pretty nearly optimal

"We cannot assume that what ended up surviving is necessarily optimal, given the conditions. Nor that brute force is the most efficient way to find a global maximum."

No, we can't assume it - but it beats out anything else we've tried in general.

If selection doesn't always reach optimal results, it tends to get pretty damn close.

This is even true for things we think we have designed: We design cars, but the cars that become most common are the ones that people ended up liking and buying more of, because they worked so well; We design buildings, but the building styles that remain are the one's that didn't fall down or otherwise fail. At some level we know there have been many failures, but we don't think about them because they are (to us) obviously bad designs - but they weren't obviously bad designs to the designers, who probably understood the subject better than most of us do.