Tweet · 19 Mar 2013, 08:28 UTC

Replying to @philarcher1

I can think of three cases in which it can be wrong (i.e. produce negative fitness) for a P to multiply (too much).