DRMacIver's Notebook

Culture is deeply contingent

Culture is deeply contingent

This is an example I got from Dave recently which I thought perfectly illustrated something I've been thinking about on and off for a while, and was most recently prompted to think about in depth by The Origins of Unfairness.

When you are walking down the street and someone is coming towards you, do you step left or right?

Well, obviously you step right.

...of course, I would say that. I'm a Londoner. If you're from a different city and you are visiting London (or, theoretically, if I were visiting your city, but as I said I'm a Londoner, we don't actually leave London).

It turns out that which direction you step is a strong cultural norm that varies from city to city. Different cities pick different directions to step in, and it doesn't really matter which you pick as long as you all pick the same one, and because most interactions are within-city there's no strong pressure to conform between cities.

A decent lens to look at this through is evolutionary game theory. Everyone is adopting a mixed strategy where they step right with some probability \(p\). When two people interact, if they both step the same way they reinforce the decision to step that way (increase \(p\) if they both step right, decrease it if they both step left), otherwise they move towards stepping the other way (the opposite). This is of course a toy model, but it captures the idea pretty well.

If you look at the population average of \(p\), the chance that a randomly selected person will step right, what you find is that under these dynamics there are three points of equilibrium, that is points where this average remains constant over time: \(p = 0\), \(p = 1\), \(p = \frac{1}{2}\). However, these three are not created equal: The \(p = \frac{1}{2}\) point has two crucial features that the others lack: 1) It's much worse. People are miscuing about half the time, and everyone gets annoyed. 2) It's unstable. Although it's an equilibrium point, if you perturb it even slightly (e.g. \(p = 0.5000001\)) the value will tend to shift away over time, eventually ending up at whichever of \(p = 0\) or \(p = 1\) is on that side (this isn't quite true - if you're close enough to \(\frac{1}{2}\) randomness can cause you to switch which side you're on and oscillate for a bit, but eventually you get far enough away and from then on it's essentially a done deal). As a result, what happens is that essentially arbitrary starting conditions get amplified into one of two simple deterministic states, where everyone adopts the same convention.

An important feature of this model is that this is only true in interacting populations, which is why we see different effects in different cities. There's no reason to expect people to pick the same left/right convention if they don't interact, because the convention is entirely arbitrary and has no significance except to be the same as everyone else's. Where people don't interact, they likely create different norms, because the norm is the result of a mix of evolutionary pressure on the strategies and the entirely arbitrary starting point they happened to be in.

A lot of culture is like this. It exists not because of any particularly good reason, but because you needed to do something and this is where you happened to end up. If culture is the way we do things around here, culture is by necessity deeply contingent.

Two consequences of this that matter:

  1. When you join a new culture, expect a lot of things to be different than you expect for no particularly good reason except that's how they happened to pan out here.
  2. If you want to change a culture, think in terms of the evolutionary aspects of it. You might be able to do it by just shaking things up and letting it slide into a new equilibrium, but this only works if there's an equilibrium where you want to be. Fighting the adaptation pressure is a losing game.