That guy on the internet was dead wrong in thinking that there was enough gene flow in human prehistory to materially interfere with local adaptation. For a polygenic trait, that would take something like a couple of percent a generation. For an adaptive Mendelian trait, you’d need more influx ( as a fraction) than the selective edge of the causal mutation.
But there’s another interesting question, which he would presumably also be wrong about: how often does even a single copy of an adaptive mutation manage to travel far, and how long does it take? I talked about this here, some time ago. But examples help.
HbS, the sickle-cell mutation, happened once about 7,000 years ago, and has spread to regions with a lot of malaria in Africa, parts of southern Europe, the Middle East, and India.
But it never got to southeast Asia or New Guinea ( until 2002) , even though there’s plenty of malaria there, and a number of local genetic defenses against malaria. Parenthetically, although HbS is found today on different haplotypes, and many people suspected that it had multiple origins, complete absence in Southeast Asia/Melanesia was always a strong hint that it had a single origin.
None of those defenses in PNG or Southeast Asia ever seem to have made it to Africa. And during the last 7,000 years, there were an increasing number of long-distance dispersal mechanisms that didn’t exist before the Holocene: lost-distance trade, ships, the slave trade, etc.