Ethics Discussions at PEA Soup: Cécile Fabre’s “War Exit,” with Critical Précis by Helen Frowe

27 Replies to “Ethics Discussions at PEA Soup: Cécile Fabre’s “War Exit,” with Critical Précis by Helen Frowe”

1. I like the article a lot, and agree with a lot of it. And Helen’s commentary is really helpful.
   Cecile, the part that I was most at odds with was the part on sunk costs and proportionality (635-7). Leave aside the redemption argument for the moment (I might say more about this later), and just focus on how we should, in general, understand the moral significance of unexpected harms that we have caused as a result of our aim to pursue good aims.
   Consider:
   TWO CHOICES:
   At t1, D has two options with the following expected values:
   1) Save 100 from death killing 10 as a side effect.
   2) Refrain from saving the 100 from death.
   Suppose that it is proportionate for D to kill 10 as a side effect of saving 100. If so, D acts evidence-relative permissibly by choosing 1).
   D chooses 1). In attempting to save the 100 lives, D accidentally kills 100. D realizes where he has gone wrong.
   At t2 he knows he can:
   3) Save the 100 killing 10 more as a side effect.
   4) Refrain from saving the 100.
   Here is how to understand one argument you make: if D chooses 3) he kills 110 to save 100. This is (fact-relative) wrong. Hence D ought to choose 4).
   But, in response, if D chooses 4) he kills 100 to save no one, which is also fact-relative wrong. Thus the fact that he will act in a way that is fact-relative wrong if he chooses 3) is no argument in favour of him choosing 4).
   The same response can be stated in terms of proportionality. If D chooses 3) he will have acted disproportionately, by killing 110 to save 100. But, in response, if D chooses 4) he will also have acted disproportionately, by killing 100 to save no one.
   His current choice situation is thus better understood as follows:
   If he chooses 3) he will kill 110 to save 100
   If he chooses 4) he will kill 100 to save no one.
   To assess what D is permitted to do, then, consider:
   ONE BIG CHOICE:
   D can either:
   5) Kill 110 to save 100; or
   6) Kill 100 saving no one;
   D, let us be clear, cannot do nothing – this is appropriate, given that in TWO CHOICES D cannot, at t2, make it the case that he has killed no one. [In case you find this hard to imagine, we can model it by imagining that D is clearly not permitted to do nothing – for example, if D does nothing a large nuclear blast will kill millions, including all of the people in the problem.]
   Furthermore, we should add that 100 of the 110 in 5) are the same people who will be killed if D chooses 6), and will be killed in an identical way at an identical time – this is also appropriate as the same 100 will have been killed in TWO CHOICES whatever D does at t2.
   In ONE BIG CHOICE, D should discount the lives of the 100, as they are doomed to be killed by D whatever D does, and in an identical way at an identical time. Thus D’s decision is, morally speaking, just the same as the decision to kill 10 to save 100. As this is ex hypothesi proportionate, D is permitted to choose 5). If I am right that TWO CHOICES is, morally speaking, like ONE BIG CHOICE, we should see the lives lost as sunk costs in TWO CHOICES. D is thus permitted to choose 3).
   You object that this gives proportionality no bite. I don’t see that. First, we acknowledge that D’s overall course of conduct from t1 was disproportionate in the fact-relative sense. We just recognize that at t2 D’s act was disproportionate in this sense whatever D now does. Secondly, at all times, D must act proportionately in the evidence-relative sense. What more bite do you want proportionality to have?
   A further objection you make is that this would imply that the lives of the 10 were protected at t1 but are not protected now (at the bottom of p.637). I can’t say that I fully understand that argument. But it is worth noting that if D is not permitted to choose 3), the lives of the 100 were protected at t1, by D having a permission that he is willing to take to protect them, but not at t2. So if I understand it, your thought cuts both ways.

2. I am deeply grateful to Helen for those thought-provoking comments. I will post my replies in three different posts, and turn to Victor’s comments on sunk costs presently.
   Helen begins the critical part of her note with the following observation: whereas I claim that jus ad bellum regulates the initiation of war, it in fact applies continually to phases of the war. On that view, we ask at any given time whether this particular phase is ad bellum just. Of a war which has just and unjust phases, we might not be able to say that it is ad bellum just overall.
   This might seem a merely verbal point. It is not. Helen is right to draw attention (as I and others have done, notably Jeff McMahan and Saba Bazargan) to the fact that wars have phases, and that those phases will elicit different moral judgements. She is also right to note that, once we accept that point, and from a reductivist point of view, we might have no choice but to claim that each individual combatant’s act of killing (or contribution to such acts) should itself be judged at the bar of jus ad bellum. And once we claim that, then the distinction between jus ad bellum and jus in bello breaks down. Far from being verbal, the point in fact undermines the usual conventional way of thinking about war ethics – in other words a good millenium of theorising.
   I am generally in favour of upsetting intellectual traditions. I agree (indeed, as Helen notes, I say in the article) that the traditional categories lack deep conceptual and normative significance. I also agree that war is no different from the use of force in other contexts, and that its constitutive acts of violence should each be assessed by the same standards as non-war acts of violence. Finally, I agree that once we decompose a war into phases, we must further decompose those phases into their constitutive acts of violence.
   However, I still think that it is useful to identify what it is exactly that we are evaluating, and that those labels help us do this without confusion. At t1, we, via our leaders or representatives, are deciding whether to go to war. We have a sense of the losses we will need to inflict or incur in pursuit of our cause; whether there are other, less costly, options than war; whether going to war now would risk jeopardising peace in the longer term. At t2, having decided for war, we – again via our leaders or representatives – are deciding whether to *continue* to go to war. Then too we have a sense of necessary losses, whether suing for peace (an alternative to war) would be less costly. But ex hypothesi, we are at war already. The clock does not turn back to t1, and when deciding at t2 to continue (or not), we must take into account precisely what has happened since. It makes sense, thus, faithfully to keep with the latin and refer to the former decision as ad (towards) bellum. It also makes sense to refer to specific acts of violence carried out by soldiers once that decision has been made as in (within) bello.
   Frowe might respond that those acts of violence themselves constitute a decision to carry on with the war, and that the category in bello is meaningless. But I don’t think that this would be quite right. For in wars which are conducted by relatively organised actors located in a chain of command, those specific acts are only partly tantamount to a decision to continue the war. To be sure, unless soldiers are willing to fight, the war cannot go on. However, in the longer paper of which this article is an extract, I distinguish between the act by which armed forces, whether as a whole or in part, surrender or withdraw from the battlefield, from the act by which a whole community is deemed no longer to be engaged in hostilities. Whilst armed forces may (at least in some cases) justly surrender or simply withdraw without the permission of their leaders, only properly authorized actors may surrender or sue for peace. For reasons given in the previous paragraph, that latter decision – whether to continue to be at war or not – is distinct from both the decision to go to war in the first instance, and from carrying on killing on the battlefield. It does warrant its own convenient expository label, originally coined by Moellendorf, of jus ex bello.

3. In the second part of her note, Helen expresses scepticism of my claim that, sometimes, a belligerent whose war is unjust at t1 might nevertheless be justified into continuing at t2. In particular, she questions my point that an unjust cause can morph into a just cause. My example is one where A goes to war against B with a view to bringing about regime change in B and replace it with a regime which will be more friendly to its own intersts. At t1, B’s regime does not provide A with a just cause for war. At t2, however, the war has resulted in the overthrow of that regime and the resulting power vacuum leads to routine rights violations. My claim is that regime change at t1 is not a just cause, but has become a just cause at t2. Frowe thinks that A in fact responds to two different causes: advancing its own interests at t1, and securing the basic rights of individuals at t2.
   I do think that there is a misunderstanding, which flows from my own lack of precision, and from lack of clarity generally in war ethics on the distinction between causes for war and a war aims. Here is what I meant to say . At t1, A wants to set up a new regime within B – not because B’s current regime violates the rights of its own citizens or of A’s citizens, but because its policies are generally inimimal to A’s interests. For example, B’s regime consistently refuses to give A preferential trade terms. At t2, when that regime has been toppled through the war and B’s citizens suffer grievous rights violations at the hands of rival factions vying for power, A *still* wants to set up a new regime. In the article I do not specify whether A’s policy is driven by its own interests, the plight of B’s citizens, or both. If A’s policy is driven by its own interests, as in at t1. Frowe is right to say that this particular policy cannot be a just cause (since no rights violations have been committed against A by B’s regime), and it is not the case that A’s unjust cause morphed into a just cause at t2. Crucially, however, if regime change can be construed as a cause for war, then to the extent that at t2 this policy would protect the fundamental rights of B’s citizens, it can be regarded as just cause (which at t1 it was not.)
   But perhaps this is wrong: perhaps regime change should be regarded as a war aim, fulfillment of which would lead succesful prosecution of the cause (better trade relations for A – not a just cause for war, the protection of those fundamental rights – clearly a just cause.) If so, I should have described the case, not as one where A’s initially unjust cause morphs into a just cause, but one where A’s initially unjust aim morphs into a just aim. I raise this largely because the vocabulary of aims has tended to fall into disuse in war ethics, to be replaced with the vocabulary of causes. But perhaps this is a mistake: perhaps we should resurrect the requirement that the aim of war should be just too.

4. The case of initially unjust wars prompts Helen to make another comment. I argue in the paper that there are differences between cases where belligerent A wrongfully brings about circumstances under which its initially unjust war becomes just (which furnishes its leaders with a justification for continuing), and cases where A’s moral situation changes as a result of factors which are beyond its control. I also argue that there is a further difference to be drawn, between cases where A’s war is unjust at t1 because it lacks a just cause, and cases where it is unjust at t1 because it breaches one or more of the other just war requirements. Those two distinctions cut across each other. Helen’s main point is that, generally, if one acquires a just cause, how one acquired it (through one’s wrongdoing, through a rightful act, or neither) does not make a difference to the permissibility of one’s continuing with the war. She also thinks that whether or not one benefits from continuing with the war is irrelevant. Now, on that very last point, I did not clarify why I thought that benefiting from an ongoing war might be relevant. I suspect that Helen is right, and that the point is distracting. I am happy to take it back. But I still think that whether or not A wrongfully brought about a now just state of affairs makes a difference, not to whether or not it may continue, but (as I note in the last paragraph of the article, before the conclusion) to the costs its troops ought to be willing to incur, relative to those they will inflict on the innocent, when continuing with the war. That said, I agree with Helen’s last point that further comparisons between wars which fail to meet different ad bellum conditions would be fruitful. If the article prompts more work in that area, it will have achieved one of its aims.

5. Thanks to Cécile for the paper and to Helen for the commentary. I wanted to raise a point about the redemptive argument. (which Victor in his comment deferred for now). Cécile argues that it faces the following dilemma: either redemption alone furnishes us with a new just cause, in which case it is tantamount to claiming that we are justified in killing enemy soldiers even though, ex hypothesi, these acts of killing are not construed as forestalling a lethal threat to the dead; or it claims that we would be justified in killing enemy soldiers because, and only if, they would thereby and at the same time prosecute their just cause, in which case the original just cause alone provides our soldiers with a justification for killing enemy soldiers. In the latter case, Cécile says, “redemption does no justificatory work at all”.
   But it seems to me that the good of redeeming the deaths of our dead soldiers (or, alternatively put, the evil of averting their unredeemed deaths) can affect the proportionality constraint in a way that makes it easier to satisfy it. Consider this example (which doesn’t involve liability at all).
   The evidence strongly suggests (at t1) that only way to save 20 is to kill an innocent ‘A’ (who is not among the 20). I kill A, but unfortunately doing so doesn’t avert the threat that the 20 face. The evidence now suggests (at t2) They will die unless I kill *another* innocent, ‘B’.
   At first, it might seem that the (evidence relative) moral calculus is the same at t1 and at t2 – save 20 at the cost of 1. But the redemptive argument suggests this is not so. At t2, I am in a position to do *more good* than at t1, even though I’d be saving the exact same number of lives. This is because at t2, I’d be saving 20 lives, and *redeeming the death* of A. Since we are in a position to do more good at t2, it is easier to satisfy the proportionality constraint. So suppose it is wrong to kill 1 *and to punch another in the face* if necessary order to save 20. Adding the value of redemption, it might indeed be permissible to do so. Notice that this does not push us to the first horn of the dilemma since punching the innocent in the face is by hypothesis necessary to save the 20.
   So my thinking here is that we can incorporate the value of redemption into the proportionality constraint in a way that makes it easier to satisfy, without claiming that redemption is a basis of liability.

6. Hi Cecile (and all). I’m still struggling to make sense of the idea that the ad/in labels are useful because they identify exactly what we are evaluating, as if we are evaluating different things when we talk of jus ad bellum compared to jus in bello. Take your suggestion that when we’re thinking about JAB, we need a sense of the losses we will inflict. How do we achieve this without thinking about the specific actions we would need to undertake in order to win – that is, the specific acts of violence carried out by soldiers? So what’s the different object of evaluation here, compared to JIB? (The natural response is that we’re judging *all* the specific actions when we talk about JAB, but then the same point about phases of the war applies – there’s no obvious sense in which we can meaningfully say that the ‘war as whole’ is just by averaging across these actions, and it’s not clear why any kind of averaging approach is morally interesting.) And the questions of whether to begin war and whether to continue war, whilst ostensibly different, surely collapse into the single question of ‘May we use force now?’. And you and I agree that this question is settled by reference to a single set of criteria – a combatant deciding whether to fight should consider whether the war is still necessary and so on, just as a leader should. So I’m still not seeing what the different objects of evaluation are, such that we need two labels to distinguish between them.

7. Quick P.S. to the above: it might be helpful if you could say what you take the conceptual significance of the categories to be. You say that they lack ‘deep conceptual and normative significance’, but above you iterate the idea that they evaluate different things, which seems to be a claim about their conceptual significance. Is there such a thing as non-deep but nonetheless genuine conceptual significance? And is this significance to be understood in terms other than the facts that they evaluate?

8. Good morning everyone. Apologies for the relatively short posts this morning (still recovering from UK elections night.)
   Reply to Victor on sunk costs
   Victor disagrees with my rejection of sunk costs. I remain unpersuaded by his forceful criticism. For a start, my argument against D killing 10 more as a side effect of saving 100 (given that he already has killed 100 while attempting to save 100) is not that so doing is fact relative wrong. In fact, I agree that both options (3) and (4) are fact relative wrong. Rather, my worry about the sunk costs argument is this – that it permits us to do more to our enemy at t2 than one would have been permitted to do at t1 in one go (or, as Kamm puts it in The Moral Target, ‘in one fell swoop). It seems to me in other words that being or not being able to inflict harm in one fell swoop cannot make a difference to agents’ fate.That view draws on luck egalitarian worries about the bearing of contingent, unchosen factors on life and death.
   At the same time, to stop with the war now on those grounds is also tantamount to allow agents to suffer harm who would otherwise have been protected from it (had we continued), on exactly the same contingent basis. Victor is right about that. But If one takes the view that harming is, other things equal, worse than allowing harm to happen, and if in this case things are roughly equal (notably the number of lives which we would destroy by continuing and the number of lives which we would allow to be lost (at the hands of our victorious enemy) by stopping), the most that we can say is that it would be morally worse to continue than to desist.
   I have a question for Victor: when, if ever, on his view, are we not morally allowed to regard lost lives as sunk costs?

9. Reply to Saba
   Saba presses me to explain more clearly why I reject the redemption argument in favour of continuing with the war. I say in the article that redemption (of lost lives) plays no justificatory work. Saba thinks that it can be incorporated into proportionality in a way that strengthens justifications for continuing with the war/killing. I should say that i now think my argument in the article is pretty weak, notably at ft 10. In particular, I now think, in line with Saba’s comment, that one need not think of redemption as a basis for liability to be killed. But even if redemption can in principle be thought of as part of proportionality and thus as bolstering all things considered justifications for war, upon closer inspection it is not clear to me that it is in fact so much of a good that it can help offset the harms of killing people when those harms are already justified anyway. To be clear, I am not denying Jeff’s claim (in ‘Proportionality and Time’) that the death of a just soldier is less bad for him if the war is won than if it had been lost. Nor am I denying that his parents may plausibly regard the fact that he did not die in vain as consolation for their losses. My worry is this. In Saba’s example, we can at t2 decide either not to save the 20 or to do so at the cost of B’s life.By doing the latter, we do more good since we not only save 20, we also diminish the barness of A’s death. But given that it is ex hypothesi permissible to kill B in the course of saving the 20, I still don’t see why it is easier to satisfy proportionality constraint than if A’s death could not be redeemed.
   To illustrate: compare the case where A is terminally ill and would have died six months later anyway, with the case where A would have lived a healthy 30 years had he not been killed in the rescue mission. Presumably, there is nothing to be redeemed in the first case. Does Saba think that there is a sense in which proportionality is better satisfied in the former than in the latter case? I don’t see it.
   (I should say that I am amongst those who believe, as indeed Jeff seems to do, that the redemption of just soldiers’ deaths has no weight at the bar of wide proportionality. But that’s another topic.)

10. Hello Cecile
    Thanks for that – UK PEA Soupers should be excused if our responses are tainted by post-election depression. It’s tempting to start a thread on alternatives to democracy!
    Here are two worries about your response.
    First. I wasn’t clear how the doing and allowing distinction is going to help here. In TWO CHOICES the 100 have been killed. As they have been killed, although accidentally, D might especially regret their deaths compared with deaths he would have allowed to occur at t1 had he not attempted to save the 100. But now he is in a position where he can save 100 killing 10. Ex hypothesi the doing and allowing distinction would not rule this out were it not for the fact that 100 have been killed. It is not clear why it makes a difference that D has killed other people in a failed attempt to pursue his goal in the past.
    Perhaps you might argue something like this: if D kills a further 100, he will have 110 killings on his ledger. So perhaps you might argue:
    If D kills, he has 110 deaths on his ledger and 100 lives saved. If he refrains from killing D has 100 deaths on his ledger and no lives saved. This would suggest that even if D may kill 10 as a side-effect of saving 100, he ought to choose 6) in ONE BIG CHOICE. I find this extremely hard to believe. The most important consideration against D killing is the right that the 10, whose deaths are not inevitable, have against D. It is hard to see how these rights are affected by what D will inevitably do to the 100.
    The second worry is that you rely on the idea that D is permitted to do more to his enemy at t2 than at t1. I wasn’t clear why this is true. We both agree that D acts fact-relative wrongly at t1, so there is no question of D’s act at t2 rendering his complete course of conduct permissible. Your suggestion, I think, relies on the implicit assumption that ‘our enemy’ has independent moral significance over the individual lives that constitute the enemy. Given your other commitments, I doubt that you are happy with this assumption. The individuals have rights qua individuals. The fact that they are part of the same group does not matter (or at least does not normally matter – perhaps things might be different where a family will be wiped out, or a community that has moral significance over and above the individuals that constitute it).
    To answer your question at the end: past accidental harms may difference when we are concerned with harms that will repeatedly be inflicted on entities that have non-derivative moral significance – most obviously individuals. For example:
    TWO CHOICES 2:
    At t1 D has two options with the following expected values:
    1) Save V from death, harming X as a side-effect to degree h.
    2) Refrain from saving V, leaving X unharmed.
    Let us suppose that inflicting h on X to save V is proportionate. D attempts but fails to save V, and in doing so inflicts h on X.
    At t2 D knows he can either
    3) Save V from death, harming X as a side-effect to degree h, so that X will have suffer 2h overall.
    4) Refrain from saving V, leaving X unharmed.
    In this case, 3) may be wrong. The fact that X has already been harmed at t1 can make a difference to his right to suffer the same harm at t2 (there are various reasons that might be offered for this view which I won’t outline now). But this should not distract us in cases where we are concerned with different individuals. If X is replaced by an unrelated individual, Y, at t2, D may save V. The fact that D has harmed X at t1 makes no difference to Y’s rights not to suffer harm for V’s sake.
    In war, on the assumption that we are concerned with the rights of individuals rather than groups who have moral significance qua groups, I doubt the significance of sunk costs. An argument in favour of taking them into account relies on the idea that the fates of the people who we will go on to kill should depend on whether we have accidentally killed others in the past in pursuit of our aims. And that seems to conflict with the luck egalitarianism that you are inclined to accept.

11. Let me have another crach at dealing with Helen’s label-scepticism. All I mean to say is that it is useful to remind ourselves that what we are evaluating once the war has started, such as acts of killing therein (and in particular how to kill) is not the same as what we are evaluating at t1, to wit a decision to start on that course of action. And it is not the same partly for reasons to do with changing circumstances as the war goes on (which we have to take into account when interpreting and assessing proportionality, necessity, etc.)
    But perhaps I am wrong to insist on this. Perhaps we should indeed abandon ad bellum, in bello and ex bello. But does Helen think that we should abandon post bellum as a useful category? For what we are justifying, after the war, are wholly different sets of actions or, rather, rights, duties, liabilities, etc: to wit, compensation, punishment, occupation, peacekeeping, etc. Even if we can still apply principles such as proportionality to those acts, they nevertheless are not the same as acts of killing and maiming. So I do think that there is a conceptual difference there. Does this make sense to Helen?

12. I think post bellum is a whole different kettle of fish, because it’s not about the justness of force, but rather what to do once we’ve stopped using force. But I am increasingly drawn to the view that ad bellum / in bello collapse into each other (and don’t even get me started on ad vim. I’m already cross about the election.)

13. Cécile — thanks for the response, especially given that you’re now defending against multiple lines of criticism from three different people.
    A few points:
    1. I agree that if the value of redemption can be included in proportionality, it won’t count for much.
    2. You worry that the value of redemption cannot help offset the harms of killing people when those harms are already justified. But in my amended example, the harms are *not* justified absent the value of redemption, since absent that value they violate proportionality.
    So, suppose that we can permissibly kill 1 to save 20, but that’s the threshold; inflicting any more harm than that would be disproportionate to the value of saving 20. (So we can’t, e.g., kill 1 and punch another guy in the face — call him ‘C’ to save the 20).
    So we kill A, and unfortunately that doesn’t avert the threat the 20 face. Now, the situation is slightly different. The only way to avert the threat is to kill B *and* punch C in the face. Recall that this combination of harms is by hypothesis too great to justify saving 20. It crosses the proportionality threshold. But we wouldn’t just be saving the 20 — we’d also be redeeming A’s life. Given the added value of redemption, maybe it’s permissible to impose slightly greater harms than the death of one innocent *if necessary* to save those 20. That is, we can permissibly kill B and punch C in the face.
    In the first attempt, killing A and punching C in the face would have been impermissible even if necessary to save the 20 since it would cross the threshold of proportionality; but now it’s fine since we bring about more value by doing, viz., we save the 20 *and* redeem the life of A. So to reiterate the harms we impose on B and C together, though necessary to save the 20, are *not* justified absent the value of redemption. Only by adding the value of redemption do those harms satisfy proportionality. (Now, a punch in the face is hardly the sort of harm we face in war, but as we increase the numbers of dead, then the value of redemption increases accordingly).
    3. I do indeed think there is something to redeem in the case where A would have died anyway. Killing an innocent infringes her rights, even if the killing deprives her of only six months (or just six seconds) of life. That being said, the less life lost, the easier it is to justify that rights infringements. But so long as there is an infringement, there is something to be redeemed.
    4. Your last parenthetical point confused me since it seems to undercut my examples, and so cannot be another topic! If redemption has no value at the bar of wide proportionality, then we cannot justify inflicting *any* amount of harm on an innocent by adverting to the lives redeemed by doing so. Yet that’s exactly what we do to C! But maybe I misunderstood your parenthetical remark.

14. To Helen: You know, I think that you are right. – and that we should dispense altogether with ad bellum, in bello, ex bello (and I agree that ad vim is really not helpful.) This has been really helpful, actually, so thank!
    To Victor: I need to think more about your first worry (in the first part of your response). As a preliminary point, though: DDA was meant to respond to your very last point (that, at the bottom of p. 637, my worry about the additional people who will be killed also applies to those whom we will not save if we continue.
    Re the second worry: I had in mind the following. If, at t1, D were to kill 110 to save 100, we would clearly say no. Yet if the 100 he accidentally kills are sunk costs and if we say that he may kill (a further) 10 at t2, we are saying that, in effect, it is morally permissible for him to have killed 110.
    I don’t see how my suggestion relies on the assumption which, you rightly say, I do not endorse, namely that merely belonging to the enemy has moral significance. But there is something to be said in favour of the claim that, the more people we kill from the enemy community, the greater the extent to which innocent members of that community will be harmed – in so far as they rely on those people for their survival (for me, an elderly person, to lose one source of help, ie my eldest child, is one thing; for me to lose my second child, is another, etc. ) This closely resembles the point you make about repeated harms, but note that it does not in any way rely on some appeal to groups.
    Now (big qualification): you might say that we are not in fact saying that. We are saying that it was fact-R impermissible to kill 100 at t1, that it is fact-R impermissible to have killed 110 overall, but that it is evidence-R permissible to kill 10 at t2 (to save 100). In the article I do not draw a distinction between fact-R and evidence-R. I implicitly assume the former. And at the bar of the former it is worse, I think, to continue at t2 than to desist. Even if that is right (which it may not be) we need to know whether D is evidence-R permitted in continuing. If he *knows* that he can save the 100 by killing 10 (which was your initial description of the case) then I think and concede that he may proceed. If he does not know but only has some reason to believe, etc., then the strength of those reasons would be decisive here in determing whether he may proceed.
    That very last point leads me to the following thought, that over time, the more often D gets it wrong and more often he tells us that, this time, he knows that he can save the 100 by killing 10, the more reason we have to believe that he will again got it wrong. So perhaps the worry about sunk costs is, in fact, a worry about the reliability of D’s decision making, such that the number of lives lost is in fact a proxy for faulty decision making. I will need to think more about that.
    To Saba: Re 1, great. Re 2: I take the point and falls back on the claim that redemption really cannot count for much. In fact: I doubt that it can count at all. It is not clear to me that we owe it to the dead to continue the job (thereby redeeming their deaths) when continuing with the job means killing more people some of who are not in fact liable to being killed. It is not clear to me either that we owe it to the loved-ones of the dead to continue with the job, given that continuing with the job means killing more (nonliable) people whose loved-ones will therefore grieve for those death. (As for my parenthetical point: by ‘another topic’ I meant the following:I took your initial post to say something like this:’ let’s assume that redemption is a good. IS best construed as a basis for liability or does it come under wide proportionality?’ My parenthetical point gestured towards saying something about the ‘if’ clause.)
    To all three of you: thank you so much for your comments. You really should not feel under any obligation to continue with this…I have found the discussion really helpful. Thank you!
    

15. Sorry for being slow to the party folks. I think Helen, Victor, and Saba, have already said much of what I would say in response to the paper, which I enjoyed, and which I think illustrates the unusual congruence in this field of topics that have both deep practical and deep philosophical importance. In particular, I think two issues stand out that are interestingly related to one another.
    The central question is how do we individuate actions for the purposes of ethical evaluation? The sunk costs problem speaks to how we individuate actions of a single agent over time. The jus ad bellum problem speaks to how we individuate actions by a number of agents either at a time or over time.
    There are a number of logical possibilities. One could argue, for example, that the proper object of ethical evaluation is an individual’s action at a time (call that synchronic evaluation). Then in the sunk costs case, the agent might have acted impermissibly at T1, but permissibly at T2. There would be no separate question of whether she had acted permissibly overall (at T1 and T2). And there’s some logic to this. After all, if we’re going to go diachronic, why stop at T1 and T2? Why not go back to all the agent’s other actions, and forward to all those—why not assess her life as a whole?
    The next possibility is individualist, but allows that sometimes we should evaluate actions together, diachronically, as part of a single campaign. When we specify hypothetical cases, it’s easy to see what counts as part of the campaign, because everything else is left out. But it would be really interesting to think about which actions by the agent get to count as part of the campaign. Is it temporally bounded? Are only actions that are aimed at the goal of the campaign to count? Or only actions that actually contribute? Only actions that she wouldn’t engage in if she were not attempting that goal?
    I think this is the view that underpins Cecile’s discussion with Victor—that we should assess the agent’s compound action at T1 and T2 as a whole. It then raises the interesting question of what one should do when one is engaged in a campaign that has already gone irredeemably wrong. That’s the issue of extrication that David Rodin in particular dwells on in his contribution to the symposium.
    (FWIW, my view on this is that sunk costs should be partly though not completely discounted, for the following reason: suppose that you’re fighting to save the lives of some group of victims, V. As more and more people die in the failed attempt to save V, their claim to aid diminishes—it becomes, in effect, used up—because there is a limit to the costs that they can expect others to bear on their behalf. I don’t think their claim ever decreases to zero, indeed it might not dip much at all—just asymptotically approach some high percentage of its original force. But I do think that it declines.)
    Anyway, the problem of how to individuate actions by an individual is compounded when we think about the evaluation of the war as a whole. It’s very interesting to me that when Cecile talks about jus ad bellum she repeatedly talks about what ‘we’ are doing, rather than about what individuals (e.g. leaders) are doing. This is a very natural way to think about the reality and morality of war—that it involves collective actions essentially, just as it involves individual actions. And it does seem that just war theory should provide space for providing holistic assessments of wars as a whole. Indeed, that’s arguably what jus ad bellum has always been about. These again can be synchronic or diachronic (as with the individual actions—we could just speak of evaluation at a time, or of campaigns as a whole). The same questions would arise here for what knits together a campaign over time as with individual action. But there would be additional questions of what knits together a campaign at a time. What makes the actions of those people a relevant component of what we do?
    So, I agree with Helen that the old Latin labels are pretty useless. But (1) I disagree that the only interesting questions to ask about the ethics of war are about the evaluation of individuals’ actions, rather than also about the evaluation of collectives’ actions and (2) I don’t think that it’s even at all obvious how we should evaluate individuals’ actions—whether synchronically or diachronically.
    In support of (1), I think it’s pretty clear that we expect wars as a whole to be proportionate, and for all the harm that they involve to be necessary. That is, the war should be fought in a way that leads to the least morally weighted harm consistent with achieving the war’s objectives. For this demand to be satisfied, we need to first say something about what counts as part of ‘a side’s war’ (e.g. Britain’s war against Germany). And I think it can be pretty easily shown that sometimes it is consistent with a side’s war being proportionate/necessary, that some of the people fighting on behalf of that side act in ways that would be disproportionate/unnecessary if considered in isolation from what others are doing (this is just a particular instance of the general phenomenon that sometimes what is collectively rational can require actions that are individually irrational for the people who take them, i.e. various forms of collective action problem).
    And notice that this is all within the ambit of fact-relative permissibility. There’s a whole further set of questions to do with evidence-relative permissibility that overlap with these in interesting ways.

16. Greetings to all.
    Cécile states: “This article argues that we must sever the ethics of war termination from the ethics of war initiation: a belligerent who embarks on a just war at time t1 might be under a duty to sue for peace at t2 before it has achieved its just war aims; conversely, a belligerent who embarks on an unjust war at t1 might acquire a justification for continuing at t 2.”A problem with this statement is that given a normal interpretation of the word “ethics”, the text after the colon does not imply, nor clarify, the text before the colon. Usually, when philosophers write books or articles on “the ethics of war,” they refer with “ethics” to some set of principles or ethical criteria (if they are, like Cécile, deontologists, otherwise they might refer to a set of virtues or to a super-principle like the maximization of happiness). Yet, Cécile applies exactly THE SAME set of principles or criteria to t1 and to t2, namely the usual ad bellum principles (“reasonable chances of success,” “proportionality,” etc.). Thus, in the usual interpretation of the word ethics, she seems not really to think that the ethics is different at the two points (and she would be right on this, in my opinion). Thus, war termination and war initiation seems not at all to be governed by different kinds of ethics. However, given the texts before and after the colon, she might use the term “ethics” in a more idiosyncratic way, not as referring to a set of principles, but rather to an ethical judgement made on be basis of a set of principles about the justifiability of a war at different points. In this idiosyncratic sense of “ethics,” her claim that the ethics at two different points in war are different (whether the two points are the points of initiation or termination or any other two points) is certainly correct – but also entirely trivial: nobody has ever claimed otherwise, and there are good and obvious reasons for this (some of which are stated in a recent article on precisely this issue by Marcus Schulzke in “The Contingent Morality of War: Establishing a Diachronic Model of Jus ad Bellum,” 2013). Thus, my charge against the main claim for which Cécile argues is that it is, depending on interpretation, either wrong (even in the light of her own views) or trivial. Thus, I agree with Helen that Fabre’s argument is only valid against a mistaken view of jus ad bellum – a view nobody seems to ever have endorsed.
    Yet, Fabre seems to think that recognizing that different phases in war “will elicit different moral judgements” will undermine the “distinction between jus ad bellum and jus in bello” – and thereby, allegedly, “a good millennium of theorizing.” However, it does not undermine this distinction at all; for example, one can quite reasonably make it along the lines suggested by Seth (I have made it along the same lines). Jus ad bellum then refers to the collective action of the community – the belligerent – and jus in bello to the individual actions of individual combatants or groups of combatants. This also seems to be the traditional view. Moreover, Cécile herself seems to be quite able to make the distinction, so it is unclear how her argument undermines it. Maybe, however, her point is not about making a conceptual distinction between jus in bello and jus ad bellum, but about their logical independence. Yet, the traditional view – the “millennium of theorizing” – is, of course – contrary to what some self-proclaimed “revisionary just war theorists” suggest – that the two are not logically independent: on what is actually the traditional view, soldiers are not allowed to participate in their community’s unjust war, at least not if they know it to be unjust (of course, I consider this traditional view to be mistaken).
    Yet, even if the basic ethics of the initiation of a war and of the termination of a war (and of all the points in between) are governed by the same jus ad bellum principles, Cecile’s argument would not be entirely trivial if she could identify a general difference in the application of these principles to these different points. What she suggests (and this is the only thing she suggests along those lines), it seems, is that at t1 one does not need to take into account what has happened in the past, while at t2 one does have to take it into account: at t2 “we must take into account precisely what has happened since,” and she seems to think that this is not so at t1. But I fail to see the difference. For example, one might have already pursued the “just aim” – or the unjust one, for that matter – with means short of war. Such means (for example a boycott) have, according to different studies, killed around half a million people in Iraq. They were also costly for people who wanted to sell their products to Iraq. Thus, if one has to take the losses (or gains) of the past into account at t2, then one must also do so at t1. At least Fabre has not provided an argument to the contrary.

17. I agree that the first sentence of the article (‘the ethics of…’) is not the right to describe what I then go on to argue for in the remaining c. 13,000 words. Your dissection of this opening sentence does not suggest to me that article’s central thesis (that waging a just war at t1 does not in itself license you to continue with the war at t2, and waging an unjust war at t1 does not itself forbid you to continue at t2) is not clearly stated. Nor does your dissection, and indeed what you then go on to say, touch on what I regard as two of the article’s main contributions, namely (a) that the ways in which circumstances change from t1 to t2 often make a difference to what belligerents may or may not do at t2 (see esp. pp. 646ff) and (b) the morality of unconditional and conditional surrender and demands for it.
    I also agree that the phrase ‘millenium of theorising’ is an overstatement. I disagree however with your imputing to me, and indeed to McMahan, Rodin, Frowe, and Coady (in other words those whom you describe as ‘self-proclaimed revisionists) ignorance of the traditional thesis regarding soldiers’ permission, or rather lack thereof, to participate in an unjust war. We are all on (published) record as acknowledging the historical roots of the various theories of the just war we are all defending. My unfortunate throw away comment in this blogpost should not override those other, and many statements.
    Finally, my point regarding sunk costs and decision-making at t1 and t2 is simply that, at t2, one may not act as if what has happened between t1 and t2 did not in fact happen. I certainly do not think, nor (in my view) give the impression of thinking, that what happened before one went to war at t1 is irrelevant when taking into account at the bar of proportionality what one may do there.

18. Uwe, my last post was a response to yours. Sorry for not being crystal clear on this.
    Seth: I apologise for the delay in responding. A characteristically rich post. I have a few speculative comments, which I will divide into a couple of separate posts. First, the distinction you draw between ‘the sunk cost problem’ and the ‘jus ad bellum’ problem does not quite work for me: the issue of sunk costs is raised both when we have to decide whether or not to go to war, and, once the war has started, whether or not to push on this or that particular part of the front, to have another go at Gallipoli or to withdraw altogether, to rescue those prisoners of war or not, and so on. The latter cases are standard in bello cases (assuming for now that the distinction between the two jura makes sense, which I know you don’t think it does but you are after all referring to one side of the distinction as, ad bellum.)

19. To Seth (c’ed): Second, i do not have a precise account of what (in your apt words) ‘knits’ individual acts of violence into a campaign, and what knits several campaigns as a war. And I agree that we do need more precise accounts, without which we cannot really make sees of the claim (along reductivist lines) ‘the war was unjust/just’. It seems to me however that looking towards the literature on direct participation might provide some help, or the seed of some help. For example, it might help us distinguish between the individual soldier who having married into a different ethnic group, kills his father in law during his regime’s genocidal campaign because he wants his farm and the soldier who does so out of commitment to the cause. Of course as a matter of fact we might not have the information we need to make that judgement, but we can still conceptually distinguish between the two cases. In other words, even if temporality matters, so do other factors, to which you point to, such as the nature of agents’ contributions, etc.
    That said, it may well be that, in the end, reductivists such as me and Helen would have no choice but to say that the claim ‘WWII was just in toto’ is simply meaningless, and that all we can do is assess whether the acts of killing carried out between September 1 1939 and August 15 1945 were morally justified. I am not yet clear as to why we should be worried about that, philosophically speaking.

20. To Seth (c’ed). Finally, and relatedly, although I do think that individual acts are the ultimate loci of moral evaluation, I also think that we can and ought to distinguish between acts which they carry out collectively (whether at the point of deciding. qua leaders) whether to wage war by deploying troops, or at the point of working out, as a platoon, whether to attack this or that village), and acts which they carry out on their own. In the former case, I want to know whether their individual contribution to/participation in the joint decision is morally justified; and I cannot know that without having a view about the moral status of that joint decision. Thus, I certainly do not think, and do not want to give the impression of thinking, that (in your words) ‘the only interesting questions to ask about the ethics of war is about the evaluation of individuals’ actions rather than…collective actions.’
    That said, I agree that evaluating individual actions is extraordinarily difficult when those actions are part of joint actions. We have no choice however but to grapple with this issue. Put differently, saying ‘we can evaluate collectives’ will not do, in so far as we want to know (in the context at hand) whether to put defendants on criminal trial: for only individuals can be so sanctioned.

21. Thanks Cecile! Agreed on all points. Much work to do for us all…

22. Cécile,
    I did not claim that your “central thesis (that waging a just war at t1 does not in itself license you to continue with the war at t2, and waging an unjust war at t1 does not itself forbid you to continue at t2) is not clearly stated.” Rather, I claimed that it is trivial and that no just war theorist has ever denied that central thesis. (If you know a just war theorist who has, please give me his or her name.) Consequently (the reason why the “central thesis” is obviously true is that your point (a) is obviously true), my “dissection” does touch on the first of your article’s “main contributions, “ namely on “(a) that the ways in which circumstances change from t1 to t2 often make a difference to what belligerents may or may not do at t2.” That thesis is of course also trivial, and again no just war theorist I know of has denied that thesis. I did indeed not touch on (b), but I am of course at liberty to choose for myself which aspects of your paper I want to comment on. Since, moreover, your treatment of (b) is based on a number of premises (for example concerning jus ad bellum, just cause, and liability) which I do not share but have already criticized at length elsewhere, it would have not made much sense to go into this here again.
    By “self-proclaimed” I mean, in line with the meaning that expression actually has, that it is not the case that some group of historians of just war theory looked at a certain group of recent just war theorists in awe and said: “Gee, they are so incredibly innovative, we just have to call them ‘revisionary’” but that, instead, you guys call yourself revisionary (I am not aware, by the way, of Coady doing this, but please correct me if I am wrong). That self-assessment would be quite acceptable if it were correct, but it is not, which brings me to the topic of ignorance. McMahan has himself admitted (I suppose that is part of what you are referring to) that due to an intervention by Gregory Reichberg he has come to realize that he has been “beaten into print” by Suárez and Vitoria – which suggests that at the time he first described his theory as “revisionary” (based on its moral inequality thesis) he might indeed have lacked knowledge of very important parts of the tradition, which, however, did not keep him from making sweeping statements about it. Moreover, that he has only been beaten by Suárez and Vitoria is, as I have argued elsewhere, somewhat of an understatement: he has, in fact, been beaten by the largest part of the tradition from Aquinas to Anscombe. As regards Frowe: she has depicted Grotius as someone who rejects the thesis of the moral inequality of combatants, but this depiction is flatly wrong. And Rodin has written a whole book arguing that the just war tradition is centrally based on the idea of self-defense – which is also flatly wrong. (This point has also recently been made against Rodin by Nigel Biggar.) To the best of my knowledge, Rodin has not retracted his assessment (but please correct me if you can). As regards you: you state in your response to my comment that your “phrase ‘millenium of theorising’ is an overstatement.” However, it is not an overstatement: it is a complete distortion.
    Moreover, if putatively “revisionary just war theorists” do not call themselves revisionary out of ignorance, then it must be for some other reason – I suspect it is the desire to paint oneself as more innovative than one actually is. However, I do not think that it is necessarily better to mislead one’s readers on purpose than out of ignorance.
    As regards your final point: good. But that only strengthens my charge of triviality, which you have done nothing to undermine.

23. Uwe, tackling every single one of your points would take one, indeed two, articles. I am afraid that we shall have to agree to disagree. Cecile

24. To Cecile:
    I was just thinking about the whole ‘can we really say that there’s a question of whether the war as a whole is just?’ point. It seems easy enough to hold the ‘disaggregating’ view when thinking about a particular combatant. We can simply look to the particular threats that he contributes to, and ask about those. That’s a simplification, but it’s a harmless one.
    But it’s a different matter when you think about noncombatants. Because we can’t usually point to any particular threat that they’ve contributed to. It’s more that they make the war as a whole possible. So for determining what they may do, and their liabilities, it seems like we need some concept of the war as a whole, or at least something as an alternative to working out which specific threats they contributed to.
    This is also going to be relevant for omissive responsibility; there might be combatants and noncombatants who are responsible for allowing the war to go ahead, who could have stopped the war, where that won’t necessarily be readily disaggregated into thinking about particular threats.

25. To Uwe:
    I’m not sure, but I think I might have coined the ‘revisionist’ term, and if I didn’t coin it, I definitely gave it currency. So let me just point out a mistake. To claim that a theory is revisionist, is simply to claim that it proposes to revise something. In this case, revisionist just war theory is in opposition to a contemporary orthodoxy, which is to say Walzerian just war theory. And I would say that revisionist just war theory earns that name, because it proposes to revise, rather than to vindicate, the ‘war convention’ with which Walzer began. It further proposes various revisions to international law (although with some hesitation).
    So: it’s a mistake to think that ‘revisionist’ means ‘completely novel in the history of thought’. That would be a silly label to appropriate. It means, instead, just the dictionary meaning: proposing revisions of something, in this case the war convention, as Walzer described it.
    There are clearly some similarities between contemporary analytical just war theory and some of the scholastics, but there are also many differences. For one thing, none of them distinguished between moral responsibility and agent responsibility; all those who thought in these terms thought that what mattered was moral responsibility (innocence vs. culpability). For another, even scholastics as enlightened as Vitoria believed that in wars against Muslims, christians need observe no restraint at all. I’m inclined to think that it’s a mistake to read historical figures as though they were writing transhistorically. Each of them wrote for their own times, with their own tools. There’s undoubtedly some interest in resurfacing parts of the just war tradition, as with anything, and it’s good that excellent scholars like Greg Reichberg and Pablo Kalmanovitz, and Danny Schwartz, are doing so. But there’s no reason to think that philosophy would be well served by everyone neurotically checking the pedigree of every idea, to see whether, read in a certain light, it might echo a sentence from a 17th century priest (writing principles for confessors) taken completely out of context. Nor is there any harm in the odd sweeping generalisation about the tradition, especially when it’s clearly just intended to provide a jumping off point for a more detailed discussion.

26. Seth
    Thanks for this. Needless to say I fully agree with your response to Uwe.
    Regarding your points about noncombatants – which I take it do not apply to leaders ordering an invasion but does apply to eg taxpayers, factory workers, etc.
    I take the point that we do not *know* which threat(s) specifically they contribute to. But our not knowing that will not affect their liability: we can say ‘we do not know which threat spefically they contribute to but we do know that they contribute to some such threat and they are liable to some harm as a result.’ So, it seems to me that we can still disaggregate.
    Further, the question at this point is ‘given that we do not know which threat they contribute to, what may we do? May we kill/harm them?’ Clearly, we cannot say ‘yes, because you contribute to this threat’. But might we not say something like this: ‘on the balance of evidence and probability, we reach the judgement that you are contributing to some unjust threats though we do not know which threats. And we may harm you to such and such degree to stop you.’ Why do we need an alternative to disaggregation, in order to reach that judgement? What we need is what in your own recent work you are beginning to offer, namely an account of decision under arduous epistemic circumstances.
    Does this make sense?

27. Hi Cecile. Yes, I think that’s going to be true for a lot of cases, fair enough. And I agree that the problems with moral decision-making under empirical uncertainty will surely persist.
    But what about if you vote for a party that has a manifesto commitment to launch an aggressive invasion, and that’s your only contribution. In that case, wouldn’t we want to say that you’re responsible, insofar as you’re responsible, for ‘the war as a whole’, whatever that should mean?