To most physicists, almost everything about evolution is puzzling. What determines how much complexity can — and has — evolved? Why is there so much biological diversity on all scales of differences? How could evolution select for future "evolvability"? Why doesn't evolution get slower and slower? One of the roles of theory is to frame questions — or at least limited versions of them — more precisely. And another is to reduce — or redirect — puzzlement by developing and analyzing simple caricature models to see what should be expected rather than be surprising. This talk will outline some recent progress in these directions, particularly by statistical physics ways of thinking.