[throwaway4889]

No one is saying evolution is all random, only mostly random. Natural selection obviously plays a role in shaping the direction of individual mutations - deleterious ones get purged relatively quickly. When those are culled, the vast majority of mutations are neutral or nearly neutral.

>Brute forcing the genome to something better is clearly possible, but if you converted all of the matter in the visible universe into DNA and mutated every strand a trillion times a second for a trillion years, it's very possible you would never even encounter a viable strand of DNA for that organism.

Yes, but consider the following. I just ran:

>>> np.random.random()

0.8867453976799686

What were the odds that I ended with this specific 18-digit float in the whole set of floats in [0.0, 1.0)? And yet, here we are.

The word you're looking for is contingency. Gould spoke much about "replaying life's tape" and how it would yield different results every time it was "replayed". Interestingly, people have actually attempted to put this thought experiment into practice at a very prosaic level [0] [1]. They find that although there are convergent patterns of adaptation, over the long term populations diverge considerably.

[0] https://science.sciencemag.org/content/362/6415/eaam5979

[1] https://sandwalk.blogspot.com/2019/09/contingency-selection-...