> As long as the difference between individuals in a single population is greater than the mean difference between populations themselves, I don't see much relevance.

This part is the Lewontin's fallacy (http://en.wikipedia.org/wiki/Lewontin%27s_Fallacy)