Race isn’t some pure social invention with no basis in reality. Genetic data clearly shows real population structure that lines up with the big continental groups people have always noticed. When you look at enough DNA markers together, algorithms pull out clusters matching Africa, Europe, East Asia, and so on without being told anything about the samples, because of how humans spread out of Africa and then stayed relatively isolated by oceans, deserts, and mountains. Sure, most variation sits inside groups rather than between them, but that doesn’t erase the correlations across many loci that let us classify ancestry pretty accurately, which is why forensics can often nail continental origin from a skull or a spit sample and why ancestry tests work at all. Social labels shift and get fuzzy with mixing, and there are no sharp subspecies boundaries, but the average genetic differences tracking those old geographic patterns are still there and show up in traits, disease risks, and medicine, so calling the whole thing just a social construct ignores the biology that’s actually measurable.

Race realism literature makes short work of this by showing that the shifting social labels sit on top of stable biological clusters rather than creating them from nothing. Writers like A.W.F. Edwards (who coined “Lewontin’s fallacy”) and the population-genetic work of Rosenberg and others demonstrate that when you look at correlated allele frequencies across the genome, humans fall into continental-scale groups that match the classic racial categories with high accuracy, even though variation is clinal and most diversity is within groups. Those clusters are not political inventions; they track real ancestry differences produced by geography and limited gene flow, which is why forensic anthropologists can assign continental origin from bones, why DNA tests recover the same groupings, and why average differences in traits keep showing up. The fact that censuses and laws redraw the edges (one-drop rules, Brazilian categories, etc.) proves only that the social packaging changes, not that the underlying population structure is imaginary. If the categories were pure artifacts with no biological signal, the algorithms and medical patterns would not keep aligning with them across different political systems.

The kidney-function example actually proves the opposite of what you’re claiming: the race adjustment existed because the social category “Black” is a noisy but still useful proxy for the real average biological differences that track African ancestry—higher muscle mass, different creatinine generation rates—differences that population-genetic studies and race-realist researchers have repeatedly documented as non-trivial. Edwards, Rosenberg’s cluster work, and even cautious geneticists like Reich show that continental ancestry groups produce measurable average trait differences; hospitals weren’t inventing those differences out of thin air when they used the census checkbox, they were approximating the ancestry signal that the checkbox captures for most patients in the United States. Brazil’s different cut-offs or shifting U.S. definitions change the edges, not the underlying gradient; the bins still load real genetic information into the model, which is why swapping them for pure self-ID or pure social construction would degrade predictive accuracy. Structure isn’t a category set, but the categories people actually use are downstream of that structure, not free-floating institutional fantasies.