Hostname: page-component-586b7cd67f-dlnhk Total loading time: 0 Render date: 2024-11-23T21:44:01.742Z Has data issue: false hasContentIssue false
  • English
  • Français

HARM PRINCIPLES

Published online by Cambridge University Press:  04 May 2015

James Edwards
James Edwards*
Affiliation:
Worcester College, Oxford, [email protected]
Get access Rights & Permissions [Opens in a new window]

Abstract

Much time has been spent arguing about the soundness of “the harm principle.” But in the philosophical literature there is no single such principle; there are many harm principles. And many objections pressed against “the harm principle” are objections to only some of these principles. The first half of this paper draws a number of distinctions between harm principles. It then argues that each harm principle is compatible with many other principles that impose limits on the law, including but not limited to other harm principles. The second half of the paper applies the lessons of the first to a number of prominent objections to “the harm principle.” That principle has been accused of a) being underinclusive; b) misrepresenting the reasons why many act-types ought to be legally proscribed; c) permitting lawmakers to treat people as mere means of achieving their ends; and d) being overinclusive. The paper argues that one harm principle survives all four objections.

Type
Research Article
Information
Legal Theory , Volume 20 , Issue 4 , December 2014 , pp. 253 - 285
Copyright
Copyright © Cambridge University Press 2015 

Access options

Get access to the full version of this content by using one of the access options below. (Log in options will check for institutional or personal access. Content may require purchase if you do not have access.)

References

1 I do not claim that the existence of distinct harm principles has gone unnoticed. But if I am correct, there are many more harm principles than is typically acknowledged.

2 I take for granted that an act is permissible if it is not wrong or if it is wrong but justified.

3 Permissive principles, then, negate candidate constraints.

4 The reference here to coercive regulation is not meant to foreclose the question of which types of regulation (or types of action) are governed by “the harm principle.” Coercive regulation is, for now, a placeholder for various possibilities. I discuss some of these below.

5 True, these rights are usually thought to be qualified—they can, that is, be infringed without being violated. My point is only that violations of these rights are impermissible.

6 Compare Gerald Gaus, The Moral Foundations of Liberal Neutrality, in Contemporary Debates in Political Philosophy (Thomas Christiano & John Christman eds., 2009), with Richard Arneson, Liberal Neutrality on the Good: An Autopsy, in Perfectionism and Neutrality (George Klosko & Steven Wall eds., 2003).

7 Compare David Estlund, Democratic Authority (2008), at 21–64, with Raz, Joseph, Disagreement in Politics, 43 Am. J. Juris. 25 (1998)Google Scholar.

8 Because those writing on the topic typically refer to “the harm principle,” I occasionally adopt their way of speaking. It should always be remembered, however, that there is no single such principle.

9 Joel Feinberg, Harm to Others (1984), at 9.

10 Id. at 11.

11 The offense principle states that “it is always a good reason in support of a proposed criminal prohibition that it is probably necessary to prevent serious offence to persons other than the actor.” Id. at 26.

12 Feinberg acknowledges this point; id. at 10.

13 John Stuart Mill, On Liberty (Elizabeth Rapaport ed., 1978), at 9.

14 On exclusionary reasons, see Joseph Raz, Practical Reason and Norms (1999), at 35ff.

15 The only reason to punish, for some who hold this view, is the prevention of further wrongdoing. For an overview of the literature, see Antony Duff, Legal Punishment, The Stanford Encyclopedia of Philosophy (Edward N. Zalta ed., 2013).

16 It is true, of course, that 1a) and 2c) have the additional properties identified above. I suppress these here for ease of exposition.

17 Andrew Simester & Andreas von Hirsch, Crimes, Harms, and Wrongs (2011), at 35.

18 This may seem inconsistent with the authors’ claim that we should accept an independent offense principle. In fact it is not. While they think there is reason to criminalize offensive wrongdoing, Simester & von Hirsch also think this reason insufficient unless the harm principle is satisfied; id. at 117–118.

19 Mill writes that his harm principle is “entitled to govern absolutely the dealings of society with the individual in the way of compulsion and control whether the means used be physical force in the form of legal penalties, or the moral coercion of public opinion” (emphasis added). See Mill, supra note 13, at 9.

20 Officially this is Feinberg's position: see Feinberg, supra note 9, at 3. Note that one might further distinguish between actions that are criminal offenses and actions that are crimes, where an action is a crime only if it is a criminal offense and there is no defense available to those who commit it. A given harm principle might be concerned either with the permissibility of the creation of criminal offenses or with the creation of crimes, or with both. I am grateful to Doug Husak and Patrick Tomlin for pointing out this further distinction.

21 As my aim here is not to evaluate the arguments themselves, I offer but a sketch of each.

22 Simmons, John, Justification and Legitimacy, 109 Ethics 739 (1999)Google Scholar, at 752.

23 One might endorse this first premise for a number of reasons. To take just one, perhaps the state has the right to φ only if its people have consented to its having that right or would consent in certain hypothetical circumstances. And perhaps people have consented or would consent to the state having the right described in the text only in cases in which their conduct aggrieves another.

24 The argument sketched in the text supports a harm principle that is a constraint. If one thinks that the list of grievances is longer—perhaps a person also has a grievance against p if p’s conduct offends that person—one will endorse a version of the grievance argument that supports a permissive harm principle.

25 Gerald Postema argues that the grievance argument, or something like it, is implicit in Joel Feinberg's work, particularly Joel Feinberg, Harmless Wrongdoing (1988), at 321–322. See Postema, Gerald, Politics Is about the Grievance: Feinberg on the Legal Enforcement of Morals, 11 Legal Theory 293 (2005)CrossRefGoogle Scholar.

26 One might, of course, endorse other versions of the grievance argument. One might, for instance, argue that all coercion is permissible only if the person coerced has aggrieved, or is about to aggrieve, another. If this argument is sound, there will be a sound harm principle in which x is any coercive act.

27 It is thus necessary to spend some time discussing these effects. The clumsiness argument itself is set out in the final paragraph of this section.

28 While it is not inevitable that if φing is criminalized, people will be punished for φing, cases where no one is punished are not typical. I thus follow Douglas Husak in assuming that one cannot ignore the morality of punishment when assessing the morality of criminalization; see Douglas Husak, Overcriminalization (2008).

29 Can one punish without intending to harm? Some argue that one cannot; see, e.g., David Boonin, The Problem of Punishment (2008), at 6–17. For present purposes, this would show only that not all criminal sentences are punishments. It would not alter the main point in the text, namely, that criminalization results in a variety of different types of harm.

30 Here I assume that one is harmed by that which diminishes one's prospects in life. Those who endorse different conceptions of harm can alter the example as required. The point—namely, that criminal sentences often harm as a side effect—remains.

31 This is the case unless, of course, they can prove a defense. That this message is sent by criminal offenses is seen by Hart. See H.L.A. Hart, The Concept of Law (3d ed. 2013), at 28.

32 Some think that p is only harmed by Z if p’s prospects in life are caused to be bad by Z, such that a very well-off person whose prospects are diminished may not have been harmed if those prospects remain fairly good. Again, my examples could easily be altered without cost to the argument.

33 Does it matter whether p’s reaction to criminalization is reasonable? Is p, who unreasonably refuses to φ, harmed by events that cause her unreasonably to refuse to φ? Perhaps not. But the types of reaction described in the text need not be unreasonable, so we need not resolve this point here.

34 Sometimes the targets of a rule are themselves constituted by the rule. I set such cases aside in the text. For discussion of the nature of rules, see Frederick Schauer, Playing by the Rules (1991); Larry Alexander & Emily Sherwin, The Rule of Rules (2001).

35 I draw here on the work of Joseph Raz; see Joseph Raz, Free Expression, in Ethics and the Public Domain (rev. ed. 1994), at 160–162. As Raz points out, it is no response to say that one's intention is not to condemn the group's way of life or that the wrongful practice is detachable from the valuable parts of that way of life. We have limited control over the social significance of our acts, and it is presumably for individuals, not the state, to decide what is integral to a particular way of life.

36 John Rawls, A Theory of Justice (rev. ed. 1999), at 386.

37 Or those otherwise morally liable to suffer harm. I focus on the narrower category of desert in the text, but the point seems to me to hold on any plausible view of liability, where p is morally liable to be harmed to some degree if p lacks a moral right not to suffer that harm.

38 If only to ensure a degree of consistency between offenders who are similarly placed.

39 One might reply that in a just legal system, punitive harms are deserved by those on whom they are inflicted and that inflicting deserved harm is an unqualified good. It might be said that the state has reasons of justice to impose said harms, not reasons to avoid imposing them for which it must struggle to compensate; perhaps conforming to the former reasons can sometimes justify criminalization even if no harm principle is satisfied. This can be doubted. Even if there are impersonal reasons to impose harm on offenders who deserve it—reasons that derive from the value of doing retributive justice—there may also be personal reasons not to impose harm on those same offenders—reasons that derive from the fact that all harm is bad for the person harmed. If this is right, even deserved punitive harms are harms the state has reason to avoid. And this makes it much less likely that our reasons to impose deserved harms could themselves defeat the reasons not to criminalize given by the other harms that criminalization brings about. For discussion of the relationship between retributivist theories of punishment and various harm principles, see Tomlin, Patrick, Retributivists! The Harm Principle Is Not for You, 124 Ethics 272 (2014)Google Scholar.

40 For present purposes, φing is criminalized only if φing is a criminal offense, and φing is a criminal offense only if those proved to have φed are liable to be convicted and sentenced in a criminal court unless they are able to prove a defense.

41 I do not claim these are the only three, but they are prominent in the literature and worth considering for that reason.

42 The distinction between 5) and 5b) is noted by others. See Holtug, Nils, The Harm Principle, 5 Ethical Theory & Moral Prac. 357 (2002)Google Scholar, at 360; David Brink, Mill's Progressive Principles (2013), at 184.

43 Whichever version of 5) we accept, satisfaction of the condition in question can be blocked by other criminal offenses that already exist in the jurisdiction. If preexisting offenses prevent harms that criminalizing φing might otherwise have prevented, criminalizing φing will not be likely to prevent harm, and 5) will not be satisfied. Of course, if we endorse a harm principle that is a constraint, those preexisting offenses must also satisfy 5). If they do not, they should be repealed.

44 One question for any defender of 6) concerns the relationship that must exist between a given action and a given harm for that action to count as harmful. Must my action have caused the harm, and in what sense of “caused”? For discussion, see von Hirsch, Andrew, Extending the Harm Principle: Remote Harms and Fair Imputation, in Harm and Culpability (Simester, Andrew & Smith, A.T.H. eds., 1996)Google Scholar.

45 Collins v. Wilcock, [1984] 3 All E.R. 374.

46 As I discuss below, there are various senses of the term harm, but I take for granted here that not all unwanted and unusual touching causes any harm. Nor does it create an unjustified risk of harm.

47 Again, this is subject to any additional conditions built into a full statement of 5) but which I temporarily suppress above.

48 These possibilities are noted by Gardner & Shute; see John Gardner & Stephen Shute, The Wrongness of Rape, in Oxford Essays in Jurisprudence: Fourth Series (Jeremy Horder ed., 2000).

49 The difference between 5) and 6) is noted by others; see Tadros, Victor, Harm, Sovereignty, and Prohibition, 17 Legal Theory 35 (2011)Google Scholar; Gardner & Shute, supra note 48; Tomlin, supra note 39.

50 For some worries about purposive harm principles, see Holtug, supra note 42, at 362–363.

51 Simester & von Hirsch, supra note 17, at 35, 47–48 (emphasis in original). HP1 is also endorsed in Gardner & Shute, supra note 48.

52 Simester & von Hirsch, supra note 17, at 35.

53 Simester & von Hirsch's discussion of so-called remote harms may also seem to suggest that they endorse an act-centered principle. Remote harm occurs when D's action leads to harm only via some “contingency,” such as a “mediating intervention” by some other person. The authors claim that cases of this type “do not fit neatly within the Harm Principle,” such that an “extended” version of that principle is required; such a principle permits criminalization of remotely harmful acts, but only where D can be held responsible for the harm; id. at 53–88. Note, however, that this worry about the type of connection needed between D's actions and particular harms is a worry only if one endorses an act-centered principle. If one endorses an instrumental principle, the relevant question is not whether the conduct in question brings about harm but whether criminalization of that conduct is likely to prevent harm. Because Simester & von Hirsch explicitly sign up to an instrumental principle, their discussion of remote harms is best interpreted as of relevance not to the harm principle but to another principle they endorse, the so-called wrongfulness constraint. I discuss this constraint in Section V.

54 Feinberg, Harm, supra note 9, at 26.

55 Husak, supra note 28, at 66 (emphasis added).

56 I leave open whether these must be other persons or whether the others in question can also be nonhuman animals.

57 Joseph Raz is one writer who endorses a wide harm principle. See Raz, Joseph, Autonomy, Toleration and the Harm Principle, in Issues in Contemporary Legal Philosophy: The Influence of H.L.A. Hart (Gavison, Ruth ed., 1987)Google Scholar, at 326.

58 Some will object that only narrow principles warrant the label “harm principle.” They may point to Mill's harm principle and to the tradition of thought that considers the harm principle a distinctively liberal principle. In brief reply, the discussion already shows that Mill's is only one harm principle among several others. Both wide and narrow principles make harm a direct determinant of the permissibility of criminalization (or legal regulation or coercion); there is thus nothing obvious about denying them the label “harm principle.” Nor are the liberal credentials of a principle obviously imperiled if it is wide. HP3 still rules out criminalization of harmless wrongs, and HP1 still requires that criminalization prevent harm. Such harm principles thus remain incompatible with many legal moralist principles, an incompatibility that gives the harm principle much of its liberalizing appeal.

59 I focus on these conceptions because I think them most plausible, though I cannot defend this here. For others, see Shiffrin, Seana Valentine, Harm and Its Moral Significance, 18 Legal Theory 357 (2012)Google Scholar; Hanser, Matthew, The Metaphysics of Harm, 77 Phil. & Phenomenological Res. 421 (2008)Google Scholar. For criticism of Shiffrin, see Tadros, Victor, What Might Have Been, in Philosophical Foundations of the Law of Torts (Oberdiek, John ed., 2014)CrossRefGoogle Scholar. For criticism of Hanser, see Thomson, Judith Jarvis, More on the Metaphysics of Harm, 82 Phil. & Phenomenological Res. 436 (2011)Google Scholar.

60 Judith Jarvis Thomson offers an improved version of this view. On Thomson's view, the relevant baseline is the position that Z prevents p from being in, by the very means by which Z affects p's position. The means by which Dr. A affects p is administration of the drug. This means does not itself prevent Dr. B poisoning the patient but does itself prevent the patient from recovering. So on this view, p’s baseline position is one of recovery, and Dr. A does harm the patient. See Thomson, supra note 59. A complex counterfactual view is also defended in Tadros, What Might, supra note 59.

61 For objections to various baselines, see Hanser, supra note 59. For an argument that whatever baseline one adopts, one ends up with an implausible version of the harm principle, see Holtug, supra note 42. I cannot address Holtug's argument here.

62 Joseph Raz, The Morality of Freedom (1986), at 414.

63 Feinberg, Harm, supra note 9, at 34ff.

64 See Raz, Autonomy, supra note 57, at 327. See also John Gardner, Offences and Defences: Selected Essays in the Philosophy of Criminal Law (2007) at 4, 30, and 244; Simester & von Hirsch, supra note 17, at 36. This view is criticized in Tadros, What Might, supra note 59.

65 Gardner, John & Edwards, James, Criminal Law, in The International Encyclopedia of Ethics (LaFollette, Hugh ed., 2013)CrossRefGoogle Scholar.

66 One such view is Feinberg's, for whom p is harmed only if one of p’s interests is adversely affected. A Feinbergian interest can be adversely affected without this diminishing the interest-holder's prospects in life (by, e.g., a trespass to land that goes unnoticed by the owner), and not everything that makes one worse off adversely affects a Feinbergian interest (including that which merely distresses, offends, or irritates). See Feinberg, Harm, supra note 9, at 45ff. For other views, see Tadros, What Might, supra note 59.

67 See Antony Duff, Answering for Crime: Responsibility and Liability in the Criminal Law (2008), at 130.

68 Indeed, it is already implicit in the claim, made in Section I, that any harm principle is only plausibly a necessary but insufficient condition of permissible coercive regulation.

69 A failure to grasp that truth also seems to be responsible for Feinberg's claim that “any plausible formulation of the harm principle” requires a special sense of harm that employs a normative baseline; see Feinberg, Harm, supra note 9, at 34. Feinberg seems to think this because he thinks that no act should be criminalized if it does not prevent violations of some moral entitlement. Even if this is so, it shows only that as well as a harm principle, we should also endorse another principle according to which actions should be criminalized only if they prevent violation of a moral entitlement. It does not show we should invent a special conception of harm.

70 As in the previous sections, I assume here that x is criminalization. It is true, of course, that the state could, in principle, condemn wrongdoers for φing without ever prohibiting φing, but it is plausible to think that the rule of law demands the prohibition as a condition of the permissibility of condemning φers.

71 HP5 states that φing is permissibly criminalized only if φing is criminalized in order to prevent harm or to prohibit what is harmful.

72 I emphasize the word “may,” because while HP6 is permissive, it need not permit lawmakers to criminalize with just any aim in mind. There may be a short list, which may or may not include the aim mentioned in 8).

73 Of course, if one endorses a normative baseline, that φing is harmful already entails that φing is morally wrong.

74 For similar thoughts on this issue, see Simester & von Hirsch, supra note 17, at 24–29.

75 There are difficult questions here about the relationship that must exist between a given act and some harm for the latter to contribute to the wrongfulness of the former. It is these questions to which discussions of so-called remote harms are trying to find answers. As Simester & von Hirsch put it, “in the context of remote harms . . . it needs to be shown why the prospect of some ultimate harm, perhaps brought about by an independent actor, makes it wrong for the defendant to act as she does.” See Simester & von Hirsch, supra note 17, at 72. If D's act is not wrong, 9) is not satisfied.

76 This is not to say that no other harm principle escapes the objections.

77 Ripstein, Arthur, Beyond the Harm Principle, 34 Phil. & Pub. Aff. 216 (2006)Google Scholar, at 217.

78 Id.

79 Duff, Answering, supra note 67, at 130, 135. For discussion of whether it could be said that A harms B, see Ripstein, supra note 77, at 218–222; Tadros, Harm, Sovereignty, supra note 49, at 48. I ignore this avenue of escape here.

80 Duff, Answering, supra note 67, at 130.

81 HP1 states that φing is permissibly criminalized only if criminalizing φing would probably prevent some harm. As mentioned, a full statement of HP1 would include additional conditions, on which I say more in Section X.

82 In a more recent paper, Duff and Marshall acknowledge both the distinction between HP1 and HP3 and that the failure to make that distinction “pervades” the relevant literature. See Duff, R.A. & Marshall, S.E., “Remote Harms” and the Two Harm Principles, in Liberal Criminal Theory: Essays for Andreas von Hirsch (Simester, A.P., du Bois-Pedain, Antje & Neumann, Ulfrid eds., 2014)Google Scholar.

83 I assume here that HP1 is not satisfied if criminalizing all trespass would probably prevent no more harm than criminalizing harmful trespass alone. This is to assume that harmful trespass is already prohibited in the legal system in question and that this prohibition itself satisfies HP1, such that the crucial question is whether criminalizing harmless trespass would probably prevent any further harm. This assumption favors my opponents, so I do not apologize for it here.

84 Or, where harmless trespass is already criminalized, the effect of its decriminalization. Some think that the fact that HP1 requires this kind of empirical calculation counts against it. I do not see why this should be so. Assessing the legitimacy of much (perhaps all) governmental activity requires that we consider its likely effects. Nor, when activities are ongoing, can we always risk finding out whether we are right about how things would change if the activity ceased. Would slashing income tax lead to an explosion of productive enterprise that enriches all or would it impoverish many? Would getting rid of the minefield on the border lead to hostilities or pave the way to better relations? Sometimes the risk involved in finding out is too great. This does not mean our calculations about what would happen are irrelevant to the legitimacy of our activities.

85 It might be said in reply that harmless trespass is permissibly criminalized even if doing so prevents no harm and that this is enough to falsify HP1. But this is far from obvious. Imagine some group of people is harmlessly trespassing somewhere at night without risk of detection by the owner, and preventing them doing so will not prevent any harm. Would it be permissible for us to take preventive steps if those steps would be highly likely to harm the group members (and their family and friends) in ways we cannot fully control, including by imposing harms they do not deserve? I doubt it but cannot argue the point here.

86 I owe this point to Doug Husak and Andrew Simester.

87 I borrow the term reactive harms from Simester & von Hirsch, supra note 17, at 47–48.

88 Duff, Answering, supra note 67, at 129 n.48.

89 Id. at 135.

90 Ripstein, supra note 77, at 218.

91 Stewart, Hamish, The Limits of the Harm Principle, 4 Crim. L. & Phil. 17 (2010)Google Scholar, at 30 (emphasis in original).

92 Simester & von Hirsch, supra note 17, at 35.

93 See Gardner, supra note 64, at 91–120.

94 There are several assumptions on which this alleged entailment depends, including a) the truth of the Gardnerian view of justification, and b) the truth of the claim that there are reasons not to criminalize, which can be defeated only by reasons to prevent harm. If these assumptions are false, HP1 does not entail HP5, and objections to HP5 are not necessarily objections to HP1. I concede the truth of the assumptions here and argue below that WO can be rebutted even with the concession.

95 Simester & von Hirsch, supra note 17, at 9, 19–20. In principle, the state could criminalize φing but never prosecute any offender. But it is plausible to think this would itself be impermissible—that the state should not pass laws it has no intention of enforcing.

96 Some would add a further condition: φing must be a wrong that is the business of all citizens and one that must be criminalized if the citizenry are to take certain values seriously. For this suggestion, see Duff, R.A. & Marshall, S.E, Public and Private Wrongs, in Essays in Criminal Law in Honour of Sir Gerald Gordon (Chalmers, James, Leverick, Fiona & Farmer, Lindsay eds., 2010)CrossRefGoogle Scholar.

97 And there will be further conditions that must also be met. I do not always add this qualification.

98 Some would wish to add that the wrongdoing must be the business of the actor. This qualification can be read into the text without affecting the argument.

99 Again, here I assume that x is criminalization.

100 Duff writes that he is concerned with “the question of what can count in principle as a good reason for criminalising a certain kind of conduct” (Duff, Answering, supra note 67, at 135) and takes the harm principle to state that the only such reason is to prevent harm. Stewart writes of a “refusal of the harm principle to recognize that, sometimes, a pure violation of rights is a good reason to criminalize”; Stewart, supra note 91, at 19.

101 John Gardner replies to Ripstein in similar vein: see Gardner, supra note 64, at 241–243.

102 As in the previous section, I assume here that HP1 is a positive principle, such that the fact that φing will probably prevent harm is a reason to criminalize. If that is so, then in cases where criminalization is justified, that reason is (part of) an undefeated reason, and lawmakers can permissibly act for it.

103 Ashworth, Andrew & Zedner, Lucia, Prevention and Criminalization: Justifications and Limits, 15 New Crim. L. Rev. 542 (2012)Google Scholar at 551.

104 Ripstein makes a similar objection when he writes that a principle such as HP1 will inevitably “run afoul of the liberal ideal of individual responsibility, which says that you cannot be prohibited from doing something that does not interfere with anyone else, simply because prohibiting you from doing it makes you or someone else less likely to commit some other genuine crime.” See Ripstein, supra note 77, at 228 (emphasis in original).

105 Gerald Cohen, Self-Ownership, Freedom, and Equality (1995), at 239 (emphasis in original).

106 Id. (emphasis in original).

107 According to 9), that φing is morally wrong is a necessary condition of the permissibility of criminalizing φing.

108 12) is endorsed by Douglas Husak; see Husak, supra note 28, at 82–91.

109 A defender of 12) need not claim that φers deserve punishment prior to the criminalization of φing; one may deserve punishment for φing partly because it is a crime. Once traffic offenses are created, others’ reliance on those laws may make my offending acts especially dangerous. Those acts may then be deserving of punishment, even though they would not have been had the traffic offenses never come into existence.

110 Whether they are right to do so is a separate question. The point here is not to argue for 9) or 12). It is rather to show that lawmakers committed to HP1 need not treat people as mere means.

111 Most obviously, if the burdens in question carry a degree of moral censure, it may be that fairness to offenders requires that judges impose only a quantum of punishment that reflects offenders’ moral blameworthiness. HP1 is perfectly consistent with this thought.

112 The objection made in this paragraph is the flip side of one of Patrick Devlin's objections to the harm principle. Devlin claims that if moral wrongness is relevant to the quantum of punishment we should impose for φing, it must also be relevant to whether we should criminalize φing in the first place; see Patrick Devlin, The Enforcement of Morals (1965), at 130–131. The present objector claims that if harm prevention is relevant to whether we should criminalize φing, it must also be relevant to the quantum of punishment we should impose for φing. My response is similar to that pressed by a number of writers against Devlin; see, e.g., H.L.A. Hart, Punishment and Responsibility (2d ed. 2008), at 8–13; Feinberg, Harmless, supra note 25, at 144–151.

113 Husak, supra note 28, at 72.

114 Duff, Answering, supra note 67, at 138.

115 Stewart, supra note 91, at 28–29.

116 Tadros, Harm, Sovereignty, supra note 49, at 50.

117 Meir Dan-Cohen, Harmful Thoughts (2002), at 152.

118 This point is well made by Husak: see Husak, Douglas, The Criminal Law as Last Resort, 24 Oxford J. Legal Stud. 207 (2004)Google Scholar.

119 Although I focus here on noncriminal alternatives, there is also the question of alternative criminal means of prevention. It is plausible to think that AMC also applies here. If some harm would probably be prevented by criminalizing φing or by criminalizing μing, it is plausible to think that it is permissible to criminalize φing only if criminalizing μing is not morally better, all things considered.

120 One is that the term proportionality refers in some legal contexts to a set of conditions including a version of AMC.

121 It might be said that we must also consider the other goods (and bads) that criminalization will bring about. It is true that the all-things-considered permissibility of criminalization requires reference to these things. But this is a separate question. If HP1 is sound, only the prevention of harm is capable of defeating the reasons not to criminalize given by the clumsily harmful nature of criminalization. PC is thus best interpreted as requiring consideration of whether this in fact occurs—whether the harm brought about by criminalization is disproportionate to the harm prevented by it.

122 What Section II calls internal punitive harms are the most obvious candidate.

123 For extended discussion of these points in a different context, see Rodin, David, Justifying Harm, 122 Ethics 74 (2011)Google Scholar.

124 For discussion, see Enoch, David, Intending, Foreseeing, and the State, 13 Legal Theory 69 (2007)CrossRefGoogle Scholar; Hosein, Adam, Doing, Allowing, and the State, 33 Law & Phil. 235 (2014)CrossRefGoogle Scholar.

125 This is one of several difficulties raised by Tadros for HP1, not all of which I can discuss here. See Tadros, Harm, Sovereignty, supra note 49.

126 Id. at 58.

127 According to AMC, it is a necessary condition of permissibly criminalizing φing that there is no alternative means of preventing harm that is morally better, all things considered. According to PC, it is a necessary condition of permissibly criminalizing φing that the harm brought about by criminalization is not disproportionate to the harm it prevents.