Dienekes has a nice post up on the follies and failures of historical population genetics. As he notes, part of the problem is that people really want to find x instead of !x, and with a statistical science that is really, really, bad.