Doing, Allowing, and the State

(late draft, please cite published version) Doing, Allowing, and the State Thanks to audiences at the University of Chicago Law School and the Centre for Values at the University of Colorado, Boulder. For very helpful written comments, special thanks to Chris Heathwood, Mike Huemer, Martha Nussbaum, Arden Rowell, and David Strauss. Adam Hosein 1. Intro The doing/allowing distinction plays an important role, as I will illustrate, in our thinking about a number of legal issues, such as the need for criminal process protections, prohibitions on torture, the permissibility of capital punishment, and so on. These are areas where there seems to be a moral difference between harms that the state inflicts and harms that it merely allows. In this paper I will argue for the importance of the doing/allowing distinction as applied to state action. I will begin explaining what exactly the doing/allowing distinction is. I will then present the familiar deontological case for the doing/allowing distinction as applied to private actions and make a prima facie case that it applies to state action also. I will then address some important recent criticisms of this view. Sunstein, Holmes, Vermeule, and others have presented influential arguments for the claim that where the state is concerned the doing/allowing distinction has no moral significance, even if it does elsewhere. I show that these arguments can be resisted. In doing so I defend some important positive theses that help us understand the state’s role in protecting people from harm, including, but not limited to, the following (some of these theses involve technical terms which I will later define): We should keep separate the doing/allowing distinction and the more familiar (in legal theory at least) action/omission distinction. Objections to using the latter to defend the relevant policies do not apply to using the former. Although the state is substantially causally involved in who gets killed in our society, its complicity in these deaths is usually limited by the intervention of other actors, and so on, in the causal chain. The state’s failing to prohibit or to prevent an action is not sufficient to count as its authorizing that action, especially where the reason for not prohibiting/preventing it is to respect rights. The special responsibilities of the state give state agents extra reasons to help individual citizens but also give them extra reasons not to harm individual citizens. So those responsibilities should not change the overall relative weight of the state’s duties not to harm and to provide aid. One can only acquire special responsibilities to do things within the constraints that are set by people’s rights not to be harmed. Thus, the special responsibilities of states do not increase the amount of harm that they are permitted to inflict, even on non-citizens. 2. The Doing/Allowing Distinction and Why it Matters There is a distinction between bad things that we do to other people and bad things that we merely allow to befall them, centrally between harming people yourself and allowing them to suffer harm. So, for instance, suppose that I take a mallet and use it to whack your toe. In this case I have done harm to you, since I have crushed your toe. Now suppose that I am standing next to you and a heavy jar, which I could’ve deflected, lands on your toe. Or suppose that I am standing next you and someone else, a third party, whacks your toe, even though I could’ve pushed them away to prevent this. In these latter cases I have allowed a harm to befall you rather than inflicted it myself. These are clear cases in which it is easy to classify my actions into doings and allowings, but there are certainly examples that are difficult to so classify. And philosophers have tried in vain for some time to give an ‘analysis’ of what it is to do something to someone rather than allow something to happen to them. That is, they have tried to figure out exactly what conditions must be satisfied for an action to count as a doing rather than an allowing of harm. While there are now some plausible analyses out there, none is without problems. So, there is some unclarity surrounding the distinction. But all the same we have a basic grasp on it that should be enough for present purposes. The ‘doing/allowing distinction,’ as it is known by philosophers, plays an important role in our common sense moral thinking. Legal theorists tend to call this the ‘action/inaction distinction’ instead, but I think this invites confusion, as I explain in Section 2, and so have stuck with the standard philosophical terms. According to that morality we have duties not to inflict harm and also duties to prevent others suffering harm (that is, duties not to allow them to be harmed). The former are often called ‘negative’ duties and the latter ‘positive.’ Common sense morality treats these duties quite differently. The difference, as Judith Jarvis Thomson puts it, is that the two types of duty have different “weight.” Judith Jarvis Thomson, “Turning the Trolley”, Philosophy and Public Affairs 36 (4) (2010): pp. 359-374; p. 360. Here is a classic example illustrating this difference: Driver: Five people are in danger of drowning. The only way to save them is to drive down to the shore and pull them out of the water. But to get there in time to save them you must run over, and thereby kill, a single person who has fallen on the road and broken his leg. This is a modified version of the case discussed in Philippa Foot, “Killing and Letting Die”, in A. Norcross and B. Steinbock (eds.), Killing and Letting Die (New York: Fordham University Press, 1994), pp. 280-289. It seems impermissible for you to run over the one so that you can get to the shore and save the five. Why? After all, you have both a duty not to harm people and a duty to save people whose lives are at risk. A good explanation of our reaction to the case is that one’s duty not to harm the single individual outweighs the duties one has to the five. Although, as I discuss below, this explanation is controversial and has been much debated in the philosophical literature. There is something morally worse about killing an individual than failing to save someone from death and so one cannot kill a person to save someone else. In fact, killing one person is morally worse than failing to save five other lives and so one cannot kill even to save five others. Some recent authors say that we do not need the doing/allowing distinction to explain our duties. See, for instance, Jonathan Quong, “Killing in Self-Defense”, Ethics 119 (3) (2009): pp. 507-537. These theorists say that you do wrong in Driver not because you harm the one but because you use the one as a means to save the five. I don’t think this is plausible because in any ordinary sense of ‘using’ you are not using the one: his presence is completely incidental to your goal of reaching the five (in fact, maybe his presence slows you down by making the path more uneven). Quong might say that you are all the same using something that belongs to him, namely, the space he occupies, and that this is wrong. In that case, let’s suppose that the one is lying near the road and the fumes produced by your car will poison him to death if you drive by. You would not be using him or anything that belongs to him, but it would still be impermissible to drive down the road for the five. The reason why is that you would be harming him. Thus, we need the doing/allowing distinction to explain our duties in Driver and similar cases. Thomson uses a similar example to make the same point in Judith Jarvis Thomson, "Killing, letting die, and the trolley problem", The Monist 59 (2) (1985): pp. 204-217. For another argument that we need the doing/allowing distinction, see Michael Moore, “Moore’s Truths About Causation and Responsibility: A Reply to Alexander and Ferzan”, Criminal Law and Philosophy 6 (2012): pp. 445–462. Now, it might be that the using/not using distinction matters in addition to the doing/allowing distinction: harming someone as a means is even worse than simply doing harm to her. If so, then there is an even stronger case against torture and so on, and the arguments I give could be easily modified to justify applying that distinction to state actions also. Thanks to the editor for suggesting that I address this issue. More generally, the moral significance of the doing/allowing distinction is that duties not to do harm have greater weight than duties not to allow harm. This difference pervades our everyday moral thinking, and the Driver case is just one example that illustrates it. To clarify though, the distinction isn’t typically thought to apply to just any way of damaging someone’s interests but, rather, a specific set of their interests including, for instance, their interests in life, bodily integrity, and so on. One’s duties are not to harm people in these particular ways, and it is harming people in these ways that is worse than failing to aid them. There is a large body of philosophical work devoted to explaining the moral difference between doing and allowing harm, too large to survey here. But the basic intuitive idea behind the distinction is fairly clear: as an agent I must take special responsibility for what I do to other people as opposed to what merely befalls them or what other people inflict on them. When I do harm to someone I treat that person worse than I treat someone whom I merely allow to die. Thus, although I have a duty not to stand by when people are being harmed, I must give greater weight to ensuring that I do not harm anyone myself. So far I’ve been talking about the doing/allowing distinction in the context of actions performed by private individuals. It also seems perfectly intelligible to speak of what the state does versus what it merely allows. For instance, when a public executioner executes someone we think that the state has killed that person. I don’t mean to suggest that it’s always, or even usually, clear whether a given event is a state doing or a state allowing. I just claim that there is such a thing as the state doing harm rather than allowing harm. Both state inflicted harm and state failures to help can be moral wrongs. All the same, we ascribe in this context too greater weight to the duty not to inflict harm. For instance, suppose that a police officer or firefighter is on duty and in the position described in Driver: she can save five people but only by running over and thereby killing one other person. Here too we think it would be impermissible to kill the one. So, we treat state inflicted harms as worse than state failures to aid. This is not surprising given our judgments about private doings and allowings. State actions, just like private actions, are performed by people, sometimes acting in concert, such as police officers and firefighters. The difference between state agents and private individuals is just that state agents act on behalf of the corporate body of the government, or perhaps the citizenry as a whole. But this difference shouldn’t, on its face, change the moral significance of their doing versus allowing harm. So, there is a good prima facie case for thinking that the doing/allowing distinction applies to state actions. The extra weight assigned to the duty that the state not inflict harm plays an important role in common moral thinking about a range of legal issues including the following Notice that I am not considering the much criticized position, sometimes associated with Supreme Court decisions such as DeShaney v. Winnebago County 489 U.S. 189 (1989), that the state has no duties to aid, only the view that its duties not to harm have greater weight. : A. Criminal process: The criminal process requirements we put in place can reduce the likelihood that an innocent person is convicted and punished but may also reduce the likelihood that the guilty are punished. A lower likelihood of being punished for a crime will, in turn, often reduce the deterrence of would-be criminals and thus increase the amount of harm people suffer through crime. So, there is probably a trade-off between protecting the innocent from state punishment and protecting citizens from crime. If we think the doing/allowing distinction matters here, then we will create policies that prioritize preventing the punishment of the innocent over preventing crime because the former harms are inflicted by the state whereas the latter are merely allowed by the state. The often cited ‘Blackstone ratio,’ which emphasizes ensuring that innocents are not punished over ensuring that the guilty are punished, perhaps reflects reasoning of this kind (though this is not the only way to justify the Blackstone ratio). The ratio was first suggested by Blackstone. See Williams Blackstone, Commentaries on the Law of England, facsimile edition with introduction by Stanley N. Katz (Chicago: University of Chicago Press,1979). B. Torture: The permissibility of state sponsored torture, in the form of ‘coercive interrogation,’ has been extremely controversial. Under some circumstances, such torture may be a means of gaining important information that helps save lives. Of course, the efficacy of torture is itself a matter of considerable controversy. If there is no difference between harming someone and failing to save them, then (all else being equal) just a few lives saved may be sufficient to outweigh the harm done by torture and thus justify it. But if not harming is more morally important than aiding, then the benefits of torturing will have to be more substantial for it to be permissible. An important question about torture is whether the wrong done to the victim is just proportional to the effects on his welfare, in which case torturing may be easier to justify than killing, or is additionally wrong because torture involves the distinctive harm of a violation of his autonomy. I set these complications aside for now. C. Capital Punishment: Some people think that the death penalty is a morally special kind of punishment, which may not be inflicted even on guilty people, perhaps because it is ‘cruel and unusual.’ But the death penalty for murder may have the effect of deterring a substantial number of murders. This means that state killings may result in fewer killings overall. So to reject capital punishment it is likely necessary (thought not sufficient) to defend the doing/allowing distinction. If there is no moral difference between the state killing someone and failing to prevent her murder then the argument against capital punishment fails. To be clear, a full defense of any of these policy choices would require more than just reliance on the doing/allowing distinction. The doing/allowing distinction is also often invoked in arguments about property and taxation. Nozick, for instance, claimed that redistributive taxation is presumptively impermissible because it involves harming one person to benefit another. See Robert Nozick, Anarchy, State and Utopia (New York: Basic Books, 1974). But this raises a special issue about the nature of property rights. Rights against bodily harm are, very plausibly, natural rights, which exist independently of state decisions. But many people argue that property rights are created by state decisions. If that’s so, then they don’t put constraints on taxation or other state interventions in the economy. Resolving that issue is beyond the scope of the current paper: for discussion see Liam Murphy and Thomas Nagel, The Myth of Ownership: Taxes and Justice (Oxford: OUP, 2005). For instance, even if it is worse to torture one innocent person than to fail to save others from similar suffering, one would have to consider whether it is similarly bad to torture a guilty person. Lee, “Deontology, Political Morality, and the State”, makes a similar point about the importance of innocence and guilt in justifying criminal procedure. See Youngjae Lee, “Deontology, Political Morality, and the State”, Ohio State Journal of Criminal Law 8 (2001): pp. 385-402. But the doing/allowing distinction seems to be at least an important component in our thinking about all these issues. It matters substantially for thinking about some important issues, then, whether we think there is a serious moral difference between state harming and state failures to aid. Adrian Vermeule, for instance, writes that seeing the insignificance of this distinction, and some related ones, is essential to recognizing the ‘War on Terror’ as “what Dewey called a ‘great movement for human liberation."' Adrian Vermeule, "A New Deal for Civil Liberties: An Essay in Honor of Cass R. Sunstein", Tulsa Law Review 43 (2008): pp. 921-931. Also, acceptance or rejection of the doing/allowing distinction marks a major fault line between different theoretical approaches to the law. Utilitarianism, and its operationalized descendent ‘cost-benefit analysis,’ gives equal weight at the fundamental level to harms done by the state and harms allowed by the state. I say equal weight “at the fundamental level” because utilitarians might well incorporate a doing/allowing distinction into their account of the heuristics we should use for everyday decision making even if they think that the fundamental moral principles determining which actions are right make no distinction between doing and allowing harm. So too do the other main strands of consequentialism, such as prioritarianism, egalitarianism, and sufficiency views. It may be that a sufficiently gerrymandered consequentialism would give different weight to harms I do and harms I allow. But I think such a position would be very unattractive. The consequentialist evaluates actions by considering how good the world would be if we performed them. It is very hard to see why a world with harms done by me in it would be worse than a world with harms that I merely allowed in it. By contrast, deontological theories give a central place to the doing/allowing distinction. Should we give different weights to bad things the state does to people and bad things it allows to befall them? Those who think we shouldn’t can offer two kinds of argument. Zamir and Medina distinguish these different kinds of argument for rejecting the doing/allowing distinction and also the intending/foreseeing distinction (I do not discuss that distinction in this paper). See Eyal Zami and Barak Medina, Law, Economics, and Morality (New York: OUP, 2010). Enoch also distinguishes these different kinds of argument with respect to the intending/foreseeing distinction. See David Enoch “Intending, Foreseeing and the State”, Legal Theory 13 (2) (2007): pp. 69-99. First, they can argue that the distinction between doing and allowing should never play any fundamental role in our moral thinking, whether we are considering private actors or the state. Many consequentialist philosophers have argued for this, claiming, for instance, that the distinction is incoherent or that it has no moral significance. For a useful survey of these arguments see the essays collected in Alastair Norcross and Bonnie Steinbock (ed.), Killing and Letting Die (New York: Fordham University Press, 1999). Though important, these arguments are controversial and, as should be clear from the forgoing, highly counter-intuitive. Second, they can concede that it is worse for a private person to do harm rather than allow it but argue that it is all the same just as bad for the state to do harm as to allow it. According to this line of thought, present in a long tradition in legal theory and currently quite influential, there are some special features of the state which put state inflicted harms and state failures to aid on a par. My focus in the remainder of this paper is on arguments of the second kind, and my response to them will be critical. I shall focus on the reasons legal theorists and philosophers have given for thinking that the doing/allowing distinction should not be employed specifically when we are thinking about the state. I’ll refer to their thesis as ‘the political insignificance of the doing/allowing distinction.’ In what follows I survey a range of influential arguments for the political insignificance of the doing/allowing distinction and find them wanting. 3. The Argument from the Costs of Rights One central argument for the political insignificance of the doing/allowing distinction focuses on the fact that protecting citizens from harms inflicted by the state requires state action and has costs. According to Sunstein, Holmes, and Vermeule, this fact shows that the doing/allowing distinction is either unintelligible or morally insignificant in the context of state action. Stephen Holmes and Cass Sunstein, The Cost of Rights: Why Liberty Depends on Taxes (New York: W. W. Norton & Company, 1999). The argument is repeated in Cass Sunstein and Adrian Vermeule, "Is Capital Punishment Morally Required? Acts, Omissions, and Life-Life Tradeoffs", Stanford Law Review 58: pp. 703-750; section II.B. Once we recognize that even to avoid harming someone the state must perform various actions, they argue, we must abandon the distinction. Now, although they draw conclusions about the doing/allowing distinction, their argument is mostly framed as an attack on the (at first sight equivalent) ‘act/omission’ distinction. So, I’ll begin by explaining what that distinction is. I’ll then explain how their argument does indeed challenge anyone who relies on that distinction as applied to the state. Finally, I’ll show that the act/omission distinction turns out to be importantly different from the doing/allowing distinction and that their argument in fact only raises a problem for the former distinction. I have been speaking about the doing/allowing distinction, but it is much more common for legal theorists, judges, and so on to characterize the relevant distinction as between ‘acts and omissions’: the difference between performing an action and not performing one. For example, if I kick someone then I have performed an action, whereas if I simply stand still as they go by then I have done nothing, or omitted to act. Similarly, the police beating someone up is a state action, whereas the police failing to respond to a distress call is a state omission. Defenders of the distinction’s moral significance say that it is worse when harm comes about due to an action rather than an omission. On the face of it the doing/allowing and act/omission distinctions are very close, and probably equivalent. They both suggest, for instance, that it is worse for the police to beat someone up than to fail to respond to a call. And the act/omission distinction is regularly used to decide the same policy issues that I discussed above. For instance, it is thought that incarcerating an innocent is worse than failing to prosecute a criminal, or prevent a crime, because the former is an action whereas the latter is an omission. Holmes and Sunstein raise a serious problem for relying on the act/omission distinction in this way. Suppose that we want to protect innocent citizens from false convictions by the state. We will need to put in place state agencies that create rules governing the arrest and trial of those under suspicion. We will need state monitoring institutions that check whether those rules are being followed. And we will need enforcement mechanisms, also put in place by the state, that are able to compel compliance when it is determined that the rules are being broken. I draw here on Holmes and Sunstein, The Cost of Rights, chap. 4. See that chapter for a more detailed discussion of this example. All of these are costly institutions that need to be run and funded by the state, so protecting citizens from false imprisonment by the state requires a substantial amount of activity on the part of the state. Holmes and Sunstein show that this is a general truth about protecting individuals from state inflicted harms: state activity is nearly always needed to protect individuals from harms inflicted by the state. For instance, they show that government involvement is needed to monitor prison guards to prevent degrading treatment of prisoners, to define what counts as an ‘unreasonable’ seizure by the police, to coerce compliance with regulations on torture by state officials, and so on. According to them, these facts about what the state must do to ensure that it does not harm anyone show that there is no difference (or no moral difference) between, for instance, the duty the state has not to falsely incarcerate anyone and the duty the state has to protect citizens from harms inflicted by other citizens. Ibid., p. 48. Sunstein and Vermeule’s “Is Capital Punishment Morally Required?” also refers to the cost of rights argument as part of an attack on the doing/allowing distinction, as does Vermeule’s “A New Deal for Civil Liberties”. Why? Well, they say, following the standard practice in legal theory, the difference between the two sorts of duty is supposed to be that the one duty requires mere forbearance on the part of the state, whereas the other duty requires state action. But we have seen that even fulfilling the duty not to falsely imprison requires a good deal of state activity. So, that duty cannot be distinguished from, say, the duty to protect people from crime. Holmes and Sunstein, The Cost of Rights, p. 48. This argument creates a serious problem for those who defend the relevant policies by relying on the act/omission distinction. However, I think it does not affect the use of the doing/allowing distinction. In the cases we are concerned with the two distinctions turn out not to be equivalent, and Holmes and Sunstein’s observations do not prevent us from distinguishing doings from allowings. According to the thesis we are concerned with, greater weight should be given to ensuring that the state does not do certain bad things to people than to ensuring that the state prevent similar bad things from happening to them. Applied to the example we’ve been considering, the suggestion is that it might be worse for the state to harm someone through false imprisonment than it might be for the state to fail to prevent harms to people that happen through crime (which it could do, let’s suppose, by relaxing the standards for criminal procedure and thus deterring more crime). Holmes and Sunstein show that government spending, and thus government action, are needed to prevent false imprisonment just as they are needed to prevent crime. However, even once we have observed this it still seems clear that when the government falsely incarcerates someone it does harm to her but it does not do harm to her when it fails to protect her from crime. It just turns out that a fair amount of spending and action is required to prevent both sorts of harm from occurring. The doing/allowing distinction suggests that we should focus more of that spending and action more on preventing one sort of harm than the other. An analogy with an example from the private sphere may make my response clearer. Sometimes I can avoid doing harm to others by simply doing nothing. If people are walking by me and I stand still, I will not cause them to trip up or harm them in any other way. But sometimes I must do things, perform actions, to stop myself doing harm to others. For instance, suppose I accidentally drive up on the sidewalk and am heading towards a pedestrian, about to kill him. To prevent my killing him I must turn the car away. In this case, I have to do something, perform an action, to stop myself harming him. But this doesn’t change the moral status of my killing him. It would still be wrong for me to kill him, or wrong to kill him just because doing so would allow me to save a couple of others. (Maybe to avoid killing him I must drive into a wall even though this will destroy my car, preventing me from saving two people I was on the way to help.) What’s wrong with killing someone is that to do so is to treat them very badly. And what makes killing someone worse than failing to save them is that killing them is a worse way of treating them. The fact that you might have to perform an action to avoid killing them doesn’t mean that killing them would treat them any less badly. The same goes for state actions: maybe the state has to do a lot to make sure it doesn’t torture anyone, but this doesn’t in any way change the fact that the state treats people very badly when it tortures them. It is thus very important that in future work legal theorists separate clearly the doing/allowing distinction and the action/inaction distinction. Holmes and Sunstein’s argument also refutes anyone who thinks that we should focus on preventing state inflicted harms over supporting state aid simply because the latter requires government spending while the former does not. They suggest that this is the rationale that many people have for endorsing the doing/allowing distinction: “The distinction gains its initial plausibility, perhaps, because it seems to track the politically more familiar contrast between small government and big government”. The Cost of Rights, p. 39. Some writers do seem to have endorsed the doing/allowing distinction on these grounds. For instance, Cranston prioritizes protecting negative rights over positive rights partly on the grounds that the former are, according to him, cheaper to protect. See Maurice Cranston, “Human Rights, Real and Supposed”, in D. D. Raphael (ed.), Political Theory and the Rights of Man (Bloomington and London: Indiana University Press, 1967). But, again, I have suggested that there is a very different rationale underlying the doing/allowing distinction. According to that rationale, the difference between doing and allowing is that, all else being equal, someone is treated worse when we harm her than when we allow her to be harmed. If it takes state spending to prevent the former as much as the latter, then so be it. 4. The Argument from Causal Involvement Often when other people are harmed we, quiet private persons, have played no role in the creation of the harm. Frequently, for instance, murders occur in Chicago which I have played no part in bringing about. This follows from a more general truth. Even if we can do something to prevent certain harms from occurring, we can often correctly say that we haven’t caused them. Sunstein and Vermeule suggest that, by contrast, it is not typically open to the state to say that it has played no substantial role in the creation of the harms that are inflicted in our society, at least if there is something the state could have done to prevent those harms from occurring. Suppose that imposing the death penalty would have the effect of deterring 18 murders for each person killed on death row. In that case, say Sunstein and Vermeule, a state that decides not to impose the death penalty is substantially causally involved in the deaths of each person that would otherwise not have been killed – just as causally involved, they say, as the state would be in the death of someone it tries and executes. This is obviously a strong claim. Can we really say that if the state didn’t adopt policies that would have prevented someone’s death, then it killed him just as it kills someone when an agent of the state gives him a lethal injection? Sunstein and Vermeule argue for this conclusion as follows. They ask us to suppose that two possible policy packages are available for reducing the murder rate. Sunstein and Vermeule, “Is Capital Punishment Morally Required?”, pp. 722-723. The first, Package A, includes policies such as the prohibition of murder, lengthy incarceration of those convicted of murder, and so on. The second, Package B, contains all of the policies included in A but also use of the death penalty. Package B, let’s assume, would result in fewer overall deaths than Package A, and policy makers know this. Whichever option policy makers choose, say Sunstein and Vermeule, the state will be performing various actions. To choose Package A over B, say Sunstein and Vermeule, would just be to knowingly pick the set of actions that will inevitably lead to more deaths. Thus, if policy makers choose A over B, we should see them as doing harm to the people whose lives would’ve been spared if B were chosen. This argument, it seems to me, faces some similar difficulties as the argument from the cost of rights. That argument showed that even for the state to avoid harming people it must do a number of things. The present argument similarly shows that a state that neglects to opt for the death penalty is doing many things, such as carrying out incarceration policies, and that if it did certain other things, fewer deaths might result. The problem with this argument is that the objection to imposing the death penalty is not that imposing it involves a greater amount of state activity but that it involves the state doing something bad to certain people. Both policy Package A and policy Package B involve government activity, but only the one package seems to involve the state doing harm to people. But what about the fact that choosing A over B inevitably results in more deaths? Doesn’t the state harm people who die for this reason? Under the relevant circumstances, the strongest thing we can say about the state’s role in those deaths is the following: if the state had acted otherwise, fewer people would have died. But the truth of this counterfactual is not enough by itself to establish that the state itself harmed those people. Suppose that in Driver I opt not to run over the one and so the five drown. In this case, we can say that if I had acted otherwise (if I had driven over the one and on to the shore) the five would have lived. Still, it seems, I didn’t kill them. So, more needs to be said if the argument from causal involvement is to be successful. One way to rescue it might be to emphasize that even if the government doesn’t itself kill people, it may still be heavily causally involved in their deaths. Here is how that argument might go. Setting aside the choice between policy set A and B, state actions in general have very large causal ramifications. This means, it might be said, that even when the state doesn’t itself kill someone, it is still heavily causally involved in the killings that occur. And this suggests that the state is responsible for those deaths. Thus, there is no moral difference between the state’s failure to prevent these deaths and the state’s killing people itself. An example might help to illustrate this argument. Consider, say, the state’s decisions about taxes on cigarettes. These decisions will heavily affect a large number of people’s consumption decisions which in turn will affect the behavior of various shop-owners, manufacturers, and so on. Even this relatively minor decision will likely cause some people to change where they shop, to change jobs (and thus cities), or to be laid off. And these effects on people’s lives will also likely affect who gets killed. The decision to, say, raise taxes will affect who decides to kill his lover because she is moving, who simply ends up in the ‘wrong place at the wrong time,’ and so on. Given these significant effects on who gets killed, the argument goes, isn’t the state just as responsible for them as for killings that it carries out itself? And so aren’t they morally equivalent? The first thing to notice about this argument is that it suggests similar conclusions about many institutions in our society other than the state. Consider, for instance, major churches, banks, and sports clubs. Decisions these institutions make will similarly have a large causal impact on how people lead their lives and thus who gets killed. There are likely many cases where an individual would not have been killed but for one of these institutions choosing a particular path. Even (non-murderous) private individuals often do things which (we would say) do not themselves harm anyone but still have effects on who gets killed. For instance, suppose that I trip someone by mistake at the airport. As a result, she misses her planned flight and, while waiting in the lobby for a different plane, is murdered. Plainly I didn’t kill this person, but without my tripping her she would not have died. If the causal argument under consideration is successful, then private institutions and individuals are responsible for the deaths that come about in these ways. One potential difference between the state and private actors which may have occurred to readers is that states have special responsibilities to protect us. I discuss this in the next section. But of course we don’t count the actions of these institutions or individuals as tantamount to (or even lesser versions of) murder, either morally or as a matter of law. There are several plausible, and complementary, explanations for why their actions don’t make them responsible for these deaths. For a helpful taxonomy of these distinctions, see Anthony Honoré, "Causation in the Law", in E. N. Zalta (ed.), The Stanford Encyclopedia of Philosophy (Winter 2010), URL = <http://plato.stanford.edu/archives/win2010/entries/causation-law/>. One is that any involvement in the deaths is relatively remote, in the sense that there are several other events intervening between the action and the death. Another explanation, most pertinent for our purposes, is that between the action and the death there is a particular kind of intervening event: the intentional actions of another agent. Some theorists say that when another individual’s choices intervene between my actions and the harm I in fact did not cause the harm, whereas others say that that the intervention merely removes my responsibility for the harm. In the text I defend the latter claim, but of course if the former, stronger, claim is true then the causal argument is still unsound. Compare H.L.A. Hart and Anthony Honoré Causation in the law (2nd ed.) (Oxford: Clarendon Press, 1985); Sanford Kadish, “Causation and Complicity: A Study in the Interpretation of Doctrine”, California Law Review 73 (1985): pp. 323–410; and John Gardner, “Moore on Complicity and Causality”, University of Pennsylvania Law Review PENNumbra 156 (2008): pp. 432–443. The deaths we are considering come about because some agent chooses to kill someone. For instance, a third party murders the person who misses her flight because of my stray leg. Because of this intervening event we attribute the murder to him, not to me. Why? Under ordinary circumstances, we consider people to be ‘agents,’ with the capacity to choose how to act. Because of this agency we hold them responsible for the outcomes of their actions. This is so even though other causal factors may have led to their actions. For instance, perhaps I was able to shoot you only because you decided to walk to work today and because someone else, an innocent, held up the train I would otherwise have taken. And maybe I was angry enough to shoot you only because my favorite sports star made an egregious fumble. Despite these other people’s contributions, moral and legal responsibility for your death would be attributed to me only. This is because I had the capacity to choose or not choose to carry out the killing and so, given that I did choose to do it, responsibility for the results falls on me and not others. It might be said that the state policies often more directly and foreseeably result in various harms occurring than in the cases I’ve been considering (even if the state isn’t itself doing harm to anyone). For instance, if the state decides not to introduce, or enforce, a requirement of using seatbelts then it is foreseeable that many more deaths will ensue. But even here it seems clear that the state is less responsible for those deaths than if it killed those people itself. Individuals have some responsibility for making their own decisions about whether to wear the belt, and this lessens state responsibility for the deaths of the seatbeltless. Of course, sometimes other people and institutions are accomplices to my wrongful behavior. But, according to standard moral and legal thought, simply having some causal impact on my actions is not enough to make you an accomplice. Furthermore, while accomplices are generally liable to the same extent as principles they seem to receive shorter sentences. This suggests that even genuine accomplices may not be doing wrong to the same degree as principles. For evidence of the shorter sentences and a defense of them, see Joseph Trigilio and Tracy Casadio, “Executing Those Who Do not Kill: A Categorical Approach to Proportional Sentencing”, American Criminal Law Review 48 (3): pp. 1371-1422. Only particular kinds of intervention make you share responsibility, such as encouraging me to perform the wrongful act or helping by deliberately providing me with the means to carry out the act. See Kadish, “Causation and Complicity”, for a discussion of exactly what it takes to be an accomplice. Thus, even though private agents and institutions can have large causal effects on who gets killed, we hold them responsible for those killings only when they are involved in certain specific ways, such as encouraging the killer. Plausibly, we should treat the state the same way and Sunstein and Vermeule give us no reason not to. Sometimes the state may, say, intervene to help one person kill another as, for instance, when the Sudanese government provided arms to the Janjaweed so that they could carry out raids in Darfur. In these cases we do, and should, hold the state responsible for its complicity in the killing. But mere causal involvement of some kind or another is not enough to establish complicity and so, in the normal case, killings should not be attributed to the state. Of course, it’s controversial exactly which causal interventions make private parties responsible for harms, either as principals or as accomplices, and this isn’t the place to argue fully for any particular theory. My central point is just that whichever familiar principles we use to limit the responsibility of private actors for harms that occur in our society should also be applied to the state. Of course, some consequentialist minded authors will argue that even private individuals and institutions have a much wider responsibility for harms than I grant here and it would follow from their view that states do also. But their position is much more radical than the one Holmes and Sunstein are trying to defend. Essentially it requires abandoning the doing/allowing distinction not just in the political context but much more generally. Sunstein and Vermeule give us no reason to accept their conclusion that the morality of state actions is distinctive. Thus, the revised version of the argument from causal involvement does not seem to be successful. There is a final more subtle argument from causal involvement that I think warrants attention here. I’ve argued that the causal impact that states ordinarily have on who gets killed is not sufficient to make them complicit in those killings. But there is one specific way in which states are sometimes involved which might suggest a greater degree of complicity. The argument I have in mind begins by pointing to one particular kind of regulation states create that makes them causally involved in the harms people suffer: their restrictions on ‘self-help.’ Sunstein and Vermeule don’t offer this particular argument, though it is gestured at in Holmes and Sunstein, “The Cost of Rights”, chapt. 5, and I think it is worth considering. Thanks to Lee Fennel for encouraging me to discuss it. States typically place serious limits on ‘self-help,’ or what individuals may do to enforce their own rights and the rights of others. For instance, states prohibit their citizens from using force to extract compensation that is owed to them, put strict limits on the defensive force that they are allowed to use against others, and prevent individual citizens from punishing rights violators. Suppose, for instance, that a divorced mother has lost custody over her child to his father, but has good reason to believe that the father is abusing him. This example is modeled roughly on the situation in DeShaney v. Winnebago County. The state does not permit her to take matters into her own hands and try to rescue the child herself. Rather, she must seek protection for the child by contacting the relevant authorities. But what if the authorities do a bad job of protecting the child (say, they fail to intervene quickly enough to prevent serious brain damage)? And perhaps the mother would have been able to do a better job of protecting the child herself if she wasn’t prevented from doing so by the state. We might say in this case that the state is complicit in the harm to the child and thus just as responsible for those harms as would be the case if a state agent had directly inflicted them. Why should we think this? Consider restrictions on self-help performed by private parties. Someone is coming after me with a baseball bat, trying to break my legs. I’m ready for them though: I can defend myself by parrying their blows with my cricket bat. But now suppose that you tie my hands so that I am unable to parry any blows and the attacker breaks my legs. Your act of tying my arms was not itself an act of breaking my legs. Yet we might all the same say that you are complicit in the breaking of my legs. Let us apply these ideas to the state. The state, like someone who ties my hands, restricts our capacity for self-help by regulating what we may do to protect ourselves from rights violations. These restrictions on self-help (sometimes) result in people being harmed. The example of hands tying suggests that the state is complicit in these harms. And this, in turn, suggests that it is just as bad for the state to fail to prevent these harms as it is for the state to harm someone directly. As should be clear, this argument does not support the conclusion that it is always just as wrong for the state to allow someone to be harmed as for the state to do harm to someone. It establishes only that doing and allowing harm is morally equivalent where the harms being allowed are the result of the state’s restrictions on self-help, and surely not all, or even many, harms people suffer are like this. For instance, some people are so disabled as to be incapable of self-help and have no caretakers willing to defend them. Also, even people who have some capacity to defend themselves are often harmed in ways they could not have defended themselves from. For instance, they are harmed by surprise attacks, by stronger and better armed aggressors, and so on. When these people are harmed it is implausible to think that state restrictions on self-help have been a significant cause of the harm. Even so, some might think the argument establishes the equivalence of state doings and allowings for a range of cases. I’ll now present two problems for this argument. First, the argument assumes, with some initial plausibility, that if I am complicit in a harm that is going to befall you, then my duty to prevent this harm has just the same weight as my duty not to harm others. But this turns out to be false. For instance, suppose that the reason the five people in Driver are drowning is that I provided a villain with a harpoon to puncture their raft. In that case I am complicit in the harm they are threatened with and have an especially strong obligation to save them. All the same, I am still not permitted to run over the one if that is the only way to save the five. My duty to prevent the five being harmed does not outweigh my duty not to harm the one even though I am morally responsible for the five being in peril. Second, the argument assumes that state restrictions on self-help are relevantly similar to private restrictions of our liberty. Both states and private agents, according to the argument, are culpable for harms that result from their restricting our liberty. This is not correct. As I have said, the state restricts our liberty by placing sanctions on the use of self-help in various contexts. These restrictions on our liberty differ importantly from the restrictions on our liberty that private parties impose on us when, say, they kidnap us. Private parties (normally) wrong us when they restrict our liberty, but the state does not, we think, wrong us by restricting our capacities for self-help. We think that individuals have no complaint against the state when it puts those restrictions on them and they ought to comply with those restrictions. Why is this? When states restrict our capacity for self-help they inevitably expose us, for the reasons I have explained, to some risk that we will get harmed in certain circumstances (in part) because of our inability to engage in self-help. How can this be justified? The familiar, and plausible, explanation is that states may fairly impose these restrictions on everyone because doing so is needed to enhance the security of each. By making itself the dominant enforcer of rights the state can ensure more consistent and impartial enforcement of rights, and this is to the benefit of each citizen. The state may fairly impose certain risks of harm on each of us in order to create an essential shared benefit. Since these risks are fairly imposed they do not wrong anyone. If the state does not wrong us by exposing us to certain risks of harm then, very plausibly, it does not wrong us if we actually suffer those harms. Thus, if the state does not wrong us when it restricts our capacity for self-help, then it does not wrong us in cases where those restrictions result in our actually getting harmed. This means that the state is not complicit in harms that come about through its restrictions on our self-help. If it were complicit in them, then it would wrong people who suffer those harms, but it does not wrong those people. Now, this argument invites the following response. It might be said at this point that, yes, states may permissibly restrict our capacities for self-help but only if certain conditions are in place. In particular, it might be said, they must accept a responsibility to protect us from harm. This suggestion is, of course, familiar from the social contract tradition. Hobbes, for instance, wrote that the members of a state trade ‘obedience’ for ‘protection,’ accepting restrictions on their liberty for the sake of greater protection from harm. Thomas Hobbes, Leviathan (New York: OUP, 1998). Given this responsibility to protect us, it might be said, it is just as bad for the state to allow us to be harmed as to do harm to us. Or, similarly, it might be said that even when the state puts permissible restrictions on our liberty it still has a heightened responsibility to protect us from harms which might results from those restrictions. For instance, suppose that the state locks up a justly convicted criminal. It seems permissible for the state to punish that person by putting them in jail (thus restricting their liberty). But having taken away that person’s liberty we think the state is required to protect them in various ways, such as making sure they are not subject to violence as an inmate, are fed and clothed properly, and so on. In the next section I will focus on the state’s responsibilities to protect and how these might affect its duties to harm and not harm. 5. The Argument from State Responsibility We’ve seen that protecting people from harms inflicted by the state has costs. We’ve also seen that what gets spent on preventing state inflicted harms could instead be spent on aiding people. So, when we take steps to prevent state inflicted harms some people lose out. Don’t these people who lose out have a complaint? Don’t their losses matter, morally speaking? Yes, I have suggested, but all the same we should give greater weight to ensuring that the state doesn’t inflict harm itself. Sunstein and Vermeule suggest that these people whom the state fails to aid should have just as strong a complaint as those that are harmed by the state because the state has a responsibility to aid them. Sunstein and Vermeule, “Is Capital Punishment Morally Required?”. They equivocate somewhat between the thesis that the state has some duties to aid, which pretty much everyone agrees with, and the much more controversial thesis that state duties to aid have the same weight as state duties not to harm. The latter thesis is suggested, for instance, by their claim that we should consider the tradeoff between state inflicted and state allowed deaths to be a pure ‘life-life tradeoff’, simply counting the overall number of lives that will be saved. It is that thesis which would require us to change our attitudes towards criminal process and so on, and I focus on it here. For instance, the state has a responsibility to protect our lives and so we have just as great a complaint if the state fails to take steps to protect us as we do if the state kills us itself. Thus, when making policy decisions, lives lost through a failure to aid should be given equal weight with lives lost through state killing. This argument was also made in earlier work by Goodin, who claimed that the state should weigh harms and benefits equally because “It is in the nature of public officials’ role responsibilities that they are obliged to ‘dirty their hands’ – make hard choices, that are wrong (or would ordinarily be wrong, or would be wrong for ordinary private individuals) in the service of some greater public good.” Robert Goodin, Utilitarianism as a Public Philosophy (New York: Cambridge University Press, 1995), p. 10. It is certainly plausible to think that the state ought to protect people’s lives (at least the lives of its citizens at any rate). Pretty much everyone will agree that the state ought to take steps to prevent our dying at the hands of murderers, and many will agree that the state ought to take steps to prevent deaths through famine, poor sanitation, and so on as well. So, the state has some duties to aid, let’s say. How does this get us to the conclusion that it is just as bad for the state to harm as to allow harm? At first sight, it simply doesn’t. Private individuals, as we have seen, have duties to help threatened innocents and so on but, all the same, those duties have less weight than their duties not to harm. The same could be said about the state. States have humanitarian duties to aid, but these can be outweighed by their duties not to harm. More needs to be said, then, for Sunstein and Vermeule’s argument to pose a problem for the doing/allowing distinction as applied to the state. What is needed is an argument that there is something special about the state’s duties to aid which might put them morally on a par with its duties not to harm, something that distinguishes the state’s duties to aid from those of typical private individuals. Here is how I think that argument could most plausibly be made. Private people, it is usually assumed, have some duties of ‘beneficence,’ or ‘charity,’ which require them to contribute aid to others. We have this duty to those in need solely in virtue of their humanity and irrespective of any special relationships we may be in with them. So, for instance, in the Driver case we began with, it is usually assumed that the five people on the track are strangers to me, and thus that my duties to them are duties of beneficence. However, it is possible to acquire other duties to aid others when one is in certain special relationships with them. The Supreme Court also sometimes adopts the language of ‘special’ relationships and responsibilities. See, for instance, DeShaney v. Winnebago County. Philosophers call duties acquired through special relationships ‘special responsibilities.’ So, for instance, if I make a promise to you to help you move tomorrow, I incur a special responsibility to you to help you then. Special responsibilities have a distinctive character from duties of beneficence. For instance, it is often thought that one can excuse oneself from giving charity today on the grounds that one has given a good deal in the past, whereas one cannot excuse oneself from keeping promises today on the grounds that one has kept a number of promises in the past. In this respect, and others, special responsibilities function more like our duties not to harm than our duties of beneficence. Most importantly for present purposes, it seems especially wrong not to aid someone where one has a special responsibility to provide them with that aid. For instance, parents have special responsibilities to aid their children, and it seems to be a great moral failing when a parent fails to protect their child adequately or neglects to provide them with basic nutrition. I thus propose that the argument at hand might be bolstered with the claim that the state has not just duties of beneficence but has special responsibilities to provide aid to its citizens. This would be a plausible claim about the state. The state makes various demands of its citizens to obey the law (including laws restricting self-help), serve in the army when needed, pay taxes, and so on, and in return citizens can demand certain benefits from the state, such as police protection and so on. Hobbes, for instance, as I noted earlier, argued that citizens exchange some of their natural liberty for the sake of the security that the sovereign can provide. See note 30 infra. and accompanying text. The special responsibilities the state has to aid its citizens make it much more plausible to think that it is just as morally wrong when the state fails to aid a citizen, say by failing to protect her life, as when it harms her, say, by punishing her for a crime she didn’t commit. Thus, according to this revised argument, the doing/allowing distinction does not bear on state actions because of the special responsibilities of the state. Is this revised version of the argument successful? As Michael Moore has recently pointed out, even where we have special responsibilities to someone it still seems more blameworthy, and thus more wrongful, to harm them than to fail to aid them. Here is his example: “In Oklahoma several decades ago a father poisoned his 8 year old son to death by giving him poisoned candy on Halloween night, claiming the son had obtained the poisoned candy from others on the son’s trick-or-treat ritual. The father did this so as to collect on a recently obtained life insurance policy on the son. Suppose the son’s mother had had nothing to do with his poisoning but, seeing the state the son was in and being in possession of an antidote to the poison, decided not to give the antidote to her son so that she too could collect on her own life insurance policy on the son. Both parents are deeply responsible for their son’s death; indeed, each should be guilty of murder. Yet isn’t the father more blameworthy than the mother?” Moore, “Moore’s Truths About Causation and Responsibility”, p. 452. We can conclude from this that it is worse to harm than to fail to aid someone, even where one has a special responsibility to her. It follows that even though the state has special responsibilities to its citizens it is still worse for the state to harm a citizen than to fail to aid her. I share, and think most will share, Moore’s judgment about his example. But even if we don’t take seriously these judgments about how to weigh duties there is a good in-principle reason to expect that the state’s special responsibilities will not change the amount of harm it is permitted to inflict. Here is why. The state’s duty to protect A may give it an additional reason to help A when she is in need. But if that's so then we should also expect that its duty to protect A gives it an additional reason not to harm A. If I am charged with protecting someone from harm, then surely I violate that duty if do something that damages her. For instance, a bodyguard can violate her duties to her employer both by failing to intervene to prevent her employer being punched but also by punching her employer herself. Similarly, we think that police officers (who are agents of the state) must take special care to ensure that they don’t harm citizens with their own stray bullets just as they must protect us from private violence. Thus, if the state’s duties to protect create additional reasons to help citizens, then they also create additional reasons not to harm them. Thus, the state’s special responsibilities to protect should not change the balance of reasons for and against inflicting harm in order to produce benefits. If it is wrong to harm A just to benefit B in the absence of any special responsibilities, then the addition of the state’s special responsibilities will not make it permissible to harm A just to benefit B. Now, so far we have been considering whether it is worse to harm someone rather than aid someone else where you have special responsibilities to both parties. What happens if you have a special responsibility to one party but not the other? Some of the most controversial legal issues about harming seem to involve cases of this kind. For instance, it is plausible to think that the United States government has special responsibilities to protect its own citizens (or at least its own residents). Does this mean that it is no worse for the United States to harm a foreigner than to fail to protect a U.S. citizen or resident from the same harm? If so, it would be much easier to justify the killing, torturing and so on of foreigners - central issues in the ‘War on Terror.’ The question is whether the existence of a special responsibility to someone can make it permissible to infringe a duty not to harm someone else, which one would not otherwise be permitted to infringe. Zamir & Medina make an objection along similar lines to the one I offer but in much less detail. See Law, Economics, and Morality, chapt. 3. All else being equal, it seems impermissible for me to stab you with my pen. Can I alter this situation by acquiring special responsibilities to someone else? Suppose someone offers me a deal: “I’ll give you tickets to the show,” they say, “if you’ll stab her with your pen.” I accept the deal, take the tickets and thus, it might seem, acquire a special responsibility to them to stab you. Can I now come after you with my bic? It still seems impermissible. Why is this? Two possible explanations come to mind. One is that while I have acquired a duty to them to stab you, this duty is still outweighed by my duty to you not to do you physical harm. A second explanation is that the exchange in fact created no special duty at all to perform that stabbing and so the action remained wrong. The second explanation seems to me the more plausible one. The natural response to the case is not to think that my reasons for stabbing, acquired through the promise, are simply outweighed by the harm I would do to you. The natural response is that I have no extra reasons for stabbing you just because I made the promise. This is compatible, of course, with thinking that I did something wrong by making the promise in the first place and that I should return the tickets, or perhaps find some other way to repay him for them. The explanation I have proposed for why it is impermissible to stab you with my pen in the example supports a more general principle about the nature of special responsibilities. The principle is that you can acquire special responsibilities to do things only within the constraints that are set by people’s rights not to be harmed. If it would be wrong for me to perform a certain act, and wrong because it would infringe someone’s rights not to be harmed, then acquiring special responsibilities cannot make that act right. Let us consider how this principle applies to the duties we are concerned with, namely, duties to aid. Suppose that I take a job as Bob’s bodyguard and thus commit myself to protecting Bob’s life. This commitment produces a duty in me to protect Bob but, on the view I have defended, only within the constraints set by other people’s rights not to be harmed. If, before taking the job, it would have been wrong for me to harm someone else, even for the sake of protecting Bob, then it would still be wrong for me to do this after I have taken on the job. For instance, suppose that (all else being equal) it is impermissible to destroy one innocent stranger’s leg just to save one innocent stranger’s life. This means that even after taking the bodyguard job it would still be impermissible for me to destroy an innocent stranger’s leg just to save Bob’s life. We can similarly say that the special responsibilities the state has to provide aid are limited by the constraints set by people’s rights not to be harmed. Someone might point out that the bodyguard’s responsibilities to protect are different from the state’s because the former are voluntarily acquired whereas the latter are not. I don’t see why that difference would be morally significant here. Furthermore, the duties of most individual state agents are voluntarily acquired. For instance, firefighters gain their responsibilities by signing contracts. States can sometimes infringe someone’s rights against being harmed in order to produce benefits for another. Just like private individuals, they have duties of humanitarianism which can justify inflicting harm where large enough benefits can be secured. But, also like private individuals, their special responsibilities do not alter the amount of benefit needed to justify inflicting harm. Let me try to further defend the principle that my response depends on. Let’s first consider a case where someone succeeds in gaining new duties to benefit someone else. Suppose, for instance, that someone, Zaida, offers me the following deal: “I’ll give you tickets to the show if you give me your ipod.” I agree to the deal and she gives me the tickets. I now have a duty to give her the ipod. How exactly did this come about? How did I arrive at this duty? Well, prior to making the deal I had a set of ownership rights over the ipod, including rights to control, use it, and so on. As the possessor of these rights, I am able to transfer them to other people, and I do so when I make the deal with Zaida. Thus, once the deal is concluded, and given that she has kept her end of the bargain, she has the rights to control the ipod, use it, and so on. And so I have the duty to give her the ipod to use as she pleases. What if the deal had taken the following form? Again Zaida proposes to give me the tickets if I give her an ipod but, unfortunately, I destroyed mine in the washing machine. However, I realize, you still have a fully functioning ipod. “Sound great”, I say to Zaida, pointing to your ipod on the table, “If you give me the tickets I’ll give you that one.” She reaches to give me the tickets. Have I acquired any duty to give her that ipod? Plainly not, it seems to me. I cannot take my offer to her of your ipod as any additional reason to steal it from you and give it to her. But I agreed to give it to her, so why doesn’t she now have a right to it? The obvious explanation is that in this case although I tried to give her rights over the ipod, I could not do so because I did not own it. It is an independently plausible principle that I am not able to transfer rights over this ipod to her because I am not the bearer of those rights. Hence, even after my conversation with Zaida, you still have a right to control the ipod. Similar reasoning seems to apply in the cases that we are concerned with. Suppose now that rather than offering her an ipod for the tickets I offer to stab you with my pen. Just as I have no rights over the ipod, I have no rights over your body. In particular, it is you, and not I, that has a right to the integrity of your body. When I offer to stab you, I am offering to trade your bodily integrity for the tickets. But, like the ipod, this is not something I am entitled to trade. Thus, even if I make an arrangement with someone to stab you, I do not thereby acquire any duty to stab you because that would be to give them something I am not entitled to give them. Finally, suppose that the deal she offers is this: she will give me the tickets if I agree to be her bodyguard for a week. To be someone’s bodyguard is, on the face of it, to do what it takes to protect her life. But does this mean that she now has a right that I do whatever it takes to protect her life? The forgoing suggests otherwise. There are rights I can transfer to her and rights I can’t. For instance, I can give her the right that I wade into a pond to help pull her out or stand in the way of a punch because I have rights of control over my own body which I can transfer to her. But I cannot give her the right that your body be punctured (including by my stabbing you) when it will help save her because I do not have rights over your body which I can transfer to her. All of this fits with our ordinary thinking about a wide range of examples. An anonymous referee points out that our reactions seem to be interestingly different when it comes to children. We often think that it is permissible to constrain the liberty of children in order to provide them with benefits. I think this is because, given their limited capacities for judgment, we don’t put as much value on the freedom of children to make their own choices and thus think that limiting their liberty is not as weighty a harm as it would be to an adult. Also, this is a case of harming someone for her own benefit, which may be much easier to justify than the actions we are concerned with, in which someone is harmed for someone else’s benefit. We ordinarily think that, say, adopting a child does not give you any extra license to harm other children in order to benefit that child. It might be more excusable but it would still be wrongful. These considerations all support the principle I proposed earlier: that one can only acquire special responsibilities to do things within the constraints that are set by people’s rights not to be harmed. Thus, the argument from state responsibilities that we’ve considered is not successful. The special responsibilities of the state do not make it permissible to violate anyone’s rights not to be harmed, except where this would be permissible anyway. What if I am wrong about the nature of special responsibilities? Perhaps someone will say that I really can acquire duties to do things that are otherwise wrong. For instance, perhaps they will say that by agreeing to become someone’s bodyguard, I really do acquire a duty to help save them even in circumstances where to do so I must wrong someone else and the wrong is not (otherwise) outweighed by the benefit to Bob. They might say of the example I gave that my being Bob’s bodyguard gives me some extra reason to do what it takes to protect him even if this involves, say, breaking someone else’s leg. If my duty to Bob is weighty enough then perhaps it really would be as bad for me to fail to protect Bob from a broken leg than it would be for me to break someone else’s leg. Furthermore, the objector might continue, since the state has special responsibilities to protect us it might also be just as bad for the state to fail to protect us as for the state to kill an innocent person that it has no special responsibilities to. Now, as I discussed earlier, I think this way of thinking about special responsibilities is mistaken. But suppose it is true that when you have a special responsibility to protect someone that responsibility gives you reason to do what is needed to protect her even in cases where this would require doing something that harms/wrongs someone else and would otherwise be wrong. It still doesn’t follow, I will argue, that it is just as bad for the state to do harm as to allow it. On the current supposition, the duties to protect that we acquire through special responsibilities must be weighed against the duties we have not to harm others. This of course doesn’t mean that our duties to protect (where acquired through special responsibilities) must have just the same weight as our duties not to do harm. For instance, my promise to do what it takes to protect your life might give me some reason to kill an innocent person if that is the only way to save your life. But that reason might still be outweighed by the reasons I have to not kill innocents. It would, in fact, be a surprising coincidence if the special responsibilities of the state put its duties to aid and not to harm exactly in balance. So, how are we to balance these supposedly competing considerations of our special responsibilities to protect and our duties to not do harm? The best we can do, it seems to me, is to consult people’s considered judgments. Sunstein, etc. take for granted that we may rely on ordinary judgments to establish the importance of the doing/allowing distinction in private (non-state) contexts, and they rely on such judgments themselves to establish claims about, for instance, the special responsibilities of the state. But if we are allowed to use such judgments to establish what duties we have then there should be nothing more problematic about using them to weigh those judgments against one another. Of course, there is disagreement in the philosophical community about the worth of ordinary judgments about morality in building theories, and I cannot enter fully into that debate here. My point is just that if we accept, as the authors I discuss do, the use of those judgments to establish what our duties are, then we ought to also accept the use of ordinary judgment to balance those duties. So let’s try it: taking into account the promise I have made to protect your life and the right I would violate by killing an innocent, we can ask whether it seems wrong, all things considered, for me to kill someone to save your life. And it does, I take it, seem wrong for me to kill the third party. So, we should conclude that the reasons I have not to harm that party are stronger than the reasons I have to do what it takes to protect your life. Despite my special responsibility to protect you it would be worse for me to harm the third party than to allow you to suffer harm. State agents seem to be in a similar situation. For instance, suppose that the police agree to provide someone with witness protection and thus acquire special responsibilities (beyond what they normally owe) to protect her. We would still think it was wrong if the police killed an innocent bystander just to protect this person’s life. So, even if we weigh special responsibilities against duties not to harm (as I argued earlier we shouldn’t), it still generally seems worse for someone to harm than fail to aid. Perhaps it will be said that there is something special about the state’s responsibilities to its own citizens: those responsibilities are so strong that it really is no worse for the state to fail to aid a citizen than to harm a foreigner. I think this suggestion is implausible on any familiar account of the source of the state’s special responsibilities. Firstly, suppose someone says state’s responsibilities to its citizens are especially strong because of how much they are expected to cede to the state, namely, a significant amount of their autonomy. Consider other contexts where people seem to have very strong obligations to others in light of what they have received. For instance, perhaps I am being paid millions of dollars to protect someone’s life, or maybe they previously saved mine and so I owe them a very strong debt of gratitude and must help them. Even in these cases it seems wrong to kill someone in order to benefit the person to whom we have an obligation. Secondly, suppose someone claims that the state has especially strong responsibilities to its citizens on account of the tight bonds of culture, identity, and so on between co-nationals. The state’s responsibilities, on this account, are grounded in what individual citizens owe each other in light of their (let’s assume) shared nationality. However tight these bonds, they are surely no tighter than the bonds that hold within the family (or, at least, within close knit families). And yet we do not think someone can kill a non-family member in order to save a family member. So, it is very implausible to think that ties of nationality would produce special responsibilities strong enough to make killing a foreigner no worse than letting a national die. 6. Conclusion I began by illustrating the basic deontological argument for thinking that it is worse for private individuals to do rather than allow harm. Assuming that this argument is sound, we saw, there is a strong prima facie case for thinking that the doing/allowing distinction also applies to the state. We have since considered some influential challenges to this prima facie case, and each of them has been found wanting. This does not mean that there is no good argument available for distinguishing the state from private agents in this context, but the currently popular ones must be rejected. We can continue to believe that just as it is worse for a private individual to do harm than allow it to be inflicted, so too it is worse for a state to do harm rather than allow it to be inflicted. We should continue to rely on the doing/allowing distinction when thinking about criminal process, torture, the death penalty, and so on. References Blackstone, William, Commentaries on the Law of England, facsimile edition with introduction by Stanley N. Katz (Chicago: University of Chicago Press,1979). Cranston, Maurice, “Human Rights, Real and Supposed”, in D. D. Raphael (ed.), Political Theory and the Rights of Man (Bloomington and London: Indiana University Press, 1967). Enoch, David, “Intending, Foreseeing and the State”, Legal Theory 13 (2) (2007): pp. 69-99. Foot, Philippa, “Killing and Letting Die”, in A. Norcross and B. Steinbock (eds.), Killing and Letting Die (New York: Fordham University Press, 1994), pp. 280-289. Gardner, John, “Moore on Complicity and Causality”, University of Pennsylvania Law Review PENNumbra 156 (2008): pp. 432–443. Goodin, Robert, Utilitarianism as a Public Philosophy (New York: Cambridge University Press, 1995), p. 10. Hart, H.L.A. and Anthony Honoré Causation in the law (2nd ed.) (Oxford: Clarendon Press, 1985). Hobbes, Thomas, Leviathan (New York: OUP, 1998). Holmes, Stephen and Cass Sunstein, The Cost of Rights: Why Liberty Depends on Taxes (New York: W. W. Norton & Company, 1999). Honoré, Anthony, "Causation in the Law", in E. N. Zalta (ed.), The Stanford Encyclopedia of Philosophy (Winter 2010), URL = <http://plato.stanford.edu/archives/win2010/entries/causation-law/>. Kadish, Sanford, “Causation and Complicity: A Study in the Interpretation of Doctrine”, California Law Review 73 (1985): pp. 323–410. Lee, Youngjae, “Deontology, Political Morality, and the State”, Ohio State Journal of Criminal Law 8 (2001): pp. 385-402. Moore, Michael, “Moore’s Truths About Causation and Responsibility: A Reply to Alexander and Ferzan”, Criminal Law and Philosophy 6 (2012): pp. 445–462. Murphy, Liam and Thomas Nagel, The Myth of Ownership: Taxes and Justice (Oxford: OUP, 2005). Norcross, Alastair and Bonnie Steinbock (ed.), Killing and Letting Die (New York: Fordham University Press, 1999). Nozick, Robert, Anarchy, State and Utopia (New York: Basic Books, 1974). Quong, Jonathan, “Killing in Self-Defense”, Ethics 119 (3) (2009): pp. 507-537. Sunstein, Cass and Adrian Vermeule, "Is Capital Punishment Morally Required? Acts, Omissions, and Life-Life Tradeoffs", Stanford Law Review 58: pp. 703-750; section II.B. Thomson, Judith Jarvis, "Killing, letting die, and the trolley problem", The Monist 59 (2) (1985): pp. 204-217. “Turning the Trolley”, Philosophy and Public Affairs 36 (4) (2010): pp. 359-374; p. 360. Trigilio, Joseph and Tracy Casadio, “Executing Those Who Do not Kill: A Categorical Approach to Proportional Sentencing”, American Criminal Law Review 48 (3): pp. 1371-1422. Vermeule, Adrian, "A New Deal for Civil Liberties: An Essay in Honor of Cass R. Sunstein", Tulsa Law Review 43 (2008): pp. 921-931. Zami, Eyal and Barak Medina, Law, Economics, and Morality (New York: OUP, 2010). 41 42