Many social norms have the format ‘You shouldn’t do X. If you do X, people should punish you. If people don’t punish you, other people should punish them’.

People mostly pick up the specific norms from observing one another and discussion about particular cases. That is, they learn what counts as X exactly and what punishment for variations of X is apt or required under what circumstances by whom.

But learning in this way means they probably learn much more about the cases that come up and get tested, and less about norms that are almost never broken, or about behavior that breaks a common norm but in an unusual way.

For instance, if someone lies, it is common to punish them by making their lying known to others, which they are expected to fear because others will distrust them. But if someone egregiously and obviously lies in public, that is in some sense more extreme yet also a more unusual case that people don’t have any practice with. The usual punishment of making their lying public isn’t available. Are observers required to make up a new punishment? Are other observers required to punish people who don’t make up a new punishment?

A bit of uncertainty here quickly causes nothing to happen. Punishing someone in a new and unusual way, going above and beyond, is much socially riskier than doing the thing that everyone expects and others are doing. And in general the less people have common knowledge of a norm, the more it can break down. For one thing, even if you know the norm, it is easy to argue that you didn’t, or that you didn’t know you were meant to police it, or you thought it was a bit different.

So society seems likely to be under-prepared for norm violators who act unusually.

I don’t know if all this is right—I just think about it sometimes, and these are my thoughts.

A lot of (the majority of?) social norms are more like tribal norms and they function more like signals to the group than anything else. If you try to interpret them as actually having an end goal purpose that has an intrinsic link to the norm itself then that leads to confusion.

In my experience what happens is that a group basically just chooses some kind of immaterial action, like say, using some specific terminology, or as _rpd below mentions holding a door. If you say/do the thing that is information about tribe membership and “normies” then use that information to make all sorts of conclusions.

If you straddle multiple tribes then you usually just end up being seen as a kind of wild card / unpredictable. I think really it‘s just an attempt at energy saving, if you are categorisable you are easier to deal with.

Interesting thoughts. I have long worried that social punishment is far less effective in the modern age than historically, because family/communitiy ties are much weaker and acceptance of diversity is much greater. Norm enforcement is VERY similar to conformity-enforcement, and may in many cases be identical, so weakening the pressure to conform is also weakening the enforcement of norms.

I do agree that modern society hasn't found many acceptable ways to punish non-criminal deviations from norms.

The general punishment for severe norm violation is expulsion from the group - you get treated as an outsider rather than an insider. This is less available when the insider/outsider divide isn't as clear. I don't think the problem is that extra-weird behavior is not understood as dangerous to a conformist in-group. It's just that many groups have lost their power to shun weirdos. Telling blatant lies is a good example: the punishment is NOT that the lies are made public, the punishment is that liars are not invited to parties or trusted to do business with. Society has forgotten that you CAN judge people by the content of their character, so this punishment is less available.

Personally, it's a tradeoff I prefer - I like freedom and diversity, and it's OK that I don't have the high level of trust that (some members of) past generations in smaller communities had.

I'd say there are two basic kinds of punishment: direct (e.g. you get punched) and ostracism (people refuse to interact with you). Ostracism does not scale well, because it requires that a greater number of people remembers to avoid a greater number of transgressors. Direct punishment requires explicit rules, and invites rule-lawyering.

I think a distinction needs to be drawn between the social "norms" that constrain behavior, and the "norms" of policing other's behavior. The expectation of honesty holds in both of your examples, but there is only a policing norm for normal lying, and no norm at all for blatant social lying.

i.e. I don't think that "norms" are the right model for responding to the untold unique ways common behavioral expectations can be violated beyond the most common.

I think many changes in social norm fashion are intended to be hard to learn. The effort to be fashionable is itself a virtue signal. And when an established norm becomes unacceptable, or vice versa, there’s an automatic population partitioning and polarization that can be used to exploit tribal dynamics. Why is holding open a door both good and evil? The point is that it is impossible to know for any given participant.