Ethics Discussions at PEA Soup: John Gardner and François Tanguay-Renaud’s “Desert and Avoidability in Self-Defense” and Jeff McMahan’s “Response,” with commentary by Victor Tadros

41 Replies to “Ethics Discussions at PEA Soup: John Gardner and François Tanguay-Renaud’s “Desert and Avoidability in Self-Defense” and Jeff McMahan’s “Response,” with commentary by Victor Tadros”

1. Greetings Victor! Your penetrating analysis has reinforced a number of things for me. Would you please clarify the “barbaric thesis” as you attribute it to McMahan, and as you suggest GRT accept? Thanks kindly.

2. Thanks to all the authors (definitely including Victor) for their really sharp and insightful work on these difficult issues.
   I wanted to explore a bit of the dialectic on responsibility. GTR wonder (p. 132) about McMahan’s claim that the degree of one’s responsibility for something could vary in proportion to the strength of one’s excuse, such that the stronger one’s excuse in doing X, the less one is responsible for doing X. Their objection to this proposal is that excuses are available only to responsible agents, such that “one needs to be responsible…before one can even have an excuse.” McMahan’s response is to distinguish between being a responsible agent and being responsible for X: yes, one must be a responsible agent (be a member of the moral community) in order to be eligible for being excused, but that’s compatible with an excuse diminishing one’s responsibility for X, proportional to the strength of the excuse.
   McMahan’s response seems right to me if we take GTR’s worry to be solely about the relation between excuses and responsible agency (which is indeed how it comes across on p. 132). But they raise the worry in the context of a discussion of McMahan’s changed, current view that one may permissibly be the target of deadly self-defense as long as one is responsible for the threat to which the self-defender responds, and even if one is not guilty — culpable — for the threat. This is what makes appeal to desert unnecessary. (I hope I’ve gotten the view right.) So insofar as one has an excuse for what one did, one may get off the guilt hook without getting off the liability hook: “even a complete excuse leaves E [the harm-causer] open to permissible self-defensive measures” (132). Their initial worry, then, is how we can make sense of a complete excuse not getting one off of the responsibility (and thereby) liability hook that permits self-defensive measures. And this is a good question: why wouldn’t a full excuse render one not responsible at all? As far as I can tell, GNT change the subject when they turn immediately to press the (different) worry that excuses presuppose responsible agency.
   But what of their original worry? Here’s what McMahan says: “[C]ulpability is a form of responsibility. Culpability for a threat of wrongful harm is just moral responsibility for that threat in the absence of a fact-relative or evidence-relative justification or any excusing conditions” (165). This raises the key issue, then: what is moral responsibility, if I can be morally responsible for causing some threat with a complete excuse, i.e., utter ignorance that what I was doing would cause it? To count as moral responsibility (MR), it must be more than mere causal responsibility, as MR implies a capacity that seemingly must implicate intentions or reasonable foresight. (One might even wonder if causal responsibility is necessary, as I could seemingly be morally responsible for omissions or for what my children or others under my charge may cause.) Is it attributability, that the action in question is somehow dependent on me qua practical agent? But that would seem to require a connection to my ends/cares/commitments/intentions in a way that a complete excuse would also completely eliminate. (Think of someone who assaults his friend after having been involuntarily intoxicated. The excuse of involuntary intoxication completely wipes out attributability here it would seem; the action is no more attributable to him qua practical agent than it would be an alter in an MPD case.)
   So my question is this: what sense of moral responsibility necessary for liability does McMahan have in mind such that it’s more than (or different from) causal responsibility but less than attributability? I fear the answer is in his book, which I haven’t read, so there may be a straightforward reply (“Just read p. X”), but I would have thought that reply would come out in the exchange with GTR, and it didn’t.

3. Thanks, Victor, for a lucid summary of these complicated arguments.
   I have a different proposal to make about the ‘avoidability argument’, one which is not considered by GTR or McMahan. It is to allow the avoidability considerations to (partly) determine the relevant degree of desert.
   Compare the two following cases, where (B) is just a version of the ‘Kill or Be Killed’ case which GTR mention at p. 121:
   (A) E threatens D in ways which D may avoid short of violent self-defensive action against E.
   (B) E threatens D in ways which D cannot avoid short of taking violent self-defensive action against E.
   Why is it assumed, as it appears to be, that E’s degree of desert is invariant between (A) and (B)? The fact that, in (A), E’s threat can be avoided by D, whereas in (A) E’s threat cannot be avoided by D, supports a morally relevant difference in the description of the threat E poses to D in the two cases. In (A), E’s threat to D can be described as presenting D with the following options: ‘Die-or-run-away’. (Imagine that D can avoid E’s threat by running away.) In (B), E’s threat to D can be described as presenting D with the following singular option: ‘Die’. E’s threat in (B) strikes me as being morally worse than his threat in (A). But then E’s degree of desert across (A) and (B) is not invariant.

4. Hello Leroy
   Here’s how I understand it.
   The barbaric thesis claims:
   BT: If D has acted wrongly it is impersonally valuable that D is harmed.
   As I understand it, there are two main versions of BT
   BT(a): If D has acted wrongly it is intrinsically good that D is harmed.
   BT(b): If D has acted wrongly it is appropriate to harm D.
   These versions come apart. BT(a) seems to imply that the harming of D is a good outcome which provides a reason for people to harm D. It seems also to imply that the harming of D is good even if D is harmed by natural causes. This, I think, is the version of BT that McMahan endorses.
   BT(b) is concerned with the actions that are appropriate in response to wrongful action. It is common to think that it is appropriate to blame people for their wrongdoing. BT(b) adds to this that it is appropriate to harm wrongdoers as well. BT(b) has nothing to say about the goodness of D being harmed by natural causes. Nature can’t act appropriately. I am not sure whether GTR endorse BT(b), but I think so.
   But perhaps Jeff, John and Francois will have different accounts of BT to offer.

5. Thanks to all the authors for some great work and thanks to all for this excellent conversation. I hope to say more with regard to the discussion of responsibility and excuse as it relates to liability in a bit, but first I wanted to give a quick reply to one of the questions Victor raises.
   Victor writes,
   “Even if the impersonal value of the person suffering were not a ‘defensive reason’, it would still have a role in determining the permissibility of self-defence. It seems hard to believe that it is intrinsically good that bad people suffer and yet that this plays no role in determining whether defending oneself against a culpable person in a way that makes them suffer is permissible. This conjunction of views seems difficult to motivate. ”
   While I tend to agree with Victor’s rejection of the “Barbaric Thesis” (nice touch on the view’s moniker, Vic – I hope it sticks), I’d like to offer a way in which one could motivate the conjunction of views that Victor describes here.
   So here is one way one could be motivated to hold this conjunction. One could hold that while there is an impersonal value that culpable person X suffers because they deserve to suffer for their wrong-doing (call this desert-suffering), that delivering that desert-suffering is only itself impersonally valuable if it is done in the right way or in the right manner or by the right actors. And that delivering that desert-suffering in the wrong manner and by the wrong means would fail to obtain the impersonal value sought in delivering the desert-suffering to the deserving person. Namely, one could hold that desert-suffering should come as the proper delivering of desert for desert sake and not as part of a co-mingled effort to defend a nonliable person (or any other good goal). And, further, as part of this, that the act of delivering the desert-suffering should be done by a given group or society, through a prior agreed upon judicial process of some kind, and so forth. And, finally, that the people to deliver such desert-suffering should not be the specific people who were harmed by culpable person X so as to ensure that the deliverance of desert-suffering was properly proportionate and, well, deserved. (That is, a felt need to avoid bias against the culpable person in delivering desert-suffering.) If one held to all of that, I can imagine that motivating the conjunction of views given above: “that it is intrinsically good that bad people suffer and yet that this plays no role in determining whether defending oneself against a culpable person in a way that makes them suffer is permissible.” The reason would be a matter of some kind of logical ordering to a morally permissible response. The first job is to determine if it is permissible to defend against person X who, if morally responsible for the unjust threat in question, is liable to bear defensive harm — even harm that makes X suffer (non-desert-suffering). Whether or not X further deserves to receive desert-suffering as part of the Barbaric Thesis is, one might hold, a decision that is not only best made later by an impartial process, but should be made later and specifically not by the nonliable person threatened/harmed by X. This at least seems to be a plausible view, if one grants the Barbaric Thesis.

6. I’m a long time listener, first time caller. Regarding David Shoemaker’s questions about McMahan’s view on responsibility.
   On his view (I believe), moral responsibility can be minimal or maximal. An individual is minimally morally responsible for an unjustified threat when that threat can appropriately be attributed to the individual’s ‘responsible agency’. What does it mean for an act to be attributable to an individual’s responsible agency? The individual, in such a case, must have made relevantly voluntary choices which, foreseeably contributed to the threat. (The criteria for foreseeability are rather weak, as I’ll point out below). Maximal moral responsibility presupposes minimal responsibility. But in addition, the individual posing the threat can be blamed or praised for the threat. Blameworthy maximal moral responsibility is just culpability.
   Your question has to do with the criteria for *minimal* responsibility, and its intersection with excusing conditions. Minimal doesn’t seem to require a connection to the individual’s ends/cares/commitments/intentions in a way that a complete excuse would completely eliminate.
   A person subject to irresistible duress, for example, is minimally responsible even if fully excused. But here is more pertinent example. Suppose I decide to go for a drive. I regularly check my car – I ensure that it is safe. And in doing so, I do as good a job as anyone can be expected to do. I also drive very safely – I go above and beyond what is required. And I am driving somewhere that I have good reason to go. But through no fault of my own – a completely random fluke in the engine – my car seizes up, and I lose control. Bad luck. The car heads for a pedestrian down a walled-off alley. The pedestrian cannot get out of the way. But she happens to have a bazooka which she can use to save her life by blowing up me and my car. In this case, I am fully excused for the threat that I pose to the pedestrian. But I am liable to be self-defensively attacked since I am minimally responsible for the threat I pose. I am minimally responsible since driving is an inherently risky activity – and I knew this, and voluntarily chose to drive anyway.
   So on McMahan’s view (I believe) the claim that a person is fully excused for a threat implies only that the person is not culpable. It does not necessarily imply that the person is absolved of all responsibility – where responsibility I understood as minimal responsible agency.
   -Saba Bazargan

7. Gerald, I think that’s a great point. I agree that there is a morally relevant difference in the descriptions of the threat E poses to D in the two cases (A) and (B). But we can’t automatically infer a difference in culpability from the fact that there is a moral difference in the two cases. You mention a morally relevant difference between the description of the two cases. But the *intensional* rather than the *extensional* description of the threat the attacker poses is also relevant to her culpability. If the attacker doesn’t know or cannot reasonably foresee that you can avoid her threat by running away or defensively counter-attacking, then her culpability is the same as it would be if you couldn’t avoid her attack — even if we assume that there is a morally relevant difference between the two cases.
   If this is right (and I’m not sure), then the answer to your question “why is it assumed, as it appears to be, that E’s degree of desert is invariant between (A) and (B)”, is that E is assumed to be unaware of the morally relevant difference between (A) and (B) — or, better: it ought to be assumed. Determining the relevance of E’s culpability is hard enough in the case where it is *invariant* between (A) and (B)… but maybe I’m getting the intended methodology of the investigation wrong here.

8. Saba: That’s all extremely helpful. Thanks. On this understanding, then, I could be responsible for violating strict liability laws, but not morally responsible, right?
   I confess to losing sight of the function of “responsibility” in this minimal sense, then. Presumably, there’s supposed to be some thought like this to explicate responsibility in the car case you give: “Well, you brought it about that the bystander was threatened.” But in what sense did I bring anything about? I got in the car and voluntarily drove it, that’s true, and as you say, I voluntarily engaged in an activity I knew bore some risks, but then this analysis easily extends to much tougher cases. Take Jane English’s example of innocent hypnotized attackers: a mad scientist has kidnapped and hypnotized a bunch of innocent people to jump out of the bushes and attack people. Are those innocents responsible in any meaningful sense for the attack? To the extent they voluntarily left their house and were, I suppose, aware that there’s always a chance that when you walk around in Philosophy Land you might be kidnapped like this, they’d be “responsible” for the attack in just the way the careful car driver is. But this just strikes me as wrong: in any plausible sense of “bring it about that,” the mad scientist is responsible for the attacks. And so, perhaps, in any plausible sense of “bring it about that,” the misfiring car is what brings about the attack in your case.
   Of course, there’s a disanalogy here insofar as the mad scientist is a distinct agent and the car isn’t, but then make the source of the hypnosis in English’s case the attackers’ suddenly mis-firing brains.
   Now I have no problem at all saying that it’s permissible to kill in self-defense in such cases, but I’ve simply lost sight of the function of saying that the attackers in each case are responsible at all where they are fully excused and there’s no longer any connection between the “action” and the agent qua agent. Hope this make sense.

9. Bradley writes (and I concur—somewhat) to offer a motivation for “the conjunction of views that Victor describes”: “…desert-suffering in the wrong manner and by the wrong means would fail to obtain the impersonal value sought… …desert-suffering should come as the proper delivering of desert for desert sake and not as part of a co-mingled effort to defend a nonliable person (or any other good goal).”
   Bradley correctly (I think) identifies at least some of the features of an ancient tradition of criminal justice, common across many cultures, that would seem to parse self-defense from punishment by virtue of the rule of law: the deserved suffering should be meted out by “a given group or society, through … judicial process”; “And, finally…” Bradley adds, the deserved suffering should be administered by disinterested persons “so as to ensure that the deliverance of desert-suffering was properly proportionate and, well, deserved.” Otherwise, we risk having Bacon’s notion of “wild justice” on the part of the self-defender.
   This strikes me as the principal reason why desert should not be among the grounds for self-defense. The fact that one deserves to suffer harm does not imply that simply anyone is permitted to inflict the suffering. Couple that with Jeff’s point that desert carries its own sufficient weight (apart from self- defense), and it seems desert should be kept out of the account for self-defense.

10. David and Saba
    In order to vindicate the difference between the Car Case and the Hypnotized Attacker case rendering the former responsible for the threat but not the latter, we need the idea that the car driver increases the risk to pedestrians by driving but the person wandering around who is then hypnotized does not increase the risk to others by wandering around. The difficulty is how to set a baseline against which the risk is increased in the former case but not the latter.
    What is more important for the debate here, though, is whether an explanation can be given why different kinds of responsibility (non-culpable risk taking and culpable threat imposition) should lead to different judgements with respect to liability without appealing to desert. I’d like to hear some accounts of this.
    BJ
    Nice attempt to rescue Jeff’s position. Many people will agree with the idea that deserved suffering ought only to be imposed by the proper institutions in the way that you suggest. I’m not sure why institutional capacity should be so important in the case of deserved suffering but not in the case of self-defence, though. Suppose that we are confident that deserved suffering will not be imposed by the proper institutions, say because there are no adequate institutions (as is the case with respect to wrongdoing during war). Why is it permissible to defend oneself where one is confident that the relevant authorities will not come to the rescue, even though there is a risk that one will act partially, disproportionately etc, but it is wrong to inflict deserved suffering oneself where one is confident that the relevant authorities will not act?
    Perhaps you might appeal to the fact that in the case of self-defence, a person will be harmed, perhaps even killed, whereas in the case of deserved suffering it is only impersonal value that is lost. But it is not obvious that we should always regard personal value as more important or more worthy of protection than impersonal value.

11. Victor, thanks for clarifying BT for me earlier. That helped a lot.
    In regard to your question to BJ: “Why is it permissible to defend oneself where one is confident that the relevant authorities will not come to the rescue, even though there is a risk that one will act partially, disproportionately etc, but it is wrong to inflict deserved suffering oneself where one is confident that the relevant authorities will not act?”
    Let me float two points of difference that seem relevant: (1) self-defense is a matter of imminent danger and it is future- directed, whereas deserved suffering is not characteristically of imminent need and it addresses past concerns; (2) the self-defender may be the only one positioned to determine the danger and to take action against irrevocable harm, whereas they are not well positioned either to determine desert to suffer or to mete out a proportionate degree of deserved suffering. Perhaps these differences are incidental, but they appear to me characteristic and morally relevant. Of course, I may be missing something in this picture of difference.
    Leroy

12. David and Victor,
    David, you asked “On this understanding, then, I could be responsible for violating strict liability laws, but not morally responsible, right?” Yes, I believe so – or better, I believe that McMahan believes so.
    McMahan has an example which, I think, captures the kind of case that David has in mind: A man’s cell phone has, without his knowledge, been reprogrammed so that when he next presses the ‘‘send’’ button, the phone will send a signal that will detonate a bomb that will then kill an innocent person. He is fully epistemically justified in thinking that pressing the button is harmless. Unlike the conscientious driver, and like the person wandering about, he does not choose to engage in an activity that has a foreseeable risk of causing serious harm. As a result, unlike the driver, the cell phone user is *not* minimally responsible – this is because using a cell phone is not a risky activity. Victor, you said that the difficulty is setting a baseline against which the risk is increased in the contentious driver, but not in the wanderer (or the cell phone) cases. But I don’t think this is necessary. We can admit (for sake of argument) that the guy who voluntarily leaves his house increases the risk to others while maintaining that the risk falls below a threshold – a threshold of risk that the contentious driver crosses. This isn’t ad hoc – driving, even at the best of times, is a risky activity, in a way that using a cell phone or walking outside are not. In the hypnotization case and the cell phone case, I think, David, you’re right in saying that the malicious individual pulling the strings from behind the scenes is the individual to which the harm ought to be morally attributed – not the cell phone user or the driver. The guy behind the scenes is the one relevantly bringing about the harm. But in the car case, the *driver* relevantly brought about the threat (not simply the car) – again, because the driver knew or was in a position to know not only that it is *possible* for driving to harm others, but that there is substantial risk of this occurring.
    Ultimately, though, I agree with Victor’s point that what is most important is providing an explanation why different kinds of responsibility (non-culpable risk taking and culpable threat imposition) should lead to different judgments with respect to liability without appealing to desert. Here is a possible answer. There are different degrees of moral responsibility – the more you have, the greater your liability to defensive harm. That seems right to me. And “culpability” as McMahan uses it might just mean “maximally high degree of moral responsibility”. We can, on this proposal, eliminate the word “culpability” altogether, and replace it with something like “maximally high responsibility” in order to avoid precisely the sort of challenge you’re raising, i.e., why culpability is relevant but not desert. I’m certainly not claiming that this is what McMahan does – but it may be a way of interpreting his view.

13. Victor,
    I had wanted to return to your comments about the implausibility of NCJ1 and NCJ2. “As McMahan suggests, taken literally, these principles seem implausible even if one accepts the barbaric thesis. They would at least need to be revised to accommodate cases where a person consents to be harmed.” (You give a compelling example.)
    It appears to me that the implausibility of the two principles runs deeper still, so that I am unclear how to salvage much from them. “Fourth, and perhaps most disturbing”, McMahan says, “by making desert a necessary condition… [NCJ1] reduces the scope of possible justifications for intentional harming to an implausibly narrow range.” (Endnote 22, considers GTR’s response: “an action… may be not permitted under one permissive norm yet still permitted under another.”)
    Examples compound the “disturbing” implication. For, NCJ1 appears to rule out:
    (a) beneficent self-sacrifice and self-inflicted harm in general, even to protect all others (this comes under your point regarding consent); e.g. a soldier shields comrades from grenade blast with her own body; a teacher stands in the doorway to protect his students from an attacker’s bullets;
    (b) measures harmful to the innocent that are intended to mitigate further harm to the innocent (this may or may not involve consent); examples abound: rescuers amputate a victim’s limb to save him from a burning vehicle, etc.
    (c) limited harm, in self-defense against well intended attacker, ignorant of threat to defender; neutral tribal villagers, living in territory controlled by unjust combatants, inflict dysentery on just combatants and unjust combatants alike to prevent destruction of crops by both sides.
    A wealth (or a flea market) of examples come to mind in which it would appear permissible—in some cases justifiable—to inflict “suffering or deprivation” on those who cannot plausibly be said to deserve the “suffering or deprivation”.
    Thanks for your insights. It is challenging material.
    Leroy

14. I side with McMahan on the major issue at stake : desert and punition should play no role in the theory of self-defense. Like McMahan, I hold that legitimate self-defense is “necessarily instrumental” (160), whereas any cogent notion of desert in this context would suggest the desirability, all things being equal, that the supposedly deserving person receive harm.
    However, I believe similar things can be said about the intentional infliction of harm, even in self-defense. If legitimate self-defense is necessarily instrumental, then the straight line to take is that one can never intend harm in self-defense. To intend harm is to seek the harm as a good in itself, quite apart from its instrumental value in averting a culpable attack. On my understanding, this is precisely the kind of reasoning which underlies the doctrine of double effect (DDE).
    McMahan, however, disputes both the reasoning and (apparently) the reading of DDE. In his response to Gardner and Tanguay-Renaud, he reluctantly concedes the “strict and literal” truth of the view that imposing a duty of compensation on a tortfeasor is not “a case of intentional harming; the harm is, rather, a side effect of ensuring that the victim is compensated.” He worries, however, that
    “if what counts as an intended effect is understood in such a narrowly restricted way, the claim that there is a special constraint against intentional harming may be vitiated. For the same claim can be made about most inflictions of harm, not only in self-defense (“Yes, your honor, I harmed the police officer—yes, all right, fatally—but only as a side effect of preventing him from arresting me”) but in other cases as well (e.g., Jonathan Bennett’s case in which a pilot bombs a city with the intention of making its inhabitants appear to be dead, though this has as a side effect making them actually dead). It seems that those (including Gardner, Tanguay-Renaud, and me) who accept the relevance of intention to permissibility must accept either a broader concept of an intended means or an understanding of the relevance of intention to permissibility that is different from that found in the Doctrine of Double Effect.” (162-3)
    To the contrary, I believe the traditional reading of DDE takes precisely the view in question. For instance, Aquinas asserts that in legitimate self-defense, the killing of one’s assailant must be praeter intentionem – beyond the intention (ST II-II, q. 64, a. 7). I think that what goes missing in McMahan’s discussion is a distinction between intentionally inflicted harm and deliberately inflicted harm. To be sure, the tradition allows that one may deliberately inflict harm in self-defense, just like Aristotle’s sailors may deliberately throw their cargo overboard during the storm. No doubt, common English usage subsumes both of these under the term “intention.” It would appear, by contrast, that the Latin “intentio” is of narrower scope: it refers to an action’s superordinate end, not to the means adopted in order to bring that end about, i.e., the killing or the dispensing with cargo. (On this see Gregory Reichberg, “Aquinas on Defensive Killing: A Case of Double Effect? pp. 362-364.) If this is correct, it might simply be misguided to attempt to make our reading of DDE produce verdicts which (as McMahan says elsewhere) “conform to our linguistic intuitions”, insofar as these are intuitions about the semantics of the English term “intention,” not about the Latin “intentio.”
    Nevertheless, one may wonder if this line of thought – even if it is cogent – would threaten to render the DDE “limitlessly permissive,” as McMahan says elsewhere. Certainly, the DDE has its share of problems. But far from being limitlessly permissive, I doubt that it would permit even the two cases that McMahan cites. Specifically, the supposition that DDE might permit these cases can only come from failing to factor in further crucial constraints involved in the DDE: (i) that the intended end itself be permissible; (ii) that the means be necessary to achieving that end (or at any rate, the least harmful means of achieving it); (iii) that the side-effect harm be proportional to the good we seek to achieve. Neither of the cases McMahan mentions satisfies these further constraints. In the first case, resisting arrest is presumably not a permissible action, and, at any rate, killing the arresting officer is not proportional to the good of escaping arrest. For the second case, exterminating an entire city’s population is presumably neither necessary nor proportionate to the end of making them “appear to be dead,” whatever the strategic importance of that end may be.

15. Victor,
    Great stuff. I thought you might say something like that in response to this notion I’m floating for imposing desert-suffering versus liability-suffering. But I think there’s an easy response for this view tied to how they would value desert-suffering.
    You write, “Suppose that we are confident that deserved suffering will not be imposed by the proper institutions, say because there are no adequate institutions (as is the case with respect to wrongdoing during war). Why is it permissible to defend oneself where one is confident that the relevant authorities will not come to the rescue, even though there is a risk that one will act partially, disproportionately etc, but it is wrong to inflict deserved suffering oneself where one is confident that the relevant authorities will not act?”
    The suggestion I’m floating here contended that desert-suffering gets its impersonal value (according to the BT), only if it is delivered in the right way, by the right means, etc. (e.g. institutionally, by disinterested parties, through a proper process, etc.). So the answer would be simply that when one is not confident that the proper institutions (what-have-you) will deliver desert-suffering, then, sadly for a holder of this view, that just means the impersonal value of desert-suffering will not obtain . The value of desert-suffering is tied to its properly being delivered.
    Contrast that with the personal value gained by just defense (and the liability-suffering to a liable person that goes along with it). In this case, the value gained is the defense of the nonliable person. So whether the liability-suffering is delivered by the police or oneself or a friend on behalf of a nonliable person, and so on — the value of defending the nonliable person still obtains.
    So its not that I’m suggesting this view must hold that personal value is always more important or more worthy of protection than impersonal value, as you think it must to avoid your objection. (Although I don’t think that’s that implausible in these kinds of cases.) But rather it’s that the view I’m suggesting thinks that the impersonal value that desert-suffering provides is only ever gained if it is carried out in the right way (e.g. by proper institutional procedures, etc.). Whereas the personal value of defense of a nonliable person obtains however it is carried out (by proper institutions designed to deliver it or by a random person coming to the rescue of themselves or another against a liable person).
    So, to clarify, your question: “Why is it permissible to defend oneself where one is confident that the relevant authorities will not come to the rescue, even though there is a risk that one will act partially, disproportionately etc, but it is wrong to inflict deserved suffering oneself where one is confident that the relevant authorities will not act?” Is answered thusly:
    It is not that one is wrong to inflict desert-suffering (as part of the reason for defense) when one is confident the relative authorities will not act because of the risk of delivering it incorrectly — whereas one can act permissibly in inflicting liability-suffering in defense in spite of the risk that they’ll act incorrectly (e.g. outside the bounds of propotionality, etc.). Rather, the idea is, one cannot ever inflict desert-suffering correctly in such a case — at all, in principle — because that’s not how the impersonal value wrought from desert-suffering obtains. For that value to obtain, it must be inflicted in the right ways, by the right authorities, etc.

16. Thanks to the editors of Pea Soup, to all those who are participating in the discussion, and above all, of course, to Victor for his characteristically illuminating and amusing comments – though I have to say that as a label the “barbaric thesis” falls slightly short of the standard he set in a comment on a paper I presented in Oxford a year ago, in which he christened and repeatedly referred to some position as “the Stupid View.”
    I’ll respond briefly to a couple of Victor’s challenges and then offer a few remarks about the subsequent discussion.
    Victor invites me to explain why I accept the barbaric thesis. I do have doubts about desert, which is one reason that I would like to be able to give a comprehensive account of the morality of self-defense and war that doesn’t rely on it. But so far at least the doubts have been insufficient to overthrow my strong intuitive sense that desert is one of the basic facts about morality. I find it compelling that people can deserve various benefits. They can deserve not to suffer, They can deserve gratitude, recognition, success, a reward, or even to enjoy a period of contentment. If that is so, it seems they can also deserve to be harmed, or to suffer. For it seems arbitrary to suppose that people can deserve only benefits and not harms.
    Victor also asks me to explain why, if there is such a thing as desert, it cannot be relevant to the justification for self-defense. Here I think our disagreement is only superficial. I don’t deny that desert can help to justify an act that has a defensive effect. Suppose that a person is liable only to X amount of defensive harm but that successful defense requires that X+N amount of harm be inflicted on him. If he earlier committed a wrong for which he deserves to suffer N amount of harm, that can make it permissible to inflict X+N amount of harm on him, with the effect of successfully defending his potential victim. Here Victor is right, in his response to Leroy, that I accept an account of desert according to which desert may be satisfied by harms that are not inflicted with the intention of giving a person what he deserves, or not inflicted by a purposive agent at all. So here I must decline BJ’s ingenious effort to rescue my view. I agree with BJ that it’s better to institutionalize the infliction of deserved harms, but that’s mainly for contingent reasons having to do with the importance of subjecting the infliction of deserved harms to stringent procedural constraints in order to prevent mistakes or abuse. (This means that my agreement with Leroy’s claim that “the fact that one deserves to suffer harm does not imply that simply anyone is permitted to inflict the suffering” is partial and contingent. We ought to have legal institutions that are authorized to inflict deserved harms. But if someone who objectively deserves to be harmed escapes those institutions, he would not be wronged by an unauthorized individual who inflicted on him the harm he deserved.)
    Notice, though, that my concession that desert can help to justify an act that has a defensive effect is like the claim that it is necessary for the justification of self-defense on a particular occasion that it will produce, as a side effect, some benefit for a third party. Perhaps the act of self-defense will also harm an innocent bystander as a side effect, so that it would be disproportionate unless that side effect were outweighed by the beneficial side effect. This doesn’t mean that the possibility of benefits to third parties must be a part of our account of what justifies the infliction of defensive harms. The same is true of desert. And I continue to think that my original point is more than just terminological. When desert makes a necessary contribution to the justification of an act that has a defensive effect, it doesn’t contribute to the justification of the act as an act of defense. This is a case in which the act cannot be justified wholly on grounds of defense but must be justified by the combination of two distinct forms of justification, one defensive, the other retributive.
    In his discussion of the question whether the harm inflicted on a person who is compelled to compensate a victim of his prior action is an intended harm, Victor raises the traditional “problem of closeness.” I think Warren Quinn was right that that problem can be avoided by revising our understanding of which intentions are morally significant rather than revising our understanding of the scope of an agent’s intention. We have been mistaken in supposing that the relevant moral constraint is on intending harm. We in fact seldom the harm itself, except when we are inflicting what we take to be deserved harms, in which case the harms are presumably exempted from the constraint. Rather, the relevant constraint applies to intending to affect innocent people in ways that we know or believe will be harmful to them. (Those who are interested in pursuing this further can follow the reference in note 21 in my response in Ethics.)
    In the discussion, I’m grateful for Saba’s responses on my behalf to David’s questions about the notion of responsibility that informs my account of liability to defensive harm. The only point that Saba makes with which I would cavil is his acceptance of David’s description of a person who violates a strict liability law as responsible (and therefore liable) but not morally responsible. Strict liability laws give people a chance either to engage in a risk-imposing activity or not to engage in it, knowing that if they do engage in it and have bad luck, they will be held liable. The combination of voluntary choice and foreseeable risk is sufficient for moral responsibility for a harm for which one can be held strictly liable, in my view. Strict liability doesn’t hold people liable for mere causation, given that the liability rules warn people in advance that the activities for which they’ll be held strictly liable are ones that impose risks on others.
    For what it’s worth, I also agree with Saba’s response to Gerald’s point. E’s desert depends crucially on what E knows or believes in the circumstances.

17. BJ and Leroy
    I find the view that the value of deserved suffering depends in a non-contingent way on whether it is delivered by institutions even more mysterious than other versions of the Barbaric Thesis. Surely it is more plausible that institutions matter because they are more likely to calculate desert accurately and to deliver it for the right reasons.
    Jeff
    Thanks for the clarifications, and I’m glad that I seem to have understood the main features of your view. I should clarify that ‘the Stupid View’, unlike ‘the Barbaric Thesis’, did not refer to one of Jeff’s views (of course, for philosophers, in contrast with political leaders, it is better to have barbaric views than stupid ones).
    Your reasons to endorse the Barbaric Thesis seem weak to me. Praise, recognition, reward, blame and punishment need not be explained by appealing to the idea that it is a fundamental principle of morality that the moral value of happiness or suffering depends on the moral qualities of the recipient.
    Also, there is no reason to think that the idea that benefits can be deserved should lead to the idea that burdens can be deserved. We might think that benefits are always good, but they are even better when the recipient is virtuous. I don’t believe this, but I think it more plausible than the view that suffering can be good.
    Consider: D wrongs V. He later realizes this. He revises his values. He also carries out all of the duties incurred as a result of his wrongdoing. He is now content. Do you still think that it would be good that he suffer? This seems implausible and barbaric. What matters, to me at least, is that wrongdoers recognise that what they have done and carry out the duties incurred through wrongdoing.
    This can also help to explain why culpability makes a difference to liability – the culpable have more powerful duties to avert the threats that the create than responsible but non-culpable threat-creators. We then need an argument why we should impose more powerful duties on the culpable without referring to desert, but that doesn’t seem to pose great challenges.

18. Leroy
    your further cases seem excellent to me, especially a) and b) (I’m not sure I fully understand c))
    NCJ needs revision, then, at least to exclude
    1) harm that is consented to for the sake of a worthwhile goal and harm to self for the sake of a worthwhile goal (these seem closely related), but also
    2) harming a person to pre-empt a worse harm being suffered (as in the amputation case).
    GTR could deal with 2) by adopting Parfit’s revisionist conception of harm from Reasons and Persons s.25 where harming a person to pre-empt a greater harm is not harmful at all. I don’t think that this is the best strategy though as I think there is good reason to use harm in its more natural sense.

19. This is a great discussion and I am sorry to post a consideration that is somewhat off track (at best). But it strikes me that there is a common ‘barbarism’ involved in both the desert and the responsibility account of liability to defensive attack which is that they both view killing as continuous with harming, not as something qualitatively different. To be killed, it seems, is simply to be harmed A LOT. But does it not make an important difference whether you argue that an aggressor – due to either desert or responsibility – should in all fairness bare some burden or that he should be killed? It does not seem accurate to say that when you kill the aggressor you put a lot of/all the burden upon him. You destroy him. Further, I must confess to having abnormal intuitions in the innocent attacker/conscientious driver and similar cases. Of course it would be rather bad to be killed by the conscientious driver, but I’m inclined nonetheless to say that killing her (with a bazooka!) would be wrong. Or at least (more modestly) I would hope not to fire the bazooka myself in such a situation (unless I knew, for example, that she would die from cancer before the end of the week). This is merely gesturing, I’m afraid. I would probably be sympathetic to GTR’s ‘reasons of humanity’ (p. 125), but unfortunately they are also just gestured.

20. Allow me to enter the discussion at the point of disagreement between Jeff and Victor regarding whether deserving benefits symmetrically requires deserving burdens. In doing so, I may only display the folly of intervening outside one’s area of expertise, but that is that chance one takes. (Although the irony of defending the person who coined the term “Stupid View” with a stupid view should give me pause.) Jeff wants to say that “it seems arbitrary that people can deserve only benefits but not harms.” Victor offers the case of a regretful person changing his behavior as a counterexample. One might argue that the regretful person would have deserved punishment but did something in the meantime that undermined that desert. In other words, had the person not revised his values, he would have deserved punishment. That still leaves the possibility of symmetry open. Is there a different reason to believe that the symmetry does not exist? One might argue that, in general, having more good in the world is better than having more bad (if I am allowed this locution). One implication of this is that adding badness has to be justified to a greater degree than adding more goodness. (There are exceptions here, as in the example of the sadist who gets pleasure from his activities, but that is covered by the qualifier “usually.”) Motivating punishment requires stronger reasons that motivating reward. So there can be some cases where one might deserve a reward for a good behavior but not deserve punishment for a symmetrically bad one. Perhaps a pedestrian example of this would be that someone deserves praise for saying something friendly to an insecure person, while someone who knowingly fails to do so does not deserve any kind of harm, even where we might blame them in some sense. If this is so, then it seems difficult to conclude just from the fact that one can deserve benefits that one can also deserve harms.
    If this is right, it would seem to reinforce Jeff’s larger point that self-defense does not require desert. If desert in the case of harm needs more motivation than in the case of praise, this would place an extra burden on the person engaged in self-defense. It is unclear to me why a person in such a position should be faced with that extra moral burden. This would be, as Jeff says in his reply to GTR, “excessively restrictive.”

21. In public debate there is a lot of confusion between the notion of desert and the functionalist notion of defense. Just remember the reasons given for killing Osama bin Laden: “Justice has been done” referred to ideas of desert and punishment, “The worlds a place that is more secure now” refers to defense or averting of a danger. And indeed, in actual human acting, reasons of defense and reasons of punishment very often go together, in an emotional mixture. But in analysis, these two concepts should be clearly held apart. And this is perfectly done by Jeff McMahan. If it is a discourse of defense (self- or other-defense), it is about how to achieve the threat going away. The thinking is instrumental. There is no place for considerations of desert. If it is a discourse of punishment, things are different. First of all, punishment has its proper place within a legal system, even if it might be very rudimentary. So there are some positive settings in it. Now, we may consider what is the proper account of punishment. We are used to giving functionalist reasons again: “Punishment is for deterrence” or “Punishment is for prevention” and so on. Victor Tadros seems to hold such a view. Punishment as an application of desert seems barbaric to him, because he defines it as harming someone without reference to some good or preferred effect (“impersonally valuable”). And this is a general way we are thinking today. But just as a sting: Kant puts it the other way around. If punishment is just for outer aims such as a change of behavior of a certain person, if it is just deterrence or prevention or things like that, we do not treat this person as a moral one. We treat him like an animal which has to be tamed or like a machine which does not work properly and has to be repaired. This is why Kant thinks that punishment has to be done according to desert.
    Victor wants Jeff to do more to explain why he accepts the ‘barbaric thesis’. Well, maybe it is just because it is not that barbaric. Desert is not just the application of harm. In effect, it has nothing to do with harming someone. It has to do with giving somebody what he deserves. And we accept this at least in the positive sense. If an opera singer has done pretty well he deserves to be applauded. If someone finds my very important key which I have lost, he deserves a reward. I think that there is a very strong moral intuition that people deserve something in certain cases. Another and a different question is, what they deserve. I think this question cannot be solved without reference to practices in a certain community or to positive legal settings. There is not an a priori answer for this.
    But what about desert in the negative sense, as a punishment? I admit that it is hard for me to grasp the idea that I do something bad to someone without positive effect. If “there is no use”. Again, people like Kant would think that giving someone what he deserves is not bad or is not harming him, even in the negative. A wrongdoer should ask for his punishment. There is some metaphysical or let’s say: ‘transempirical’ assumption in this. This leads us to the fundamentals of morality itself. But there is a strong impulse that people deserve punishment in certain cases. But to explore this I cannot give any reasons, but just hint to our feelings. It not about “saying”, its about “showing”. Let’s assume that Anders Breivik was really ill when he killed all these boys and girls, and therefore he cannot be punished for this, because he was not responsible. People seem to accept this because he will be locked in anyway, as a means of prevention. But would people accept that the doctors say: “He is not ill anymore, he is no threat, he won’t do this again, there is no use in locking him away anymore” and a judge says: “Well then, he can go free.”
    Thanks for posting and reading my comment.

22. Hello everyone.
    On the exchange between Gerald Saba and Jeff re the degree to which E’s desert depends on what E knows or believes. Recall Gerald’s initial thought (quoted verbatim):
    (A) E threatens D in ways which D may avoid short of violent self-defensive action against E.
    (B) E threatens D in ways which D cannot avoid short of taking violent self-defensive action against E.
    Gerald suggests that we should not regard E’s desert as invariant. Sabba’s explanation for that suggestion (with which Jeff agrees) is that it really matters whether E knows, in (A), that D has an escape route. The thought is this: suppose that E culpably attacks D, and that he does not know (nor can be reasonably expected to know) that D can save his life by running into a dark alleyway. The fact that D has that option does not affect E’s desert: the latter is the same in this variant of (A) as it is in (B). Contrastingly, if E knows, in (A), that D has an escape route, then he deserves to be killed to a lesser degree than in (B). I find it puzzling. We may hold of course that D ought not to kill E in (A) (eg flowing from a general prohibition on inflicting unnecessary harm). But this strikes me as compatible with the view that E deserves to be killed defensively in both cases (assuming that he is equally culpable, etc), irrespective of what he knows of D’s defensive options.
    On the counterintuitive implications of NCJ1 – mentioned by Jeff in his Ethics reply, and Leroy above. I take the point re consensual/self-inflicted harm. It would be very easy however for GTR to revise NCJ1 accordingly, and restrict its scope to harm to others, as well as harm which is not intended as a means to provide a benefit to the person being harmed (to respond to Leroy’s second counterintuitive implication.) The only serious one seems to me to be the third (NCJ1 would thus disallow lesser evil arguments in favour of the deliberate killing of eg justified attackers or innocent bystanders. Victor: does that clarify it for you?)
    Finally, a theme running through Jeff’s comment is the putative asymmetry in some sceptical arguments about desert between harm and benefits: if it is appropriate (as it surely is) to say that agents sometimes deserve some benefits, then (Jeff says) we need to know why they cannot be held sometimes to deserve some harms. Victor’s worry (in his initial comments here) is that appealing to desert in the case of harm seems to commit one to the barbaric thesis, whereby it is impersonally good (sometimes) to inflict harm – whereas (implicitly) we do not worry about it being good (sometimes) that agents should be benefitted. In this thread, however, Victor distinguishes between two variants of the BT, as follows:
    BT(a): If D has acted wrongly it is intrinsically good that D is harmed.
    BT(b): If D has acted wrongly it is appropriate to harm D.
    The worry is BT(a). But if we go for BT(b), then we have less reason to reject appeals to desert in the case of harm, which in turn suggests that we can more easily accept appeals to desert both in the case of harms and in the case of benefits. That said, BT(b) does not seem barbaric at all to me – at least on common readings of ‘appropriate’, to mean ‘justified’ or ‘permissible’ – so much so that I doubt that it brings much to the table (in discussions of desert.)

23. Cecile,
    Thanks for this. I think that GTR intend lesser evil cases to be governed by principles of comparative justice. They might hold that lesser evil harming is all things considered permissible but prima facie wrong. This seems to cover Leroy’s (c) cases, if I understand them correctly, in a plausible way. In the original comments I outlined some reasons to doubt the idea that lesser evil cases always involve prima facie wrongdoing, though.
    On desert, I should probably clarify: BT(b) is intended to indicate that it is appropriate to harm D even where doing so does no good. We normally think this about blame: it is appropriate to blame a wrongdoer for his wrongdoing even if doing so has no instrumental benefit. Some might doubt this about blame. Let’s suppose that we do not. The question is whether the same thing can be said about harming D.
    Obviously, it is permissible to harm D to achieve certain ends (in my view, it is permissible to harm D as a means to achieve some end only if D has incurred an enforceable duty to serve that end). We can use the language of desert here if you like, but it is common, for the sake of clarity, to reserve the language of desert for non-instrumental reasons to give a person some benefit or cost. I take it that this is what GTR and McMahan, in different ways, mean by desert. Do you find the view that there is a reason to harm wrongdoers without any further purpose, but simply because it is appropriate in itself, barbaric? I do.
    Bernard
    Unlike many Kantians I think that it sometimes is consistent with respect for a person’s moral status to harm a person as a means to the good. This is permissible when the person has an enforceable duty to serve the good. Kant’s views about punishment were pretty mysterious and terrible. Some Kantians (Allen Wood, for example) think that these views do not fit naturally with the rest of his moral philosophy. Without being any expert on Kant, I tend to think that is right.
    Todd
    The reason why I focused on the repentant person who has fulfilled his duties is that it is often permissible to harm a person to achieve recognition or to enforce duties incurred as a result of wrongdoing, and both recognition and fulfilling duties will often be painful, perhaps in the latter case necessarily so given that the scope of a wrongdoer’s duties might be set not only by the magnitude of good that he does but by the costs that he bears in bringing about the good (in other words a wrongdoer has not fulfilled duties incurred as a result of his wrongdoing until he has borne a certain cost). If that is right, we may typically confuse our desire to see that wrongdoers recognize that they have done wrong and fulfill duties incurred with the barbaric thesis. If you think that repentance undermines desert, though, we need another way to achieve this. Perhaps we could think about whether a wrongdoer who cannot achieve recognition nor fulfill duties incurred through his wrongdoing has an independent reason to harm himself for no further purpose.
    Imagine a wrongdoer on a bountiful desert island. He cannot achieve recognition for psychological reasons and he cannot fulfill duties to anyone whilst isolated in this way. Does he have a reason to lock himself up in a cave and prevent his own escape, or perhaps to flagellate himself? BT(a) seems to imply that he has reasons of this kind, and perhaps BT(b) does as well, assuming that there is no reason to exclude the wrongdoer himself as the person to execute appropriate suffering. Needless, by now, for me to evaluate this proposal!

24. I would side with McMahan on the issue at stake here: desert and punition should play no role in the theory of self-defense. Like McMahan, I hold that legitimate self-defense is “necessarily instrumental” (160), whereas any cogent notion of desert in this context would suggest the desirability, all things being equal, that the supposedly deserving person receive harm.
    However, I believe similar things can be said about the intentional infliction of harm, even in self-defense. If legitimate self-defense is necessarily instrumental, then the straight line to take is that one can never intend harm in self-defense. To intend harm is to seek the harm as a good in itself, quite apart from its instrumental value in averting a culpable attack. On my understanding, this is precisely the kind of reasoning which underlies the doctrine of double effect (DDE).
    McMahan, however, disputes both the reasoning and (apparently) the reading of DDE. In his response to Gardner and Tanguay-Renaud, he reluctantly concedes the “strict and literal” truth of the view that imposing a duty of compensation on a tortfeasor is not “a case of intentional harming; the harm is, rather, a side effect of ensuring that the victim is compensated.” He worries, however, that

    “if what counts as an intended effect is understood in such a narrowly restricted way, the claim that there is a special constraint against intentional harming may be vitiated. For the same claim can be made about most inflictions of harm, not only in self-defense (“Yes, your honor, I harmed the police officer—yes, all right, fatally—but only as a side effect of preventing him from arresting me”) but in other cases as well (e.g., Jonathan Bennett’s case in which a pilot bombs a city with the intention of making its inhabitants appear to be dead, though this has as a side effect making them actually dead). It seems that those (including Gardner, Tanguay-Renaud, and me) who accept the relevance of intention to permissibility must accept either a broader concept of an intended means or an understanding of the relevance of intention to permissibility that is different from that found in the Doctrine of Double Effect.” (162-3)

    To the contrary, I believe the traditional reading of DDE takes precisely the view in question. For instance, Aquinas asserts that in legitimate self-defense, the killing of one’s assailant must be praeter intentionem – beyond the intention (ST II-II, q. 64, a. 7). I think that what goes missing in McMahan’s discussion is a distinction between intentionally inflicted harm and deliberately inflicted harm. To be sure, the tradition allows that one may deliberately inflict harm in self-defense, just like Aristotle’s sailors may deliberately throw their cargo overboard during the storm. No doubt, common English usage subsumes both of these under the term “intention.” It would appear, by contrast, that the Latin “intentio” is of narrower scope: it refers to an action’s superordinate end, not to the means adopted in order to bring that end about, i.e., the killing or the dispensing with cargo. (On this see Gregory Reichberg, “Aquinas on Defensive Killing: A Case of Double Effect? pp. 362-364.) If this is correct, it might simply be misguided to attempt to make our reading of DDE produce verdicts which (as McMahan says elsewhere) “conform to our linguistic intuitions”, insofar as these are intuitions about the semantics of the English term “intention,” not about the Latin “intentio.”
    Nevertheless, one may wonder if this line of thought – even if it is cogent – would threaten to render the DDE “limitlessly permissive,” as McMahan says elsewhere. Certainly, the DDE has its share of problems. But far from being limitlessly permissive, I doubt that it would permit even the two cases that McMahan cites. Specifically, the supposition that DDE might permit these cases can only come from failing to factor in further crucial constraints involved in the DDE: (i) that the intended end itself be permissible; (ii) that the means be necessary to achieving that end (or at any rate, the least harmful means of achieving it); (iii) that the side-effect harm be proportional to the good we seek to achieve. Neither of the cases McMahan mentions satisfies these further constraints. In the first case, resisting arrest is presumably not a permissible action, and, at any rate, killing the arresting officer is not proportional to the good of escaping arrest. For the second case, exterminating an entire city’s population is presumably neither necessary nor proportionate to the end of making them “appear to be dead,” whatever the strategic importance of that end may be.

25. IMPORTANT:
    Dear All,
    A number of people are having trouble posting their comments. Josh Glasgow informs me that there is a word limit on comments imposed by TypePad and over which we have no control. So if you have a long comment, try dividing it into two as Endre Begby has. Also, given the trouble that people are having, it is advisable to compose your post in a word processor and then paste into the TypePad comment box. That way you won’t lose your work. We apologize for the technical difficulties. If nothing seems to work, then try contacting the editors (email all four of us) and we’ll post the comment for you. See the About section above for our names.

26. Here are our email addresses:
    Douglas.Portmore@asu.edu
    dshoemak@tulane.edu
    DanBoisvert@uncc.edu
    joshuamglasgow@gmail.com

27. Like Jeff, I hold that retribution (and desert) do not contribute to the justification of act of defense as an act of defense. Unlike him, I also hold that retribution does not contribute to the justifications of acts at all. It seems to me that, if Jeff holds that retribution can so contribute, then he should focus his theory of self-defense on the question of when acts of defense *do not wrong the attacker*. This would provide, I think, a clearer picture of what Jeff’s view imply in concrete cases (and would be sensitive to retribution).
    I’m here assuming that, if retribution is relevant to the permissibility of actions, then it is relevant to whether individuals are wronged (e.g., owed apology or compensation). (By contrast, the benefit of saving a million lives may be relevant to the permissibility of killing an innocent person, but it is arguably not relevant to whether that innocent person is wronged.) Perhaps, however, Jeff, thinks that retribution is not relevant to whether an individual is wronged?

28. Cecile,
    You mentioned that you are puzzled about Gerald’s suggestions — why think that E is more culpable in (B) than in (A), assuming that E is aware of D’s options in both cases? Why think that E deserves to be killed more in (B) than in (A)? This is a worry that I had. (I hope this is your worry – if it isn’t then this post should be read as addressing a worry related to yours). My own response to this worry is that the degree of wrongful risk knowingly imposed on a victim is relevantly to culpability. Consider the two cases:
    (A*) E wrongly throws at D a dart the tip of which is coated with a poison that has a 20% chance of killing D. This is the most potent poison E could find. E throws the dart with the malicious intention to kill D.
    (B*) E wrongly throws at D a dart the tip of which is coated with a poison that has a 95% chance of killing D. This is the most potent poison E could find. E throws the dart with the malicious intention to kill D.
    And suppose E is aware of all the relevant facts in both cases. Now there is a sense in which E is an equally bad person in both cases. After all, in both cases E is maliciously doing his utmost to kill D. But we shouldn’t infer from this that E’s culpability is the same in both cases. Mutatis mutandis for Gerlad’s original case (with the caveat that E is aware of all the relevant facts): E maliciously imposes a lesser risk on D in (A) than in (B), since in (A) D is less likely to be killed in (A). Though E is doing his utmost in both cases, the fact that he knowingly imposes a lesser risk in A diminishes his culpability relative to B.
    But what grounds the claim that E’s culpability differs between the two cases? We might say that what you know about the risk you impose places constraints on the degree of culpability you bear. If I know that I’m imposing just a 20% risk of death, then this delimits my culpability – even if I would impose a greater risk if I could, and even if I really hope that the 20% comes to pass.
    Here is a more complete explanation: the object of culpability is different in the two cases – in A* the known object of culpability is that of imposing a 20% risk of death, and in B* the known object of culpability is that of imposing a 95% risk of death. In fact, to ask whether E’s culpability is the same in the two cases is ambiguous between asking whether the *degree* of culpability is the same, and whether the known *object* of culpability is the same. We can admit that the degree is the same (100% culpable!) but that their objects differ. And a person who is 100% culpable for a imposing a 20% risk of death might be thought of as less culpable than someone who is 100% culpability for imposing a 95% risk of death. (Again, this assumes that E is aware of all the relevant facts in both cases).
    Gerald’s original claim (with the caveat that E is aware of all the relevant facts) gave me pause; it wasn’t obvious to me that even with the caveat E’s culpability would differ between (A) and (B). But the above ruminations brought me back around to Gerald’s view (with the caveat).

29. Victor,
    Thanks for your generous “intervention” (as we say in Continental philosophy), and for your patience and acumen throughout this thread. Since we are largely in agreement, let me just clarify my position vis-à-vis yours. My reservation is not with your position, call it a rejection of the Symmetry View, but with your original example. I think one could argue that the example leaves open the possibility (to those who would like one) of holding on to the Symmetry View. Since I reject that view, I don’t need that possibility myself. My point then was to try to back up your rejection of the Symmetry View from a different angle.
    Turning to your response, achieving recognition and enforcing duties seem to me distinct from desert; I think you recognize this point by contrasting both with the barbaric thesis. If we move then to the desert island example, my view would commit me to the possibility that there would be no reason, based on desert (moral as opposed to island), to sit in a cave. And I accept that. I don’t want to make the stronger claim that the wrongdoer definitely does not deserve to harm himself. That would require a more fine-grained context to consider. Rather, my point is that if we take on board the view that, in general, more goodness is better than more badness, and following that (but perhaps not following FROM that) that motivating punishment requires more justification than motivating reward, then if we want to tie desert to self-defense we make justifications for self-defense “excessively restrictive.” This would be a rejection of the Symmetry Thesis (deserving benefits entail deserving burdens), but in keeping with Jeff’s larger point.

30. Victor
    A quick clarification. I wrote that “there would be no reason, based on desert…” That was a mistake. I should have said, “there MIGHT be no reason…”

31. Victor,
    I was tempted to jump on Bernard’s Kantian trolley before it derailed. There is, however, further insight in his comments. Bernard is surely right in observing that self-defense is characteristically instrumental: it aims at thwarting the threat. The question as to what the attacker deserves Bernard thinks “cannot be solved without reference to practices in a certain community or to positive legal settings.” This calls to mind your response to BJ and me.
    You write, “I find the view that the value of deserved suffering depends in a non-contingent way on whether it is delivered by institutions even more mysterious than other versions of the Barbaric Thesis. Surely it is more plausible that institutions matter because they are more likely to calculate desert accurately and to deliver it for the right reasons.”
    Where you see mystery I find social order. (I think Bernard might concur.) Do you intend that vigilantism is morally equivalent to due process? There is no such thing as desert simpliciter. I think we can agree that desert entails something owed to someone for something done to or received by someone. I should think desert of any kind must be grounded in some notions of human social relations, from which a prior consideration of quid pro quo is derived that governs what is deserved, by whom, and for what reason, no matter whether the desert is positive or negative, whether moral, criminal, or commercial, etc. I grant that in the case where I owe my neighbor some compensation for damage done to his property when the tree I felled landed on his car—a sort of chainsaw massacre—he deserves to be paid by me directly. But the thief that stole the birdbath from my yard does not deserve punishment by me. No doubt, had I caught him, I might have finagled compensatory service from him—perhaps he (she? I may be looking for the wrong “man”) would have agreed to clean out my gutters rather than to go before the magistrate.
    Thus far, it is a contingent matter that I may not punish, since we can imagine having organized a small society—perhaps a tribal entity of limited size, in which the victim metes out the deserved punishment directly. As you correctly point out, the prudential advantage of, say, legal institutions, for administering punishment (and, for that matter, deciding issues of tort) is to improve the accuracy of measuring desert. I don’t think we can extrapolate much beyond these simple cases, without invoking an abstract authority as conceptually necessary.
    (continued in next comment)

32. (Leroy resumed from previous comment)
    For, at a more complex level—even before we consider self-defense in war—it is only the socio-political structure that can instantiate the several variables in the complex quid pro quo. This is not a contingent matter. The mystery to me is how one can make coherent philosophical sense of any notion of desert without conceptual dependence upon some quasi-institutional socio-political order—I grant that in some societies it may be much less “formal” than we expect in modern cosmopolitan society. I do not mean to suggest that there must be a “moral” institution to determine what is morally deserved. However, social-political institutions assume limited moral responsibility (e.g. to administer justice) nevertheless.
    Now consider the more germane case of desert in war. Recalling Jeff’s concern in Killing in War to distinguish between just combatants and unjust combatants (e.g. on the basis of fault under jus ad bellum to which the soldiers fighting for an unjust cause are party). There are many layers of moral engagement to consider. At one extreme, Adolf Hitler one might regard as having deserved to suffer harm for instigating a terrible war and for perpetrating heinous atrocities against humanity. Apparently he committed suicide. Does it not matter morally that he deprived the world of the opportunity to have rational, principled judgment determine his guilt and to have meted out his punishment, on behalf of humanity? If it does not matter that Hitler committed suicide, then it would not matter if an enraged Allied guard were to have assassinated Hitler, had he been taken prisoner. In regard to desert, the manner in which Osama Bin Laden died appears morally less disturbing than the manner in which Hitler died; and the hanging of Saddam Hussein, in consequence of a trial by law, would seem even less disturbing morally. Even a legal positivist can agree. These figures are morally blameworthy on grounds of jus ad bellum and of jus post bellum.
    The “mysterious” aspect of all this is that, as a deep and abiding feature of what I call the cultural dimension of war (to extend a notion of Michael Howard’s), we insist upon the importance of a formal, military armistice, followed by a grand settlement in accordance with international law. This is part and parcel of jus post bellum (of which the founding of the United Nations is an example). The Nuremberg Trials were a function of jus post bellum. If it doesn’t matter morally who metes out deserved harm at the individual level, why should it matter (as it seems to do) at the grand level.
    Whew! Thanks for the fire under my seat.
    Kind regards, Leroy

33. Todd – thanks for the clarification.
    Leroy – thanks for tat. Of course, I don’t endorse desert claims at all. I was only suggesting that if thought desert was important I would think that whether institutions deliver it is relevant only for contingent reasons (my sense is that GTR and McMahan will agree with this).
    Criminal justice is motivated in other ways in my view – its functions are a) to protect citizens against crime, primarily through deterrence; and b) publicly to affirm the rights of victims and criticize or condemn offenders for what they have done. In the Hitler case, the second function, at least, was problematically undermined.
    The institutional question does raise an issue for GTR. Suppose that we accept BT. Even if we accept that state institutions have a primary role in delivering desert only for contingent reasons, where those institutions are in place it might be argued that no-one else ought to act on desert-based reasons. This might pose a problem for GTR. Here is an argument
    1) GTR hope to explain the asymmetry between attackers and defenders through desert.
    2) Whether defenders can act on desert-based reasons depends on the existence of institutions which might deliver desert more accurately and effectively than defenders.
    3) The asymmetry between attackers and defenders does not plausibly depend on whether such institutions exist.
    4) Therefore desert cannot explain the asymmetry between attackers and defenders.
    I’m not sure that this argument is successful, but it is worth thinking about.

34. Consider the following cases:
    Republican Guard: In a just war of self-defense against an aggressor, your military unit is on the way to taking aggressor’s capital city and forcing the surrender of the rotten dictator who rules the aggressor state. You are confronted by a battalion of 5,000 Republican Guards, elite volunteer troops loyal to the dictator. You have two ways to defeat these Republican Guard troops. You can call in an airstrike that will incinerate every one of them, or you can direct tank fire at their front lines, causing several hundred enemy casualties, and mass disarray among the rest of the enemy battalion, leading to their prompt surrender. Either way you will be able to take the capital city by the next day. Your own expected casualties are zero if the airstrike option is chosen and P1 if the tank fire is chosen.
    Draft Army: Same as the Republican Guard case except that the enemy troops are all drafted members of the regular army.
    Most non-pacifists would say that in a just war you are permitted to kill both enemy draftees and enemy volunteer soldiers. But surely the Republican Guard troops bear greater culpability than the draftees and are more deserving of being killed in self-defense. Or, to put it the other way around, the excuses for the draftees are much greater than for the Republican Guard troops. For many non-pacifists, there is a value of P1 such that they would consider it permissible to call in the air strike on the Republican Guard, but wrong to call in the air strike on the draftees, even though doing so avoids the deaths of P1 members of your unit.
    Cornered Murderer: The police are searching for X, a vicious murderer. One police officer with gun drawn has X cornered, when X takes out a knife and charges the officer. The officer could aim to kill X or aim to wound him, though the latter involves some risk, P2, to the officer.
    Cornered Hallucinating Attacker: The police are searching for Y, an innocent person into whose drink a villain has dropped a hallucinogen, causing Y to become deranged and kill someone with a steak knife. One police officer with gun drawn has him cornered, when Y takes out the knife and charges the officer. The officer could aim to kill Y or aim to wound him, though the latter involves some risk, P2, to the officer.
    There are some values of P2 that would justify shooting to kill both X and Y. Likewise, there are some values of P2 such that shooting to kill wither X or Y is impermissible. But it is likely that there are some values of P2 for which shooting to kill X is justified but not Y.
    In both sets of cases, what may permissibly be done depends on a proportionality calculation that weighs the harm to aggressors (whether culpable or nonculpable) against the benefit of reducing harm or risk of harm to those defending against the aggressor. The fact that we make different proportionality calculations in Republican Guard versus Draft Army and Cornered Murderer versus Cornered Hallucinating Attacker seems plausibly to be the result of the different culpability of the aggressors in the different cases.
    Thus, it seems to me that GTR are mistaken when they say that if there is a proportionality restriction in D’s right to self-defense “as seems likely, then what E deserves is not its metric, nor a plausible candidate to be part of its metric. D’s self-defensive steps in such a case need to be proportionate to something else.” (p. 133).

35. The discussion seems to have fizzled out but I’ll add one more comment for the sake of future historians who may comb back through these archives.
    (1) I agree with part of Todd’s response to Victor: namely, that a person can through reform, atonement, and reparation cease to deserve to be harmed. It’s better when it happens this way. The infliction of deserved suffering should, if possible, be a last resort.
    Victor later implies that he thinks it absurd that a person should inflict deserved harm on himself. But there is a tradition of thought according to which this kind of penance or atonement is entirely appropriate. I’m not convinced that it’s wrong. There was an instance late in his life when Dr. Johnson returned to the market square in his home village and stood in the rain all day as a means of atoning for some much earlier failure to help his father in that same market square. That does not seem absurd to me; rather, I find it quite moving. This is in the spirit of Bernhard’s remark that “a wrongdoer should ask for his punishment” (though I’m skeptical of the further suggestion that deserved punishment is not an all things considered harm – if it weren’t it wouldn’t really give the wrongdoer what he deserves).
    (2) I share Morten’s sense that the moral significance of killing is not entirely continuous with that of inflicting lesser, nonlethal harms. But it’s difficult to identify precisely what the difference is. One problem is that all of us have prudential equivalences between the certainty of a some nonlethal harm and a certain probability of death. After all, we routinely accept some small probability of death for the sake of a single benefit. And the badness of death is really nothing more than the loss of a certain range of benefits.
    (3) I made a few notes while reading Cécile’s comment but found later that Saba had anticipated all I wanted to say and went well beyond it with the distinction between the degree of culpability and the object of culpability – I was definitely out-sophisticated by that.
    (4) While I appreciate the subtlety of the position that Endre sketches on the relevance of intention to permissibility, I remain unconvinced. As I understand him, Endre wants to say of my man who has killed a police officer to avoid being arrested that although he didn’t harm the officer intentionally, he did harm him deliberately. I’m not sure whether “deliberately” is supposed to be equivalent to “knowingly,” in which case there’s no special constraint against it, or whether there is supposed to be a hierarchy of objectionableness from knowingly harming the innocent, through deliberately harming the innocent, to intentionally harming the innocent. In any case, Endre suggests that one can deal with the case of my killer by noting that his end is impermissible and the harm inflicted disproportionate. It seems true that in my case the intended end is impermissible, though even that might be disputed, as it may depend on exactly how the end is specified or described. But that is merely a contingent feature of the case. Change it to lopping off the head of the person in front of you at the sports event to get a better view of the field. There the end is permissible but the means disproportionate. Again, though, intention isn’t doing any of the work. What was thought to be the main element of the Doctrine of Double Effect – intention – seems to have become almost superfluous. In most cases the work is going to have to be done by the proportionality clause. But what makes the Doctrine of Double Effect a distinctive view is that it holds that the proportionality restriction is more stringent in the case of harms that are intentionally inflicted. It seems to me that Endre is in danger of embracing an emasculated doctrine according to which most judgments of proportionality in the infliction of harm on innocent people are going to be judgments about proportionality in the infliction of unintended harms. This is a weaker constraint than the Doctrine of Double Effect has traditionally been thought to provide.
    I should mention, in closing, that François Tanguay-Renaud has been prevented from joining the online discussion by a family emergency. I know he was eager to participate and regrets being unable to do so.
    Thanks again to everyone for such a stimulating discussion.

36. If you have trouble posting your comment, please contact the editors and we will make sure your comment is posted as soon as possible. We apologize for the technical difficulties. For some reason beyond our understanding, comments are getting marked as spam for no good reason. We’re trying to work with our host to resolve the issue. Thanks for your patience.

37. First time contributor. Thanks to the editors for asking me to contribute. Sorry to be so late in doing so.
    I have to confess that I have read much but not all of what proceeds this, so my apologies if I repeat something that has already been discussed.
    I want to offer a friendly suggestion to Jeff. This is offered in response to what I think is the strongest point made by GTR, who rightly focus (p. 132) on the peculiarity of treating SD as justified with reference to liability, if that is connected to responsibility, when a person is also excused—-as is the case when confronting an innocent aggressor (IA).
    What I suggest is that Jeff has run together, under the heading of “liability,” two justifications for targeting an attacker. One is the case of a culpable aggressor, who is at most partially excused. He is indeed liable to be attacked in virtue of his actions. Moreover, proportionality would allow, in essence, an infinite number of such attackers to be harmed–as long as the individual harm in each case is proportionate to the threat–to save an innocent victim.
    That case really should not be assimilated, however, to the case of the IA. A different justification is called for there. I suggested in a paper on abortion and innocent threats–one that Jeff read and liked long ago–that the best way to think of that case is that the IA is morally burdened by being in a state that he or she should not want to be in. As a result, his or her claim not to be harmed should, in the balance, be weaker than the claim of the innocent victim (IV) not to be harmed. The IV, or a third party, has a right, then, to inflict as least as much harm on the IA as he or she would inflict on the IV if doing so is necessary to save the IV from that harm. But at the same time, the proportionality calculus must look much different.
    Here’s an analogy with a trolley switch case. If the trolley is headed to one person, that is the problem of that person first. He or she is not permitted to turn it onto another to save his or her life. (He or she might be excused, fully or partially, for doing so, given the pressures of the moment.) And a third party should try to prevent him or her from turning what is his or her burden onto another.
    The same is, I suggest, true of an IA. Being the bearer of a threat is morally analogous to having the threat pointed at one, like a runaway trolley. A third party should help an IV avoid that threat, even if it means causing as great a harm to the IA. But just as a trolley may be turned from two onto one-—or so I believe-—so may two IAs have claims not to be harmed that outweigh one IV’s claims not to be harmed.
    Notice, this has nothing to do with being responsible as an aggressor, as opposed, say to being a mere projectile, as an innocent threat (IT) is. Proportionality should work the same for both. What matters is only that the IAs or ITs are rights bearers. And the question in all cases is how strong those claims should be in the balance of competing claims that determine who, in the end, has a right to act in self- or other-defense.
    That said, I think Jeff is absolutely right, for the reasons he has given, to think that desert is generally irrelevant to SD, as it both does not explain why SD is permissible against innocent aggressors, nor does it explain why what may be done against culpable aggressors normally far outstrips the punishment they deserve.

38. Jeff
    On Dr Johnson: even if BT is true, Dr Johnson can’t possibly have deserved to stand in the rain all day long without a hat on for failing to attend his father’s market stall as a boy fifty years before. This idea in itself strikes me as fairly absurd. If that is what he thought, the fact that he continued to feel deeply guilty about what he did as a child is tragic and probably shows the awful power of some hideous religious ideas on the human psyche.
    Perhaps there are better ways of understanding Johnson’s conduct, though, that don’t appeal to the intrinsic goodness of him suffering. Antony Duff, in his account of punishment, suggests that a period of suffering can help to structure a person’s grasp of their wrongdoing. It seems a shame that some people need to suffer to achieve this, but it may nevertheless be a fact about some people’s psychology that suffering is the best way of achieving recognition, understanding, reconciliation etc. Even on this explanation, which is more attractive than BT, Johnson’s case seems pathological and tragic.

39. Coming rather late to this discussion, I have the feeling that most of the points that interested me have been covered, but I would like to make a brief comment on one of Victor’s objections to what he politely calls “the barbaric thesis”. His own position really deserves a name as well so I suggest, in a purely descriptive spirit, “the wimpish thesis”.
    In support of the WT, Victor imagines a case where some wrongdoer has “revised his values” and has “carried out all of his duties incurred as a result of his wrongdoing” and thinks it “implausible and barbaric” that there should be any good in his suffering at all for his wrongdoing. One might, as Todd May argues, hold that reform of this kind cancels the desert of punishment. There is, after all, no reason why something “valuable in itself” should always be chosen or remain valuable all things considered in certain contexts. Pleasure is often regarded as a good in itself (as pain is an evil in itself) but can often be foregone in order to achieve other goods (just as pain can endured for good purposes) or even vitiated by its context as in malicious pleasures. Again, one might question whether some wrongdoer who simply reforms (“revises all his values”) has done enough to acquit him or herself of the wrong they have done. Consider the Nazi death camp murderer who has escaped to South America and subsequently lives an exemplary life but has never suffered the least inconvenience for his crimes. Surely he deserves some negative life dividend for what he has done?
    Of course Victor adds to his story the bit about “the duties incurred” by the wrongdoing, but, as he seems to admit, this requires some unpacking that doesn’t refer to desert. He thinks this is easily done but I’m somewhat skeptical about that. I think one such duty is plausibly social accountability for the crime. Here, the talk of “suffering” can slant the discussion too much towards “barbarism”. Certainly punishment involves deprivation and sometimes suffering is the appropriate term, but Victor gets more plausibility for his barbarity accusations by citing such things as flagellation and self-imprisonment. One can favour the view that (serious) wrongdoers deserve negative outcomes without endorsing the barbarities of capital punishment, flogging, extreme solitary confinement, or even many forms of imprisonment. If the desert island chap imagined by Victor discovers one of his victims cast away on the island and thinks he owes the victim nothing by way of recompense or satisfaction for the crime, then it seems to me he is wrong.
    As for Dr. Johnson (cited by Jeff), the issue isn’t whether Dr. Johnson’s reaction is pathological or extreme but whether something of this kind, though perhaps less dramatic, is appropriate by way of desert. Knowing Victor, I’m surprised that he finds spectacular and unconventional displays pathological!
    Tony Coady

40. Tony,
    thanks for the comment. I am quite wimpish.
    The real disagreement is not about whether it is permissible to make wrongdoers suffer at all. It is about whether it is permissible to make wrongdoers suffer for no further end. Nothing in your comment indicated that you endorse this idea. For example, I agree that it may be permissible to harm wrongdoers to force them to provide an account of their wrongdoing to their victims, to encourage them to recognise that what they have done is wrong, to compensate and so on. You seem to run these ideas together with desert and I want to separate them (as, I think, do GTR and Jeff).
    I also think that the duties incurred by wrongdoers extend beyond compensation in a number of ways. I also think that the scope of these duties might be set not only in reference to whether sufficient good is obtained but also the magnitude of cost on the offender. For example, suppose that D seriously wrongs V. D could fully compensate V at little cost to himself (however ‘full compensation is to be understood). D, in this case, may owe V more than full compensation. It is permissible to harm D for the sake of further benefits to V.
    What I don’t believe is that it is permissible to harm wrongdoers simply for the sake of harming them and with no further end in sight. It is that thesis that I label barbaric.

41. Many thanks to Jeff for the response. By way of a quick follow-up, I want to note that my initial comment was concerned to rebut the claim that the DDE (on the interpretation that I favor) would be “limitlessly permissive.” I take it we now agree that this is not the case. However, Jeff still worries that my interpretation would “emasculate” the DDE, and offer “a weaker constraint than the [DDE] has traditionally been thought to provide.”
    I take it that “weaker” here means “more permissive.” If so, I’m not convinced. This matter could only, I suppose, be decided by considering cases. And, crudely put, I have yet to see a case which my interpretation would permit but Jeff’s would not. I could be wrong, of course, but my rough guess is that my interpretation will be exactly as restrictive as Jeff’s, once the constraints of proportionality and necessity are correctly factored in.
    But this reliance on proportionality and necessity is also of concern to Jeff. He voices the worry that, on my account, “intention isn’t doing any of the work” and has become “almost superfluous.” To the contrary, I would insist that intention is doing a lot of work in my account: pertinently, it renders strictly impermissible all acts involving intentional infliction of harm.
    I think this gets us to the heart of the matter. Jeff appears to worry that intention isn’t doing any *real* work in restraining action on my account, since agents can always redescribe their actions in ways that would not implicate them in intentional harming. In this way, the restriction on intentional harm is effectively bypassed, and the proportionality and necessity constraints are left to do all the work.
    This kind of thinking is widespread. It can be traced back at least to Nagel’s “War and Massacre,” where it is argued that the DDE is vulnerable to certain “casuistical problems,” in that its application “depends on one particular description of the act, and the reply might be that the means used against the guerrillas is not: killing everybody in the village—but rather: obliteration bombing of the area in which the twenty guerrillas are known to be located. If there are civilians in the area as well, they will be killed as a side effect of such action” (p. 131).
    This certainly is a worry for the public assessment of actions. I am not convinced, however, that it threatens the underlying moral principle. The tradition from which the DDE stems is staunchly realist about intentions: agents can redescribe their actions all they want, but it won’t change what they actually intended. And the moral principle applies to the actual intention, not to the averred intention. If this problem of public assessment threatens my interpretation of the DDE, I cannot see how it would fail to similarly threaten any other account which (i) holds that intention is important to permissibility, yet (ii) allows that intentions cannot be gauged directly from the outward characteristics of the act.