1. As these remarks suggest, my focus here is on criminalization and punishment in municipal legal systems.

2. Moore, Michael, A Tale of Two Theories , 28 Crim. Just. Ethics 27, 37 (2009)Google Scholar.

3. Douglas Husak, Overcriminalization (2008), at 82 (emphasis in original).

4. Id. at 78.

5. Husak, Douglas, Reservations About Overcriminalization , 14 New Crim. L. Rev . 97, 103 (2011)Google Scholar.

6. Husak, supra note 3, at 82.

7. Husak, supra note 3, at 197; Moore, supra note 2, at 36.

8. Moore, supra note 2, at 36.

9. Id. at 32.

10. Id.

11. For defense, and references to other defenders, see Boonin, The Problem of Punishment (2008), at 12–23.

12. Husak, supra note 3, at 83; Moore, supra note 2, at 31.

13. Justified actions, on the other hand, are not wrongful. Moore explicitly accepts this. See Michael Moore, Placing Blame (1997), at 673–674.

14. Husak, supra note 3, at 83.

15. Id. at 78.

16. “May” because, on an alternative view, officials have the power to punish people who have committed crimes whether or not procedural conditions are met. Those conditions determine whether exercising that power is legally permissible, not whether the power exists at all.

17. It is worth briefly noting that the proposal's truth is far from obvious. There certainly appear to be criminal sentences that fail to satisfy the conditions I earlier labeled (i) and (ii). One candidate is the sentences imposed in many legal systems on so-called dangerous offenders. These sentences are to be imposed for the reason that doing so will “protect the public.” It is true that such sentences are properly imposed only on condition that B was convicted of a crime. But satisfaction of this condition is not the reason for B's sentence. B is to be deprived of liberty not because B violated some norm, but because B is a dangerous individual against whom others need to be protected. It is true, of course, that B's criminal act is taken to be evidence of B's dangerousness. But it remains the case that it is the dangerousness—B's propensity to act in certain ways in the future—not the norm violation—something B did in the past—that provides the court's reason for sentencing B. It is far from clear that if a legislature creates liability to such sentences—and only to such sentences—it has not criminalized.

18. I am grateful to Patrick Tomlin for discussion of this point.

19. The rules of the road are the classic example. Once we have rules of this kind, failing to follow them will often be to violate “antecedently existing moral norms against risking harm to others.” So it will often be morally wrongful to fail to follow the rules. See Moore, supra note 13, at 73; Husak, supra note 3, at 103–119. Husak doubts that are many cases in which the law makes a moral difference of this kind. But he does not deny that there are some.

20. This is not, of course, true of all act-types. But it is true of some. Killing is an obvious example.

21. Joel Feinberg, Harm to Others (1984), at 14–15.

22. This claim might be defended in several ways. I will mention just one: we already saw that criminalization confers power on state officials; the risk of this power being misused makes criminalization wrongful if there is no reason to criminalize.

23. Husak, supra note 3, at 136ff.

24. Husak, Douglas, Polygamy , in Criminalization: The Political Morality of the Criminal Law (Duff, R.A., Farmer, Lindsay, Marshall, S.E., Renzo, Massimo & Tadros, Victor eds., 2014)Google Scholar.

25. Husak, supra note 3, at 103.

26. Id. at 100.

27. Id. at 102.

28. One might ask why expert manufacturers would not simply be exempted from liability. There are two ways this might be done: (i) granting licenses to experts, or (ii) granting experts a defense to liability. There are reasons not to pursue either option. Licensing schemes are costly to set up and administer. An expertise defense reintroduces the problem of error mentioned in the text.

29. Duff, R.A., Crime, Prohibition and Punishment , 19 J. Applied Phil . 97 (2002)Google Scholar.

30. Heath Wellman, Christopher, Rights Forfeiture and Mala Prohibita , in The Constitution of the Criminal Law (Duff, R.A., Farmer, Lindsay, Marshall, S.E., Renzo, Massimo & Tadros, Victor eds., 2014)Google Scholar.

31. Assuming, as we are throughout, that Husak and Moore are correct that (W) and (D) are conditions of permissible punishment. Husak describes expert manufacturers of the kind I have described as epistemically privileged. Husak, supra note 3, at 155–156. They know that their conduct does not create the risks that justify imposing criminal liability on others. He appears to think that making these offenders liable to punishment is consistent with the punishment thesis, because it is sometimes the only way to further substantial state interests, such as reducing harm done to consumers. But this does not establish that epistemically privileged offenders are permissibly punished, because it does not show that their conduct is morally wrongful or deserving of punishment, and Husak claims that (W) and (D) are both conditions of permissible state punishment.

32. Consider the following example. Some crimes—theft, sexual assault, and murder, for instance—can be committed only if B has certain intentions. What should intention mean for these purposes? Should B be held to have intended X only if B acted in order to bring X about? Or should it also be possible to find that B intended X if B appreciated that X was a virtually certain result of her actions? If this should be possible, should appreciation of this kind be sufficient for intention? Or should there be some additional hurdle that must be cleared before B can be held to have intended X? If so, what should that hurdle be? And what about cases where B appreciates that X would be virtually certain if only B were more competent, or appreciates that X is highly likely to occur, or intends to make X highly likely? Should cases of this kind ever—and if so, under what conditions—count as cases of intention? It is far from clear that these questions can be (confidently) answered by appeal to the norms that pick out theft, sexual assault, and murder as prelegal moral wrongs. For the reasons given in the text, we have reason to craft criminal offenses that draw a clearer line. We might, for instance, require that B either acts in order to bring X about or appreciates that X is virtually certain to result from her actions. To do this, however, is to make offenders out of some people whose conduct clearly is morally justified, and, equally clearly, is not deserving of punishment. If throwing one's baby out of an attic window is the only way to save her from an approaching fire, the fact that one knows death is virtually certain to result does not make one's conduct wrongful or culpable. The suggested definition of intention, however, has it that such a parent intends death: it suppresses the morally salient facts that make the parent's conduct morally permissible.

33. In Nicklinson, [2013] EWCA Civ 466 (UK), the Court of Appeal explicitly stated that there is no such general defense. Re A (Children), [2000] 4 All ER 961 (UK) was held to be an exceptional case that did not suggest otherwise.

34. Devlin v. Armstrong, [1971] NI 13.

35. Either because it is the lesser of two evils between which the offender had to choose, or because it is a reasonable response to a nonimminent threat of violence.

36. Moore seems to have something like this argument in mind. He writes that it “follows from retributivism” that only morally wrongful conduct should be criminalized, and that this follows because retributivism “limits punishment away from the non-deserving as much as it demands punishment of the deserving.” See Moore, supra note 2, at 31.

37. Husak, supra note 3, at 78.

38. This second audience comprises Hart's “puzzled” or “ignorant” persons. See H. L. A. Hart, The Concept of Law (3d ed. 2013), at 40.

39. By, for instance, making it the case that conformity protects conformers against abuse, and reduces the risk of harm to others. See the pharmaceutical example discussed in Section II.

40. Husak might have in mind a moral need to punish φers once φing is criminalized. But he does not make clear what this moral need would be. At one point, he writes that for the state not to punish some class of criminalized conduct would be for the state to “neglect its own proscriptions” (Husak, supra note 3, at 78). Now φing is proscribed if and only if there is a legal duty not to φ. For Husak, a proscription is part of the criminal law only if judges have a legal power to punish those who breach the aforementioned duty. It is not clear why failing to exercise the power entails that one is neglecting the duty, or why this failure is always morally problematic.

41. As I claimed above, the list also gives potential offenders better guidance about what counts as an offense. Fewer offenses might be committed as a result, and fewer people might be punished. So even if the proportion of punishments that is impermissible in the fact-relative sense turned out to be larger, the total number of such punishments might still be smaller if we opted for the authoritative list.

42. One might, of course, endorse a mixed view. But that makes no difference here.

43. It is worth repeating that an act may be wrongful even if excused. So preventing people who believe it is necessary to use force from using force that is in fact unnecessary is still preventing wrongful (though excused) acts.

44. All else being equal, it is better to prevent someone from being harmed or wronged than to allow them to be harmed or wronged and then hold the harmdoer or wrongdoer responsible later.

45. The nonabolitionist may, of course, claim that there are other effects of criminalization that render it impermissible. I turn to some possible effects below.

46. I do not intend to wear out my shoes by walking around in them. But I know it will happen.

47. We might do so because the result is one of our ends, or because it is our chosen means of bringing one of those ends about. Some claim that we also intend results that we know will happen, and that are sufficiently “close” to those we take to count in favor of our actions. The idea of “closeness” is notoriously difficult to unpack. But on any view, criminalization and punishment are too distant. They are different acts, typically carried out by different agents, at different points in time.

48. For this distinction, see Walen, Alec, Transcending the Means Principle , 33 Law & Phil. 427 (2014)Google Scholar.

49. To be clear, it is not enough to make punishment permissible in this sense that the punisher mistakenly believes that tokens of some act-type are wrongful and deserving of punishment. Permissible punishment is imposed for act-types that are believed to be, and actually do, satisfy (W) and (D).

50. Consider, for instance, the list of aggravating factors drawn up by the Sentencing Council for England and Wales: https://www.sentencingcouncil.org.uk/explanatory-material/item/aggravating-and-mitigating-factors/.

51. True, many offenders will have no intention to kill. But we already saw that prosecutors and judges may exercise discretion to ensure that these offenders are not punished. And we saw that even if some offenders will be impermissibly punished, it does not follow—at least for the nonabolitionist—that criminalization is by that very token impermissible.

52. Moore, supra note 13, at 28.

53. Assuming, with Moore, that to deserve punishment one must have acted wrongfully.

54. For a range of challenges, see Victor Tadros, The Ends of Harm (2012), at 60–87.

55. I already mentioned that this is true in English law.

56. One might reply that the motivational thesis can be weakened. It might be permissible to punish people for φing if a certain percentage of those who φ will be deserving of punishment. It is not clear, however, that this suggestion can be made to work. What percentage is required? Fifty? Seventy-five? And how can we know when we criminalize what the percentage of deserving offenders will turn out to be? Everything depends on the dispositions of legal subjects, as well as on the scenarios in which they find themselves. It is not clear that the motivational thesis, so understood, can usefully figure among the conditions of permissible criminalization.

57. By over- and underinclusive here, I mean over- and underinclusive relative to the limits set by (W) and (D). The safe possession offense described above is an overinclusive offense. The defense of self-defense described above is an underinclusive defense.

58. See supra note 52.

59. Recall the distinction, drawn in Section I, between offenses and crimes. An action is a crime only if it is both a criminal offense and no defense is available to those who so act.

60. Committing murder in self-defense is not wrongful. Committing robbery under duress is not deserving of punishment.

61. Ratios of the kind discussed in the text are often used to defend particular standards of proof. For criticism of this reliance in the absence of empirical evidence, see Laudan, Larry, Is It Finally Time to Put ‘Proof Beyond a Reasonable Doubt’ Out to Pasture? , in The Routledge Companion to Philosophy of Law (Marmor, Andrei ed., 2012)Google Scholar.

62. The most well-known defender of this view is Antony Duff. For a recent statement of his views, see Duff, R.A., Relational Reasons and the Criminal Law , in 2 Oxford Studies in Philosophy of Law (Green, Les & Leiter, Brian eds., 2013)Google Scholar.

63. And the rest can be offered by B at the sentencing stage.

64. It is frequently claimed that to criminalize φing is, inter alia, to threaten to punish those who φ. See, e.g., A. P. Simester & A. Von Hirsch, Crimes, Harms and Wrongs (2011), at 6; Tadros, supra note 54, at 269; Andrew Ashworth, Positive Obligations in Criminal Law (2014), at 33.

65. Berman, Mitchell, Blackmail , in The Oxford Handbook of Philosophy of Criminal Law (Deigh, John & Dolinko, David eds., 2011)Google Scholar, at 37.

66. This will be the case if it is worse for someone to be impermissibly harmed than harmed permissibly.

67. Or that A believes to be unwelcome.

68. Joseph Raz, The Morality of Freedom (1986), at 36.

69. William Edmundson, Three Anarchical Fallacies (1998), at 113.

70. Id.

71. Greenawalt, Kent, Criminal Coercion and Freedom of Speech , 78 Nw. U. L. Rev. 1081, 1098 (1983)Google Scholar.

72. I assume here that if one commits oneself to a course of action, it is rational to pursue that course even if one's commitment does not successfully create reasons to pursue it. This is to follow Parfit in distinguishing between that which it is rational to do, and that which one in fact has reason to do. See Derek Parfit, On What Matters (2011), ch. 5.

73. Similar conclusions to these are reached in Lamond, Grant, Coercion, Threats and the Puzzle of Blackmail , in Harm and Culpability (Simester, A. P. & Smith, A. T. H. eds., 1996)CrossRefGoogle Scholar.

74. Or what the speaker claims to be such reasons.

75. Raz, Joseph, The Problem of Authority: Revisiting the Service Conception , 90 Minn. L. Rev. 1003, 1018 (2006)Google Scholar.

76. Or is claimed by the speaker to exist independently.

77. Which is not to say that it would be impermissible to punish for these reasons. That is a further question.

78. Even Feinberg, who is more careful than most, sometimes falls into this trap. He begins Harm to Others by claiming that his topic is liberty-limiting principles that apply to criminalization. But he soon writes that his interest is in whether “criminal sanctions are special enough to require their own liberty-limiting principles.” Feinberg, supra note 21, at 24. Criminalization is one thing; criminal sanctions are another.