[deadbeeves]

There are crappy combination padlocks that can be picked by turning the combination wheels and pulling on the shackle. When you've turned a wheel into its unlocking position the shackle will shift open slightly, signalling that you've cracked that digit. So a padlock with 3 digits could be opened in at most 30 moves, instead of 1000.

When you have a selection mechanism, the randomness of mutations becomes kind of moot. Each step of the way that produces a new adaptation only has a handful of optimal (or optimal-enough) solutions. Now, suppose that we're an underwater species that's in the process of developing light-sensitive cells, and from our present genome there are 2^64 possible genes that will produce a good protein for the transduction step (of converting light into some other form of energy). Do we need to mutate 2^64 times to find the "correct" gene? No, all of those 2^64 genes are "correct". Our descendants should not fall into the trap of thinking that because they got gene #5541741487894936799 that it was a special outcome. They could have just as easily gotten gene #5541741556614413535.