Why Towns Conform
Imagine a group with a social norm against spitting on the sidewalk. If this group is very small, then everyone should know everyone well, and any one act of spitting will have only a small influence on how people think about the spitting person. A consistent habit of spitting might cost you, but any one spit would not. If the group is very large, there is also little risk that any one spitting will result in an informal social sanction. You’ll probably never meet the strangers who see it again, and they probably don’t know each other, so why should anyone make a fuss? At an intermediate scale, however, spitters should fear that any one act of spitting will produce a widespread rumor about this act, making folks who know them only moderately avoid them after hearing this rumor. Why deal with someone if you have other options and the main thing you know about him or her is negative?
In general, social norms are enforced via two key informal mechanisms:
When norms are usually followed, rare violators are often undesirable in objective ways. They may lack intelligence or self-control, for example. So people avoid violating such norms to avoid sending bad signals about themselves.
Meta-norms often require observers of norm violations to punish violators, such as by refusing to associate with them. This includes observers of a failure to punish a failure to punish, and so on.
These two mechanisms play out differently on three different social scales:
Foragers only interacted with a hundred or so others, all of whom they know in great detail.
Farmers lived in larger social networks of roughly thousands of folks near enough by to matter. This is small enough for rumors to tell most everyone about big norm violations, but too big for everyone to know everyone well.
Today we live in communities so big that, outside of smaller networks of neighbors or coworkers, rumors only reliably tell everyone about extreme norm violations. Informal rumors will not tell most people you deal with about your norm violations.
These two norm enforcers seem to work best at intermediate social scales. Signaling discourages norm violations best when people that matter tend to hear about norm violations, but know little else about violators. At a smaller scale one norm violation will add only a small amount to what observers know about that person, and at a larger scale observers will probably not have heard about the norm violation. But inbetween, observers will prefer to avoid someone when they know little else besides one bad sign.
Meta-norms to punish non-punishers also work best at an intermediate social scale. At a very small scale, when few observers see each violation, observers can coordinate to avoid the meta-norm of punishment; “let’s not and say we did.” Punishment can be expensive, after all. At a very large scale, you many care little about the opinions of those who happen to see you fail to punish a non-punisher. But at intermediate scales, a single bad signal can induce a strong shunning reaction. Why take a risk on a near stranger with a big negative strike against them?
The fact that norms are enforced best at an intermediate social density helps explain why higher-density farmers had stronger social norms than lower-density foragers, and yet even higher-density modern folk have reverted back to a weaker forager-like level of norm enforcement.
Am I the only one to whom this post smacks of this description of phlogiston theory? This is just my knee-jerk reaction talking here, but it feels like you're stretching a little too hard to make this essay work.
While I think the degradation at the high end of the population density scale is adequate, I propose a different - though perhaps equally-forced - mechanism for the transition from low to medium:
As a poor Forager, you do what you need to do to survive; everyone understands this, and there are very few social norms at all except those that really, really matter. People in survival situations often have trouble behaving rationally because they're not comfortable with instantly shirking a huge number of social norms in their shift to an extremely low available population - a theme often played-with in fiction during e.g. a zombie crisis, or being lost in a desert.
As a Farmer in a moderate environment, you are afforded ways to attempt to elevate yourself socially. Ornamentation becomes a way to specifically signal to what degree one need not concern oneself with mere survival. Ornamentation can be long fingernails and gold; it can also be the accumulation of etiquette. In addition to this, you are afforded more leeway for behaviors which serve only purely psychological purposes and are objectively quite inefficient in terms of maximizing food energy available.
This effect at the lower shift and the one in the article at the higher shift can be considered two unconnected forces with ramifications at vastly different scales. As an example, I present the arbitrary norm-observance function to population variable x: f(x) = 1000x - x²
What I prefer about this explanation is that it more cleanly fits into the supposition that we are Forager people by default, that is to say preferentially, where we only favor forming and favoring norms when we have enough food to not be desperate to survive, but still low-enough populations that the anonymity effect doesn't kick in. Under the OP model, on the other hand, we are Farmer-type norm-followers by default, but simply empathize too strongly with others at low populations. This requires additional explanation for why we appear to prefer to be Farmers but are Foragers at low populations, prefer to be Farmers and are so at medium populations, and prefer to be Foragers and are so at large populations. If you venture such explanations, it would be conscientious of you to affix a dotted line with a label reading "APPLY OCCAM'S RAZOR HERE".
Key question is whether farmers have 'stronger' norms, or simply 'more' norms. 'More' seems a good candidate: higher productivity -> more resources left for norm industry.
My strong bayesian is foragers have at least equally 'strong' norms - belief in curses, black magic, evil eyes, at least equal propensity for vendettas, honor-killings, etc.
Probability-adjusted punishment would actually predict 'stronger' norms among less dense people. I think the modern city is more a matter of lowered consequences to violation as a result of higher incomes.