It has become a habit of public argument to describe disputes over free speech as a contest between those who believe in the right and those who do not. This framing flatters both sides and clarifies nothing. Pressed with a concrete example -- a false claim that a named private citizen has committed a crime, a speech that names a specific person and calls a crowd to attack them within the hour -- almost no one, on either side of the supposed divide, actually maintains that such speech must be protected simply because it is speech. The absolutist position, stated plainly, has almost no real adherents; what looks like a disagreement over principle is usually a disagreement over where a line already conceded to exist should be drawn.

Three categories of restriction command something close to consensus, and examining why clarifies the deeper question. Direct incitement -- speech calling for imminent violence against an identifiable target, with a real prospect of being acted on -- is restricted because the harm is immediate, concrete, and not mediated by any further choice on the part of a rational listener weighing the claim. Defamation is restricted because a false factual claim about a person is not an idea open to debate but a falsehood capable of being definitively established as such, and one that inflicts a specific, locatable injury on a specific person. Both categories share a feature worth naming: the restriction targets not the viewpoint expressed but a narrow, identifiable harm that exists independently of whether the underlying opinion is popular, unpopular, true, or false.

The harder case, and the one that generates most contemporary argument, is speech that causes no such narrow, locatable harm but that a great many people nonetheless find offensive, humiliating, or corrosive to a group's dignity. Here the case for restriction weakens considerably, not because the offence is not real, but because the harm is no longer separable from the content of the opinion itself. To restrict speech because it offends is, in practice, to restrict it because of what it says, and a rule that operates that way cannot be applied without someone deciding, opinion by opinion, which offence is serious enough to warrant suppression and which merely regrettable. That decision, history suggests repeatedly, tends over time to be made by whoever currently holds power, and to be made in that power holder's own interest rather than in any neutral, principled way.

This is not an argument that offensive speech is costless, or that its targets have no legitimate grievance. It is an argument that the machinery built to suppress it is far harder to confine to genuinely dangerous cases than its proponents generally admit, and that a state trusted with the power to decide which opinions are too offensive to permit is a state that will, sooner or later, use that power against opinions it merely dislikes rather than opinions that cause any locatable harm at all. The line between direct incitement and mere offence is not always crisp at its edges, and reasonable people will continue to argue about exactly where a specific utterance falls. But the categories themselves are not arbitrary: one asks whether a concrete, identifiable harm exists independent of the opinion's content, and the other does not, and that distinction, however imperfectly applied in hard cases, remains the most defensible ground on which the boundary of a genuinely free society can be drawn.