You include a good countervailing point in your comment, which is that our experiments are largely confined to what we think are ideal conditions. Evolution (or

You include a good countervailing point in your comment, which is that our experiments are largely confined to what we think are ideal conditions. Evolution (or, whatever you want to call this abiotic Darwinian selective process) searches the space of possible designs randomly, which means it can happen on "improvements" that would be hard to predict a priori.