Do we have good intuition or data about which operations need speeding up most? Everybody always assumes it's BINARY_ADD, but much Python code isn't actually numeric, and binary operations aren't all that common. (For example, IIRC at my previous employer the hottest code was some kind of web template expansion and another bottleneck was marshalling RPC requests and results.)

Optimizers usually evolve to be excellent at optimizing the benchmark du jour -- what benchmark do we want to use?