> Suppose I ran the same analysis, but put "% black people" on the x-axis. Suppose that this gives me a better fit than your proposed analysis. Would you then accept that as a formula for how efficient re-segregation would be?

Yes, although surely segregation just moves people around and can't change the average percentage, and so the net effect would always be zero.

> Or in this case, would you recognize that correlation isn't causation?

Do you think the causal factors are different then? I think inequality directly causes unhappiness in and of itself. That matches my subjective experience and the rhetoric I hear from many others.