You are all over this thread putting words in people's mouths instead of discussing what they are actually saying. You are even painting an entire class of study with a broad brush in a guilt-by-association way. Are you sure that's actually helping your case and the conversation here?

> These studies have known drawbacks in that they serve as a rallying point for all sorts of discrimination masquerading as high-minded discourse, and we're yet to see a benefit so far, so the burden of proof here is on you.

You're arguing for deliberate ignorance because reality is uncomfortable.

> At least part of the problem here is given two children / job-applicants / defendants / whatever, who are absolutely identical in every way except for a specific cosmetic phenotype, what possible justification is there for treating them differently based on correlation between that phenotype and some other characteristic over a large population?

If two people are identical in every observable way, and that single observable way indicates a strong difference between them, why wouldn't one differentiate based on that single observation? They are, after all, indifferentiable in every other way.

Indeed, mathematically speaking if one ignores the single useful datum you posit, then one will make worse decisions: if you have no other data and just flip a coin between greens and blues in your thought experiment a few posts back, then you will choose worse than if you bias your coin towards greens. That's just a fact; whether you choose to ignore it based on the discomfort it causes you is up to you.