Telling Groups Apart Is Not the Same as Explaining Them
2026-08-11 · AI-generated insight
These two notes look like combatants in the same war, but read carefully, they are answering different questions — and noticing the difference clarifies what the race debate is actually about.
Winegard and Carl make a statistical point against Lewontin: small differences at any single genetic locus can aggregate into robust population structure when you look at correlations across many loci. Their face analogy is elegant — nose size alone can't sort men from women, but whole faces can, 95 percent of the time. Conclusion: human populations are biologically distinguishable, so race is not a mere social fiction at the level of classification.
But notice what Boas, as Degler describes him, was actually claiming. Not that populations are physically indistinguishable — Boas was a physical anthropologist who measured skulls for a living and knew perfectly well that groups differ in appearance. His radical move was about causation of social outcomes: observable differences in how societies function are products of different histories, not different biology.
The face analogy, ironically, supports Boas as much as it rebuts Lewontin. Yes, you can classify faces by sex with high accuracy — but the features doing the classifying (nose size, jaw shape) tell you nothing about the mental or social capacities of the person behind the face. Classifiability and functional significance are independent properties. A trait cluster can be statistically real and causally trivial for the outcomes we care about.
This suggests the perennial argument persists partly because the two sides talk past each other. One side hears "race is biologically real" as a claim about innate mental differences; the other hears "race is a social construct" as a denial of visible population structure. Boas's position — real physical variation, historical explanation of social variation — is compatible with everything in the Winegard critique. The genuinely contested question is not whether populations can be told apart, but whether the traits that tell them apart carry any weight in explaining civilizational outcomes. Winning the classification argument settles nothing about that.