That's my guess too, but my argument is less that this arises from a specific place of malice or intention, and rather this phenomena comes from an uncritical a

That's my guess too, but my argument is less that this arises from a specific place of malice or intention, and rather this phenomena comes from an uncritical approach to handling the dataset. There is, for instance, no reason not to group all the CJK data as "Asian" for the sake of similarity with the rest of the ethnicities. But in uncritically reflecting the dataset, it introduces downstream effects that could potentially be amplified in later work.