If you get to pick the sample universe like that, the argument
becomes a tautology, doesn't it? Pick a sample set ("scientists")
that you associate with "genius", and in which there are more
men, and therefore higher populations at the tails of the curve,
and QED, more male geniuses. In other words, there are more men
who are good at the kind of thing men are good at, which I will
call "genius".
In order for the more-outliers argument to be meaningful, you have
to be plotting a distribution of some objective measure. Then we
can begin to argue about whether the measure is in fact objective,
and what evolutionary and cultural forces may explain it.