The Schelling segregation model is meaningless

We are not always aware of our external costs and benefits. Even if we have zero knowledge of them, over time, with low enough cost of relocation, populations will segregate on the basis of their externality profile.

The Schelling or Sakoda model provides simulations which show this process at work. But, the model is meaningless. What matters is externalities not attitudes or biases or 'segregation' per se. Why? If externalities are out of whack with attitudes, it is attitudes which get pruned by economic processes.

Furthermore, the existence of arbitrage between discoordination games will cause segregation to break down in any case. Thus we can expect the poshest segregated clubs to have more diversity. This works the other way as well. The most racist places with the worst attitudes too will have the most diversity.