[snovv_crash]

The problem isn't the clustering. The problem is choosing which individual points belong to which clusters, because every time people have mixed-race (or mixed-mixed-race) or whatever kids, which has been happening for millennia, a new 'cluster' gets invented.

Basically, there are no nice separations between the clusters, there are just denser and sparser regions in the feature space, so assignment of a point to a single cluster doesn't make sense.

[dekhn]

it's almost a textbook case for unsupervised learning. Like, almost as if you could train a manifold that separated out people by genetic history (much of which was super-regionalized, and did lead to large phenotypic differences that clustered spatially)... and in that manifold "mixed-race" people would probably fall on points between two well-separated clusters.

In fact there are entire scientific papers (mainstream) doing exactly this. We've learned all sorts of interesting things, like the total genetic diversity in africa is larger than the african/eurasian diversity split.