Back to Document View

LexisNexis™ Academic


Copyright (c) 2001 Georgetown Law Journal
Georgetown Law Journal

March 2001

89 Geo. L.J. 605

LENGTH: 30534 words

ARTICLE: Beneficence, Law, and Liberty: The Case of Required Rescue

Liam Murphy*



* Professor of Law and Professor of Philosophy, New York University. I have benefited from the comments of members of workshops at Columbia Law School and the University of California, Berkeley (Boalt Hall), and from discussion with Helen Hershkoff. Paul Chevigny and Mark Geistfeld provided extremely helpful written comments on an early draft. The support of the Filomen D'Agostino and Max E. Greenberg Research Fund of New York University School of Law is gratefully acknowledged.

SUMMARY:
... After all, the specific legal duty that has been at the heart of this debate is a duty to rescue a person in distress when this can be done at little cost or danger, something that most of these same lawyers assume that almost all people would in any case do. ... Moore is thus content to employ the traditional idea of a "presumption of liberty" in this context, which is simply that infringement of negative liberty is always regrettable just because freedom from coercion is always valuable, and so any criminal duty must have enough to be said in favor of it to outweigh the loss of liberty it brings. ... The paradigm agent-neutral duty would be a (positive) duty of beneficence; paradigm agent-relative duties would be the (negative) duties not to violate rights. ... I think it fair to assume that Macaulay operated with a legal moralist view of the criminal law. ... Therefore, though there are, as I have said, good utilitarian reasons to limit any legal duty of beneficence to the kind of minimal duty found in Rhode Island, that does not mean that a utilitarian supporter of such provisions can ignore Macaulay's problems. ... It appears then that if we accept the optimizing principle of beneficence as part of our account of the aim of the criminal law, we will nevertheless restrict criminal duties of beneficence to duties to rescue with minimal demands and minimal sanctions for violation. ...  

TEXT-1:
 [*605]  INTRODUCTION

English-speaking lawyers continue to be resistant to the idea of positive legal obligation, whether civil or criminal, and in this they differ from lawyers in Continental Europe and Latin America. n1 This is in one way surprising, because English is the language not only of the common law but also of utilitarianism, a normative theory that explains all of morality and political justice in terms of positive obligation--the single positive obligation to benefit people as much as possible. n2 But perhaps it is precisely this greater exposure to utilitarian thought that has made English-speaking lawyers more sensitive to the uncomfortable implications of the legal enforcement of positive duties. The main normative ground offered for the rejection of positive legal duties has been that they constitute excessive interference with individual liberty, and perhaps only those who take seriously the idea of truly extreme positive moral or political duties, such as are implied by utilitarianism, would make this rather dramatic objection. After all, the specific legal duty that has been at the heart of this debate is a duty to rescue a person in distress when this can be done at little cost or danger, n3 something that most of these same lawyers assume that almost all people would  [*606]  in any case do. If this were the most intrusive duty that came to mind when thinking about positive duties it seems unlikely that great revulsion at the infringement of individual liberty would naturally arise in one's breast. But if much more extensive positive duties come to mind, such as Lord Macaulay's imagined duty to travel across India to provide medical services, n4 then a concern with liberty is easier to understand. It is noteworthy, indeed, that in explaining his rejection of a criminal duty to render assistance for the Indian Penal Code of 1837, Macaulay paid special attention to the fact that Edward Livingston had included such a duty in his 1833 draft penal code for Louisiana. n5 In including such a provision, Livingston was following Bentham. n6

This negative connection with utilitarianism is partly speculative, and there are of course other aspects of the history and structure of the common law that would figure in an explanation of its traditional antagonism to positive duties. n7 But my focus in this Article is normative; I concentrate on what has been offered by way of normative defense of the traditional view by judges and legal theorists, and in this connection the influence of utilitarian normative thought is not unlikely to have been important. n8

The main argument in defense of the traditional view, as I have said, has been that positive legal duties threaten the common law's traditional deference to individual liberty. n9 I will argue that this avowed concern with individual liberty  [*607]  is disingenuous, or at any rate mistaken. Positive duties as such do not raise a significant concern about liberty in particular. What they do raise, for some of us at least, is the potential for serious material cost--serious diminution of our welfare or well-being. n10 This was Macaulay's main objection to a legal duty to rescue, and none of the many defenses of such duties that have followed Macaulay have adequately addressed it.

This is not to say that the common law's aversion to positive legal obligations is justified, but rather to say that the possibility of very costly legal duties presents the most compelling ground for such aversion. As there are no easy  [*608]  solutions to the problem of cost, it is not surprising that we have not managed to improve on Macaulay's discussion.

Insofar as legal theory is concerned with the normative justification of the imposition of civil or criminal liability, it can neither ignore the underlying problem of cost, nor avoid it in an unprincipled way by drawing arbitrary lines. Moreover, it will not be sufficient for legal theory to offer purely practical explanations of why costly positive legal duties would not make sense. A full defense of positive legal duties must ultimately appeal to a fundamental normative principle that is itself plausible; and plausible in general, not just as a criterion for torts and the criminal law, but for all areas of government policy and indeed personal conduct apart from the law as well. For example, it will not do to defend minimal positive criminal duties by way of utilitarian argument, noting that there are good practical utilitarian reasons for keeping the costs of such duties low, without addressing the problem of the apparently unacceptable levels of sacrifice utilitarianism requires in the realm of personal conduct. n11

Though legal theory must address the problem of the costs of positive duties as it arises generally in moral and political theory, this does not mean that it must embrace a particular solution to that problem. I will argue that all plausible views on the underlying normative problem converge in supporting the legitimacy of the kind of minimal criminal duty to rescue that is typical in Continental Europe, and is becoming quite popular in the United States. This "overlapping consensus" among plausible normative views is, I argue, sufficient to justify legal duties to rescue. n12

It may seem that, despite the attention it has received from legal theorists, the issue of legal duties to rescue is in practical terms rather trivial. Most opponents of legal duties to rescue are quick to insist that the person who fails to perform an easy rescue is a "moral monster," but they also note, as I have said, that such monsters are rare. This low level of need for a legal duty to rescue is sometimes said to strengthen the case against introducing it. n13 But it could also be thought to show that the whole debate is somewhat pointless, because so little turns on it in practice.

 [*609]  It is a mistake to assume that only those legal questions that promise significant immediate practical payoff are important. A good reason for the traditional common law aversion to legal duties to rescue may also be a good reason to resist positive legal duties of all kinds. n14 The implications of such a conclusion are potentially of enormous practical significance, especially when we move from the issue of the positive duties of individuals to the positive (constitutional) duties of government. n15 One writer has even taken the extreme step of claiming that the general absence of personal liability for failure to rescue is a ground for rejecting the constitutional legitimacy of transfer payments and "welfare obligations." n16

So it is the broader significance of antagonism to positive legal duties that largely justifies our interest in the special case of duties to rescue. n17 Here we have a simple example of a highly contested positive legal duty where the case in favor is clear. We can learn a lot about the idea that positive legal duties are in principle undesirable by focusing on this example. More generally, legal duties to rescue provide an excellent case study for thinking about the appropriate  [*610]  methodology for moral and political argument about what the substantive content of the law should be.

In any case, it is simply not true that the issue of the legal duty to rescue is unimportant in practice. It is thought to be so because people "don't need the law" to incline them to provide emergency assistance where it can be done at low risk. n18 But even if it is right to assume that it is a rare person who lacks the minimal level of benevolence needed to move her to perform an easy rescue, this line of thought ignores the fact that strong contrary inclinations will often be present in such situations. For example, in the well-known Kitty Genovese case, it is asserted that a fear of "getting involved" kept people from calling the police. n19 The point is clearest in the internationally most famous instance of alleged failure to rescue--that of the Princess of Wales and the paparazzi. One does not have to believe that if the photographers had, as it was initially alleged (the charges were later dropped), failed to call for the emergency services when the Princess of Wales was dying in front of them, they would have shown themselves to have been totally devoid of any benevolent impulses. n20 These photographers had an enormous incentive not to waste any time calling for assistance: As they knew very well indeed, photographs of the Princess were worth extraordinary amounts of money.

With the possibility of strong self-interested reasons pushing in the opposite direction, it is clear that an assumption of widespread minimal benevolence does not establish the practical irrelevance of a duty to rescue. Now, of course, it is true that not even France's very stringent criminal duty--which, as all the world must now know, provides for a sanction of up to five years in prison and a fine of around $ 66,000 (500,000 FF) n21 --can eradicate failures to rescue from  [*611]  French soil. n22 But so too, of course, are legal duties not to murder ineffective in very many cases. The point is not that no one in a jurisdiction imposing a legal duty to rescue will ever fail to perform an easy rescue, but rather that the case for such a duty does not turn on the prevalence of imaginary moral monsters who are entirely indifferent to human life or who like to see people die just for the fun of it. All too familiar human motivations can rather easily overcome whatever degree of concern for human life we would like to think is distributed among people generally.

I begin my argument by introducing the traditional debate about the duty to rescue and canvassing the structural and doctrinal issues that such duties raise for criminal law in Part I.A. and tort law in Part I.B. I argue that in the criminal context there is no serious structural problem raised by the enactment of a duty to rescue. Matters are less straightforward in tort law, but the structure of tort doctrine may also be sufficiently flexible to accommodate such a duty. The real problems in this area are therefore normative. I discuss a variety of normative objections to the duty in Part II, and argue that the traditional focus on individual liberty is misplaced and misleading. This leaves the problem of setting principled limits to the interference with individual well-being by way of legal institutions. Part III explains the scope of this problem for legal theory; Part IV argues that any plausible view on the matter will justify the kinds of minimally sanctioned criminal duties that have been enacted in a small number of U.S. states. n23 The desirability of civil liability for failures to rescue, it turns out, is less clear.

I. THE DUTY TO RESCUE AND THE COMMON LAW

In common-law countries, liability for failing to assist a person in immediate peril, even though one could do so at little danger or cost, is exceptional both at tort and at criminal law. Indeed, liability for any kind of omission or failure to act is exceptional. As an initial question, we must examine whether this legal preference for commission over omission--misfeasance over nonfeasance--is  [*612]  made necessary by central structural features of legal doctrine. This Part begins by noting the exceptions to the general rule of no liability for omissions in criminal law.

A. CRIMINAL LAW

1. Commission by Omission

In the celebrated case of People v. Beardsley, n24 a man had left for dead a woman, not his wife, whom he had observed taking morphine and camphor tablets before she fell into a coma. The Supreme Court of Michigan overturned his conviction for manslaughter, observing: "The fact that this woman was in his house created no such duty as exists in law and is due from a husband towards his wife . . . . Such an inference would be very repugnant to our moral sense." n25 After quoting these same sentences, Graham Hughes has this to say:

To be temperate about such a decision is difficult. In its savage proclamation that the wages of sin is death, it ignores any impulse of charity and compassion. It proclaims a morality which is smug, ignorant and vindictive. In a civilized society, a man who finds himself with a helplessly ill person who has no other source of aid should be under a duty to summon help, whether the person is his wife, his mistress, a prostitute or a Chief Justice. The Beardsley decision deserves emphatic repudiation by the jurisdiction which was responsible. n26

The State of Michigan has found it possible to resist Professor Hughes's eloquence these past forty years--Beardsley has not been repudiated.

The facts of Beardsley raise the issue of crimes of commission by omission. n27 The contrast here is with genuine crimes of omission, discussed in the next section, in which a statute directly imposes a positive duty to act--such as the duty to stop to render assistance when involved in an automobile accident. n28 In a crime of commission by omission, the charge is not that the accused violated an explicit duty to act in a particular way, but that an omission attracts liability for a crime of commission. The doctrinal link that renders an omission sufficient for liability for a crime of commission is the existence of some independent duty to perform the omitted act. The trial court in Beardsley erred, however, in holding that the defendant's commonsense moral duty to render assistance was  [*613]  sufficient for these purposes; as the Michigan Supreme Court noted, the independent duty in question must be a legal duty. n29

This remains the standard account of the doctrine, n30 but as Professor Hall long ago pointed out, it is not terribly informative. n31 Not just any independent legal duty will be sufficient to attract liability for any given offense. n32 Instead, a series of particular legal duties has been recognized by courts as sufficient to trigger liability for a failure to act. Such duties may be found in statute or contract, or they may be entirely creatures of the courts. Thus, one basis for liability for commission by omission is failure to perform some statutory (not necessarily criminal) duty to act. If a person breaches a statutory duty to render assistance at the scene of an accident and the victim dies, this failure to perform the legal duty to assist may be the basis of a prosecution for manslaughter by way of the doctrine of commission by omission. n33 An example of commission by way of failure to perform a contractual duty is a railroad gateman omitting to lower the gate to warn motorists of an approaching train. n34 The range of duties to act that have been judicially recognized--without the involvement of legislatures or any explicit undertaking by the defendant--is wide, but most of them are based on the existence of some special relationship between the defendant and the victim, such as the relationship of husband and wife, parent and child, and others. n35

The emptiness of the slogan that commission by omission requires an independent legal duty and not just a moral duty should be evident. The duty of husbands to render assistance to wives became an independent legal duty sufficient to ground prosecution for homicide when the courts began to treat it as such sometime in the nineteenth century. As Hughes explains, the slogan

is a relic of the principle worked out in the nineteenth-century manslaughter cases that the common-law duty to care for children could be used as a base on which to found a homicide conviction if death ensued through neglect, but  [*614]  that this duty could not be extended into a more general duty to aid those in distress where the status relationship of paternal support was absent. n36

In its current usage, the slogan simply means that a duty will be deemed merely "moral" when a court finds insufficient grounds for commission by omission liability to attach. This aspect of the jurisprudence of criminal omissions has provided courts with what Hughes calls "a facade behind which to hide the progress of the law." n37 This is unfortunate from Hughes's perspective because, "as often happens with such protective devices, the refusal to acknowledge the reality of the progress has impeded the advance." n38

Though I share Hughes's perspective, I find more troubling the rather neglected fact that the control of judges over the expansion of liability for commission by omission raises serious problems of legality. n39 Even when the independent positive duty is imposed by statute, its sufficiency for liability for commission by omission is typically a decision for the courts. As Professor Fletcher notes, most discussions of liability for omissions are either, in the pro-liability camp, "concerned about condemning the injustice of not punishing immoral omission," or, in the anti-liability camp, concerned about protecting individual liberty; hardly anyone is concerned about "the issue of legality in letting courts roam freely about our moral sentiments." n40

I will not dwell on the issue of legality in this article. I raise it here because it is one reason why criminal liability for omissions would be better based on explicit provisions to that effect, rather than by piggybacking on crimes of commission. Repudiation of Beardsley should take the form of an explicit statutory provision requiring the positive act of rendering assistance, rather than further extension of liability for commission by omission.

2. Omission

Though provisions criminalizing failure to assist are the norm in Continental Europe and Latin America, in the common-law world they are currently found in only a few U.S. states and Australia's Northern Territory. n41 Rhode Island's provision is exemplary:

 [*615]  Any person at the scene of an emergency who knows that another person is exposed to, or has suffered, grave physical harm shall, to the extent that he or she can do so without danger or peril to himself or herself or to others, give reasonable assistance to the exposed person. Any person violating the provisions of this section shall be guilty of a petty misdemeanor and shall be subject to imprisonment for a term not exceeding six (6) months or by a fine of not more than five hundred dollars ($ 500), or both. n42

Similar provisions are found in Minnesota and Vermont. n43 Though the Rhode Island "duty to assist" provision appears innocuous enough, it is this kind of minimal duty to render aid that has been at the center of the debate about positive legal duties. Some of the reasons offered for opposition to such duties have been based in features of the structure of Anglo-American criminal law doctrine. I consider those objections here, leaving the normative objections to Part II.

That there should be structural/doctrinal objections to duties to rescue is prima facie somewhat puzzling, as such duties are hardly unique in providing for liability for omissions. I have mentioned the ubiquitous duty imposed on motorists that they stop when involved in an accident; equally ubiquitous are duties imposed on parents to provide for the minimal needs of their children n44 and on income earners to file tax returns. n45 In one enlightened common-law country, Australia, all resident citizens have the duty to enroll with the electoral office and to vote in federal, state, and local elections. n46 But despite the familiarity of these types of positive criminal duty, it is undeniable that their existence sits somewhat uncomfortably with some general doctrinal principles of criminal law. n47 As criminal liability for omissions came to the common law slowly and rather recently, this appearance of conflict is perhaps not so surprising after all. n48

We have already considered the legality problems raised by crimes of commission by omission. Statutory provisions imposing liability for omissions directly obviously raise no such problems. But these provisions do raise concerns, among scholars at least, with respect to at least three other general requirements of the criminal law: the act requirement, causation, and mens rea. Furthermore,  [*616]  potential structural concerns arise in situations of multiple victims and multiple potential rescuers.

Criminal liability is said to require an (voluntary) act. n49 What then do we say about omissions? It is not helpful to defend the conceptual propriety of regarding omissions as kinds of acts. n50 It is true that the word "omission" seems out of place in the context of some failures to act, such as my current failure to do some act that has not even crossed my mind; there is thus some plausibility to the view that omissions are a distinct subcategory of non-actions. Nevertheless, the concept of omission is clearly not robust enough in our legal and moral practice to carry much normative weight on its own. Even if we were comfortable thinking of omissions as kinds of acts, they would clearly be very special kinds of acts, and the substantive question would immediately arise whether these special kinds of acts were appropriately criminalized. Conceptual intuitions about what counts as an omission will not be sufficiently reliable to help us. We will need to define omissions as non-actions with certain characteristics, and then we will have to evaluate the significance of those characteristics.

The most sensible approach is to interpret the act requirement as aiming not so much at the distinction between doing and not doing, but at the distinction between thought and action. n51 Thus it is the criminalization of thought alone that the act requirement rules out. A crime must consist in either an act or a failure to act. This is the approach of the Model Penal Code. n52 Michael Moore takes a slightly different approach, defending a strict interpretation of the act requirement--what is required is indeed an act, in the usual sense of a willed bodily movement. n53 He then treats omissions as exceptions to the act requirement.  [*617]  This reflects Moore's view that punishment for action is the standard case while punishment for omission is non-standard and requires special justification. Because the difference between Professor Moore's approach and that of the Model Penal Code turns on a particular normative view about the proper purpose of criminal law, n54 I postpone further discussion of Moore's view until the next Part.

The best way to think about the causation requirement mirrors the best way to think about the act requirement. Thus it is pointless to attempt to defend a philosophical account of causation such that we can say that, for example, my omitting to save a drowning person caused his death. n55 Causation is an ancient and fascinating philosophical topic, but it is unlikely to be relevant to substantive criminal law. This is fortunate, as the issue is so puzzling, and disagreement about it so persistent, that we would be in poor shape if the substantive question of liability for omissions required prior conclusions on the correct philosophical theory of causation. n56 Luckily, there is no need for lawyers to settle on one particular philosophical understanding of causation; this is evident enough from the vague and avowedly non-metaphysical notion of proximate causation that is at the heart of the legal causation requirement. At the doctrinal level, indeed, there is no problem with saying that omissions are causes: They are "actual" or but-for causes, and whether they pass the proximate causation test is a question not so much of the theory of causation but of the proper limits of criminal liability n57 --the very substantive question that we are investigating for the case of omissions. If a legal theorist proposes that the limits of criminal liability should be set in part by asking whether the defendant caused the harm according to the correct philosophical--as opposed to legal--theory of causation, then the appropriate response is to examine that philosophical theory of causation and see whether its claimed normative significance is plausible. It should be  [*618]  remembered, however, that if omissions turn out not to be causes, the upshot will not be conservative--the well-established crimes of commission by omission will turn out to be outside the scope of legitimate criminalization. n58

The only genuine doctrinal puzzle raised by criminal liability for omissions lies with the requirement of mens rea. Professor Hughes argues that satisfaction of the mens rea requirement for crimes of omission puts pressure on the traditional maxim that "ignorance of the law is no excuse." The maxim, Hughes writes, "ought to have no application in the field of criminal omissions, for the mind of the offender has no relationship to the prescribed conduct if he has no knowledge of the relevant regulation. The strictest liability that makes any sense is a liability for culpable ignorance." n59 Though this seems to overstate the problem, Hughes has identified an important distinction between crimes of commission and crimes of omission with respect to mens rea.

Professor Hughes criticizes Glanville Williams for offering this test of the intentionality of an omission: "If the defendant had been asked at any time, while the omission was continuing, 'Are you doing so-and-so?' (which the statute makes it his duty to do), would the true answer based on the facts . . . be: 'I am not'? If so, the omission is intentional." n60 Hughes points out that although he will say that he is not climbing Mount Everest if asked while at home eating an orange, it does not follow that he is, at the moment asked, intentionally not so climbing. n61 But the problem here raised seems not to be unique to omissions. n62

I would prefer to cast Hughes's point as follows. To conclude that an action or its omission was intentional we require that the agent was aware that the description made relevant by the criminal provision applied to her action or omission; for a conclusion that the action or its omission was reckless or negligent, we require that the agent knew or ought to have known that it was likely that the relevant description applied to her act or omission. n63 Where the agent performed an action, we can typically apply these tests without inquiring into the state of mind of the agent with respect to the existence of a legal duty. Where the agent omitted to do something, by contrast, we may often be unable to apply these tests without such an inquiry. When I break the window of the house and crawl through the resulting hole, I am aware that the description  [*619]  "breaking and entering" applies to my act even if that phrase isn't literally before my conscious mind. I don't need to know that I have a duty not to break and enter before I know that is what I am doing--as opposed to, say, removing immediate obstacles to my progress in a northerly direction. And if, upon leaving the restaurant, I pick an umbrella from the basket at random, I can be said to consciously disregard a substantial and unjustifiable risk that I am taking someone else's umbrella, and thus said to be reckless with respect to the relevant description, even if I have no knowledge of the prohibition on theft.

With omissions, on the other hand, it is often the case that the only thing that could connect the mind of the agent to the relevant description is her state of mind with regard to the legal duty to perform the omitted act. Hughes gives the example of the pharmacist who does not know about the legal duty to register the sale of poisons. In such a case, Hughes says, "liability should depend upon the culpability of his ignorance." n64 The point here is that whether the pharmacist's failure to register the sale of a poison was reckless or negligent depends entirely on whether the pharmacist knew or ought to have known that there was a good chance that there was a duty to register sales. There is simply no other way that the description, "failing to register the sale of a poison," could become linked in the appropriate way to the agent's state of mind. No doubt the pharmacist also failed to dance a jig in celebration of the sale; for "failing to dance a jig" or "failing to register" to be relevant descriptions, something is required to pick them out from the crowd of possibilities. Thus also, if a convicted felon fails to register as required by a city ordinance, while having no inkling of any such regulation or of the likelihood of its existence, it is implausible to say that the failure was intentional or reckless--though it might be arguable that it was negligent on the ground that felons ought to consider the possibility of special regulations governing their movements. n65

The asymmetry between actions and omissions here is clear, and it is due to the simple fact that at any one time there are so many potentially legally relevant things that I am not doing. But the asymmetry is not as stark as Hughes suggests. Clearly, some omitted actions are sufficiently salient to justify saying that the omission was intentional without any knowledge of the duty to act: If I fail to rescue a child I know to be drowning in front of me, for no reason other than that I cannot be bothered to get out of my chair, it is not at all unnatural to say that I intentionally failed to prevent the drowning. n66 Furthermore, some prohibited actions are not of sufficient salience, under the relevant description,  [*620]  to count as intentional absent knowledge of the prohibition. If a city were to pass an ordinance (without debate or advertisement) prohibiting people strolling in the city park from intentionally coming within fifteen feet of the river (for some reason, perhaps having to do with hatching birds, not apparent to the lay person), and a person were to stroll within the proscribed area, the only way that we could resolve whether the person did so with the required mens rea would be to inquire whether she was aware of the ordinance--and this remains the case even if, had the person been asked whether she was within fifteen feet of the river, the stroller would have replied that, as it happened, she was.

An inquiry into the agent's state of mind with respect to the existence of the legal duty is typically not necessary to establish mens rea when a person has acted in some way, because the descriptions of that act that are likely to be legally relevant will be familiar enough in ordinary life--the issue of knowledge, recklessness, or negligence with respect to the description can be solved without such an inquiry. With certain highly unusual descriptions, however, the inquiry may be necessary. So the problem Hughes discusses, though much more significant in cases of omission, is not unique to them. The moral to draw from his insight is that mistakes of law, in the sense of ignorance of the existence of the legal duty, n67 will negate the required mental state much more commonly in the case of offenses of omission. But we do not need to go further and say that the entire requirement of mens rea is inapplicable in the case of omissions. n68

A final pair of possible structural objections to duty to rescue provisions are more practical than doctrinal; they concern the possibility of situations of multiple victims or multiple potential rescuers. But in fact neither situation is problematic. In the case of multiple victims, a rescuer complying with our exemplary Rhode Island provision would continue to render assistance to as many people as she could, so long as this involved no peril to her and so long as the cost sustained remained reasonable; beyond that, she is acting above and beyond the call of legal duty. Where there are multiple potential rescuers, it is not absurd to think that if none of them does anything to help the victim, each of them could be held criminally liable. n69 It may not be practical to prosecute all such people, but selective apprehension and prosecution, so long as the selection criterion is legitimate, should raise no greater problem than does selective  [*621]  apprehension and prosecution of speeding motorists. n70 And if one person out of the crowd does act, there is no problem with the liability of the others--the victim no longer needs any help, and so they are no longer under any legal duty. n71 It may be thought that there is a puzzle about who should make the first move in this kind of case, n72 but there is not. Utilitarian wisdom, but also simple common sense, tells us that the person best placed to help (in terms of effectiveness and cost) should be the one. Of course, if the best-placed person does not act, the legal duty does not exonerate the others. What is required is that, taking into account everything that is going on around you (including the behavior of others) you render reasonable assistance if you can. Needless to say, in many cases people lack information about exactly who is best placed to help, who next best, and other factors that would help them decide whether to act. What reasonable people do in such circumstances is react to events as they unfold; one takes the first steps to render assistance, ready to back off if it becomes apparent that there is a superior rescuer at hand.

It could be said that this response to what is essentially a practical objection is speculative. To the contrary, the objection itself ignores a rather long history of experience with legal duties to rescue. The population and the courts of France, to take just one example, have managed to cope with criminal and civil duties to rescue for some fifty years. If the problem of multiple potential rescuers were serious enough to warrant rejection of the duty to rescue, one would expect some supporting evidence to have emerged by now. n73 It is somewhat astonishing to read practical objections such as these that take no account of the existence of a world in which the impossible has come to pass. n74 In any case, such objections also apply to current common-law doctrine. If twenty school teachers all failed to throw a rope to a drowning elementary school pupil at the school rowing regatta, would our courts be unable to find a way to impose liability? n75

B. TORTS

Outdoing the criminal law's Beardsley, tort law offers us Osterlind v. Hill, n76 Buch v. Armory Manufacturing Co., n77 and many more "unappetizing" n78 cases  [*622]  besides. In Osterlind, the stock example of failing to effect the easy rescue of a drowning person came to life. The villain of the piece had rented a canoe to the evidently drunk victims. As the victims drowned, the villain, boat and rope close at hand, sat smoking on the dock. In Buch, a boy injured himself by putting his hands in the defendants' machinery. Chief Justice Carpenter had this to say:

The defendants are not liable unless they owed to the plaintiff a legal duty which they neglected to perform. With purely moral obligations the law does not deal. For example, the priest and Levite who passed by on the other side were not, it is supposed, liable at law for the continued suffering of the man who fell among thieves, which they might, and morally ought to have, prevented or relieved. Suppose A., standing close by a railroad, sees a two year old babe on the track, and a car approaching. He can easily rescue the child, with entire safety to himself, and the instincts of humanity require him to do so. If he does not, he may, perhaps, justly be styled a ruthless savage and a moral monster; but he is not liable in damages for the child's injury, or indictable under the statute for its death. n79

Unlike Beardsley, both Osterlind and Buch have effectively been overruled, but on narrow grounds that apply only to the duties of landlords. n80 Chief Justice Carpenter's example of the child on the tracks remains apt. His rhetoric about law and morality is, however, even less helpful than the similar slogan used in the context of criminal offenses of commission by omission. The question before him was whether a legal duty to protect the boy existed; it is obviously empty to say that if a duty is purely moral, and thus not also legal, it is not legal. Perhaps he meant to say that moral duties are not ipso facto legal duties, but this hardly seems worth mentioning.

A person may be liable in tort for failing to provide aid where that failure is part of a positive course of action already undertaken, such as the failure to blow the whistle on the part of a train driver; n81 here the failure to act is counted as part of a wider pattern of misfeasance--negligent or otherwise tortious operation of the train. Liability for genuine nonfeasance depends on the existence of some relationship between the agent and the victim; landholders, carriers, hosts, and various others may be liable for not taking positive steps to prevent harm or provide aid. n82 The authors of Prosser and Keeton on Torts announce that decisions such as Osterlind and Buch are "revolting to any moral  [*623]  sense." n83 They write that the sorry state of the law is due to the "difficulties of setting any standards of unselfish service to fellow men, and of making any workable rule to cover possible situations where fifty people might rescue one." n84 The first of these difficulties does indeed go to the heart of the problem, but the second difficulty seems illusory. If all fifty fail to rescue the one, then, as in the criminal case, they are all liable (though in this case we would add that the plaintiff can recover only once and that the enforced-against tortfeasor may seek contribution from the others). n85 Likewise, the concern that, where there are multiple potential rescuers, no one will know what is required of her can be answered, as in the criminal case, by appeal to commonsense norms.

The situation of multiple potential victims is also tractable in the same way as for criminal liability: Any tort duty would presumably be one to take reasonable steps to render assistance--those duties to aid that are currently recognized take this form--and at a certain point the costs of continuing to rescue may become unreasonable. n86 A very particular kind of situation involving multiple victims does, however, seem difficult under tort doctrine. This is the case of multiple victims and a potential rescuer who fails to act, but with the added twist that if the rescuer had acted, she would have been able to rescue only some of the victims. We may suppose that a ferry has capsized in the harbor and fifty people are floundering in the water. The villain has a motorboat that could save as many as ten people before they drown, though not more; yet the villain sits smoking on the dock. The potential rescuer has failed in her duty, but to whom? Who has suffered the damage necessary for tort liability? In such a case the victims' survivors could argue that the villain's failure removed a chance of their decedent's survival, thus drawing an analogy with cases where conduct that fails a but-for test for causation of harm nevertheless increased the risk of that harm. n87 Once more, such a situation could certainly arise under existing  [*624]  doctrine that imposes a duty to act on, say, the captain of a ship in respect of her passengers; the problem is not unique to a general duty to rescue.

We can conclude then that there is nothing in the structure of common-law tort doctrine to block adoption of a general duty to rescue. n88 And indeed, a recent path-breaking decision of the New South Wales Court of Appeals, upholding a damages award of over three million Australian dollars against a physician who had refused to come to the assistance of a stranger, resulting in that person's permanent quadriplegia, proves that it can be done. n89 Whether this is a development to be welcomed, let alone extended beyond the specific  [*625]  category of medical practitioners, is another matter. Like several other commentators, I believe that the case for civil liability for failure to rescue is weaker than the case for criminal liability. n90 The reasons for this will be discussed in Part IV.

II. LAW AND LIBERTY

A. MILLIANS

Though all agree that the person who fails to effect an easy rescue is a moral monster, there is a good deal of disagreement about why. At the level of moral theory, there should be no room for disagreement: The duty to rescue is a duty of beneficence, a positive duty requiring people to benefit others, even total strangers. n91 Most defenders of legal duties to rescue are anxious to avoid this natural characterization, however. n92 The reason for this will emerge as we begin to consider the normative objections that have been raised against a legal duty to rescue.

All normative objections to a legal duty to rescue depend upon at least one, but sometimes two, kinds of normative premises. One premise of any argument that there is a general reason of political morality to resist legal duties to rescue must be some view about the appropriate aims of the law. More specifically, whether explicitly stated or not, there will be a premise about the positive aims of the particular branch of the law under discussion. The second kind of premise, which may or may not be present in any given argument, is a claim about the negative constraints that apply to the pursuit of the specified aim. n93 Thus, for example, one might hold that the aim of the criminal law is to promote the social good and also accept, as a constraint on legitimate pursuit of that aim by government, a certain set of individual rights.

My aim in this Part is not to critique the various premises of the arguments discussed--that would be too great a task--but rather to show that these premises do not lead to the conclusion that legal duties to rescue are illegitimate  [*626]  or obviously unwise. I will begin with the criminal law, and with a theorist who is gratifyingly explicit about the premises of his arguments.

In Harm to Others, the first volume of his four-volume work, The Moral Limits of the Criminal Law, n94 Joel Feinberg presents an elaboration and defense of what he sees as the central thrust of John Stuart Mill's essay, On Liberty. n95 As a utilitarian, Mill held that the aim of the law was to promote happiness, or well-being. n96 As a constraint on all attempts to promote general well-being by means of coercion--either by the state or by informal social pressure--Mill presented his harm principle, stating that "the only purpose for which power can be rightfully exercised over any member of a civilized community, against his will, is to prevent harm to others." n97 The harm principle was not, for Mill, a departure from utilitarianism, but rather a specification of what utilitarianism, properly understood, implied about interferences with people's liberty. n98

As H.L.A. Hart emphasizes in his discussion and defense of Mill's position, there are two strands to the harm principle: the rejection of paternalism and the rejection of what Hart calls "legal moralism," which is the view that a legitimate aim of the law is the enforcement of morality for its own sake. n99 Both strands are evident in Mill's sentences immediately following the one just quoted:

His own good, either physical or moral, is not a sufficient warrant. He cannot rightfully be compelled to do or forbear because it will be better for him to do so, because it will make him happier, because, in the opinions of others, to do so would be wise, or even right. n100

As Hart points out, the two strands of Mill's position do not stand or fall together; Hart is himself less concerned about resisting paternalism than legal moralism. n101 For us, however, the crucial point is that these are the main two strands of Mill's position and that the use of the word "harm" is not meant to mark out an important distinction between harm and benefit. This seems clear enough from the fact that Mill, like Bentham before him, explicitly endorses a  [*627]  legal duty to rescue. n102 Mill states that "in all things which regard the external relations of the individual, he is de jure amenable to those whose interests are concerned, and if need be, to society as their protector." n103 He goes on to say that the decision whether to enforce a person's positive responsibility to other persons depends largely on "expediences," and his discussion makes clear his view that the enforcement of duties of beneficence is in general less expedient than the enforcement of negative duties not to cause harm. n104

It is therefore rather surprising that Feinberg, who is an enthusiastic supporter of a criminal duty of easy rescue, should find it so important to cast the duty to rescue as a duty to prevent harm--as opposed to a duty to confer a benefit. n105 The key contrast for Feinberg here is not that between a positive and a negative duty, for he is willing to construe the harm principle as allowing for the enforcement of positive duties to prevent harm rather than just negative duties not to cause harm. n106 Instead, he thinks it important for the legitimacy of a duty to rescue that it be understood as the positive duty to prevent harm rather than the positive duty to confer a benefit.

One of Feinberg's concerns is that if the moral duty to rescue is understood as a duty of beneficence, then rescuees are merely beneficiaries and they cannot be said to have a right that the rescuer take action. n107 In this connection he discusses Mill's version of the traditional distinction between perfect and imperfect duties. For Mill, a perfect duty is one that correlates with some particular person's claim, or right. n108 Thus my duty not to kill correlates with each person's right not to be killed. A duty of beneficence, by contrast, is open-ended--on any plausible account, I can fulfill my duty while many people remain unbenefited. These unbenefited people are not wronged by me--I do not violate their rights--for it is in the nature of imperfect duties that I must make some choices between equally worthy beneficiaries. Furthermore, if I fail entirely in my duty and assist no one, I violate no one's rights even though I have acted wrongly. This feature of duties of beneficence does cause some difficulties in the tort context, as we saw. n109 But Feinberg is concerned about it because on his account of the harm principle, harms are violations of rights. n110 On Feinberg's relaxed view of rights, just about any setback to a person's  [*628]  interest can count as a violation of a right, so his insistence that criminal prohibitions are legitimate only when they protect individual rights is clearly not motivated by some sense that rights protect especially important interests. His concern is rather that for each crime there must be an identifiable victim, lest we countenance illiberal "victimless crimes." n111 But this is strange; as Feinberg himself acknowledges, there are many crimes for which there may be no identifiable victim--violation of a criminal pollution prohibition is one obvious example. n112 The phrase "victimless crime" is most naturally understood to refer to crimes condemned as paternalistic, such as sexual conduct between consenting adults or the use of narcotics, but this is not the problem with the duty to rescue.

The more important source of Feinberg's desire to understand a duty to rescue as a duty to prevent harm rather than as a duty to benefit is his apparent concern that the latter understanding opens the door to legitimate legal duties of beneficence beyond the rescue context. n113 This too could seem a strange concern, because it could seem that rescue situations are distinctive enough that commitment to the duty to rescue need imply nothing about what other legal duties of beneficence may be legitimate. However, it is extremely difficult to find a plausible normative basis for distinguishing rescue contexts from other contexts in which one person can benefit another. We can seek to define a rescue situation, or at least offer paradigm examples, but it is very hard to justify treating these situations as distinctive in any normatively relevant way. This issue is discussed in some detail in the next Part; I mention it here to show that one apparently easy way out is not available to Feinberg.

If Feinberg is right, then the significance of the issue of legal duties to rescue is much less than many of us, especially opponents of the duty, have supposed. Feinberg wants to present the moral duty to rescue as sui generis. This would mean that we can support its legal enforcement without signing on to the idea that it is in general legitimate to coerce one citizen to benefit another. But opponents of the legal duty to rescue are right on this issue--one cannot support the enforcement of the duty to rescue without admitting the prima facie legitimacy of legally-enforced beneficence generally. As I said in the introduction, it is this wider implication of the legitimacy of legal duties to rescue that largely explains the importance of the topic.

On an ordinary understanding, A fails to benefit B when A could, but does not, perform an action that would make B better off. On this ordinary understanding, a failure to rescue is a failure to benefit. Feinberg insists, however, that our ordinary understanding of the verb "to benefit" is ambiguous in an important way. If A fails to halt some deterioration in B's well-being and bring B back up to B's "normal baseline," then, Feinberg says, this is not really a case of failing  [*629]  to benefit, but rather a case of harming, or at least failing to prevent harm. n114 Genuine--that is, mere--benefits are those that "advance another's interest to a point beyond his normal baseline." n115 If someone fails to render a genuine benefit, that person has not harmed or failed to prevent harm, she has merely failed to benefit. The point of all this is to give an account of "mere benefit" that marks off rescue contexts as involving something morally more serious than the mere opportunity for A to benefit B. n116 The idea is that if B's life was proceeding just fine prior to some sudden turn for the worse, then A's failure to help B back to normal life is not a mere failure to benefit, but rather a failure to prevent harm (which is morally much more serious). By contrast, we must assume, where B has suffered a slow and steady decline, but is about to die (perhaps of starvation), any benefit from A would count as a genuine--that is, mere-benefit; for in that case B has no normal baseline to be returned to. The discontinuity that appears in the classic rescue case, and that Feinberg appeals to in his account of failing to prevent harm, is lacking.

Feinberg is of course free to define terms as he likes, but his idea that the failure to aid is morally much less serious in the latter case than in the former strikes me as very implausible. Why are the transient and unexpected crises of the moderately well-off more worthy of our attention than the chronic and predictable dangers faced by a badly-off person--even a badly-off person I may confront with my own eyes? n117 I doubt that Feinberg would be happy to have the point put this way, but this is the position to which his attempt to take the duty to rescue out of the realm of controversy leads. n118

 [*630]  We can conclude that a duty to rescue is what it appears to be: a duty of beneficence. We can further conclude that this fact should cause a Millian liberal no concern. Legal enforcement of a duty to rescue is clearly not paternalistic. Nor need it be motivated by a concern to enforce morality for its own sake--the case for a criminal duty to rescue does not rest on legal moralism. To be clear about this point, we must not confuse the fact that a legal duty to rescue would be the enforcement of a moral duty with the legal-moralist claim that the aim of the enforcement of the duty to rescue is to enforce morality. It is obvious enough that most of the criminal law enforces moral duties; the Millian liberal insists that the purpose of this enforcement is to benefit society, not to enforce morality for its own sake. n119

One might wonder at this point where liberty enters into Mill's liberalism as I (following Hart) have described it. The answer is that for Mill, the value of liberty, or more precisely of "negative liberty" (the absence of coercive interference) n120 lies in its contribution to a person's well-being. Human beings who make their own decisions about how to live, who (as Joseph Raz puts it) make their own lives, can be said to be autonomous. Mill values negative liberty as a contribution to autonomy, which he believes to be a central aspect of people's well-being. n121

Negative liberty contributes to autonomy just because coercive interference reduces autonomy. The most obvious way in which coercion reduces autonomy is by restricting those basic requirements of an autonomous life which we may refer to collectively as a person's positive freedom--the availability of a full range of options and the ability to make reasonable choices among them. n122  [*631]  Thus we can say that interference with my negative freedom matters because it reduces my positive freedom, which in turn impairs my autonomy and thus reduces my well-being.

To bring these points to bear on Mill's harm principle, we can see that paternalistic coercion may be said both to reduce my positive freedom and to undermine my autonomy more directly by expressing contempt for my ability to create my own life. As Mill puts it, "He who lets the world, or his own portion of it, choose his plan of life for him, has no need of any other faculty than the ape-like one of imitation." n123 Mill's opposition to legal moralism is also connected to the value of autonomy insofar as autonomy is valued as an aspect of human well-being. As we have said, on Mill's account the purpose of the law is to promote human well-being. Human well-being is not promoted by the enforcement of morality for its own sake, unless acting morally is intrinsically valuable for a person. Mill obviously rejects this view. n124 Most straightforwardly, we can see that coercion contrary to the harm principle is bad because it reduces a person's negative liberty (and thus potentially reduces her autonomy) n125 while doing nothing to promote the interests of anyone else. But a duty to rescue can legitimately be enforced precisely for the reason that, though it does diminish the negative liberty of the person coerced, it promotes the interests of others. Of course, if the promotion of the interests of others actually achieved by such a provision were outweighed by the loss in autonomy caused by the interference with negative liberty, such a provision would make no sense for a Millian liberal. But as a legal duty of easy rescue would interfere with liberty only minimally, it is not surprising that Mill felt it unnecessary to defend his support for it. n126

Given that there is no reason whatsoever for a Millian liberal to question the legitimacy of the legal enforcement of a duty of easy rescue, what is surprising is that so many people, in a Mill-dominated tradition, have thought that such a duty would indeed involve an unacceptable interference with liberty. One plausible explanation, mentioned at the start of this Article, is that the utilitarianism of our tradition has alerted many people to the possibility that legal duties of beneficence a good deal more stringent than a duty of easy rescue will also survive the balancing test of liberty lost versus interests of others served. n127 Given that there are many people in any given country who are capable of being  [*632]  benefited to a significant degree, it may turn out that this balancing test is satisfied even when the losses to the people coerced--if not of liberty, then at least of money--are very great. I return to this issue in the next Part.

Another important explanation is that our tradition is not just Millian; it is also, to a significant degree, libertarian or "Lockean." n128 And the libertarian understanding of the right to liberty raises entirely different problems for the duty to rescue. Before turning to that other strand of liberalism, n129 however, we need to discuss the views of a prominent opponent of criminal duties to rescue who, though he disagrees with Mill on the proper aims of the criminal law, is broadly Millian in his understanding of the value of liberty. This will also allow us to investigate a third explanation for the otherwise surprising opposition to legal duties to rescue--many people believe, contrary to what I have asserted, that positive duties in their very nature pose rather significant threats to individual liberty understood in Millian terms.

In his recent book, Placing Blame, Michael Moore defends the legal moralist view of the criminal law: The purpose of the criminal law is to enforce morality. He reaches this position by way of his defense of a retributivist theory of punishment. n130 He thus departs radically from Mill's utilitarian account of the proper aim of the criminal law as well as from the broader textbook account of that aim as the prevention of social harm--or, as I would rather put it, the promotion of the social good. n131 But Moore does believe that there are countervailing considerations that should temper the aim of enforcing morality. One important consideration of this kind is the general value of negative liberty. n132

Moore's idea is not that individuals have a right to be free of legal coercion. Moore accepts Ronald Dworkin's argument that any such right, so often overridden  [*633]  by legitimate law as it obviously is, is too weak to be worthy of the name. n133 In so doing, Moore also embraces Dworkin's understanding of the term "right," an understanding that is very different from that of Feinberg mentioned above. Dworkin uses "right" in the stronger and more standard sense of a constraint on what may be done to a person even for the sake of promoting general welfare. n134 Rights in this strong sense thus correlate with negative duties whose moral importance cannot be reduced to a concern with the promotion of human well-being, or indeed even with the minimization of rights violations. As Robert Nozick illustrates, if you have the right not to be assaulted, I am duty bound not to assault you even for the sake of preventing several other assaults to other people. n135 In discussions of rights that follow, I too will have in mind this strong sense of the term. n136

So Moore accepts that there is no moral right to negative liberty. The conflict between liberty and the aim of the criminal law is rather a conflict between two values; the value of enforcing morality and the value of negative liberty. Moore is thus content to employ the traditional idea of a "presumption of liberty" in this context, which is simply that infringement of negative liberty is always regrettable just because freedom from coercion is always valuable, and so any criminal duty must have enough to be said in favor of it to outweigh the loss of liberty it brings. n137 Freedom from coercion is valuable, according to Moore, because positive liberty--by which he means strictly the range of options available to a person--has intrinsic value. n138 Moore does not go on to defend the value of positive liberty in terms of its contribution to autonomy, but in holding that the value of negative liberty lies in its instrumental contribution to positive liberty, he remains squarely in the Millian tradition. n139

 [*634]  Moore believes that negative criminal duties will generally have a better chance of overcoming the presumption of liberty than positive duties. This, for Moore, explains and justifies the traditional position on the duty to rescue in common-law countries and the reluctance to embrace positive criminal duties of any kind. There are two mutually enhancing aspects to the argument. First, Moore claims that negative moral duties are generally more important, or weighty, than positive moral duties, and thus the legal moralist case in favor of their criminalization is stronger. Second, he claims that positive duties generally interfere with liberty more than do negative duties. n140

Neither of these claims seems at all plausible to me. It is a well-known assumption of commonsense morality that it is worse to cause some harm rather than allow it to happen; thus it is worse to drown a baby than to allow it to die. n141 But it does not follow that all negative duties are morally more stringent than all positive duties. The duty not to lie is obviously less stringent than the duty to save a drowning child at low cost, or, indeed, to take care of one's own children, even at high cost. n142

Responding to this point--mindful especially of the strong positive duties people are commonly taken to have toward their children--Moore suggests that the relevant distinction may be that between agent-relative and agent-neutral duties, the former being more stringent than the latter. This widely-used distinction, originally formulated by Thomas Nagel, can be drawn in various ways. n143 I find the following account congenial: agent-neutral duties give all of us the same aim; agent-relative (or "deontological") duties give different agents different aims. n144 The paradigm agent-neutral duty would be a (positive) duty of beneficence; paradigm agent-relative duties would be the (negative) duties not to violate rights. n145 But some positive duties, such as the duty of a parent to take care of his own children, are also agent-relative. Moore's suggestion is that while the negative/positive distinction roughly marks out the more stringent from the less stringent duties, the distinction between agent-relative and agent-neutral duties does so more accurately.

There are several problems with Moore's shift to the distinction between agent-neutral and agent-relative duties. In the first place, the moral duty to rescue (which, as a duty of beneficence, is an agent-neutral duty) is widely  [*635]  thought by the critics of legal duties to rescue to be very stringent indeed (recall the moral monsters). Second, the shift of focus to the contrast between agent-relative and agent-neutral duties makes it quite clear that the negative/positive distinction in itself carries no weight on the question of the stringency of moral duties. This is an uncomfortable result for a theorist who has put so much weight on the importance of the act requirement in criminal law. Indeed, because the distinction between agent-relative and agent-neutral duties does not track action and inaction, we are left with no account of why we should believe the former to be more weighty than the latter--Moore simply asserts that this is the case. n146 Without some account of why we should take this view, legal moralists must rely on case-by-case commonsense intuitions about which particular moral duties are, in fact, more important.

Moore is actually prepared to grant that the positive (and agent-neutral) moral duty to perform an easy rescue is more weighty than some negative (and agent-relative) duties which are nevertheless criminalized. This is not fatal to his position, he believes, because there is always the other aspect to his argument--that concerning the impact on liberty. In discussing Samuel Freeman's claim that there is much greater reason for the legal moralist to criminalize failing to rescue a child than stealing her purse, n147 Moore grants that the moral duty to save may be more important than the moral duty not to steal, yet insists that the "liberty differential is still there." n148 But the claim that "it diminishes liberty less to prohibit theft than it does to require life-saving activities" n149 is preposterous. Recall that for Moore the importance of negative liberty lies in its contribution to positive liberty, or the range of available options for choice. As the opportunity to effect an easy rescue never arises in the vast majority of people's lives, the thought that there is a serious interference with people's opportunities for choice here is so evidently bizarre n150 that it calls for some explanation. One possibility, once again, is that Moore and others who make this claim must have other, more demanding, duties of beneficence in mind--the utilitarian duty always to perform that act which will maximize expected aggregate well-being does indeed appear to be very invasive of positive liberty. Be that as it may, the fact remains that it is not credible to say that the Rhode Island provision constitutes more than a de minimis interference with the liberty of Rhode Islanders.

The claim that a duty not to steal is in general not very invasive of liberty is also clearly false. Only well-off people who rarely seriously contemplate the option of theft would ever be tempted to think as much. There is, indeed, an obvious and simple political point to be made about the claim that negative  [*636]  duties as a class interfere with liberty less than positive duties as a class: The richer one is, the more plausible this will seem; the poorer one is, the less plausible it will seem. If we focus our minds on the options of a destitute and uneducated person in the United States, and think of the entire range of negative criminal duties lined up against her, the suggestion that she remains free of significant interference with her liberty, because, thank goodness, she has no positive duties imposed on her, should strike us as absurd.

It is important to remember that I make these responses to the second aspect of Moore's argument without challenging his view that negative liberty is always valuable because its infringement leaves us with a smaller range of options for choice. Thus, as against two of his critics, I agree with Moore that the value of liberty is not dependent on my desiring the options in question, and agree also that there is value in being able to choose to do things that may be wrong or worthless. n151 Furthermore, though I am inclined to think that positive liberty is nevertheless in some sense more valuable the more the relevant options matter (either subjectively or objectively), n152 we can for the sake of argument grant Moore that this is not so. The rejection of the idea that positive duties generally interfere with liberty less than negative duties does not at all depend on disagreement with Moore's conception of liberty; it depends rather on two simple points about the way that different duties impact on a person's opportunities for choice.

To repeat and expand on these points, the first is that the extent to which a positive duty will interfere with liberty--diminish the number of options for choice--depends crucially on the pervasiveness of the duty's application to a person's life. The popular claim that positive duties are terribly detrimental to liberty, and much more so than negative duties, seems to be based on the thought that while a negative duty merely cuts off one option, a positive duty cuts off all options but one. But this abstract way of putting the point is very misleading precisely because it focuses on the single moment when the duty must be performed, rather than a person's life as a whole. Life is full of positive duties that hardly interfere at all with our total range of options for choice. A positive duty to take out the garbage once a week is very different from the positive duty to become a garbage collector. Likewise, a duty to perform an easy rescue would reduce most people's opportunity sets not at all; a duty to spend every day standing by the river watching for trouble would of course be very detrimental to liberty, but we shouldn't confuse the two duties.

The second point is the more important of the two, as it reveals in stark form the ideological nature of the traditional aversion to positive duties as opposed to negative duties. One of the things that most determines a person's set of possible options is the amount of resources at her disposal. The familiar  [*637]  negative duties prohibiting attacks on person and property are extremely detrimental to the positive liberty of those who lack resources; indeed, the coercive prohibitions of theft and violence are far more restrictive of the liberty of the destitute than any positive legal duty anyone has ever seriously proposed.

The rejection of the claim that negative duties are less detrimental to liberty than positive duties obviously has force beyond the context of Moore's legal moralist theory of the criminal law. It applies to any account of the criminal law that includes, as a constraint on whatever aim the criminal law is taken to have, a concern for the value of negative liberty conceived of as a means to positive liberty.

B. LIBERTARIANS

To find some general reason of political morality to oppose positive legal duties, we therefore need to turn from Millian liberalism to the libertarian version of rights-based liberalism that in the English-speaking world derives primarily from standard interpretations of Locke. n153 The central claim of libertarian political theory that touches on our topic is that all persons have a right to negative liberty. Here, freedom from coercion is not understood as a value, much less an instrumental value in the service of positive liberty, but rather, simply, as a natural right. Indeed, many writers in this tradition emphasize that in their view the commitment to negative liberty implies no commitment at all to any form of positive liberty. n154 What makes freedom from interference at the hands of other human beings so important? Why is it prima facie so much worse if I am prohibited from proceeding along a road by rocks that you put there rather than by ones that fell? The idea is simply that a special wrong is done when people coerce other people. Freedom is a right, and rights, in the strong sense of the term that we are using, are all about constraints on what people can do to one another. n155

Thus the heart of the libertarian account of law's legitimate content is the constraint that individuals' rights must not be violated by the legal regime--especially not individuals' rights to be free of coercion. Of course, it is obvious to everyone that coercion is, as Hayek says, "unavoidable," for the very reason that coercion or its threat is necessary to prevent coercion. n156 This leads straight to the idea that it is no violation of my right not to be coerced for me to be coerced into not violating someone else's right not to be coerced. But libertarians usually go much further and allow that coercion is legitimate for the sake of protecting the whole range of rights that are generally taken for granted in a  [*638]  market economy--notably a right against all forms of aggression, a right to private property (as somehow identified via principles of acquisition and transfer), n157 contractual rights, and the right to be free of fraud. n158 Thus libertarians believe that coercion, and rights infringement generally, is justified for (and only for) the sake of protecting a certain familiar list of individual rights. When it is recalled that on the account of rights we have been employing, and which is employed by contemporary libertarians, there is no such thing as a positive right to aid, it is not hard to see how libertarians reach the conclusion that "the state may not use its coercive apparatus for the purpose of getting some citizens to aid others." n159

The most prominent opponent of a tort duty of easy rescue among contemporary legal theorists has been Richard Epstein. In his early writings on the topic, the main thrust of his opposition was that affirmative legal obligations constituted an illegitimate interference with individual liberty--understood along libertarian lines. n160 As such, the rejection of duties of easy rescue came with radical implications. Because one of the rights on the libertarians' standard list is the right to private property, redistributive taxation turns out to be illegitimate because it infringes on the right to private property while protecting no other right. In his book Takings, Epstein does not shrink from this conclusion. n161 Nevertheless, Epstein has long believed that there are areas of the law that cannot be made compatible with libertarianism and must instead be subjected to  [*639]  a kind of supplementary utilitarian analysis. n162 Furthermore, in Takings, Epstein explicitly rejects Nozick's thoroughgoing libertarianism on the ground that he cannot see how the state can be justified if all forced exchanges are ruled out. n163 Once rights can be infringed for the sake of the general good, however, the cat is out of the bag, and it not surprising that shortly after the publication of Takings Epstein announced that the traditional natural rights view he had been espousing was defensible on utilitarian grounds. n164

Even though Epstein no longer holds to libertarianism at the foundational level, an inchoate form of the libertarian view would appear to be widespread among opponents of legal duties to rescue n165 and thus it is important to discuss it a little further. One point has already been alluded to: If one argues against legal duties to rescue on libertarian grounds, one had better be prepared to reject as illegitimate whole swathes of familiar legal regulation. A possible way of containing these radical implications would be to tamper with the libertarian's typical list of rights. Many of the most radical implications of the doctrine flow from the supposed absolute moral right to property; if this right were removed from the list, redistributive taxation would come out as legitimate while forced labor and legal duties to rescue would remain illegitimate, for violating the right to negative liberty. Given the difficulties facing accounts of an absolute moral right to private property, and given the overwhelming normative implausibility of the libertarian's traditional opposition to any form of redistribution away from market outcomes, this would seem a promising way to save the core idea of this form of liberalism. However, it is not so easy to deradicalize libertarianism. Under this view, noncontractual positive duties to act will remain illegitimate; the duty to file a tax return will be rejected even if automatic withholding  [*640]  for the sake of redistribution will not. And there are of course all the other noncontractual positive duties we have mentioned along the way--to register for the draft, to "rescue" one's aged parent, among others; all these legal duties will fall along with the duty to rescue. Libertarianism remains radical so long as it insists that the only permissible ground for coercion is to prevent the coercee from infringing the negative rights of another.

The fact that libertarianism is a radical view does not disprove it, of course; it merely makes clear the high stakes of any libertarian argument against the duty to rescue. Whether libertarianism is a plausible political theory is obviously not an issue we can settle here. What is important to note, however, is that such a discussion would focus on, first, the plausibility of the libertarians' list of rights, and, second, the claim that these rights can justifiably be infringed only in order to prevent the coercee from infringing the rights of others. n166 This description makes it clear that despite libertarian rhetoric, the evaluation of libertarianism has little to do with the question of whether one is in favor of more or less liberty; it has rather to do with a very particular focus on the moral force of certain rights. Whether it is therefore appropriate to call libertarians "rights-fetishists" depends upon your point of view, but it does seem right to say that libertarianism is a misleading label. n167

To see this more clearly, consider the libertarian slogan that liberty may be limited only for the sake of liberty; this suggests that the whole of political morality can be understood in terms of negative liberty. n168 As Sidgwick characterized the view one hundred years ago (commenting optimistically that it was "now perhaps somewhat antiquated"), "all natural Rights, on this view, may be  [*641]  summed up in the Right to Freedom; so that the complete and universal establishment of this Right would be the complete realization of Justice,--the Equality at which Justice is thought to aim being interpreted as Equality of Freedom." n169 The trouble is simply that so much of what the libertarian list of rights protects has nothing to do with freedom--especially not on the negative conception of freedom which libertarians insist is the only conception they mean to employ. n170 The most obvious, and politically most important, illustration of this once again concerns the right to private property. In what sense is the prohibition of theft a case of coercion for the sake of protecting liberty? But the point holds even for the right against aggression, which cannot be equated with the right to be free from coercion. n171 For example, it is not in the least clear why prohibiting A from poisoning B's water so as to cause B occasional headaches is a case of coercing A for the sake of protecting B's liberty; it is not, indeed, even clear why prohibiting A from killing B is a case of limiting liberty for the sake of liberty. n172 If there is a single conception of freedom animating libertarian political morality at all, it is obviously a strongly moralized conception. n173 The idea must either be that any violation of B's rights is by definition a restriction of B's liberty, and/or that whenever A is prohibited from violating B's rights A's liberty is by definition not thereby restricted. n174 But such definitions  [*642]  are clearly ideological because the resulting notion of freedom bears almost no relation to either negative freedom or any version of positive freedom.

This concludes our survey of the various versions of the claim that duties of easy rescue constitute an excessive degree of interference with liberty. We have found no version of the claim to be plausible. It is true that libertarian political morality, unlike Millian liberalism, does provide a reason to reject legal duties to rescue (along with all noncontractual positive duties)--such duties are said to violate our rights. But evaluation of libertarian theories of political morality is, as I have said, beyond our scope. As such theories have implications far beyond the issue of the duty to rescue and yet are also--to my mind at least--deeply implausible, this is not such a great loss to our evaluation of the case against legal duties to rescue. Nevertheless, it must be acknowledged that the discussion that follows will be beside the point for committed and consistent libertarians. What even they must admit, however, is that their opposition to legal duties to rescue is not grounded in a concern with individual liberty.

C. CORRECTIVE JUSTICE

With the exception of Epstein's libertarian opposition to tort duties to rescue, the arguments so far discussed in this Part have concerned the criminal law. Arguments in the tort context fit within the structure we have been using: There must be some premise about the aim of tort law and perhaps also a premise about the constraints on legitimate pursuit of this aim. In a general sense, we might expect the case for criminal duties to rescue to face stronger opposition, simply because criminal conviction is a more serious matter than a judgment of civil liability. This is not necessarily so, however. Indeed, it is the potential severity of a finding of civil liability for a failure to rescue that leads some people, myself included, to question the wisdom of a tort duty to rescue. n175

Furthermore, there are a range of views about the proper aim of tort law, and not all of them allow for a duty to rescue. If the aim of tort law is to promote general well-being, then the desirability of a duty to rescue depends entirely on the question of whether, all things considered, such a duty would promote general well-being. One might also adopt a legal moralist conception of tort law according to which civil liability is simply a secondary way of enforcing  [*643]  morality; if so, there should also be no principled bar to the duty. n176 Matters are otherwise if the preferred account of the aim of tort law appeals to some notion of corrective justice--here understood as an independent, normative account that is meant to be independently appealing, rather than as an interpretation of the normative structure of tort law as it now is. Most theories of corrective justice do place special emphasis on misfeasance as opposed to nonfeasance and so would reject civil liability for failure to rescue. n177 As with the libertarian theory of political morality, evaluation of the various possible accounts of corrective justice is beyond my scope here. n178 Therefore all I can do is acknowledge that there are specific normative views concerning individuals' responsibility for each others' fates that rule out requiring a non-rescuer to pay damages to the victim.

Like adherents of libertarian theories of political morality, adherents of various theories of corrective justice rule out legal duties of beneficence as a matter of principle. Such views therefore do not reach the issues that will occupy us for the remainder of this paper, which concern the difficulties that arise once legal duties of beneficence get a foot in the door. Very many of us, however, think that the aim of tort law should be to promote the social good, and most of us take this same view (or else the legal moralist view) of the criminal law. Thus I think most of us need to take seriously the following problems. n179

 [*644]  III. THE (UNAVOIDABLE) PROBLEM OF DEMANDS

In a few pages in his notes to the Indian Penal Code, Macaulay presents in lucid form the most serious problem raised by duties to rescue. The problem has two parts. The first is that it is very difficult to find defensible normative grounds on which to distinguish rescue situations from nonrescue situations. As a descriptive matter, rescue situations are sudden, unexpected emergencies involving severe suffering or threat to life. But Macaulay rightly wondered how, once duties to render aid in such situations are accepted, it could be consistently maintained that the rich have no duty to save beggars in Calcutta from slow but certain death by starvation. n180 The second problem is that in rescue and nonrescue situations alike, there is the potential for enormous loss to the benefactor. This second problem is sometimes referred to, following Macaulay's own words, as the "line-drawing problem." n181 I call it the problem of demands. A duty to render aid is, as I have said, a duty of beneficence. The duty of beneficence embraced by utilitarians imposes demands with no limit: We must go on benefiting others until the point where further benefits will burden the donor as much as they will benefit the donee. For even moderately well-off people, compliance with this duty would make for a radical change of lifestyle. But if this optimizing duty of beneficence is to be rejected as obviously absurd in virtue of the demands it makes, what more reasonable duty of beneficence should we embrace?

Before expanding on these problems, we need to answer a crucial question. As presented, Macaulay's problems are problems of moral/political theory. What is their relevance to the issue of the legitimacy or desirability of legal duties to rescue? Why do we need to be able to have a complete account of the morality of beneficence before we can conclude that there is no objection to at least the kind of minimal legal duty of beneficence found in Rhode Island?

The first point to make is that even if these two problems of moral theory were irrelevant to legal theory, strictly analogous problems would be relevant. Suppose a legal theorist were to hold that the criminal law aims to promote the social good and that criminal duties such as the Rhode Island provision do this and therefore should be enacted. Such a theorist must still answer this question: Why stop with minimal duties to rescue? Why not enact more demanding legal duties of beneficence as well? It will not do to reply that more demanding duties may or may not be called for, but because they seem to be controversial, we can ignore them. Despite what some defenders of legal duties to rescue say, this  [*645]  problem is not one that can be solved by careful drafting. n182 In the absence of any principled basis for drawing the line of legally required beneficence at a particular point, the legitimacy of any actual rescue provision, no matter how minimal, will remain in doubt, and concerns about later extensions of the legal duty of beneficence will remain. n183

So at the very least problems analogous to Macaulay's must be dealt with in any principled legal-theoretic defense of legal duties to rescue. But I want to make a stronger claim than that. Legal theory itself must confront Macaulay's two problems as problems of moral theory. This claim is significant, because it renders inadequate one possible principled basis for a strict limit on the extent of legally required beneficence. As we will see in Part IV, there are very important practical reasons why it would most likely not make sense, even for a utilitarian, to support general (that is, not limited to rescue contexts) and highly demanding legal duties of beneficence. Thus, a utilitarian defender of Rhode Island-style duties to rescue could claim to have the very best kind of reason to support the enactment of only minimally demanding and rescue-specific legal duties of beneficence. Minimal duties, focused in particular on emergencies, would do more good than harm, but general and more demanding legal duties of beneficence would do more harm than good. A practical explanation for limiting duties to rescue, however, will not be adequate if a legal theorist must also be able to explain why, on her view, people do not face extremely demanding moral duties of beneficence. The practical reasons that recommend only limited legal duties of beneficence do not apply in the moral sphere.

Why does a defense of legal duties of beneficence require a plausible account of moral duties of beneficence as well? First, suppose we take the legal moralist view of the criminal law (for the remainder of this section it will not be necessary to distinguish between the criminal law question and the tort question). On that view, the purpose of the law is to enforce morality, so it is abundantly clear that the question of moral theory is a question for lawmakers as well. I think it fair to assume that Macaulay operated with a legal moralist view of the criminal law. And it was his lack of comfort with the idea of enforcing one particular duty of beneficence while not being able to say why other more stringent duties of beneficence should not, on pain of inconsistency, also be enforced, that led him to exclude the duty to rescue from the code. So any legal moralist who wishes to be able to give a principled defense of legal duties to rescue needs to take Macaulay's problems seriously.

The more important and interesting point, however, is that the theorist who holds that the purpose of the law is to promote the social good must also take Macaulay's problems seriously. Legal theorists working in a normative vein provide, naturally enough, normative arguments about what the content of the  [*646]  law should be. But in reaching their conclusions about the preferred content of certain specific legal rules, they often do not ask whether the fundamental principles to which their arguments appeal are plausible in general. In particular, they do not ask whether those fundamental principles, used in some specific context of institutional design, are plausible in other areas of institutional design; nor do they ask whether those principles are plausible as principles for personal conduct. The most obvious example of this phenomenon is found in the frequent avowal of utilitarianism as the correct account of the aim of promoting the social good, and thus the correct principle to guide institutional design, by legal academics who would certainly reject the idea that they must live their lives by the extremely demanding utilitarian criterion of personal conduct. n184

There is, I believe, no justification for the assumption that principles that govern specific questions of institutional design need not be plausible generally, including at the level of personal conduct. The issue of the connection between norms for institutional design and norms for personal conduct is important and controversial, and I can do no more than briefly set out my own view. n185

Traditionally, moral and political philosophers believed that the very same fundamental normative principles governed both institutional design and personal conduct. Thus, for example, though classical utilitarians such as Bentham were primarily interested in institutional design, this focus was due more to a belief in this issue's practical importance and to a skepticism about human motivation than to some sense that the principle of utility did not also apply directly to people. n186 The idea that different principles govern on the one hand institutional design (the subject of justice) and on the other hand personal conduct (the subject of morality) is due to Rawls. This dualist view of political morality, as I call it, is a central theme of both A Theory of Justice and Political Liberalism. But I see no good reason to follow Rawls here, and one very good reason not to. n187 The main problem with dualism emerges when we ask what connects just institutions to actual people. There must be such a connection:  [*647]  Even if different normative principles govern institutional design and personal conduct, some normative principle must nevertheless connect institutions to people--institutions do not get to be just all of their own accord. Rawls posits a "natural duty of justice"--people have a duty to sustain and promote just institutions. n188 This provides the necessary normative link between people and institutions, but it also brings out what I believe to be the deep implausibility of the dualist approach. On this view, the obligations people have with respect to matters of justice are always and only to perfect institutions, even when the aims of such institutions could be better promoted in some other way. But I believe we have little interest in institutions for their own sake; we are interested in them primarily for what they can do. If, in a given situation, trying to make institutions more closely correspond to some criterion of justice will achieve less than some direct attempt by people to, say, promote well-being (if that is the criterion for institutional design), then people should pass institutions by. The overwhelming practical importance of institutions in achieving the aims of justice in the typical case should not blind us to the fact that what matters to us, ultimately, is not whether our institutions are just, in the sense that they achieve our aims, but rather simply the extent to which those aims are achieved, however that might be done.

Thus, whatever fundamental normative principles we use to evaluate institutions must also be used to evaluate ourselves. We cannot breezily evaluate legal institutions such as tort law or the criminal law with the utilitarian criterion without thinking about the implications of that criterion in the realm of personal conduct. If the utilitarian criterion is correct, it is correct generally; we do not have some fetishistic attachment to utilitarian institutions and no direct concern with the promotion of human well-being independently of institutional design. Therefore, though there are, as I have said, good utilitarian reasons to limit any legal duty of beneficence to the kind of minimal duty found in Rhode Island, that does not mean that a utilitarian supporter of such provisions can ignore Macaulay's problems. If the only way to justify a legal duty to rescue is by appeal to a fundamental principle that we reject as unacceptably demanding as a principle of personal conduct, the case for the legal duty must fail.

This first reason illustrates why a legal theorist who holds that the aim of the law is the promotion of the social good cannot ignore Macaulay's two problems. It is rather fundamental, and does not turn on any particular commitments of political morality. A second reason why the problems may be unavoidable is more specific and substantive. A theorist who holds that the aim of the law is the promotion of the social good might hold that there is nevertheless an important constraint on the pursuit of that aim: The law cannot promote the social good by requiring people to sustain sacrifice beyond what they are morally required to  [*648]  sacrifice voluntarily. n189 This would be a genuine constraint on the law, so that, if it is accepted, legal duties of beneficence that are more demanding than the duties of beneficence found in the best view of morality would be illegitimate.

Having shown the importance of Macaulay's problems for the issue of legally required beneficence, I return now to setting out those problems in greater detail. The first problem is that of explaining why beneficence should be required in rescue situations but not for the sake of meeting desperate needs that are mundane and chronic. n190 None of the criteria that one might suggest to mark out rescue cases as generating special and especially stringent obligations of beneficence seems on reflection to carry any weight. In his recent book Living High and Letting Die, Peter Unger devotes considerable attention to this issue. He considers nine possible factors that might mark out rescue cases as having special normative significance, and rejects all of them. n191 The force of Unger's discussion comes not just from direct reflection on the plausibility of the various possible criteria, but also from his ability to construct, for each factor, a case that seems to prompt the wrong intuitive response. For example, to those who suppose that physical proximity is what underlies the stronger obligation to help in a rescue case, Unger offers the case of a motorist who, rather than confronting a bleeding victim on the side of the road, hears over his CB radio that a twenty-mile detour will take him to a bleeding victim. Unger notes that people he has asked tend to feel that it would be just as wrong to fail to rescue in both cases. n192 But if a rescue situation is marked out by the factor of proximity, this is the wrong answer.

Unger's eventual conclusion is that in most of the cases where we readily accept a stringent obligation to meet the needs of others, those needs are very conspicuous to the person in question. n193 He further claims that while "our basic moral values" support a stringent duty of beneficence for rescue and nonrescue cases alike, our "futility thinking"--roughly the thought that anything we can do to help will be a mere drop in the ocean--blinds us to this obligation in most cases. In cases like the typical rescue case, however, the needs of the victim are  [*649]  sufficiently conspicuous to us to break the hold of the futility thinking. n194 The upshot of this is that there is nothing morally special about rescue cases; the widespread assumption to the contrary is due entirely to the psychological effect that a dramatic confrontation with a particular person's needs has on our futility thinking.

I find Unger's exhaustive argument against the existence of a distinct normative category comprising rescue cases convincing. n195 At least, we can be confident that there is no simple or obvious way to mark out a distinct principle of beneficence that concerns rescue situations only. n196

In any case, even if we could explain the special moral significance of rescue cases (thus disposing of Macaulay's first problem), Macaulay's second problem, the problem of demands, would remain. This problem is most obvious for a general requirement of beneficence (that is, one not limited to rescue contexts), but an open-ended duty to rescue could also generate extreme demands. We can ignore any duty to rescue that requires an agent to sustain expected burdens greater than the expected benefit of her act. But even a duty that requires a person to act so long as the expected benefit is substantially greater than the expected burden to her can yield extreme demands. Thus a boat owner who would not normally take a fifty-fifty risk on her life by going out into the storm would nonetheless be required to do so by such a duty if there is the same chance of rescuing all of the ten people from the capsized yacht.

Rescue situations can also generate extreme demands of other kinds, not just threats to the rescuer's life. Suppose that a group of astronomers plan an  [*650]  observation that must take place at sea at a particular date and time, and spend many hundreds of thousands of dollars in preparation. Just as the astronomers are about to take their readings, they receive a Mayday signal; if they rescue the sailors in distress, the project will come to nothing. n197

Clearly, an unlimited duty to rescue can generate extreme demands. Nevertheless, even more dramatic demands of beneficence emerge with principles of beneficence that are not restricted to rescue contexts. The utilitarian's optimizing principle of beneficence requires of ordinarily well-off people such enormous sacrifice that it is widely regarded as absurd. n198 There is no need to dwell on the details of the sacrifices that the world's moderately well-off must sustain before further beneficial efforts would do no more good than harm. But one point must be emphasized. It is not plausible to say that the losses that would flow from compliance with the optimizing principle of beneficence are all, or even most importantly, losses in liberty. n199 The problem of demands is simple: Compliance with the optimizing principle of beneficence would have a terrible impact on the agent's resources (including time) and thus on her well-being. n200 It is true that if one were required to give all one's money away, this would greatly reduce one's positive freedom; but we must remember that if positive freedom matters, it matters just because it is an aspect of human well-being. The demands of utilitarianism fall heavily on all aspects of well-being, including the more mundane, but nevertheless central, goods of leisure and pleasure. Thus a claim that utilitarianism must be rejected because it interferes so heavily with freedom in particular must depend on the idea that the most important element of human well-being is the extent of the range of choices for action--either as an intrinsic good, or as a means to autonomy. Because this idea is so controversial, it is misleading to describe the costs to complying agents solely in the lofty language of liberty. The degree to which one makes one's own life surely does matter; but it is not the only thing that matters.

What, then, is to be done? If commitment to a duty to rescue brings with it a commitment to a general moral requirement of beneficence, and if the most  [*651]  straightforward general moral requirement of beneficence is the optimizing requirement of the utilitarians, it would seem that commitment to legal duties to rescue comes at the price of embracing the allegedly absurd demands of that requirement. What is needed, obviously, is an account of a general principle of beneficence that does not impose absurd demands. This, it turns out, is an extremely complicated matter. All I will do here is lay out some possibilities. If one of these approaches to the morality of beneficence is plausible, it may be possible to defend legal duties of beneficence without committing oneself to an implausible personal morality.

The first possibility is to hold tight with the utilitarians' optimizing principle of beneficence, extreme demands and all; with this approach we simply deny that a moral principle making such extreme demands is absurd and must be rejected. This extremist view is not unrepresented among philosophers. n201 A second (and more popular) view responds to the extreme demands of the optimizing principle of beneficence by embracing a nonoptimizing principle--one that gives out at a certain point when the demands get too great (we can call this a "limited principle of beneficence"). n202 Though natural, this response faces severe problems. n203 In the first place, Macaulay's worry about the basis on which the line is drawn remains. It is hard to come up with criteria for the limit to required beneficence that do not seem simply to track our natural tendency to think that the way we live now cannot be too bad. More fundamentally, it is very hard to know how we are to assess the demands of a moral requirement. What baseline should be used? In particular, how are we take into account the fact that, though I might be worse off if I take some beneficent action than if I do not, I am still very much better off, overall, because of the beneficence of my community that is expressed in the form of social and legal institutions that do me so much good? Once we take into account how much good comes to us from the compliance of other people it is hard to see any moral principle as demanding, in an overall sense, at all.

In the face of these problems, it is worth exploring a third possibility: The problem with the demands of the optimizing principle of beneficence is not that they are extreme, but that they are, in a special way, unfair. n204 For it is  [*652]  characteristic of the optimizing principle that in situations of partial compliance, where not everyone is acting as a good optimizer, those who are complying must take up the slack left by the others. Thus, a complying person may have to do more than she otherwise would not just because of natural disasters, for example, but also because others are not doing what they ought to do. This requirement that one shoulder not only one's own share of what should be a collective effort but also the shares of others can be regarded as unfair. If this is right, what we can call a "collective principle of beneficence," according to which people must sustain only that amount of sacrifice they would be required to sustain if everyone were doing their part, becomes attractive. n205

The collective principle of beneficence faces its own problems, of course. Most simply but also most fundamentally, not everyone agrees that the reason why the demands of the optimizing principle of beneficence are absurd is that they are unfair in situations of partial compliance. Even under full compliance, the optimizing principle can make very high demands, and this can still seem objectionable. At the same time, in some contexts doing one's fair share may not seem to be enough.

I have introduced the three responses to the problem of demands--the hold-tight utilitarian response, the response of the limited principle of beneficence, and the response of the collective principle of beneficence--not to defend one of them over the others, but to lay out our options. Fortunately, when we turn back to the question of the desirability of legal duties to rescue, each of the three moral views yields the same verdict: Rhode Island-style duties to rescue may be desirable but more extensive legal duties of beneficence are very unlikely to be.

IV. AN OVERLAPPING LEGAL CONSENSUS AMONG PLAUSIBLE MORAL VIEWS

A. A CRIMINAL DUTY

I favor the solution to Macaulay's problem of demands that is offered by the collective principle of beneficence. But this difficult issue of moral theory does not need to be finally resolved before we can justify some legal duties to rescue. As I have argued, legal theory must address the problem of demands as it emerges in moral theory because the fundamental principles we use to guide institutional design must also be plausible in the realm of personal conduct. Thus if there were no plausible alternative to the utilitarian's optimizing principle of beneficence, defense of Rhode Island's criminal duty to rescue would require the acceptance of that principle's truly radical implications in the personal realm. But there are alternatives to the optimizing principle, and at  [*653]  least one of them, the collective principle, seems to me to be plausible. And both of the two alternatives agree with the utilitarian verdict that legally required beneficence should be restricted to criminal duties to rescue with strictly limited demands. As luck would have it then, legal theory can to a certain extent side-step Macaulay's problems after all.

It is not hard to see that a general criminal duty of beneficence (one not limited to rescue contexts) would make no sense, even for utilitarians. n206 A general duty of beneficence requires the promotion of overall well-being. It is obvious enough that if the state has this aim, the best way to discharge it is by setting up institutional structures that transfer resources, provide education and health care, and other social services--not by criminally enforcing "a thousand points of light." Of course if the state does not promote general well-being, a moral principle of beneficence will require (better-off) people to do so. n207 But our question is now not what people should do, morally speaking; our question is whether there should be a legal requirement of general beneficence. On the view that the aim of the criminal law is to promote the social good, the state must obviously take a holistic view of its responsibilities--it should not try to achieve through the criminal law what can be better achieved in other ways. And on the legal moralist view too, if the state can in effect discharge people's responsibilities of beneficence for them via the tax and transfer system or other mechanisms, and indeed do a better job of it than people could themselves, then there is no call for enforcing individuals' duties of beneficence. n208 The general point is that overall well-being is best promoted through the organized and coordinated efforts of well-trained people. n209 This holds true for most emergency situations as well--that is why we have police forces and fire departments. The only kind of situation in which the involvement of the ordinary civilian is indispensable, even under an ideal set of institutions, is one calling for immediate attention by whomever happens to be both aware of the problem and able to act. n210 Often, all that is required in such cases is that the bystander summon trained experts, but because we cannot and do not want to have policemen on every corner, this summoning function is very important indeed.

These considerations suggest that criminal duties of beneficence should, like the Rhode Island provision, be limited to emergency scenes--to situations  [*654]  where action by a particular private citizen will make all the difference. But though emergencies are the only general type of case where the involvement of private individuals is indispensable, it is surely true that individuals can do a lot of good in other contexts too, even under ideal institutions. Should not a utilitarian therefore favor a mopping-up legal duty of beneficence--with the defense available that no liability attaches where a given beneficent course of action would not have been necessary if the state had been doing its beneficent job as well as is possible? Such a legal duty does not require people to do what could better be done by the state, but it is objectionable nevertheless for the way in which it would operate in practice. A mopping-up duty would put an inordinate informational burden on individuals, who would constantly need to be assessing whether there is good to be done that could not, ideally, be better done by the state. Restricting the application of the duty to cases in which a person has actual knowledge that some cost-justified benefit can be conveyed would reduce these burdens, but the determination of whether the benefit could, ideally, have been better handled by the state would still be extremely burden-some. n211 And a rule that is costly for each individual to comply with is also costly for the state to administer. These costs of deliberation and investigation, not to mention the costs of people's apprehension about possible failure to fulfill such an obtuse duty, must be taken into account by a utilitarian. n212 In addition, insofar as positive freedom contributes to a person's well-being, we must remember the presumption of liberty: All criminal duties come at the basic cost of curtailing people's range of choices. If these costs are outweighed by the benefits achieved, there is no utilitarian objection to the mopping-up duty. It seems very doubtful, however, that the aggregate costs would be less than the aggregate good such a criminal duty might achieve. n213

We thus have two considerations that govern the shape of desirable criminal duties of beneficence. Such a duty should apply only to situations where the involvement of individuals is indispensable; and it should specify such situations in advance, rather than requiring people to determine the matter for themselves. The duty to rescue is a criminal duty of beneficence that fits this bill. But is it the only such duty? Granting the need to limit the duty to emergency situations, what is the reason for further limiting it to situations that involve grave suffering or the threat of death? From a utilitarian perspective,  [*655]  there is no magic in the magnitude of the benefit conveyed on each occasion. Whether the legal duty of beneficence should be extended to cover emergency cases involving moderate need, or even to situations where the destruction of valuable property can be prevented, depends only on whether the extra benefits of such an extension outweigh any additional burdens it would bring. It seems to me that this test would not be met because the extension would yield a great benefit only if it applied to a large number of cases. n214 But even if it did apply to a large number of cases, the additional administrative costs and costs to positive liberty would likewise be large. Thus, the net benefits of extending the duty seem likely to be small at best. n215 Any assessment of the benefits and burdens of a particular legal rule is in large part speculative; a sensible rule of thumb would be that the costs of legal regulation should be incurred only where the expected net benefit is substantial. The reason for moderate confidence in the overall net benefits of the duty to rescue is that it has the potential to bring out a very great net improvement in overall well-being in each situation to which it applies.

So far we have seen that practical considerations support limiting legal duties of beneficence to rescue contexts even if we accept the optimizing principle of beneficence. These considerations apply mutatis mutandis to our two alternative principles of beneficence. Matters are less straightforward when it comes to the issue of demands. As we saw, rescue cases can give rise to extreme demands even where the expected benefit of the rescue is very great. Should the criminal law require a person to take a fifty-percent chance of losing her life for the sake of a fifty-percent chance of saving twenty? To address this question, we need to take each of our three principles of beneficence in turn. If we take a legal moralist approach to the criminal law or accept the constraint against legally required sacrifice beyond what a person is in any case morally required to sustain, then the upper bound of the legitimate demands of legally required beneficence will lie wherever the relevant moral principle of beneficence puts it. n216

 [*656]  Let us again start, then, with the implications of the optimizing principle of beneficence. Like most other actual legal duties to render aid, the Rhode Island provision requires only "reasonable" assistance, and specifies that this will never involve danger or peril to the rescuer. n217 As a matter of statutory interpretation, it is not plausible to read this latter limitation such that a rescue attempt is required only if it would involve no risk whatever of physical harm. Even walking to the edge of the dock to toss the life preserver entails some risk to my life (I could slip and fall off, hitting my head on the rocks). In any event, a sensible legal duty to rescue will require rescue attempts so long as only a minimal risk of serious harm is involved. n218 This means that the risk that a good swimmer will get a cramp and drown if she swims out into the calm lake to rescue a floundering child would not be sufficient to justify inaction. But an average swimmer (as opposed to an off-duty, professional surf lifeguard) should not be required to head off into rough surf to attempt to rescue someone caught in a rip tide.

From a utilitarian perspective, it is clear enough that people who have no special training, and thus cannot count on any substantial probability of succeeding in a rescue attempt, should not be encouraged, let alone required by the criminal law, to take substantial risks with their own lives. n219 Of course, it may be that a particular boat owner can count on roughly a fifty-percent chance of success, and so, if there are twenty people to be saved, the optimizing principle of beneficence would, as a moral matter, require her to take the chance. The utilitarian legislator's first-best result would be a provision that requires such a person to act without encouraging dangerous rescue attempts by the rest of us. It is not clear, however, that this can be done. A provision could specify that rescue attempts are required so long as "the effort, risk, or cost of acting is disproportionately less than the harm or damage avoided." n220 Whether a particular person is liable for nonrescue would then turn on an investigation of her actual chances of success and (perhaps) her awareness of those chances. But though courts in tort cases deal routinely with the idea that more can be expected of some people than of others, such an approach seems very unwise in the current context. As noted, it is not for the best to encourage people with no particular skills to take unknown (to them) levels of risk with their own lives; yet, the suggested language would do just that. It would be possible to require particular classes of rescuers to sustain a higher level of risk, but this also seems  [*657]  unnecessary because the relevant classes of people are likely already to be on the job when the opportunity to attempt a rescue arises. n221

So much for the risk of physical harm. Equally difficult is the issue of all other kinds of loss. The drafters of the Rhode Island provision take the standard approach of not specifying what "reasonable" non-physical cost might amount to. Is this what utilitarianism recommends? A utilitarian rescuer must weigh the value of the chance of saving a life or preventing or alleviating severe suffering against some very different kind of value, such as a simple loss of money. A utilitarian lawmaker understands that this kind of cost-benefit analysis cannot reliably be done on the spot. Because the criminal law should not encourage wasteful rescue expenditures, the utilitarian lawmaker aims to encourage potential rescuers to attempt a rescue only in those cases where the expected benefit clearly outweighs the expected loss. Because it is doubtful that any useful specific criteria could be extracted from the range of possible cases and written into the law, this aim could not be better served than by asking the trier of fact to determine whether "reasonable" efforts were used. n222 Even if further criteria were available, making the duty more complicated would greatly weaken its educative/expressive role, which may be more important than its role as a deterrent. The presence of such a duty in the criminal law reminds us of our responsibilities toward other people, even total strangers, and the more qualified the duty, the less powerful this message. n223

Triers of fact can get things badly wrong, and potential rescuers can know this. The loose reasonableness standard, coupled with ex ante mistrust of triers of fact, could thus seem to encourage rescue efforts that are not cost-justified. This potential problem provides a further reason to limit the duty to rescue to cases of severe suffering or threat to life, because in these cases it is hard for an individual to expend too much in the way of material resources. Moreover, in the criminal context the prosecution's high burden of proof should allow people to be less mistrustful of triers of fact than may be warranted in the civil context. n224

It is clear that even an ideal trier of fact might decide that a person ought reasonably to permit the ruin of his million dollar yacht in order to rescue some drowning people, and utilitarians would have no grounds for regret. n225 Thus,  [*658]  even if I am right that utilitarianism would recommend a legal duty to make only reasonable efforts, this may still leave us with a legal duty of beneficence that some will find too demanding.

As it happens, there is a utilitarian answer to this problem. Saul Levmore has plausibly argued that the most efficient incentive package for legal duties to rescue combines a (small) sanction for failure to rescue with a small reward for rescuing as required. n226 By a small reward, Levmore means one that does no more than compensate the rescuer for her expenses. n227 Such is the arrangement in France and Germany. n228 The upshot of this result is that even in those cases where reasonable rescue is very costly for the rescuer, an optimal legal regime would reimburse those costs. Most states' crime victims' compensation statutes already provide for compensation for rescuers in the context of violent crimes. n229 And in the non-crime context, existing principles of restitution may already provide for such compensation where the rescue is carried out pursuant to a legal duty such as the Rhode Island provision. n230 If so, providing for compensation would require no further legal innovation--unless it was thought, as seems plausible, that the compensation should flow not from the person rescued, but from the community at large by way of a statutory compensation scheme. n231

 [*659]  I will not rehearse the details of Levmore's argument about the optimal size of the rescue "carrot"; but we should briefly address the issue of the rescue "stick"--the penalty for violation. On a legal moralist approach, the penalty for violation should depend on our judgment of the severity of the moral wrong done by the person who fails to assist. n232 One of the problems with the commission by omission approach to failure to save a life is that it makes it impossible for the law to be sensitive to the difference between the moral wrongness of killing versus letting die--a commonsense distinction that many moral philosophers take very seriously. n233 A statutory duty to rescue, however, can treat letting die as far less serious than homicide, and existing provisions do just that. The maximum penalty for failure to assist in Vermont is a fine of $ 100, Rhode Island's maximum of $ 500 and six months imprisonment is at the high end in the common-law world, n234 and the norm in Europe and Latin America is for a penalty of either a fine or no more than a year's imprisonment. n235

But we do not have to be retributivists to have a reason to distinguish between the sanctions appropriate for positive harming as opposed to failure to aid. Though the incentives for failing to aid can be great, as in the case of the Princess of Wales, it is generally true that the payoff available from doing  [*660]  positive harm will be much greater--simply because one can do harm all on one's own and thus choose to harm where it is most profitable. Similarly, a person motivated by great hatred and anger will rarely be able to satisfy his desire by merely letting harm befall someone. Thus, in a deterrence analysis, the sanction for failing to harm should generally not be as severe as the sanction for harming. Indeed rather minimal sanctions would seem to be adequate, and this is what we observe. n236

It appears then that if we accept the optimizing principle of beneficence as part of our account of the aim of the criminal law, we will nevertheless restrict criminal duties of beneficence to duties to rescue with minimal demands and minimal sanctions for violation. I have argued, however, that such a principle of beneficence is very implausible in the realm of personal conduct. Thus we must examine whether our two alternative principles of beneficence, the limited principle and the collective principle, also support this kind of minimal legal duty to rescue. Of course neither principle will support a legal duty to rescue that is more demanding than the utilitarian duty. What needs to be shown, to satisfy legal moralists and those who embrace a constraint against imposing sacrifice beyond what a person is morally required to make independently of law, is that this minimal legal duty to rescue is not more demanding than is compatible with the collective and limited principles of beneficence.

As a matter of personal morality, the operation of the limited principle and the collective principle in rescue contexts turns out to be a very complicated matter. n237 But the main point is simple: Because rescue contexts are not normatively special, and because under both principles the sacrifice required for the sake of benefiting others is not unlimited, there will be some situations in which these principles will not require a person to perform even a very easy and cheap rescue that is certain to succeed. This may seem to ruin the case for legal duties to rescue; the only principles of beneficence that seem to have a hope of being plausible generally do not in the end support a blanket legal requirement to rescue, no matter how minimal.

Here again the option of compensation makes all the difference. Remembering that more-than-minimal risks of physical harm would not be required, even by a legal duty based on the optimizing principle of beneficence, the provision for compensation of material expenses means that those who comply with the legal duty to rescue sustain no significant net sacrifice at all; the problem of demands is, in the end, sidestepped at the level of institutional design. n238

 [*661]  Thus plausible principles of beneficence do appear to converge in recommending legal duties to rescue that require minimal risk-taking and that provide for both compensation and a moderate sanction for violation. n239 Needless to say, this conclusion is not irresistible. For one thing, I have not considered the possibility that there are ancillary bad effects of criminal duties to rescue that, together with the costs already discussed, outweigh the good these duties would do. This does not seem to me to be likely, but I will not investigate the issue here. n240 As already noted, I do not pretend to have conclusively established that the benefits of a legal duty to rescue substantially outweigh even the costs so far considered. All these arguments involve speculation; their conclusions can only be regarded as more or less plausible. My main aim has been to uncover the structure of the legal, moral/political, and practical issues that underlie the debate about the duty to rescue and to outline some plausible options available  [*662]  to us. And what I do claim to have shown is that if there is a reason to oppose all criminal duties to rescue, it will not be because such duties are offensive to plausible considerations of political morality. So long as compensation is provided, all plausible principles of beneficence support such duties and no plausible arguments grounded in individual liberty count against them. Rather, it will be because an exhaustive evaluation of the benefits and burdens of criminal duties to rescue leaves us insufficiently confident that such duties yield a clear net benefit. That almost all the countries of Continental Europe and Latin America should be wrong about this practical issue, and the handful of common-law countries right, seems to me prima facie dubious. n241 But it could be so.

B. NO TORT DUTY

The reasons for thinking that tort liability for failure to aid is less desirable than some form of criminal liability can be quickly stated. n242 The advantage of criminal liability lies in the flexibility of the sanction. If we could assume that no one would ever violate it, a tort duty to make cost-justified rescue attempts would be unobjectionable from the point of view of social costs. n243 But of course, some people would violate such a civil duty to rescue, irrational though that may be. n244 And such tortfeasors would be liable for full compensatory damages. There are two problems with this.

Just as one does not have to be a full-fledged retributivist to believe that some criminal sanctions that would be appropriate on a pure deterrence rationale must be rejected as too severe (no matter how the deterrence calculations come out, drawing and quartering would not be an acceptable punishment for tax dodgers), one need not deny that the aim of tort law is the promotion of social good in order to believe that legitimate tort liability has its limits. Indeed, whatever the correct account of the normative basis of tort law might be, the prospect of a  [*663]  person who failed to respond to an emergency situation being liable for damages in full compensation for the death of the victim seems likely to be highly unappealing. n245 The recent Australian decision of Lowns v. Woods n246 awarding three million Australian dollars in damages against a doctor whose nonfeasance led to a child (not his patient) becoming permanently disabled may seem tolerable only because the defendant was a physician--protected, no doubt, by liability insurance.

Apart from considerations of proportionality, there is the simple fact that, assuming that the aim of tort law is to promote general well-being, full compensatory liability may well be excessive from an instrumental perspective. First, the availability of either rescue or (at least a chance of) full compensation may encourage foolish and highly risky activities by potential rescuees. n247 Second, as full compensatory liability is almost certainly more than is required to deter nonrescue for most people, the costs incurred in bringing about the transfer of the loss from a victim to a nonrescuer are costs without benefit. Which brings out the main point: Unlike civil liability, criminal sanctions can be adjusted to achieve the desired deterrence effect. Once criminal liability is in place, there is simply no need for additional civil liability. n248 In the absence of any grounds of corrective justice for requiring the nonrescuer to compensate the victim--an absence that is pronounced in contemporary theories of corrective justice n249 --the only possible ground for imposing tort liability would be that criminal liability has special disadvantages that outweigh its immeasurably greater flexibility. n250

It is notable that the conclusion we have reached, in favor of limited criminal duties to rescue but wary of civil liability, has some support from the situation in Continental Europe, where civil liability for failure to rescue is much less common than criminal liability. n251

CONCLUSION: LAW, MORALITY, AND STATE

Criminal duties to render aid of the kind that are now increasingly popular in the United States face no objection on the ground of their interference with liberty, and moreover may be desirable, all things considered. The problem of finding a plausible moral principle of beneficence must be addressed by defenders  [*664]  of legally required beneficence but it need not be finally solved, because it turns out that all the plausible options would support appropriately limited criminal duties to rescue. The reasons why more extensive legal duties of beneficence are not appropriate on any of the plausible views about the morality of beneficence are practical: State institutions can do a better job of promoting well-being than can individuals in all but emergency contexts, and even in those contexts there are good reasons not to legally require people to sustain a substantial risk of death or injury attempting to aid others.

In conclusion, I would like to make two observations that turn on the fact that the reasons why only strictly constrained legal duties of beneficence are defensible and are, indeed, essentially practical. First, we may have here one of those areas of normative thought where the law misleads us about political morality. The fact that it would be absurd to enact a criminal duty to promote social welfare generally may lead us to conclude, wrongly, that the duties people have in respect of beneficence are minimal or optional, especially outside rescue contexts. This would be a serious mistake. n252 It is true that when we think about what kind of criminal duty of beneficence would make sense we must consider how the state might better promote welfare via some other institutional means: In an ideal society, very little may end up being the responsibility of individuals, especially not on pain of criminal sanction. But it is also true that if the state fails to set up the appropriate institutions, then moral requirements of beneficence apply directly to people, regardless of the law. The fact that an ideal state would make it unnecessary for people to be that much concerned, in their daily lives, about social welfare does not mean that we need not be concerned about social well-being in our actual, nonideal situation. And so, as I have said, the fundamental normative principles to which we appeal when trying to describe ideal legal institutions had better be ones we can live with in our actual, nonideal world.

The second observation returns us to the issue of the positive responsibilities of government. If I am right, there is no fundamental objection of political morality to the very idea of positive legal obligations. But notice the difference between the case of people and that of government. In the case of people, we saw that the problem of finding a reasonable moral principle of beneficence did not have to be finally resolved by legal theory because all the candidate principles yielded the same recommendation on the issue of legally required  [*665]  beneficence. When we turn to states, however, the practical reasons for accepting only strictly constrained legal requirements of beneficence disappear. At least, the nature of the practical arguments changes dramatically, and the practical potential for the state to do good is obviously vastly greater than is the case for individuals. Thus, when we are addressing the responsibilities of states, the question of which principle of beneficence we accept as a matter of political morality cannot be avoided. Whether the state's duties of beneficence to its members should be constitutionalized is another matter. But it must not be pretended that the fundamental question of what a decent community must do to promote the well-being of its members need not be reached--either because there is some intrinsic incompatibility between the idea of positive state obligation and the basic values of our legal culture, or because, practically speaking, putting beneficence into the law would never make any sense.

And so we see that acceptance of legal duties to rescue, and thus of some requirement of beneficence, does indeed open the can of worms that some commentators have been so keen to keep closed. n253

FOOTNOTES:
n1 For the common law, see discussion infra Part I. For Continental Europe and Latin America, see Alberto Cadoppi, Failure to Rescue and the Continental Criminal Law, in THE DUTY TO RESCUE: THE JURISPRUDENCE OF AID 93 (Michael A. Menlowe & Alexander McCall Smith eds., 1993); id. at 97-100 (providing a brief history of legal duties to rescue in ancient Mediterranean and European cultures and in modern Europe); F.J.M. Feldbrugge, Good and Bad Samaritans: A Comparative Survey of Criminal Law Provisions Concerning Failure to Rescue, 14 AM. J. COMP. L. 630, 632-34 (1966); Gilbert Geis, Sanctioning the Selfish: The Operation of Portugal's New 'Bad Samaritan' Statute, 1 INT'L REV. VICTIMOLOGY 297, 297-300, 301-02 (1991); Aleksander W. Rudzinski, The Duty to Rescue: A Comparative Analysis, in THE GOOD SAMARITAN AND THE LAW 91 (James Ratcliffe ed., 1966); Martin Vranken, Duty to Rescue in Civil Law and Common Law, 47 INT'L & COMP. L.Q. 934, 938-39 (1998).

n2 Utilitarianism, which holds that people are required to promote overall well-being (either of humans or of all sentient life) is thus one form of consequentialism, the theory that holds that people are required to optimize the overall good (on some account of the good, which may include more than well-being). The version of utilitarianism popular among economists is narrower still, as it understands human well-being exclusively in terms of preference satisfaction. This preference-satisfaction version of utilitarianism should not, in turn, be confused with either the Pareto or Kaldor-Hicks criterion of efficiency; these efficiency criteria are employed by economic analysts of law in an attempt to guide institutional design without making the interpersonal comparisons of utility required by any version of utilitarianism. See, e.g., JULES COLEMAN, MARKETS, MORALS AND THE LAW 95 (1988).

n3 See infra Part I.

n4 THOMAS BABBINGTON MACAULAY, Notes to the Indian Penal Code, in THE WORKS OF LORD MACAULAY 314-15 (Lady Trevelyan ed., 1900).

n5 Id. at 315-16; 2 EDWARD LIVINGSTON, Code of Crimes and Punishments, in COMPLETE WORKS 126-27 (1873).

n6 "In cases where the person is in danger, why should it not be made the duty of every man to save another from mischief, when it can be done without prejudicing himself, as well as to abstain from bringing it on him?" JEREMY BENTHAM, THE PRINCIPLES OF MORALS AND LEGISLATION ch. XVII, 1.XIX (1789). In a footnote, Bentham offers these excellent examples:

A woman's head-dress catches fire: water is at hand: a man, instead of assisting to quench the fire, looks on, and laughs at it. A drunken man, falling with his face downwards into a puddle, is in danger of suffocation: lifting his head a little on one side would save him: another man sees this and lets him lie. A quantity of gunpowder lies scattered about a room: a man is going into it with a lighted candle: another, knowing this, lets him go in without warning. Who is there that in any of these cases would think punishment misapplied?

Id. at 323 n.1.

n7 See infra Part I. Cadoppi very plausibly suggests that a significant factor is the lack of large scale codification. Cadoppi, supra note 1, at 115-16.

n8 The strange juxtaposition of the popularity of utilitarianism and the unpopularity of positive legal duties is illustrated by James Barr Ames in his well-known 1908 lecture. See James B. Ames, Law and Morals, 22 HARV. L. REV. 92 (1908), reprinted in THE GOOD SAMARITAN AND THE LAW, supra note 1, at _. Ames, having declared roundly and without argument that "the law is utilitarian," id. at 17, and having noted that a duty to rescue a stranger from peril, even at little cost, is unknown in the common law, has no trouble concluding that the law ought not "remain in this condition," id. at 20.

n9 See Robert L. Hale, Prima Facie Torts, Combination, and Non-Feasance, 46 COLUM. L. REV. 196 (1946). Hale writes:

Perhaps judicial reluctance to recognize affirmative duties is based on one or both of two inarticulate assumptions. One of these is that a rugged, independent individual needs no help from others, save such as they may be disposed to render him out of kindness, or such as he can induce them to render by the ordinary process of bargaining, without having the government step in to make them help. All he is supposed to ask of the government is that it interfere to prevent others from doing him positive harm. The other assumption is that when a government requires a person to act, it is necessarily interfering more seriously with his liberty than when it places limits on his freedom to act--to make a man serve another is to make him a slave, while to forbid him to commit affirmative wrongs is to leave him still essentially a free man.

Id. at 214. A recent articulation of the second of these assumptions, from high common-law authority, is found in Lord Hoffmann's speech in Stovin v. Wise:

There are sound reasons why omissions require different treatment from positive conduct. It is one thing for the law to say that a person who undertakes some activity shall take reasonable care not to cause damage to others. It is another thing for the law to require that a person who is doing nothing in particular shall take steps to prevent another from suffering harm from the acts of third parties . . . or natural causes. One can put the matter in political, moral or economic terms. In political terms it is less of an invasion of an individual's freedom for the law to require him to consider the safety of others in his actions than to impose upon him a duty to rescue or protect.

1996 C.A. 923, 941 (appeal taken from Court of Appeal). For Lord Hoffmann's "moral" point, see infra note 84.

A nonjudicial source on the same issue is Sheldon Richman:

The American philosophy of jurisprudence doesn't permit the government to impose positive obligations on citizens. America was founded on a bedrock of inalienable individual rights. Under that theory, each person is the owner of his life and has no positive legal obligations to others that are enforceable by government except those that are voluntarily accepted. All that government can require of you is that you abstain from violating the rights of others by subjecting them to force or fraud. Thus, even if you believe that morality requires you to help an accident victim, it's not enforceable by government . . . . People might find Good Samaritan laws reasonable because they believe people of good will should help others in distress. But where individual rights are respected and government power is limited, good will cannot be enshrined in the law. It would undermine freedom.

Sheldon Richman, You Can't Legislate Goodwill, CHRISTIAN SCI. MONITOR, Oct. 2, 1997, at 19 (author is identified as "vice president of policy affairs for the libertarian Future of Freedom Foundation in Fairfax, Virginia"); see also Eric Mack, Bad Samaritanism and the Causation of Harm, 9 PHIL. & PUB. AFF. 230 (1980) ("In our law, there's no duty to rescue someone or save someone's life . . . . Our society is based on the right and sanctity of the individual." (quoting Allegheny Court, Pennsylvania Judge John P. Flaherty in McFall v. Shimp, 10 Pa. D. & C.3d 90 (Allegheny County 1978))). For further discussion, see infra Part II.

n10 In seeing this issue of costs or demands as the central normative problem associated with legal duties to rescue, I am in agreement with Michael Menlowe. See Michael A. Menlowe, The Philosophical Foundations of a Duty to Rescue, in THE DUTY TO RESCUE: THE JURISPRUDENCE OF AID, supra note 1, at 5. Another discussion similar in approach to my own is Samuel Freeman's. See Samuel Freeman, Criminal Liability and the Duty to Aid the Distressed, 142 U. PA. L. REV. 1455 (1994).

n11 See infra Part III.

n12 The expression "overlapping consensus" is John Rawls's. See JOHN RAWLS, POLITICAL LIBERALISM (1993).

n13 See Richard A. Epstein, Rights and "Rights Talk," 105 HARV. L. REV. 1106, 1118 (1992) ("Imposing affirmative legal duties necessarily increases the total level of coercion, public and private, in society. The risk of excessive government power should not be run because, in the good Samaritan context, self-help remedies are usually effective."); Marc A. Franklin & Matthew Ploeger, Of Rescue and Report: Should Tort Law Impose a Duty to Help Endangered Persons or Abused Children?, 40 SANTA CLARA L. REV. 991, 1004 (2000); William M. Landes & Richard A. Posner, Salvors, Finders, Good Samaritans, 7 J. LEGAL STUD. 83, 127 (1978); Melody J. Stewart, How Making the Failure to Assist Illegal Fails to Assist: An Observation of Expanding Criminal Omission Liability, 25 AM. J. CRIM. L. 385, 432 (1998).

It is true that reported cases are rare, to put it mildly. The only reported case concerning a prosecution solely under a duty to assist provision I am aware of is State v. La Plante, 521 N.W.2d 448, 452 (Wis. Ct. App. 1994) (upholding constitutionality of provision imposing duty to assist crime victims).

n14 It is, of course, not the case that there are no positive legal duties of any kind in common law jurisdictions. See infra Part I. The question is whether there is a general normative reason to treat their introduction with caution.

n15 See Jackson v. City of Joliet, 715 F.2d 1200 (7th Cir. 1987) (Posner, J.); see also DeShaney v. Winnebago County Dep't of Soc. Servs., 489 U.S. 189 (1989). For critical discussion, see Susan Bandes, The Negative Constitution: A Critique, 88 MICH. L. REV. 2271 (1990); Frank I. Michelman, The Supreme Court, 1968 Term--Foreword: On Protecting the Poor Through the Fourteenth Amendment, 83 HARV. L. REV. 7, 13-14 (1969) (arguing that the government has an affirmative obligation under the Constitution to provide assistance to the badly-off). For discussion of positive duties under state constitutions, see Helen Hershkoff, Positive Rights and State Constitutions, 112 HARV. L. REV. 1131 (1999). I do not in this Article attempt to engage with the further legal issues raised when we move from the case of individuals to that of government. For discussion of some of those issues, see generally Barbara Armacost, Affirmative Duties, Systemic Harms, and the Due Process Clause, 94 MICH. L. REV. 982 (1996); David Currie, Positive and Negative Constitutional Rights, 53 U. CHI. L. REV. 864 (1986); Hershkoff, supra.

n16 Richard Epstein writes:

The fundamental problem in a system of welfare is that it conflicts with the theory of private rights that lies behind any system of representative government . . . . The legal theory that recognizes no obligation to rescue a stranger in imminent peril cannot generate, let alone nourish, a system of transfer payments and welfare obligations.

RICHARD EPSTEIN, TAKINGS 318-19 (1985). The connection with the takings clause is explained by Epstein. Id. at 322 ("The basic rules of private property are inconsistent with any form of welfare benefits."). Epstein has recently recast his earlier, rights-based opposition to duties to rescue in utilitarian terms. See infra text accompanying note 163. This wholly undermines the direct argument from the lack of a duty to rescue to the rejection of transfer payments because a utilitarian will have to evaluate each issue on its merits. See Symposium, Richard Epstein's Takings: Private Property & the Power of Eminent Domain, 41 U. MIAMI L. REV. 49, 179-91 (1986) [hereinafter Richard Epstein's Takings]. Epstein does, however, make broadly utilitarian points against the public welfare system in Takings. EPSTEIN, supra, at 320-23.

n17 See Theodore M. Benditt, Liability for Failing to Rescue, 1 LAW & PHIL. 391, 396 (1982). In addition to the effect on thinking about the legal duties of the State, there is the effect law-school discussion of the issue of the duty to rescue may have on the attitudes of public officials. See MARY ANN GLENDON, RIGHTS TALK: THE IMPOVERISHMENT OF POLITICAL DISCOURSE 80 (1991).

n18 See, e.g., Epstein, supra note 13, at 1118.

The recent admitted actions and subsequent statements of Mr. David Cash may make such remarks less common in the future. David Cash admitted seeing his friend Jeremy Strohmeyer begin the assault on a seven year old girl that eventually led to her death in a Nevada casino on May 25, 1997. Mr. Cash is reported to have said: "I'm not going to get upset over somebody else's life. I just worry about myself first." Don Terry, Mother Rages Against Indifference, N.Y. TIMES, Aug. 24, 1998, at A1. This case provided the motivation for Nevada's duty-to-report provision referred to infra note 23.

n19 See generally A.M. ROSENTHAL, THIRTY-EIGHT WITNESSES (1964).

n20 See, e.g., Marlise Simons, French Magistrates Clear Photographers in Death of Diana, N.Y. TIMES, Sept. 4, 1999, at A2. Mohammed al-Fayed lost his first appeal of the decision to drop the charges but has announced his intention to take his appeal to the Superior Court. Al-Fayed to Appeal Ruling over Death of Diana and Dodi, THE HERALD (Glasgow, Scotland), Nov. 1, 2000, at 14.

n21 C. PEN. art. 223-6. The sanction provided for in this provision is a good deal harsher than similar provisions in Continental Europe and Latin America--the norm is a fine and not more than one year's imprisonment. The reason for the exception is that the French penal code is unique in having no separate provision providing for offenses of commission by omission. See Cadoppi, supra note 1, at 109. See generally infra Part I.A. The French provision was part of a proposed reform of the French penal code in 1934 but was first enacted by the Vichy regime in 1941; it was included in the Penal Code in 1945. See Andrew Ashworth & Eva Steiner, Criminal Omissions and Public Duties: The French Experience, 10 LEGAL STUD. 153, 156-57 (1990). Ashworth and Steiner provide a helpful summary of the main types of cases in which the provision has been invoked. Id. at 158-60.

n22 "It is obvious that Diana, Princess of Wales, Dodi Al Fayed, Henri Paul, and Trevor Rees Jones were no safer from their peril in Paris with its duty-to-assist law than they would have been in any place without such a law." Stewart, supra note 13, at 435.

n23 Currently, duty-to-aid provisions are in force in Vermont, VT. STAT. ANN. tit. 12, 519 (1973), Minnesota, MINN. STAT. 604.01(a) (1998), and Rhode Island, R.I. GEN. LAWS 11.56-1 (1994). Hawaii, HAW. REV. STAT. ANN. 663-1.6 (Michie 1995), Massachusetts, MASS. GEN. LAWS ANN. ch. 268, 40 (West 1990), Washington, WASH. REV. CODE ANN. 9-69.100 (West 1998), and Wisconsin, WIS. STAT. ANN. 940.34 (West 1996 & Supp. 1997), have provisions imposing duties either to aid crime victims or report an ongoing crime to the authorities. Florida, FLA. STAT. ANN. 794.027 (West 1992 & Supp. 1998), and Nevada, NEV. REV. STAT. 202-882 (1999), have duty-to-report provisions restricted to certain sexual offenses. In recent years, duty-to-aid bills have been considered in Hawaii, H.R. 885, 20th Leg. (Haw. 1999) and Illinois, H.R. 2612, 91st Gen. Ass. (Ill. 1999). A bill that would enact a duty to report the witnessed sexual assault of a child is still, at the time of writing, under consideration in New York, Senate 12, 223d Leg. (N.Y. 2001). For an account of the incident that led to the Washington bill, see Stewart, supra note 13.

n24 113 N.W. 1128 (Mich. 1907).

n25 Id. at 1131.

n26 Graham Hughes, Criminal Omissions, 67 YALE L.J. 590, 624 (1958).

n27 Hughes takes this expression from the French, see id. at 598; it is now widely used, see, e.g., GEORGE FLETCHER, RETHINKING CRIMINAL LAW 422 (1978).

n28 See, e.g., N.Y. VEH. & TRAF. LAW 600 (McKinney 1996) ("hit-and-run" statute). See generally infra notes 45-46 and accompanying text.

n29 Beardsley, 113 N.W. at 1129. The Model Penal Code provides for liability for omissions in two circumstances: (1) where the law defining an offense so provides; or (2) where the duty to act is "otherwise imposed by law." MODEL PENAL CODE 2.01(3)(b); see also JOSHUA DRESSLER, UNDERSTANDING CRIMINAL LAW 91 (1995); WAYNE R. LaFAVE & AUSTIN SCOTT, JR., CRIMINAL LAW 202 (1986).

n30 See, e.g., LaFAVE & SCOTT, supra note 29, at 203.

n31 JEROME HALL, THE GENERAL PRINCIPLES OF THE CRIMINAL LAW 193-94 (1960).

n32 Id.; see also Hughes, supra note 26, at 620.

n33 See LaFAVE & SCOTT, supra note 29, at 204-05.

n34 See id. at 205. George Fletcher makes the plausible point that in many cases it is not so much the contract (in the technical sense of an agreement enforceable under contract law) as the undertaking to provide assistance that creates the relevant duty. FLETCHER, supra note 27, at 615-16.

n35 See, e.g., State v. Miranda, 715 A.2d 680 (Conn. 1998) (upholding conviction for assault in the first degree on basis of defendant's omission to protect child of live-in girlfriend from abuse).

Note that not all the duties--omission of which may give rise to criminal liability for a crime of commission--are in themselves sufficient to ground tort liability. See W. PAGE KEETON ET AL., PROSSER AND KEETON ON THE LAW OF TORTS 376 (5th ed. 1984).

n36 Hughes, supra note 26, at 620.

n37 Id. at 621.

n38 Id.

n39 See FLETCHER, supra note 27, at 628-31; PAUL H. ROBINSON, CRIMINAL LAW 195-99 (1997); George P. Fletcher, On the Moral Irrelevance of Bodily Movements, 142 U. PA. L. REV. 1443, 1449 (1994) [hereinafter Fletcher, On the Moral Irrelevance]. See generally Douglas Husak, Does Criminal Liability Require an Act?, in PHILOSOPHY AND THE CRIMINAL LAW 60 (Antony Duff ed., 1998).

n40 Fletcher, On the Moral Irrelevance, supra note 39, at 1449. The problem of legality is said to be one reason why French law allows liability for commission by omission in only a very narrow range of cases. See Ashworth & Steiner, supra note 21, at 155.

n41 See Cadoppi, supra note 1, for a recent list of European and Latin American provisions, along with translations. For U.S. provisions, see supra note 23. For Australia's Northern Territory, see Criminal Code Act 155 (N. Terr. Austl.), which provides for a sanction of up to seven years imprisonment for a "callous" failure to rescue. See also infra note 234.

Steven Heyman argues that in addition to the offense of "misprision of felony" (failing to report a felony), the early common law imposed a criminal duty to prevent felonies. See Steven J. Heyman, Foundations of the Duty to Rescue, 47 VAND. L. REV. 673, 685-90 (1994).

n42 R.I. GEN. LAWS 11-56-1 (1994).

n43 MINN. STAT. 604.01(a) (1998); VT. STAT. ANN. tit. 12, 519 (1973).

n44 See, e.g., N.Y. FAM. CT. ACT 413 (McKinney 1998).

n45 I.R.C. 6012 (West Supp. 1999).

n46 Commonwealth Electoral Act, 1918, 101 (Austl.).

n47 See LaFAVE & SCOTT, supra note 29, at 193.

n48 See Hughes, supra note 26, at 590-97 (history of criminal omissions); see also Heyman, supra note 41, at 682-90.

n49 See generally MICHAEL S. MOORE, ACT AND CRIME: THE PHILOSOPHY OF ACTION AND ITS IMPLICATIONS FOR CRIMINAL LAW (1993); Husak, supra note 39.

In what follows, I treat the act requirement as distinct from the requirement of voluntariness. See generally DRESSLER, supra note 29, at 71-77.

n50 See John Kleinig, Criminal Liability for Failures to Act, 49 LAW & CONTEMP. PROBS. 161, 162-67 (1986). Contra Christopher H. Schroeder, Two Methods for Evaluating Duty to Rescue Proposals, 49 LAW & CONTEMP. PROBS. 181, 183-84 (1986). See generally MICHAEL MOORE, PLACING BLAME 262-77 (1997) (listing references).

n51 For a different approach that turns on the notion of control, see Husak, supra note 39.

n52 MODEL PENAL CODE 2.01(1). For discussion see DRESSLER, supra note 29, at 69-71 (though the tone of Dressler's discussion of omissions suggests that he is attracted to Moore's approach, discussed infra notes 53-55 and accompanying text); LaFAVE & SCOTT, supra note 29, at 195; ROBINSON, supra note 39, at 175-87.

n53 See MOORE, supra note 49, and his replies to critics in Placing Blame, MOORE, supra note 50, chapter six and accompanying references. See also Husak, supra note 39.

One of Moore's critics, Bernard Williams, doubts that much of substantial relevance to the criminal law could turn on the philosophy of action. Bernard Williams, The Actus Reus of Dr. Caligari, 142 U. PA. L. REV. 1661 (1994). I agree. For the related discussion of causation, see infra text accompanying notes 56-59. Part of Moore's response to Williams in Placing Blame, MOORE, supra note 50, at 253, is that his argument starts with the moral claim that acts are central to criminal liability, and only then proceeds to the investigation of the essence of acts. But it is a strange view that asserts the moral significance of acts--whatever they on further investigation may turn out to be. How do we know whether they are morally significant if we don't know what they are? Note that Moore's argument is different from Derek Parfit's use of metaphysics in normative theory. Parfit believes that his discussion of the metaphysics of personal identity has relevance for ethical theory, but here the argument proceeds by showing that the question of personal identity (what makes me, me, and you, you), which is central to a range of ordinary moral views--and which in ordinary life we think we understand well--turns out to be badly understood, resulting in the revision of some of our ethical views. See DEREK PARFIT, REASONS AND PERSONS pt. 3 (1984). Moore's argument is different in that we do not, in ordinary life, operate with an uncontroversial account of what is an act, and (not surprisingly given the first point) the normative relevance of acts to criminal liability and punishment is itself not uncontroversial.

n54 See MOORE, supra note 49, at 34 ("Perhaps . . . [the] criminal law is mistaken.").

n55 See, e.g., John Harris, The Marxist Conception of Violence, 3 PHIL. & PUB. AFF. 192 (1974); John Kleinig, Good Samaritanism, 5 PHIL. & PUB. AFF. 382 (1976); Arthur Leavens, A Causation Approach to Criminal Omissions, 76 CAL. L. REV. 547 (1988). Alison McIntyre provides an excellent critical discussion of this approach. See Alison McIntyre, Guilty Bystanders?: On the Legitimacy of Duty to Rescue Statutes, 23 PHIL. & PUB. AFF. 157 (1999). Christopher Schroeder succinctly states the essential point: "If failures to act can be causes, the issue of assigning legal responsibility for these causes is not yet resolved." Schroeder, supra note 50, at 190 (emphasis added).

n56 See also the related discussion of Williams's critique of Moore, supra note 53.

n57 See DRESSLER, supra note 29, at 167; LaFAVE & SCOTT, supra note 29, at 278; Hughes, supra note 26, at 627-31. The Model Penal Code does away with the traditional doctrine of proximate causation and explicitly replaces it with tests of culpability. See MODEL PENAL CODE §§ 2.03(2)(b), 2.03(3)(b); DRESSLER, supra note 29, at 174-75.

n58 See HALL, supra note 31, at 195 (noting that the issue of causation is never considered to be a difficulty in the well-recognized cases of commission by omission); Husak, supra note 39 (arguing that the notion of control would do a better job of capturing the underlying substantive point of the causation requirement).

n59 Hughes, supra note 26, at 602.

n60 Id. at 604 (quoting GLANVILLE WILLIAMS, CRIMINAL LAW 40 (1953)).

n61 Hughes, supra note 26, at 604.

n62 See HALL, supra note 31, at 200. Hall's argument for this point is somewhat different from that which follows in the text. See also FLETCHER, supra note 27, at 424-25.

n63 What I am calling the "relevant description" is, in the terms of section 1.13(9) of the Model Penal Code, a combination of specified "conduct" and "attendant circumstances"; for relevant discussion, see ROBINSON, supra note 39, at 232-38.

n64 Hughes, supra note 26, at 602.

n65 These are the facts of Lambert v. California, 355 U.S. 225 (1957), a case which turned on the constitutional issue of the notice required by due process; the majority judgment held that in the specific circumstances of the case, a strict liability application of the requirement to register violated due process. It is not at all clear what principle motivated the decision in Lambert. See Hughes, supra note 26, at 618-19. The case is rarely discussed and has not, in federal courts, been the basis of the constitutional invalidation of any other provision. See DRESSLER, supra note 29, at 152-54.

n66 See LaFAVE & SCOTT, supra note 29, at 208.

n67 See generally LaFAVE & SCOTT, supra note 29, at 405-10.

n68 Hughes, supra note 26, at 605-06. Jerome Hall argues that the relevant distinction is whether the crime is petty or not: Ignorance of the law may (should be able to) undermine mens rea in the case of petty offenses of both omission and commission. HALL, supra note 31, at 200-01, 403. But the seriousness of the offense, though correlated with the salience of the relevant description to the agent, is not the same thing. Consider an offense of failing to remove exposed asbestos from one's building: If the duty was established before there was public knowledge of the hazards of asbestos, the failure to act may lack mens rea and yet be a very serious offense.

n69 This is the approach taken in Continental European jurisdictions with criminal duties to rescue. See Feldbrugge, supra note 1, at 641.

n70 See Hughes, supra note 26, at 628; A.D. Woozley, A Duty to Rescue: Some Thoughts on Criminal Liability, 69 VA. L. REV. 1273, 1290-91 (1983).

n71 See Feldbrugge, supra note 1, at 641 (citations of European case law on this point).

n72 See the comment by Richard Epstein in Richard Epstein's Takings, supra note 16, at 184.

n73 See generally Ashworth & Steiner, supra note 21; Andre Tunc, The Volunteer and the Good Samaritan, in THE GOOD SAMARITAN AND THE LAW, supra note 1, at 43.

n74 "The short and simple answer to the it-won't-work objection is to point out where and for how long it has been working." Woozley, supra note 70, at 1290.

n75 See Pirkle v. Oakdale Union Grammar Sch. Dist., 253 P.2d 1 (Cal. 1953) (stating standard of care required of an officer or employee of a public school).

n76 160 N.E. 301 (Mass. 1928)

n77 44 A. 809 (N.H. 1898).

n78 KEETON ET AL., supra note 35, at 375 n.22. Another favorite of commentators is Yania v. Bigan, 155 A.2d 343 (Pa. 1959), where the defendant somehow incited the victim, a visitor on defendant's land, to jump into a deep trench with water at the bottom; the defendant then did nothing to help the victim, who drowned.

n79 Buch, 44 A. at 810.

n80 See Pridgen v. Boston Hous. Auth., 308 N.E.2d 467, 477 (Mass. 1974) (overruling Osterlind); Castonguay v. Acme Knitting Mach. & Needle Co., 136 A. 702, 705 (N.H. 1927) (overruling Buch).

n81 Southern Railway v. Grizzle, 53 S.E. 244 (Ga. 1906).

n82 See KEETON ET AL., supra note 35, 56.

n83 Id. at 376.

n84 Id. The quotation from Lord Hoffmann, mentioned supra note 9, continues:

A moral version of this point may be called the "why pick on me?" argument. A duty to prevent harm to others or to render assistance to a person in danger or distress may apply to a large and indeterminate class of people who happen to be able to do something. Why should one be held liable rather than another?

Stovin v. Wise, 1996 C.A. 923, 941.

n85 See Ernest J. Weinrib, The Case for a Duty to Rescue, 90 YALE L.J. 247, 262 (1980) (arguing in support of a tort duty to rescue; Weinrib later came to believe that such a duty is incompatible with the fundamental normative structure of tort law, see discussion infra note 88); see also Saul Levmore, Waiting for Rescue: An Essay on the Evolution and Incentive Structure of the Law of Affirmative Obligation, 72 VA. L. REV. 879, 936 (1986); Richard W. Wright, The Standards of Care in Negligence Law, in PHILOSOPHICAL FOUNDATIONS OF TORT LAW 274 (David G. Owen ed., 1995). Levmore also discusses a different (and rather unreal) problem of what to do where only one person fell under the duty to rescue and failed to act as required, but it is not known which person that is. Levmore, supra, at 934-35. On contribution generally, see KEETON ET AL., supra note 35, 50.

n86 See KEETON ET AL., supra note 35, at 377.

n87 See Herskovits v. Group Health Coop. of Puget Sound, 664 P.2d 474 (Wash. 1983) (finding a fourteen percent reduction, from thirty-nine percent to twenty-five percent, in decedent's chance for survival sufficient evidence of causation to allow jury to consider possibility that physician's failure to provide timely diagnosis of decedent's illness was the proximate cause of his death). This doctrinal maneuver has so far been limited to cases of medical malpractice. The more general issue of casting omissions as causes is not substantially different--that is, equally unproblematic--in torts than in the criminal law. See supra text accompanying notes 55-58; see also KEETON ET AL., supra note 35, at 265.

n88 By this I merely mean that the existing conceptual and doctrinal structure of tort law is flexible enough to accommodate such an innovation. Having once favored it, see Weinrib, supra note 85, Ernest Weinrib now opposes a tort duty to rescue on the ground that this would be false to the essential features of tort law as a normative system, one of which is that there is no liability for nonfeasance, see, e.g., Ernest J. Weinrib, Understanding Tort Law, 23 VAL. U. L. REV. 485 (1989). In his later writing, Weinrib has noted:

The misfeasance requirement is essential if tort law is to represent the ordering intrinsic to doing and suffering. The requirement signals that consequences do not attract obligations except as they arise out of the risk inherent in a particular act. Suffering by the plaintiff that is independent of the defendant's doing has no significance for tort law. Accordingly, no liability lies for failure to prevent or alleviate suffering. The defendant's duty reflects the correlative morality of doing and suffering, and thus arises only from the potency of injury in the defendant's act. The plaintiff's unilateral need for assistance, no matter how urgent, falls outside the relationship of doing and suffering. The exclusion of liability for nonfeasance ensures that the normative implications of a tortious act depend not on the undesirability of plaintiff's suffering as such but on the suffering being the materialization of a risk inherent in the defendant's action.

Id. at 517.

We can grant that the misfeasance requirement is essential "if tort law is to represent the ordering intrinsic to doing and suffering." We can even grant that tort law, as it has in fact been developed in Anglo-American law, reflects that ordering. But why should tort law continue to represent that ordering?

Let us first leave aside the issue of whether a general duty to rescue could legitimately be introduced by way of judicial innovation. It may be that on the best theory of adjudication this would be illegitimate because too great a departure from precedent. But we would still need to ask whether the fact that tort law has represented the ordering intrinsic to doing and suffering gives us any reason to resist legislative enactment of a general tort duty to rescue. There are two possible reasons. The first is that we accept a normative theory of corrective justice that expresses the ordering of doing and suffering, and we believe that the aim of tort law is to enforce corrective justice. This is clearly one part of Weinrib's motivation for his rejection of tort duties to rescue. I will return briefly to such arguments infra Part II.C. But Weinrib's main concern would seem to be that introducing liability for nonfeasance into tort law would render it incoherent, which would in itself undermine its justification as a coercive regime. See ERNEST J. WEINRIB, THE IDEA OF PRIVATE LAW 38-45 (1995) [hereinafter WEINRIB, PRIVATE LAW]. I cannot discuss this aspect of Weinrib's position here.

n89 See Lowns v. Woods, AUSTL. TORTS REP. P81-376 (1996). For critical discussion of the decision, see Les Haberfield, Lowns v. Woods and the Duty to Rescue, 6 TORT L. REV. 56 (1998). For a more positive view, emphasizing the special social role of physicians, see Natalie Gray & James Edelman, Developing the Law of Omissions: A Common Law Duty to Rescue?, 6 TORTS L.J. 240, 243 (1998).

n90 See GLENDON, supra note 17, ch. 4; Benditt, supra note 17, at 410-18; Anthony D'Amato, The "Bad Samaritan" Paradigm, 70 NW. U. L. REV. 798 (1976); Franklin & Ploeger, supra note 13, at 1008-09.

n91 Leslie Bender argues for recognition of a tort duty to rescue as an expression of the "ethic of care." See Leslie Bender, A Lawyer's Primer on Feminist Theory and Tort, 38 J. LEGAL EDUC. 3, 33-36 (1988) (citing the feminist approach in CAROL GILLIGAN, IN DIFFERENT VOICE (1982)). Contra Linda C. McLain, "Atomistic Man" Revisited: Liberalism, Connection, and Feminist Jurisprudence, 65 S. CAL. L. REV. 1171, 1238-42 (1992) (criticizing Bender's approach, and noting that the ethic of care and beneficence are not the same thing).

n92 See JOEL FEINBERG, HARM TO OTHERS ch. 4 (1984); McIntyre, supra note 55, at 177; Woozley, supra note 70, at 1293-99. Weinrib's early defense of a tort duty to rescue provides an important exception. See Weinrib, supra note 85, at 266; see also PATRICIA SMITH, LIBERALISM AND AFFIRMATIVE OBLIGATION 33-43 (1998).

n93 A different methodology can be found in Weinrib's work. Both his early defense of a tort duty to rescue and his later position that rejects such a duty turn on accounts of the overall coherence of private law. See WEINRIB, PRIVATE LAW, supra note 88, at 46; Weinrib, supra note 85, at 268-79.

n94 FEINBERG, supra note 92, at 15.

n95 JOHN STUART MILL, ON LIBERTY (1859) [hereinafter MILL, ON LIBERTY], reprinted in JOHN STUART MILL, THREE ESSAYS (Richard Wollheim ed., 1975).

n96 See generally JOHN STUART MILL, UTILITARIANISM (George Sher ed., 1979) (1861).

N97 MILL, ON LIBERTY, supra note 95, at 15.

n98 Id. On the next page, Mill writes: "I forego any advantage which could be derived to my argument from the idea of abstract right." Id. at 16. Whether Mill is right that utilitarianism sets the limits of interference with liberty at the prevention of actions that harm others is dubious. Feinberg himself believes that the prevention of actions that cause serious offense to others is a legitimate aim of the criminal law, and that the harm principle and the "offense principle" together make up the "liberal position" on the moral limits of the criminal law. FEINBERG, supra note 92, at 26.

n99 H.L.A. HART, LAW, LIBERTY AND MORALITY 6 (1963).

n100 MILL, ON LIBERTY, supra note 95, at 15.

n101 HART, supra note 99, at 30-34.

n102 MILL, ON LIBERTY, supra note 95, at 16-17.

n103 Id. at 17.

n104 Id.

n105 See FEINBERG, supra note 92, at 130-50; see also Woozley, supra note 70, at 1293-99 (a similar view). Feinberg's view is consistent with David Lyons's interpretation of Mill. See David Lyons, Liberty and Harm to Others, in RIGHTS, WELFARE, AND MILL'S MORAL THEORY 102-03 (1994).

n106 See FEINBERG, supra note 92, at 165-71. Feinberg also argues, in the alternative, that failures to prevent harm can be cast as causes of the harm. Id. at 171-85. I find this a futile strategy. See supra text accompanying notes 55-58.

n107 See FEINBERG, supra note 92, at 131-34.

n108 See MILL, supra note 96, at 48-49.

n109 Supra text accompanying note 87.

n110 FEINBERG, supra note 92, at 109-14, 144.

n111 Id. at 143-48.

n112 Id. at 227-32; see also id. at 117-18 (discussing Feinberg's concept of a victim).

n113 Id. at 135-43; see also SMITH, supra note 92, at 46-74.

n114 FEINBERG, supra note 92, at 136-39.

n115 Id. at 138-39.

n116 This is not quite accurate. There are cases where A brings B back to his normal baseline that Feinberg would want to count as a genuine benefit--for example, when A hands over to B the amount of money he has just lost at the races. In that case, we measure benefit against a baseline of B's post-loss condition. Why? Because A has no duty to benefit B in this case. Feinberg in effect simply defines all those cases where A can return B to his "normal" baseline and where, as it intuitively seems, A owes B a duty, as failures to prevent harm rather than failures to benefit. For effective criticism of this aspect of Feinberg's view, including the point that it may take Feinberg close to legal moralism, see McIntyre, supra note 55, at 174-81.

n117 For very similar criticism of Feinberg on this point, see 2 F.M. KAMM, MORALITY, MORTALITY: RIGHTS, DUTIES, AND STATUS 24 (1996). Kamm asks us to suppose, in the drowning child case, that

the child had been born to its mother while she was in the pool, so that it had never been on dry land, and it was drowning from the time it was born. Would the fact that we would not be bringing it back to status quo ante be relevant in deciding whether it had a human right to aid?

Id.

n118 For related criticism, though not specifically directed at Feinberg, see PETER UNGER, LIVING HIGH AND LETTING DIE 42-45 (1996).

There is another way, more straightforward than Feinberg's, in which failure to rescue could be seen as a failure to prevent harm--we could cast it as a failure to prevent a deterioration in a person's well-being. On a broad sense of benefit, such a failure is also a failure to benefit; we could, however, draw a distinction between failure to prevent deterioration (call that failure to prevent harm) and failure to improve a person's well-being (call that failure to benefit). The relevant baseline is simply the victim's condition at the time the potential rescuer first has an opportunity to act. This distinction is more natural than Feinberg's and is congenial to tort law, but it would not suit Feinberg's purpose. On this account, I could walk by an accident victim suffering extreme pain that, we can imagine, will not get any worse (though it may take a long time to get better), and fail to rescue the victim. The proposed distinction would capture only those "rescue cases" where intervention could prevent further disaster. Note that the Rhode Island provision refers to a person who "is exposed to, or has suffered, grave physical harm." R.I. GEN. LAWS 11-56-1 (1994). The suggested alternative account would also include too much in that a general duty to prevent all deteriorations in well-being (harms) would require us to assist all people on a downward slope, not just those involved in emergencies.

n119 See generally Kent Greenawalt, Legal Enforcement of Morality, 85 J. CRIM. L. & CRIMINOLOGY 710 (1995). Greenawalt discusses criminal duties to rescue. Id. at 713-15. For a non-Millian liberal discussion of the enforcement of morality, see RONALD DWORKIN, TAKING RIGHTS SERIOUSLY 240-57 (1977).

n120 For the distinction between negative and positive liberty, see ISAIAH BERLIN, Two Concepts of Liberty, in FOUR ESSAYS ON LIBERTY (1969). Many different accounts of both positive and negative liberty have been proposed since Berlin's essay was first published in 1959. Furthermore, the notion of coercion that the definition of negative liberty in the text employs is itself capable of various importantly different construals. See ROBERT NOZICK, SOCRATIC PUZZLES 15-45 (1997); ALAN WERTHEIMER, COERCION (1990). For our purposes we can leave this complication to one side.

n121 See MILL, ON LIBERTY, supra note 95; JOSEPH RAZ, THE MORALITY OF FREEDOM (1986) (especially chapters one, fourteen, and fifteen). Mill's word for what I, following Raz, call autonomy, is "individuality." Raz's book develops this aspect of Mill's view, while nevertheless departing from Mill in several important ways.

n122 I follow Raz's account. See RAZ, supra note 121, at 408. Obviously, if one values negative liberty for its contribution to autonomy, then one will also value positive liberty directly. Id. at 148-57 (discussing the various different ways in which coercion interferes with human autonomy). Raz's discussion makes it clear that this question is more complicated than many have supposed.

n123 MILL, ON LIBERTY, supra note 95; see also RAZ, supra note 121, at 155. However, Raz's own version of liberalism is not rigidly anti-paternalistic. See id. at 422-23.

n124 Note that Raz, for reasons largely turning on his conception of political authority, does not resist the idea of legal moralism. See RAZ, supra note 121, at 414-15.

n125 Raz rightly points out that not all interferences with negative liberty reduce autonomy. Id. at 409-10.

n126 See supra text accompanying note 102.

n127 See Richard Epstein, A Theory of Strict Liability, 2 J. OF LEGAL STUD. 151, 198-99 (1973); Schroeder, supra note 50, at 192-93.

n128 I use quotation marks because the tendency of contemporary libertarians to appeal to Locke for historical support, see, e.g., ROBERT NOZICK, ANARCHY, STATE, AND UTOPIA (1974), seems tendentious at best, see A. JOHN SIMMONS, THE LOCKEAN THEORY OF RIGHTS (1992) (especially chapter six).

n129 I do not mean to suggest that the two kinds of liberal theory that I discuss here, the Millian and the libertarian, exhaust the liberal possibilities. Most importantly, my discussion ignores the contemporary egalitarian liberalism most significantly developed by John Rawls in A Theory of Justice (1971) and Political Liberalism (1993). This is because there is nothing in the Rawlsian account of liberty that raises problems for the duty to rescue. See infra note 134.

n130 Moore writes that:

Retributivism, when combined both with the principle of legality and the insight that law as law does not even prima facie obligate citizen obedience, yields the legal moralist theory of proper legislative aim: all and only moral wrongs should be prohibited by the criminal law, for the reason that such actions (or mental states) are wrongful (or culpable) and deserve punishment.

MOORE, supra note 50, at 754; see also id. pt.1 (Moore's defense of retributivism); id. at 661-65 (argument from retributivism to the legal moralist account of the appropriate legislative aim in criminal law).

n131 See, e.g., DRESSLER, supra note 29, at 69; LaFAVE & SCOTT, supra note 29, at 10. We can take "social good" to include a variety of different goals, such as the promotion of human well-being, the protection of rights, and others; what it excludes, at least in the usage I will adopt, is the goal of enforcing morality for its own sake.

n132 MOORE, supra note 50, at 746-50.

n133 See DWORKIN, supra note 119, at 266-78; MOORE, supra note 50, at 749. Moore does believe that there is a right to liberty, roughly corresponding to the right to privacy that has been recognized under the U.S. Constitution. See MOORE, supra note 49, at 763-77. As this narrower right is not relevant to the issue of positive criminal duties, I will not discuss it here.

n134 See generally DWORKIN, supra note 119; RONALD DWORKIN, A MATTER OF PRINCIPLE (1985). Dworkin's view falls with Rawls's into the category of liberal theory I described, supra note 129, as "liberal egalitarian." Neither theory recognizes a general right to "liberty," and neither offers a Millian account of the value of liberty in terms of human well-being; both are rights-based liberal theories that reject a blanket right to liberty and as such provide no liberty-based grounds to resist legal duties to rescue. For Rawls, see especially JOHN RAWLS, The Basic Liberties and Their Priority, Lecture VIII, in POLITICAL LIBERALISM 289 (1993).

n135 NOZICK, supra note 128, at 28-33. This makes it clear that, on this understanding of rights, we cannot say that they protect especially important aspects of well-being, such as, in this example, the value of not being assaulted by another person. For this weaker understanding of rights, which is compatible with utilitarianism, see generally Amartya Sen, Rights and Agency, 11 PHIL. & PUB. AFF. 1 (1982).

n136 This understanding of the concept of rights seems to me the most useful for philosophical discussion. It is obviously not the only possibility. See RAZ, supra note 121, at 165-216, 245-66.

n137 MOORE, supra note 50, at 749.

n138 Id. at 747.

n139 Moore believes that negative liberty is also a means to certain other goods, in addition to positive liberty. See id. at 747-78.

n140 Id. at 278.

n141 This distinction has been embraced by the U.S. Supreme Court in cases concerning physician-assisted suicide. See Vacco v. Quill, 521 U.S. 793, 793-94 (1997). For a broadly sympathetic discussion of Moore's appeal to the greater weight of negative duties, see F.M. Kamm, Action, Omission, and the Stringency of Duties, 142 U. PA. L. REV. 1493 (1994).

n142 See Freeman, supra note 10, at 1463-64.

n143 See THOMAS NAGEL, THE POSSIBILITY OF ALTRUISM (1970); THOMAS NAGEL, THE VIEW FROM NOWHERE 152-53 (1986) [hereinafter NAGEL, THE VIEW].

n144 See PARFIT, supra note 53, at 27, 143.

n145 The duty not to violate rights is agent-relative precisely because it operates as a side-constraint. See supra text accompanying note 135. If the duty were to minimize rights violations, rather than not to violate rights, it would be an agent-neutral duty.

n146 MOORE, supra note 50, at 280.

n147 See Freeman, supra note 10, at 1463.

n148 MOORE, supra note 50, at 279.

n149 Id.

n150 See, e.g., SMITH, supra note 92, at 43; Freeman, supra note 10, at 1478-79; McIntyre, supra note 55, at 182.

n151 For Moore's discussion of the criticisms of Samuel Freeman and George Fletcher, see MOORE, supra note 50, at 281-82.

n152 See, e.g., RAZ, supra note 121, at 13.

n153 For criticism of the standard libertarian interpretation of Locke, see SIMMONS, supra note 128; Freeman, supra note 10, at 1465-67; Heyman, supra note 41, at 699-703.

n154 See, e.g., F.A. HAYEK, THE CONSTITUTION OF LIBERTY 16-19 (1960).

n155 The first sentence of Anarchy, State, and Utopia reads: "Individuals have rights, and there are things no person or group may do to them (without violating their rights)." NOZICK, supra note 128, at ix.

n156 See HAYEK, supra note 154, at 21.

n157 See NOZICK, supra note 128, at 150-82.

n158 On fraud, see HAYEK, supra note 154, at 143-44.

n159 See NOZICK, supra note 128, at ix. It is quite clear that in rejecting positive rights to aid, or "charity," contemporary libertarians depart from Locke. See SIMMONS, supra note 128, at 307-52.

The core of the libertarian position, as I have said, is a constraint on the legitimate content of law: Individuals' rights must not be violated. This leaves open the (positive) aim of either criminal law or torts. Nozick, following Locke, holds that at least part of the positive aim of law is the enforcement of morality. See NOZICK, supra note 128, at 10-11. Locke puts the point by saying that individuals have the right to enforce the law of nature. See 2 JOHN LOCKE, TWO TREATISES OF GOVERNMENT ch. 2 (1698). For discussion, see SIMMONS, supra note 128, at 121-66. Libertarians need not, however, embrace this legal moralist aim. They could hold that the point of law is solely to promote the overall social good. For libertarians, this might be understood in terms of the protection of rights (although that is not an especially plausible goal--it does not seem to be worse to die in a fire than to be murdered). Be that as it may, government pursuit of any conception of the social good is constrained, on the libertarian view, by the prohibition on violating the rights of citizens. The only occasion on which coercion does not violate rights is when it is for the sake of preventing the coerced person from violating rights. Thus, we reach the same result without invoking legal moralism.

n160 See Epstein, supra note 127, at 198. Epstein's libertarianism is more fully set out in a later article. See Richard Epstein, Causation and Corrective Justice: A Reply to Two Critics, 8 J.L. STUD. 477 488-89 (1979) [hereinafter Epstein, Causation and Corrective Justice]. "I attach a good deal of importance to the 'natural' set of entitlements that I think are generated by a concern with individual liberty and private property rights. In most cases I think that these rights are deserving of absolute protection and vindication." Id. at 488.

n161 EPSTEIN, supra note 16, at 306-29 (chapter entitled "Transfer Payments and Welfare Rights"). "The basic rules of private property are inconsistent with any form of welfare benefits." Id. at 322.

n162 See Richard A. Epstein, Nuisance Law: Corrective Justice and Its Utilitarian Constraints, 8 J. LEGAL STUD. 49, 74-79 (1979) [hereinafter Epstein, Nuisance Law]; see also Epstein, Causation and Corrective Justice, supra note 160, at 487-90.

n163 EPSTEIN, supra note 16, at 331-38.

n164 See Richard Epstein's Takings, supra note 16; Richard A. Epstein, The Utilitarian Foundations of Natural Law, 12 HARV. J.L. & PUB. POL'Y 713 (1989). Epstein had listed mutual benefit as a condition for deviation from the libertarian constraint on law-making: "Presence of implicit in-kind compensation from all to all that precludes any systematic redistribution of wealth among the interested parties." Epstein, Nuisance Law, supra note 162, at 79. This does not make the departure from libertarianism any less real, because (paternalistic) coercion for the sake of benefiting the coerced is clearly a violation of individual rights on the libertarian account, and the coercion is not itself for the sake of preventing the coerced person from violating rights. What it shows is that Epstein's bedrock moral conviction is that compulsory redistribution against a baseline of market outcomes is wrong; all else flows from that.

In this Article, I will not discuss the outlandish claim that utilitarian moral theory somehow dovetails with libertarianism. For criticism from a libertarian, see Eric Mack, Comment, A Costly Road to Natural Law, 12 HARV. J.L. & PUB. POL'Y 753 (1989). Epstein replied to Mack's criticism. See Richard Epstein, Postscript: Subjective Utilitarianism, 12 HARV. J.L. & PUB. POL'Y 769 (1989); see also Larry Alexander & Maimon Schwarzschild, The Uncertain Relationship Between Libertarianism and Utilitarianism, 19 QUINNIPIAC L. REV. 657 (2000).

n165 See the quotation from Sheldon Richman, supra note 9. The libertarian opposition to legal duties to rescue is, in my experience, very popular among law students.

n166 See Thomas Nagel, Libertarianism Without Foundations, 85 YALE L.J. 136 (1975). There are, in fact, several different arguments to the libertarian conclusion, each of which requires separate evaluation. For a very useful survey, see WILL KYMLICKA, CONTEMPORARY POLITICAL PHILOSOPHY 95-160 (1990).

n167 See PHILLIPE VAN PARIJS, REAL FREEDOM FOR ALL 15 (1995). Van Parijs offers this example to make his point:

Think of an island which happens to be owned . . . by one of its inhabitants. Providing it is difficult or expensive enough to leave the island, the owner can impose on the other inhabitants any condition she fancies. If they are to be allowed to earn their livelihood, they may have to work abysmally long hours, for example, or give up their religion, or wear scarlet underwear. On a libertarian account . . . such a society would not cease to be free.

Id. at 14.

n168 See, e.g., JAN NARVESON, THE LIBERTARIAN IDEA (1988). The first sentence of chapter one of his book reads:

At the outset let us say that 'Libertarianism,' as the term is used in current moral and political philosophy . . ., is the doctrine that the only relevant consideration in political matters is individual liberty: that there is a delimitable sphere of action for each person, the person's 'rightful liberty,' such that one may be forced to do or refrain from what one wants to do only if what one would do or not do would violate, or at least infringe, the rightful liberty of some other person(s).

Id. at 7. Friedrich Hayek writes as the first sentence of chapter one of his book: "We are concerned in this book with that condition of men in which coercion of some by others is reduced as much as possible in society." HAYEK, supra note 154, at 11.

n169 HENRY SIDGWICK, THE METHODS OF ETHICS 274 (7th ed. 1907).

n170 For a very clear account of the equivocation between different senses of liberty in the libertarian slogan, see MOORE, supra note 50, at 744.

n171 Nozick calls the constraint against aggression "the libertarian side constraint," see NOZICK, supra note 128, at 33, whereas I have said the core libertarian right was the right to be free of coercion. It is important to note that Nozick himself does not trade in the misleading rhetoric about liberty I discuss in this paragraph.

n172 See H.L.A. Hart, Rawls on Liberty and Its Priority, 40 U. CHI. L. REV. 534 (1973), reprinted in READING RAWLS 244 (Norman Daniels ed., 1975).

n173 See generally three articles by G.A. Cohen: Capitalism, Freedom and the Proletariat, in THE IDEA OF FREEDOM: ESSAYS IN HONOUR OF ISAIAH BERLIN 9 (Alan Ryan ed., 1979); Illusions About Private Property and Freedom, in ISSUES IN MARXIST PHILOSOPHY 223 (J. Mepham & D. Ruben eds., 1981); Are Freedom and Equality Compatible?, in ALTERNATIVES TO CAPITALISM 113 (J. Elster & K.O. Moene eds., 1989); see also Shelly Kagan, The Argument from Liberty, in IN HARM'S WAY: ESSAYS IN HONOR OF JOEL FEINBERG 16 (Jules L. Coleman & Allen Buchanan eds., 1994).

It is not the case, of course, that only libertarians favor a moralized conception of liberty. Indeed, Jeremy Waldron defends a legal duty to rescue against the liberty objection by appealing to such a conception. See Jeremy Waldron, On the Road: Good Samaritans and Compelling Duties, 40 SANTA CLARA L. REV. 1053, 1082-84 (2000).

n174 A fine illustration of the moralized conception of liberty and its dangers for the champion of negative liberty is provided by this passage from Richard Epstein:

A person does not exercise his liberties when he kills or enslaves another; he does not vindicate his property rights when he steals from another. If he is restrained from these actions by another, he cannot claim a loss of liberty, but only the loss of an ability to act to which he was never entitled. Liberty is best understood as freedom from force and falsehood, not as a maximization of the things which are under one's disposition and control. And the principle (at least as a matter of corrective justice) is the same whether the infringements of the rights of another are great or small.

Epstein, Causation and Corrective Justice, supra note 160, at 489. The first two sentences make it clear that it is individuals' rights, not any recognizable notion of freedom, that is at the center of the account. In the next sentence, however, Epstein reverts to the familiar assertion that it is negative and not positive liberty that matters. In doing so, however, he all but contradicts the first two sentences. Though the idea that veracity promotes liberty requires a moralized conception of freedom, Epstein undermines the moralized view when he says that liberty should be understood as freedom from force or falsehood simpliciter--as he has just stated that, on his view of liberty, the person who is forced not to steal has not had his liberty infringed. On the moralized conception of freedom Epstein announces in the first two sentences, he can at most claim that liberty is freedom from illegitimate force not, obviously enough, a particularly informative or uplifting idea. The moralized conception of freedom returns to center stage with the focus on rights in the final sentence.

n175 See infra Part IV.

n176 This may be the correct way to understand Weinrib's view in The Case for a Duty to Rescue. See Weinrib, supra note 85.

n177 See supra note 88 (discussing Weinrib's later view); see also JULES COLEMAN, RISKS AND WRONGS (1992); Stephen Perry, Risk, Harm, and Responsibility, in PHILOSOPHICAL FOUNDATIONS OF TORT LAW, supra note 85, at 321. The corrective justice approach could be cast either in legal moralist terms (the aim of tort law is to make sure that tortfeasors fulfill their duties of compensation), in terms of the social good (the aim of tort law is to vindicate the rights of tort victims), or a combination of the two.

n178 My own view is very close to that expressed by Jeremy Waldron. See Jeremy Waldron, Moments of Carelessness and Massive Loss, in PHILOSOPHICAL FOUNDATIONS OF TORT LAW, supra note 85, at 387.

n179 One alleged normative problem with legal duties to rescue that I have not mentioned in this section is that such duties would reduce opportunities for morally worthy acts. There are at least three different claims that have been made along these lines.

First, Epstein makes a point that appears to derive from a mistaken understanding of Kant's idea that actions done from inclination have no moral worth. On Epstein's version of this idea, if inclination (to avoid legal sanction) is present, an act in accordance with duty lacks moral worth, whereas on Kant's account a person could still act from duty in such a case. See Epstein, supra note 127, at 200; Weinrib, supra note 85, at 266. Second, the claim is sometimes made that it is important to preserve the realm of the morally supererogatory and that what is legally required cannot be supererogatory. This claim seems to depend on some (controversial) assumption of a standing moral obligation to obey the law. In any case, legal duties to rescue would hardly exhaust the realm of the supererogatory. Third, there is Landes and Posner's claim that enacting a legal duty to rescue would lead to less acts of altruism because the possibility of legal liability would lead potential altruists to avoid sites where they may have the opportunity to rescue. Landes & Posner, supra note 13, at 120-22. The claimed substitution effect that this point depends on is dubious. See infra note 243 and accompanying text. Apart from that, the point has some plausibility--Aristotle remarked in The Politics that one reason to favor the institution of private property is that it affords greater scope for the virtue of generosity or liberality--so long as one accepts what I do not: that acting morally or displaying virtue has value in itself. See also Woozley, supra note 70, at 1292-93.

n180 See MACAULAY, supra note 4, at 316. In the relevant passage, Macaulay is specifically responding to the provision in Livingston's code that required rescue when this could be done "without personal danger or pecuniary loss." Id.

n181 Id. at 496; see also SMITH, supra note 92, at 47 (referring to the problem as the "slippery slope argument").

n182 See, e.g., FEINBERG, supra note 92, at 156; Woozley, supra note 70, at 1299.

n183 See Epstein, supra note 127, at 197-201; Schroeder, supra note 50, at 193-94; see also Weinrib, supra note 85, at 267-68.

n184 See, e.g., Louis Kaplow, A Fundamental Objection to Tax Equity Norms: A Call for Utilitarianism, 48 NAT'L TAX J. 497 (1995). For a critique along the same lines as that sketched in the text, see Ronald Dworkin's discussion of the economic analysis of tort law. RONALD DWORKIN, LAW'S EMPIRE 285-301 (1986). Though I agree with Dworkin's criticism of the utilitarian argument he constructs in those pages, I disagree with his own view, which turns on a non-instrumental division of private and public responsibility, id. at 299, and which is therefore a version of the dualist view I criticize. See infra text accompanying notes 185-89.

n185 For the full argument, see Liam Murphy, Institutions and the Demands of Justice, 27 PHIL. & PUB. AFF. 251 (1998).

n186 For a canonical statement of the utilitarian position on this issue, see SIDGWICK, supra note 169, at 457-59; see also id. at 87-88 (a note on Bentham).

n187 Note that my argument is against dualistic theories of justice only, which are distinct from (what Waldron aptly calls) theories of politics, which address "the nature and principled basis of political choice on matters of justice and right" in the face of disagreement about these matters; democratic theory is largely a theory of politics in Waldron's sense and it is, indeed, all about institutional design. JEREMY WALDRON, LAW AND DISAGREEMENT 1-4 (1999).

n188 JOHN RAWLS, A THEORY OF JUSTICE 115, 333-37 (1971).

n189 See ROBERT H. MYERS, SELF-GOVERNANCE AND COOPERATION 59-68 (1999); Larry Alexander, Scheffler on the Independence of Agent-Centered Prerogatives from Agent-Centered Restrictions, 84 J. PHIL. 277 (1987); Robert H. Myers, Prerogatives and Restrictions from the Cooperative Point of View, 105 ETHICS 128 (1994).

n190 My account of this problem draws on LIAM MURPHY, MORAL DEMANDS IN NONIDEAL THEORY 127-33 (2000).

n191 UNGER, supra note 118, ch. 2; id. at 53-54 (listing the nine factors). Unger himself does not use the description "rescue case."

n192 See UNGER, supra note 118, at 34-35 ("The CB Radios"); see also Woozley, supra note 70, at 1289 ("Sometimes it happens that a ship's Mayday signal is picked up only by a ham radio operator halfway around the world; we expect him to report it as quickly as he can . . . ."). For criticism of Unger's argument, see F.M. Kamm, Rescue and Harm, 5 LEGAL THEORY 1 (1999).

Waldron defends duties to rescue from "general apprehensions about the law requiring abstract benevolence," by appeal to the factor of proximity. Waldron, supra note 173, at 1075, 1097-1103.

n193 UNGER, supra note 118, at 28-29, 77-79.

n194 Id. at 77-78.

n195 Unger's argument is not uncontroversial, and a full discussion cannot be attempted here. For criticism, see F.M. Kamm, Faminine Ethics: The Problem of Distance in Morality and Singer's Ethical Theory, in SINGER AND HIS CRITICS (Dale Jamieson ed., 1999) [hereinafter Kamm, Faminine Ethics]; Kamm, supra note 192.

n196 For arguments in defense of a distinct normative category comprising rescue cases, see SMITH, supra note 92, ch. 3; Kamm, Faminine Ethics, supra note 195, at 198-202; Kamm, supra note 192, at 19-20.

One common argument in favor of the special normative status of rescue situations is that such situations could be the subject of a mutually beneficial ex ante agreement. See NARVESON, supra note 168, at 242-44; D'Amato, supra note 90, at 805-08; Freeman, supra note 10, at 1480-81; Eric H. Grush, The Inefficiency of the No-duty-to-rescue Rule and a Proposed "Similar Risk" Alternative, 146 U. PA. L. REV. 881 (1998); Richard L. Hasen, The Efficient Duty to Rescue, 15 INT'L REV. L. & ECON. 141 (1995). But such an agreement could only be hypothetical, and it is not clear why hypothetical bargains should have any moral force at all. See DWORKIN, supra note 119, at 150-52. Furthermore, the argument requires strict unanimity, and this seems dubious. Take a well-guarded billionaire who never swims or takes personal risks; such a person's resources might well be called on in emergencies (imagine some variation on Unger's case, "Bob's Bugatti," UNGER, supra note 118, at 136) while his chances of needing a stranger's assistance are close to zero. See, e.g., PARFIT, supra note 53, at 109. Lastly, we may note that if the argument from mutual benefit does succeed, it would surely be a wonderful coincidence that the agreed upon duty was precisely the minimal duty to rescue and not one requiring significant sacrifice in some circumstances. I take this point from some remarks attributed to F.M. Kamm in Samuel Freeman's article, see Freeman, supra note 10, at 1480 n.96; Freeman's response to Kamm's point (essentially that demanding general duties of beneficence are, indeed, too demanding) seems irrelevant to the mutual benefit argument.

n197 I owe this example to Shelly Kagan. See also H.M. Malm, Liberalism, Bad Samaritan Law, and Legal Paternalism, 106 ETHICS 4, 17-19 (1995). The costs of rescue may be especially high if the rescuer's ethical sense requires her to take responsibility for the saved life. See BOUDU SAUVE DES EAUX (Sirius 1932) (film by Jean Renoir).

n198 See, e.g., RICHARD B. BRANDT, A THEORY OF THE GOOD AND THE RIGHT 276-77 (1979) ("Act utilitarianism makes extreme and oppressive demands on the individual, so much so that it can hardly be taken seriously; like the Sermon on the Mount, it is a morality only for saints.").

n199 I here disagree with Schroeder, supra note 50, at 192-97.

n200 See also MURPHY, supra note 190, ch. 2. Note that extreme demands can come from parts of morality other than beneficence, including such paradigm negative duties as the duty not to kill, but such duties are never accused of being absurd because extemely demanding (no one says that it is absurd to ask me to forego the million dollars I would gain if I killed you; many say it is absurd to expect a millionaire to give a million dollars to famine relief). See Shelly Kagan, Does Consequentialism Demand Too Much?, 13 PHIL. & PUB. AFF. 239 (1984); F.M. Kamm, Supererogation and Obligation, 82 J. PHIL. 118 (1985). This factor raises significant complications that I will simply ignore in this Article. For discussion, see MURPHY, supra note 190, chs. 3, 5.

n201 See SHELLY KAGAN, THE LIMITS OF MORALITY (1989) (especially chapter ten, "Extraordinary Morality"); UNGER, supra note 118 (especially chapter six, "Living High and Letting Die Reconsidered: On the Costs of a Morally Decent Life"). Some philosophers who officially embrace utilitarianism try to avoid the problem of demands by arguing, in effect, that most people are not really in a position to do much good for anyone else. See, e.g., Frank Jackson, Decision-Theoretic Consequentialism and the Nearest and Dearest Objection, 101 ETHICS 461 (1991).

n202 See THOMAS NAGEL, EQUALITY AND PARTIALITY (1991) [hereinafter NAGEL, EQUALITY]; NAGEL, THE VIEW, supra note 143, at 189-208 (chapter entitled "Living Right and Living Well"); SAMUEL SCHEFFLER, HUMAN MORALITY (1994); SAMUEL SCHEFFLER, THE REJECTION OF CONSEQUENTIALISM (rev. ed. 1994).

n203 See MURPHY, supra note 190, chs. 3-4.

n204 See CHARLES FRIED, RIGHT AND WRONG 130 (1978); MURPHY, supra note 190, ch. 5; PARFIT, supra note 53, at 30-31; HENRY SHUE, BASIC RIGHTS: SUBSISTENCE, AFFLUENCE, AND U.S. FOREIGN POLICY 114-19 (2d ed. 1996); Dan W. Brock, Defending Moral Options, 51 PHIL. & PHENOMENOLOGICAL RES. 909, 912 (1991); L. Jonathan Cohen, Who Is Starving Whom?, 47 THEORIA 65 (1981); Carlos Rosenkrantz, Igualitarismo y Libertarianismo: Politica no antropologia, 7 REVISTA DEL CENTRO DE ESTUDIOS CONSTITUTIONALES 193 (1990).

n205 For details, see MURPHY, supra note 190, chs. 5-7.

n206 See FEINBERG, supra note 92, at 158-59; Freeman, supra, note 10, at 1467-68; McIntyre, supra note 55, at 181-82; Menlowe, supra note 10, at 48; Weinrib, supra note 85, at 272-73, 292.

n207 A crucial point not acknowledged by the authors cited in the previous note. See, e.g., Freeman, supra note 10, at 1467 ("There is in fact little that we as individuals can do to help the destitute (except in emergencies, and then the relevant duty is the duty to aid the distressed). The problem of destitution is largely one of background justice . . ."). Contra Murphy, supra note 185, at 278-84.

n208 See Thomas Nagel's discussion of the "moral division of labor." NAGEL, EQUALITY, supra note 202, 53-62.

n209 This is not to deny the obvious fact that each of us is especially well-placed to promote the interests of a certain small group of intimates.

n210 See sources cited supra note 206.

n211 Stipulating, as an element of the offense, that the defendant must have known that this was not the case would render the duty largely inert in practice. Proof beyond reasonable doubt that a person knew that she could have conveyed a cost-justified benefit to someone, the need for which, as she also knew, would not have been eradicated under ideal institutions, would seem to be a very difficult matter. The only cases in which it seems plausible to have no reasonable doubt about the defendant's knowledge of these two facts are, indeed, rescue cases.

n212 Though perhaps not by the utilitarian legal moralist. On such a view, these kinds of costs could be an acceptable price to pay for the sake of enforcing individuals' moral duties.

n213 It seems likely that such a duty would, in any case, be unconstitutionally vague. See the discussion of vagueness in the context of Wisconsin's duty to aid provision in State v. La Plante, 521 N.W.2d 448 (Wis. Ct. App. 1994).

n214 What about a specific duty to call the fire brigade if one knows that a major art museum is burning down? This could potentially produce a great benefit even if it only once applies to an individual's situation, so long as that individual is the only person in a position to raise the alarm. The more important the property, however, the less likely it is that this condition would be met, especially under ideal institutions.

n215 There is also this further problem. One of the aspects of rescue situations that makes them easy to identify is that they involve severe suffering or peril. It is not as easy to know whether one is confronted with emergencies more broadly conceived. Stipulating, as an element of the offense, knowledge that one is in a situation of the relevant kind does not fully solve the problem here, because a person may be apprehensive about the way in which this mental state requirement would be applied in practice and thus find herself wasting time and energy searching for emergencies in his neighborhood. For a similar point supporting the limitation to cases of severe suffering or threat to life, see infra text accompanying note 224.

n216 If we are not legal moralists and we reject that constraint, then there would presumably be no limit to legitimate legal demands on people, even if we accept either the limited or collective principle of beneficence as our principle for both personal conduct and institutional design. The prima facie implausibility of such a view shows the prima facie plausibility of the constraint against forcing sacrifice beyond what a person is morally required to sustain voluntarily. See generally sources cited supra note 189.

n217 See supra text accompanying note 42.

n218 See Feldbrugge, supra note 1, at 636-38 (discussing European approaches to this issue).

n219 Note that this point does not depend on the assumption that the aim of the criminal law is to promote the social good, because a legal moralist utilitarian holds that if it makes no utilitarian sense for a particular person to act, there is no moral requirement that she act.

n220 See Wallace M. Rudolph, The Duty to Act: A Proposed Rule, in THE GOOD SAMARITAN AND THE LAW, supra note 1, at 243.

n221 In this respect professional rescuers--surf lifeguards, firemen, search and rescue personnel, and others--are unlike medical practitioners.

n222 For some worries about leaving such decisions to jurors, see Malm, supra note 197, at 21-22 (arguing that it is illiberal to allow jurors to decide which losses count as reasonable).

n223 One reason for opposition to criminal duties to rescue may be precisely an awareness of the expressive effect of the lack of such a duty: In lacking such a duty, our law expresses the idea that we have no responsibilities to mere strangers.

n224 Cf. Jason S. Johnston, Punitive Liability: A New Paradigm of Efficiency in Tort Law, 87 COLUM. L. REV. 1385, 1395-96 (1987) (explaining how raising the burden of proof in civil liability cases reduces the risk of inducing too much investment in care). We have here a further reason--additional to those set out infra Part IV.B.--for preferring a criminal to a tort duty to rescue.

n225 See Unger's example, "The Yacht." UNGER, supra note 118, at 63-67.

n226 See Levmore, supra note 85, at 882-94. Levmore draws on the analysis of Landes and Posner. See Landes & Posner, supra note 13.

n227 Larger rewards seem likely to provide incentives for potential rescuers to create demand for their own services. See Levmore, supra note 85, at 886-87.

n228 See id. at 913-17; see also John P. Dawson, Rewards for the Rescue of Human Life?, in THE GOOD SAMARITAN AND THE LAW, supra note 1, at 63-87.

n229 See Desmond S. Greer, A Transatlantic Perspective on the Compensation of Crime Victims in the United States, 85 J. CRIM. L. & CRIMINOLOGY 333, 350 (1994).

n230 Under current law, a nonprofessional rescuer is generally not entitled to restitution even of his out of pocket expenses. See RESTATEMENT OF RESTITUTION 116(a) cmt. a (1937) (referring to comment c of section 114; "intent to charge" a requirement for recovery); see also RESTATEMENT (SECOND) OF RESTITUTION 3 cmt. c (Tent. Draft No. 1, 1983) ("It is virtually a precept that the merit of emergency salvage efforts is, and should be, its own reward."). The existence of a criminal duty to act would presumably rebut this presumption of gratuitousness. See generally 2 GEORGE E. PALMER, THE LAW OF RESTITUTION 10.3, at 369-70 (1978 & Supp. 1999); Ross A. Albert, Comment, Restitutionary Recovery for Rescuers of Human Life, 74 CALIF. L. REV. 85 (1986); John W. Wade, Restitution for Benefits Conferred Without Request, 19 VAND. L. REV. 1183 (1966).

For discussion of compensation for rescuers from a "liberal-communitarian" perspective, see Heyman, supra note 41, at 746-50.

n231 Adjudication of this question of where the compensation should come from would require, inter alia, discussion of the normative basis of the law of restitution; it is beyond our scope here. For general discussion of restitution in the rescue context, including discussion of the civil law doctrine of negotiorum gestio, see Hanoch Dagan, In Defense of the Good Samaritan, 97 MICH. L. REV. 1152 (1999). For discussion of the possibility of a state compensation scheme for rescuers, see Norval Morris, Compensation and the Good Samaritan, in THE GOOD SAMARITAN AND THE LAW, supra note 1, at 136-39. One reason why a state compensation scheme would be preferable is that it would allow greater certainty and flexibility with respect to the amount of compensation, a difficult issue under restitution doctrine. Suppose that a reasonable rescue nevertheless turned out to be fatal for the rescuer. Should the rescuee be liable for this loss of life? For discussion of this problem, with focus on a German case on point, see Dawson, supra note 228.

n232 Note that a legal moralist "utilitarian" would not object to the compensation of rescuers' expenses if that were for the best in utilitarian terms: on such an approach what matters is that people do their utilitarian duty and suffer retribution for not doing so, not that they suffer costs for doing so.

n233 See Cadoppi, supra note 1, at 117 (stressing this advantage of duty-to-rescue provisions and the early--early nineteenth century--appreciation in Continental Europe of the distinction between liability for omission and liability for commission by omission); McIntyre, supra note 55, at 187-90. For philosophical discussion of the general significance of the distinction between killing and letting die, see the articles collected in Killing and Letting Die. KILLING AND LETTING DIE (Bonnie Steinbock & Alistair Norcross eds., 2nd ed. 1994); see also KAMM, supra note 117, at 13-200.

In most Continental European jurisdictions, the duty-to-rescue statute cannot be used to ground homicide by way of commission by omission. See Cadoppi, supra note 1, at 95; Feldbrugge, supra note 1, at 648-52. The very fact that U.S. duty-to-rescue provisions specify specific and minimal sanctions should block their use in a commission-by-omission prosecution as a matter of statutory interpretation, but it would seem that an explicit provision to that effect would be wise; for a suggestion along these lines, see Jay Silver, The Duty to Rescue: A Reexamination and Proposal, 26 WM. & MARY L. REV. 423, 436 (1985) (proposing rescue statute providing that "nothing contained in this Act shall alter existing law with respect to liability for criminal homicide"). In State v. Cabral, 810 P.2d 672, 677 (1991), the Hawaii Intermediate Court of Appeals prospectively approved the use on retrial of Hawaii's duty-to-assist provision as the basis of a homicide charge. This had not been done in the original trial; it was apparently not done in the retrial either. See State v. Cabral, 883 P.2d 638, 639 (Haw. Ct. App. 1994).

I cannot in this Article evaluate the appropriateness of commission by omission liability generally. One obvious point is that the existence of a relationship is likely to increase culpability. For discussion, see generally FLETCHER, supra note 27, at 581-634; Glanville Williams, Criminal Omissions--The Conventional View, 107 L.Q. REV. 86 (1991).

n234 The Australian Northern Territory's provision, Criminal Code Act section 155, provides for up to seven years imprisonment, but requires that the failure to aid was "callous," which has been interpreted to mean "more than normal" intent in Salmon v. Chute, 4 N.T.L.R. 149 (Austl. N. Terr. 1994). This still leaves matters rather unclear; for a brief discussion see John T. Pardun, Good Samaritan Laws: a Global Perspective, 20 LOY. L.A. INT'L & COMP. L. REV. 591, 594-97 (1998). The geography of the Northern Territory (extremely sparsely populated with a very harsh environment) greatly heightens the need for private efforts to assist, for example, stranded motorists.

n235 See supra note 21.

n236 This further reduces the seriousness of the problem of the possibility of excessive expenditures due to mistrust of triers of fact. See supra note 224 and accompanying text.

n237 See MURPHY, supra note 190, ch. 7. I argue there that the strong commonsense moral condemnation of a nonrescuer is perhaps based more on implicit judgments of bad character than on--what seems to be the case--judgments of wrong action. The upshot is that failure to rescue will suggest an undesirable set of motivations even where it is not strictly wrong.

n238 If this is right, and I am also right that at the level of personal morality there may not be a general requirement to perform easy rescues, but rather a requirement to develop the kind of motivations that would lead one to be strongly disposed always to perform easy rescues, see supra note 237, then the upshot would be that as a requirement on action, legal duties to rescue can be defended even though they are not in the end part of the morality of personal conduct. This is exactly the reverse of what is generally supposed.

n239 It is true that the provision of compensation renders the issue of the demands of the legal duty to rescue moot at the level of individuals who must act on the duty. But could not the limited nature of the demands of the limited and collective principles of beneficence undermine the case for the legal duty to rescue when they are applied at the level of institutional design as accounts of the aim of promoting the social good? I assume that the answer is no, though establishing this for our two principles would be complicated and involve speculation. What needs to be established is that legal duties to rescue could be part of an overall scheme, support of which does not burden any member of the community (by way of the cost of funding the scheme) more than is justified by each principle.

n240 It may be worth just mentioning some problems that have been raised. There is the problem of the officious intermeddler: the busybody who believes, for example, that the existence of a criminal duty to aid makes it his business to supervise the neighbors' child-rearing practices. See Weinrib, supra note 85, at 281 (for the particular example I thank Christopher Eisgruber). This does seem to be a possible problem, but it is hard to see why this problem could not be solved by careful drafting; the Rhode Island provision specifies that there must be an emergency that exposes someone to grave physical harm. It is not plausible to interpret "grave physical harm" to include the effects of spanking or playing in the sun without a hat.

A more serious problem concerns the potential for prosecutorial abuse. In brief, the concern is that prosecutors have been given further means for the extraction of guilty pleas and cooperation with the prosecution. See Svetlana Kornfeind, Deputization and the Statutory Duty to Aid (May, 1997) (unpublished paper, New York University Law School) (on file with author); see also Daniel B. Yeager, A Radical Community of Aid: A Rejoinder to Opponents of Affirmative Duties to Help Strangers, 71 WASH. U. L.Q. 1, 36-38 (1993). I do not know how serious a problem this is, but to the extent that it is, it also seems to be solvable by drafting. A duty-to-notify bill that was considered (and then abandoned) by the Florida legislature in 1999 shows how it might be done. Subsection (2) of that bill provided:

This section may not be construed to apply to a person either who is prosecuted as a principal in the first degree or an accessory after the fact to an offense, or who is prosecuted for attempting, soliciting, or conspiring to commit an offense, when the prosecuted offense has occurred in the course of the same criminal conduct, transaction, or episode as the criminal offense which exposes the victim to bodily injury.

H.B. 69, 1999 Reg. Sess. (Fl. 1999).

Finally, a recent essay by Eugene Volokh raises the possibility that duty to report provisions would have the bad effect of deterring witnesses who failed to report at the time of the crime but who might otherwise have cooperated with the police at a later date. See Eugene Volokh, Essay, Duties to Rescue and the Anticooperative Effects of Law, 88 GEO. L.J. 105 (1999).

n241 It is worth noting that the different countries of Continental Europe did not simply inherit their legal duties to rescue en masse from some common source--Roman Law or the Napoleonic Code. Rather, such duties were introduced independently and at different times in different countries. The earliest was Russia in 1845, but in most countries such duties were first enacted after World War II. Feldbrugge, supra note 1, at 630-31.

n242 There is a danger that courts will find a civil duty just in virtue of the existence of the criminal duty. See KEETON ET AL., supra note 35, 36; Franklin & Ploeger, supra note 13, at 1010-17. Once again, one would hope that sensible statutory interpretation would prevent this, but as it may not, an express provision to the effect that mere violation of the criminal duty to rescue does not ground civil liability would be desirable. Hawaii's provision requiring aid for crime victims excludes civil liability for violators. HAW. REV. STAT. ANN. 663-1.6(c) (Michie 1995).

n243 The doubts expressed by Professors Landes and Posner on this score seem unwarranted. See Ian Ayres, A Theoretical Fox Meets Empirical Hedgehogs: Competing Approaches to Accident Economics, 82 NW. U. L. REV. 837, 841 (1988); Richard L. Hasen, The Efficient Duty to Rescue, 15 INT'L REV. L. & ECON. 141, 146 (1995); Levmore, supra note 85, at 890-91 (discussing Landes and Posner, supra note 13, at 119-27); Wright, supra note 85, at 272; see also Epstein, supra note 127, at 190 (noting that the no-duty-to-rescue rule seems inefficient according to the Carroll Towing formula for negligence).

n244 Under a cost-justified rescue rule, the burden of required rescue will always be less than the loss to the victim that rescue would prevent, and thus it will always be cheaper to sustain the burdens of rescue than to pay compensatory damages.

n245 See Franklin & Ploeger, supra note 13, at 1009; Wright, supra note 85, at 274. There are many cases where full compensatory liability for misfeasance can seem disproportionate to fault as well--as for example when a mildly negligent moment leads to disastrous consequences. On this issue, see Waldron, supra note 178. The upshot is that my objection to tort liability for failure to rescue may not be neutral with respect to existing features of tort law. Nevertheless, the issues are distinct, as the objection in the text applies even to an explicitly intentional tort of failure to rescue.

n246 AUSTL. TORTS REP. P81-376.

n247 See D'Amato, supra note 90, at 802-04.

n248 See id. at 809.

n249 See supra Part II.C.

n250 See supra note 240 and accompanying text.

n251 See Rudzinski, supra note 1, at 111-15.

n252 One danger of Weinrib's methodology, see, e.g., Weinrib, supra note 85, at 275, 293; sources cited supra note 88, is that it encourages this mistake, see GLENDON, supra note 17, at 89-108 (an illuminating discussion of this kind of mistake). Glendon seems to believe that the mistake is ineradicable in the United States, GLENDON, supra note 17, at 102, and therefore suggests that authors of legal opinions, especially on constitutional questions, should be careful to remind readers that the legal question and the question of political morality are different. See also Mark Kelman's discussion of "conflating legal solubility with the existence of a problem." MARK KELMAN, A GUIDE TO CRITICAL LEGAL STUDIES 275-79 (1987). Also to the point is the book's next section, on "the synthetic individualist tradition." Id. at 279-84.

n253 See EPSTEIN, supra note 16, at 318-19; Epstein, supra note 127, at 199-203.