Pretty and non-pretty people don't form historically separated populations,

They don't have to be physically separated to be reproductively separated. I think there is some segregation on attractiveness, but not that much.

attractiveness isn't known to be correlated with numerous non-superficial genetic differences the way race is (e.g. genetic diseases).

Of course it is! There is a huge correlation with health, often revealed through things like parasite and disease resistance.

Those points are true, but I stand by my advice. I don't believe the difference in the amount of evidence required is tremendous, and it is a natural tendency among humans to underestimate this amount in any case.

I made it up to be commensurate - it has no visible basis in evidence. I could try to reverse-engineer the details of my thought-processes, if you like.

Only if you also assume that there are many robust factors with long histories contributing to do the following:

1) Encourage you to think of pretty people as an out-group

2) Strongly bias you towards considering garden-variety not-niceness in pretty people indicative of sociopathy, while doing no such thing about garden-variety not-niceness in ugly people

3) Prompt pretty people to act more sociopathic in a variety of circumstances due to psychological factors working on them

4) Make examples of pretty sociopaths dramatically more accessible in media and public cached thoughts than examples of ugly sociopaths or pretty non-sociopaths

Those variables are not changed in this hypothetical - the only variable that changes is that I propose that "pretty people are more likely to carry genes for sociopathy" in all seriousness, rather than counterfactually.