Earlier quoted context omitted.
Using name is a reasonable way to optimize figuring out someone's race. The question is: why are we trying to figure out someone's race? If you're talking about reducing effort, isn't the easiest thing to just not ponder the question of someone's race? People wouldn't bother to try and answer that question if they didn't care about what the answer was. Now, the fact that people care is not in and of itself evidence o…
That's roughly right in a moral sense, but in a Bayesian sense I don't think it's unreasonable. If race (or background or country of origin, etc.) and some attribute (education, wealth, etc.) are correlated, and names and race are correlated, it's intuitive that names and that attribute are likely correlated. It's a huge problem when you reverse the process and say, "Despite evidence of having X attribute, I'm going…
A lot of those assumptions that people make, while statistically justified, will never be rebutted in individual instances because it's socially impractical to do so.