>I would count 20 years, possibly less, as a generation during
>most of that time, which would push it to 500 generations

That's what I've always intuitively thought, too.  However, I say 300
because all of the material I've read that gives the '10,000-year total
generations for humans' figure reports it roughly as such, and I just
assumed that there was an underlying mechanism (mathematical or otherwise)
beyond my immediate grasp that lengthened the supposed period between
generations, and never bothered to investigate for details.