The closest case I could imagine for this in humans are the australian aboriginals and papuans. You don't have a continuous gradient south-north between them and asian populations (assuming the Out Or Africa as true), but a sharp cut dated 50K years ago. Europeans are genetically closer to an asian than asians to papuans or australians.
Lets imagine a small group of engineers that develope a bad software pattern in full isolation. They don't know better, so they work in that way for years. Now suppose they get in contact again with the rest of the world. How could their bad pattern expand to the rest of the world engineers if it is inferior? The ones that addopt it would be selected against, the ones who doesn't would be selecter positively.
For it to expand it had to have at least something positive (unless you do a Chengis Khan). This is what happened to sickle cell disease, which is a bad mutation but gives you resistence to malaria, so the sickle cell disease is common in Africa. Meanwhile, African Americans have much less of the mutation, because in the USA there is no malaria, thus the gene is being slowly removed from the population.
https://onlinelibrary.wiley.com/doi/full/10.1111/exd.14142
When you analyze if attractiveness or height gives you a genetic/population advantage, you cannot stop at "they are promoted". They have to have more children, and the children have to also have genetic advantages.
For example, if you have a pale skin in Sweden 20.000 years ago, your vitamin D metabolism is better, so of your 5 children 4 reach adulthood healthy. Meanwhile, your black neighbor had 5 children and only 2 reached adulthood, and they have rickets. In a few generations, the skin gets white among the population.
But if you both lived in Africa, 4 of your 5 kids developed melanoma at young age. They either die or get horribly scarred. Meanwhile, your neighbor still has 4 healthy kids that never get sunburns. The amount of sunlight is so high that D vitamin is not an issue.