[davisYC]

you can infer sizes from genetic diversity ('effective pop')

for the out of africa group i've seen sizes from 100 to 10,000. i used 1,000 to 10,000 as a likely moderate number, and the range also points to the census size being larger

we can assume that conditional on the size of OoA the other groups were way larger (10x or more). otoh, the main thing i did not explore is that i am now believing that africans themselves especially agriculturalists are 'compounds' of various bottlenecked groups (why west africans are closer to eurasians than they are to khoisan)

[AlotOfReading]

It's worth noting that effective population size != actual population size. The former is simply a lower bound on it. If the population is structured enough and in the right ways, the actual size can be orders of magnitude higher. This has been an issue with several supposed bottlenecks within African populations.

[davisYC]

the rule of thumb for mammals is Ne is 30% of census size.

i think this is probably an overestimate tbh (assumes poisson distribution of offspring).

[AlotOfReading]

I'm not sure that's super applicable to humans. Doesn't it also assume panmixia and non-overlapping generations?

Either way, my understanding is that all indications point to middle and late Pleistocene populations being highly structured and "unusual" as far as standard assumptions go.