As conceived by J.S. Mill, the harm principle holds that "[t]he only purpose for which power can be rightfully exercised over any member of a civilised community, against his will, is to prevent harm to others."(1) This essay concerns a narrower formulation of the principle, the claim that the criminal law should never punish conduct that does not harm unconsenting third parties. What I have to say about the harm principle as a limit on the justifiable reach of the criminal law likely holds, in spades, for any version of the principle that immunizes harmless conduct from civil or social sanctions, but I here advance no such claim. My concern is with the old and brilliant debate over morals legislation.(2)
The harm principle holds a central place in liberal theories of criminal law, and indeed in liberal political theory generally. Joel Feinberg goes so far as to say that "[l]iberalism, as I have understood it, is the view that" harm and offense to others exhaust "the relevant reasons for state coercion by means of the criminal law."(3)
My thesis holds that the harm principle is profoundly deficient as an expression of the limits that should restrict society's resort to its most coercive instrument. The harm principle's institutional deficiencies derive directly from describing the criminal law's proper limits in terms of the consequences of individual conduct which is to be immunized against punishment. Serious protection of individual liberty depends, I suggest, on limiting resort to the criminal sanction along procedural or jurisdictional lines, rather than according to the substance of private conduct.
The argument proceeds through four steps. First, I point out that, given the peculiar mission of constraining the popular impulse to punish, the harm principle should not be evaluated as a purely normative description of the criminal laws a just or good society would adopt. Instead, the harm principle should be evaluated as an appeal to individual actors to internalize a norm of tolerance, or as an external restraint to be respected even though not internalized. Second, I suggest that the harm principle is an unlikely candidate for an internalized norm of toleration. This is because the justifications for the harm principle run either too deep or too shallow. By that I mean that perfectionist arguments stressing the value of individual choice and the risk of official error fall short of categorically excluding punishment of harmless wrongdoing, while anti-perfectionist rights theories depend on premises that will never persuade a broad consensus in a free society.
Third, I suggest that the harm principle cannot fulfill the role of an external constraint on political choice. The improbability of justifying the harm principle as an internal norm of course makes it unlikely that any polity will adopt the principle as a constitutional limit on that polity's discretion to punish harmless wrongdoing. But even if the harm principle were realized as a constitutional constraint on state power to punish, the application of an immunity for private conduct based on its apprehended consequences would not provide a serious limitation on the actual practice of punishment. The idea of harm is too vague, too dependent on baseline assessments of private rights, too open to long chains of causal speculation, and too catastrophic in its categorical judgments to give liberty much practical protection.
Liberals should turn instead to content-independent limits on penal legislation. At the legislative level, I suggest a supermajority requirement for penal statutes, and a sunset provision terminating all such enactments unless readopted once a decade. At the judicial level, I propose investing constitutional courts with the power to insist upon even-handed enforcement of the criminal law. These approaches require some constitutional definition of criminal, as distinct from civil, sanctions. They …