Punishing what harms no one else

The following is a plausible Liberal principle:

1. It is only appropriate punish that which harms someone or something else, or is intended or sufficiently likely to do so.

(If one thinks that exposing someone to a sufficient probability of harm is itself a harm, one can simplify this. Note also that what counts as sufficiently likely is relative to the degree of punishment and degree of harm. Doing something that has a one in a million chance of causing me a hangnail is probably not deserving of punishment, except maybe of the most trivial sort, but doing something that has a one in a million chance of blowing up New York may well deserve serious punishment—cf. Parfit on small chances.)

I shall argue against this principle. Recall Mill's very plausible insistence that:

2. Being subject to social opprobrium is a kind of punishment.

(And one often would rather pay a hefty fine than be subject to social opprobrium, so it can be a heavy punishment.) Now observe:

3. Some irrational beliefs are appropriately subject to social opprobrium even though they harm no one else, are not intended to harm anyone else and are not sufficiently likely to do so.

For instance, consider someone's really crazy conspiracy theoretic beliefs which were formed irrationally, out of a desire to be different, rather than out of an honest investigation of the truth, and suppose that this is someone whom no one is likely to believe, and hence someone harmless in these beliefs. Or consider the racist beliefs of someone who is too prudent to ever act on them because she does not wish to risk social disapproval.

Therefore:

4. It can be appropriate to punish something that harms no one else, and is neither intended nor sufficiently likely to do so.

Now, one can get out of this consequence if one makes some sort of a communitarian assumption that no man is an island, that one person's irrationality is a constitutive part of the community's being thus far irrational, and is eo ipso harmful to other members of the community even if they do not themselves follow this irrationality, since now they are made to be participants in a community that exhibits this irrationality. But if one allows such "extended harms", then the principle (1) becomes uninteresting. Likewise, if one brings in "extended harms" to God, where God is said to be harmed in an extended sense provided that one acts against his will.

Could one turn this around and make it an argument for tolerance of irrationality? This would involve insisting on (1) and concluding that harmless irrationality should not be the subject of opprobrium. Yet such opprobrium seems to be an important part of what keeps us rational, and it seems obviously appropriate, especially when the irrationality is a result of the agent's moral failings.