probability that a person will be harmed is normally irrelevant to the gravity of one's wrongdoing if one succeeds. For this reason, arguments that Lazar offers in favour of the significance of the magnitude of risk that one person imposes on another are irrelevant to wrongness of intentionally killing noncombatants.
Consider the following circumstances: (a) X and Y are equally likely to be innocent. (b) D intentionally kills both X and Y with independent acts and without justification. (c) When he shot, D had a 20% chance successfully killing X and an 80% chance of successfully killing Y.

VICTOR TADROS
Lazar argues: (1) Killing an innocent person is worse the higher the probability, when you acted, that your action would kill an innocent person.
(2) When D shot X he had a 20% chance of killing an innocent person; when D shot Y he had an 80% chance of killing an innocent person. (3) Therefore it was morally worse for D to shoot X than Y.
This argument seems mistaken. To remove distractions about a), suppose that X and Y are innocent, and D knows this. D intends to kill both X and Y. The fact that D had a lower probability of succeeding in what he intended to do does not make his killing of X less bad.
To see why, let us see why the magnitude of risk is relevant in side-effect cases. Suppose that D kills X as a side-effect of ving. The badness of D's act plausibly depends on the probability that X would be killed, as I have suggested in the case of reckless driving. The badness of killing a person by reckless driving depends on how reckless the driving was, and that depends on the probability that the driving would cause harm. Here is why. Suppose that D drives home for some good reason -say to get home to watch the football. It counts against his doing this that he might kill X as a side-effect. How much it counts against his doing this depends on the probability that X will be killed. His reason for action -watching the footballcounterbalances this badness. If the chance that X will be killed is very small, D's reason is sufficiently powerful to justify his driving. Driving, in this case, is not wrong from the evidence-relative point of view. D can offer the fact that X might well not have been killed to explain why his driving was not wrong in this sense.
As the magnitude of the risk increases, the reason that he offers is outweighed by the risk, and his driving becomes wrongful; how seriously wrongful it is plausibly depends on the extent to which the countervailing reason weighs against the reason in favour, and that depends on the magnitude of the risk imposed. Now return to the case where D intends to kill X. D cannot normally offer the fact that X might not have been killed as a justification for executing his intention. The fact that he was trying to kill X normally makes the fact that he might have failed irrelevant. He cannot appeal to this fact, as it was a fact that he was trying to VULNERABLE AND DEFENCELESS avoid bringing about. This is why the punishment that a sniper who murders his victim receives does not depend on how far he was from his target, or whether he was a good shot. As he was trying to kill the target, the fact that he had a high chance of failing is irrelevantit cannot help to justify his act. This is one reason why we cannot do away with intentions in mens rea, and simply rely on recklessness. 14 Were we to do so, we would be inclined to exonerate intentional killers with low levels of competence.
There is an exception to the rule that risk does not matter in the case of intentional killing -cases in which the person does not form the intention to kill because she is motivated to kill, but for some other reason. In such cases, the person can hope that she fails, consistently with her intentions, and she can get credit for this depending on the probability of failure. Consider: Duress: Boss threatens D that if D does not try his hardest to kill X Boss will harm D. Boss can determine with perfect accuracy how hard D is trying to kill X.
It seems easier for D to justify trying to kill X if D knows that he is inept than if he is skillful. The more likely he is to succeed if he tries, the harder it is for him to justify trying. And if he does kill X, and it was wrong for him to try, the gravity of D's wrongdoing is greater if he was more likely to succeed. But this is a rare case in which D is motivated to try to kill X, but he is not motivated to succeed. Just as our normative reasons to try and our normative reasons to succeed can come apart, 15 so our motivating reasons to try and our motivating reasons to succeed can come apart. As D does not wish to succeed in killing X, he can appeal to the chance that X will survive in an overall justification of his decision to try to do so. This exception, though, does not vitiate the general principle that the probability of failure is irrelevant to the gravity of intentional wrongdoing. This helps us to understand why the risk that a person is not liable to be killed is significant to the gravity of wrongdoing where the risk of success or failure in killing is not. Suppose that D kills X believing that X might be a wrongful attacker. Suppose, also, that the chance that X was a wrongful attacker was too low to justify D's act from the evidence-relative point of view. Plausibly, the lower the probability that X was a wrongful attacker, the greater the gravity of D's wrongdoing.
We can now see why. The possibility that X was liable to be killed can help to justify D's act. The higher the probability, the better the justification. When D's act was unjustified, the higher the probability that X was liable, the less serious was D's wrongdoing. This is because it was not part of D's plan to kill a non-liable person.
Rather than Risky Killing, I think we should support: Risky Killing II: If D wrongly kills V, and he does not kill V intentionally, the gravity of D's wrongdoing is proportionate to magnitude of the risk that D's act kills innocent others. If D wrongly kills V intentionally, the gravity of D's wrongdoing is proportionate to the magnitude of the risk that V was not liable to be killed.

C. Do Non-combatants Face Greater Risks of Death?
Lazar might argue that even if I am right, Risky Killing II provides some support for Moral Distinction. He might claim that Risky Killing II makes it worse to kill non-combatants than combatants as a sideeffect, because of the greater risk that is imposed on them. And it is worse to kill non-combatants than combatants intentionally because of the greater risk that they are not liable to be killed. The idea that non-combatants are less likely to be liable to be killed is plausible for the unjust side. It is less often true for the just side. This is for the reason noted above: neither combatants nor noncombatants on the just side are liable to be killed. In WWII, for example, it is not very plausible that it was less seriously wrongful for Nazi combatants to kill Allied combatants on the grounds that they were liable to be killed. Hence, this aspect of Lazar's view provides much stronger support for Moral Distinction II than Moral Distinction.
Lazar's other idea -that non-combatants are exposed to greater risks if they are attacked than combatants -might be thought to apply to both sides in the war. However, I doubt that the relevant facts uniformly, or even regularly, track the combatant/non-combatant distinction.

VULNERABLE AND DEFENCELESS
First note that Risky Killing II can support Moral Distinction only if risk distribution matters independently of expected outcomes. We can see this clearly from the fact that the overall magnitude of a risk that one inflicts on a population depends on the overall expected outcome. For example, consider a population of two {A and B}. Hold constant the expected outcome as one death. This is consistent with a range of risk distributions: 100% risk to A and 0% risk to B; 100% risk to B and 0% risk to A; 50% risk to A and 50% risk to B; 25% risk to A and 75% risk to B, and so on. If only expected outcomes matter, and risk distribution does not independently matter, Risky Killing II supports any action that reduces the death toll, regardless of who is killed. If killing non-combatants reduces the death toll, and risk distribution does not matter, Risky Killing II supports killing non-combatants.
Perhaps Lazar might argue that killing combatants rather than non-combatants regularly reduces the death toll. This is not very plausible. Consider a war that is fought between sides with different military capacities, as most wars are. Successful kills by combatants on the weaker side against combatants the stronger side will tend to prolong the war. The weaker side will lose anyway; they will just take longer to do so. This will substantially increase the death toll. Killing non-combatants, in contrast, typically achieves nothing, or may even strengthen the resolve of the stronger side leading to a swifter resolution, reducing the overall death toll.
So Lazar must rely on the idea that risk distribution matters independently of the expected outcome of a person's risk-taking actions. He thinks that it does. Were this true, it might support the view that it is sometimes wrong to inflict harm on non-combatants rather than combatants, even when doing so reduces the death toll overall: this is so if the conditional risks that non-combatants face are higher than the conditional risks that combatants face. Lazar argues that non-combatants do face higher conditional risks of this kind.
However, whether this is true depends on how the non-combatants are harmed. To see this note that targeting a large group of people, where only some of these people will be harmed, does not impose a high risk of death on any particular person. Now suppose that the just side can end the war either by attacking combatants or non-combatants. Randomly targeting a large non-combatant popu-lation will be effective in ending the war -this might be because the civilians will be terrorized, or it might be because the civilian infrastructure that is necessary to support the war will be easily undermined. It may well improve the risk distribution to randomly target civilians in the large population when compared with attacking combatants on the battlefield in the conventional way: the former act imposes a low risk of death on all members of a large population, where the latter imposes a high risk of death on particular identified individuals.
But attacks on non-combatants often impose small risks on each member of a large population: aerial bombardment of major cities is the obvious example. Lazar's view seems to support this practice over targeting combatants on the battlefield. I have some doubts about the moral significance of risk distribution -if it is significant, I doubt that it is very significant. If we have a choice between killing a particular person, or killing a person at random from a group, we do not have a strong reason to prefer the latter to the former. Lazar thinks otherwise. Either way, Risky Killing II does not favour Moral Distinction.

D. Defencelessness
In addition to his views about vulnerability, Lazar thinks that it is worse to kill the defenceless than those who are not defenceless. Defencelessness is bad, he claims, because the defenceless are dominated, in Philip Pettit's sense. 16 Lazar also thinks that because defencelessness is bad in itself, it is especially wrong to kill or harm the defenceless.
I doubt that it is a bad thing to be defenceless, in Lazar's sense, or to be dominated in Pettit's sense. Consider: Happy Island: X and Y live together on a desert island for the whole of their lives. They are very fond of each other, live a terrific life together full of joy, laughter, and cooperation, fully respecting the importance of each other's valuable options. Both are very robustly disposed to respect the importance of each other's valuable options, so there was never any real chance that they would wrongly restrict these options. Both die fully contented with their lot, celebrating their good fortune to have been abandoned with such great company. X could have completely restricted Y's valuable options without sanction and Y could have done the same to X without sanction -each could do so by harming the other in a way that prevented retaliation, and there was ample opportunity for each to do this had they so wished. Neither even thought of making a threat to the other, given how well-disposed they were to each other. X and Y are each very vulnerable (in Lazar's conditional sense of vulnerability) and defenceless. Each also dominates the other in Pettit's sense. If X threatens Y, Y is certain to be harmed, and Y is completely unable to defend himself; and vice versa. It seems plausible that X and Y are each very vulnerable and defenceless, regardless of whether Lazar's account of these things is correct. But these facts about X and Y seem in no way morally problematic. They do not dominate each other, in any morally important sense of the word, and they have no reason to regret the power that each has over the other. 17 Given this, it would be surprising if it were worse to kill the defenceless. And, indeed, I doubt that it is. Consider: Island Kill: D is flying over Happy Island, on which X and Y live, more or less in the circumstances in Happy Island. X is an expert with an anti-aircraft gun that is on the island. Y is not. D's plane has been hijacked by evil terrorists, who threaten D that if he does not kill either X or Y, D and all of the passengers on D's plane will be killed. If D fires at either X or Y, X will fire his anti-aircraft gun at the plane, and there is a chance that this will avert the attack. D has an equal chance of killing either X or Y if he targets that person.
Lazar implies that it would be worse for D to shoot Y than X. X and Y are equally vulnerable to attack by D, but only Y, not X, is defenceless. This is because X is able to defend himself, whereas Y is reliant on X to defend him. I don't find this worse.
Let us, though, examine Lazar's arguments for his conclusion. He claims that D ought to attack X rather than Y because if D attacks Y, he makes Y beholden to X's whims -X's decision whether to defend Y. If D attacks X, in contrast, he does not make X beholden to Y's whims.
But when X is excellently disposed either to defend himself or to defend Y, the fact that X could decide differently seems irrelevant. True, if X decided not to protect Y, Y would have a greater risk of being killed. In this way, Y depends on X. But that seems unimportant. Depending on others in this way is not a bad thing when one has every reason to be very confident that one can do so. It is a standard feature of complex cooperative societies that we all depend on others to do the tasks that are necessary to advance our freedom and well-being. This is not a regrettable fact about such societies. It might even be something to celebrate.
Here is a second idea of Lazar's: how well our lives go depends not only on 'how things actually turn out, but on what could have happened if things had gone differently'. 18 Lazar does not defend this claim, and it doesn't seem true. Suppose that I contract some fatal disease and die. Is it really true that how well my life goes depends on whether I could have avoided the disease, or on whether I would not have died in some nearby world? I doubt that it is.
Overall, neither vulnerability nor defencelessness support Moral Distinction. Either they fail regularly to track the distinction between combatants and non-combatants, or they seem false.

VII. CONCLUSION
As usual, my comments are critically focused. Although I think that his main conclusion -Moral Distinction -is false, many of Lazar's arguments are successful in supporting a narrower principle: Moral Distinction II. And Moral Distinction II is an important principle -it is the principle that ought to guide those with just aims.
Of course, many people without just aims think that their aims are just. And this affects the general legal rules and social conventions that we should favour. Perhaps we should tell people that something like Moral Distinction is true because that would have good effects, given that people can't tell wrong from right in war. But that does not make Moral Distinction true, and I doubt that it is.

OPEN ACCESS
This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/ by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.
School of Law University of Warwick, Coventry, UK E-mail: v.tadros@warwick.ac.uk