In this context it would be interesting to see how the current practices in affirmative actions would change under different metrics. For example, what if affir

In this context it would be interesting to see how the current practices in affirmative actions would change under different metrics. For example, what if affirmative action was mostly calculated in terms of a combination of local Gini coefficient and yearly income. Would it in practice cause the same "re-leveling" of the playing field? In the same line of thought there was a nice blog post on the (temporarily deceased Slate Star Codex) about how country of origin of ancestors in white Americans of European descent have strong statistical effects on income/class today. the same for African immigrants. From this perspective the question becomes "can we improve the effect of affirmative action on fighting racial social segregation by using non-race based metrics?"