> Whereas racial categories like these are socially constructed, and do not correspond all that well with genetic population groupings.
If you look at genetic population groupings, the gap between populations that the layman would identify as "black" and "white" is as wide as the Sahara Desert. Literally so, if you put these population groupings on a map.
It's true that, if you look at New World, non-indigenous populations, things get more complicated. But genetic populations still correlate very highly with geography and ethnicity. The layman'a notion of "blackness" or "whiteness" may be flawed, but it's not completely incoherent.
EDIT: Some replies suggest that I may have been unclear. The social grouping into race is obviously not random, but based on perception of visible morphological attributes. However the categorization that results is probably not the same categorization that you would get if you looked at the whole genotype and tried grouping humans together based on genetics. One basic fact of human genetics should guide you here: there is more genetic variability within any socially constructed "race" than between "racial" groups. So claims that common racial categories correspond to big differences in genotypes don't make sense. Whether there are races at all in humans is something that biologists disagree about, with most at the moment saying "no". Whether race is useful to medical diagnosis is also contentious. The human species is unusually genetically uniform. Most animals that all look about the same to us, like chimpanzees, have (in that case) about three times our genetic diversity.