"Here's one: sickle cell anemia."

Yep, that's a good example. I'm actually a carrier (I have sickle cell trait). I became a carrier because my father (African) is also a carrier but my mother (Caribbean) isn't. Both of them look just as black as each other. My sister, thanks to the vagaries of genetics, isn't a carrier. Even though she would tick the same box for "race" as me.

If I wanted to have a child, there'd be no way to know by "race" whether my prospective partner was also a carrier (if both parents are carriers there's a high chance the child will have anemia). The vast majority of black people don't have sickle cell anemia or trait and about 20% of the people who *do* aren't black. Sickle cell trait is a (fairly rare) medical reality, not a racial one. So, in this case, simply asking people whether they have sickle cell trait would be far more useful than asking them whether they have any genealogy from sub-Saharan Africa.

I'm not trying to argue that there aren't physiological realities regarding people from different parts of the world. I'm just saying that we can't meaningfully flatten these out by the concept of race. Melanoma risk, for example, is largely related to how dark somebody's skin is, which is broadly a function of how many of their ancestors lived near the equator. But at what point as we move away from the equator do people become a different "race"? What are the lines of latitude that separate black people from brown people from white people?

Yes, people with lighter skin are more likely to be at risk for skin cancer. Yes, people from sub-Saharan Africa are more likely to have sickle cell trait/anemia. It's useful to know these things. But while "race", by which I think you really mean skin colour, *feels* like a useful shorthand for people's differences, it almost never maps them accurately when you're trying to be even slightly precise.