Larry Moran with a good post about what many scientists still get wrong about natural selection:

Most people think that beneficial alleles will always become fixed in a population. That’s one of the most important misconceptions about natural selection and it’s a shame that it was left out of the article.

This misconception is behind much of adaptationist thinking. To them it seems to be sufficient to postulate a benefit, no matter how small, and it automatically follows that the entire population/species will acquire it. The reality is that such adaptionist thinking requires two separate components: (1) the existence of a possible beneficial allele and, (2) the demonstration that the postulated benefit is of sufficient potency to lead to fixation with a high probability.1

There’s one other misconception that’s missing. Many people think that natural selection only occurs when the environment changes. This is formally equivalent to a belief that, in a stable environment, all species become perfectly adapted so that no further adaptation can take place. There’s no evidence to support this concept. It requires that most species are sitting at the top of an adaptive peak.