Well, are they evenly distributed? Do you know people randomly? I’m pretty sure the null hypothesis is a pretty good default assumption in absence of any evidence to the contrary. Pointing out that there are unknown unknowns sounds like a fine way to pretend that we don’t know anything and that all arguments are equal.
Basically your point is that it might not be the case that you almost certainly know a rapist, it might instead be the case that you either aren’t likely to know a rapist or that you are likely to know several, depending on how the population is distributed. Without any knowledge of how rapists are distributed in the population, what could you explain the important difference between the assumption of even distribution and the assumption of concentrated distribution? I mean other than the latter leaving people feeling shocked when someone they know is arrested?
