[jknoepfler]

One also expects the most variance at the tail-end of the success curve, because the sample size is so small. Give it a few centuries and we'll have some data about that part of the spectrum of success stories.