> But the combinations of amino acid sequences they would have to search through in order to find the correct shape is so large
Evolution isn't looking for that sequence.
It is looking for any change in sequence with a positive payoff, and in the meantime constantly diversifying sequences with similar outcomes, creating more opportunities for serendipity.
Every large animal is born with mutations. So we are also quite robust to spreading the search, running multiple experiments at a time, taking small risks with genes not quite as good, which will get weeded out quickly when combined with other weaker genes, but in the meantime cast a wider net for meshing with another gene that complements it.
So yes, in any given species with a nontrivial population, millions or billions of genetic variations are being explored at any point in time. We are nothing like carbon copies of each other, differing by just a couple checkmarks.
This is a radical speed up. Just as sexual recombinatory reproduction is. Evolution today operates with vastly more efficient genetic environment, structures and systems than what early life did.
Tractable statistics do no justice to how biology works and all the paths it searches. I am not knocking formal statistics at all, just noting that past one or two step events, the layered statistics of chemistry, genes, gene clusters, epigenetics, populatoin dynamics of complex creatures in their complex environments, etc. are not going to be tractably modelled.
Measurable sometimes for sure, but not symbolically characterizable or calculatable.
> It is looking for any change in sequence with a positive payoff
The entire reason I used the example of the combination lock was so that hopefully you would understand situations in which "any change in sequence with a positive payoff" is a nonsense concept. You either get the correct combination or don't get any positive payoff whatsoever.
When you are trying "millions or billions" of genetic variations are being explored at any point in time, as you point out, you are off by so many orders of magnitude in terms of the search space that you need to explore that the problem is intractable.
You may as well tell me that you can guess a random password that I come up with because you have a way to test millions or billions of combinations at a time. That's great, but I can come up with a random password that it would take you from now until the end of the universe to guess even if you guess quadrillions of possibilities at a time because I can create a password that would require you to test 10^84 possibilities.
Evolution isn't looking for that sequence.
It is looking for any change in sequence with a positive payoff, and in the meantime constantly diversifying sequences with similar outcomes, creating more opportunities for serendipity.
Every large animal is born with mutations. So we are also quite robust to spreading the search, running multiple experiments at a time, taking small risks with genes not quite as good, which will get weeded out quickly when combined with other weaker genes, but in the meantime cast a wider net for meshing with another gene that complements it.
So yes, in any given species with a nontrivial population, millions or billions of genetic variations are being explored at any point in time. We are nothing like carbon copies of each other, differing by just a couple checkmarks.
This is a radical speed up. Just as sexual recombinatory reproduction is. Evolution today operates with vastly more efficient genetic environment, structures and systems than what early life did.
Tractable statistics do no justice to how biology works and all the paths it searches. I am not knocking formal statistics at all, just noting that past one or two step events, the layered statistics of chemistry, genes, gene clusters, epigenetics, populatoin dynamics of complex creatures in their complex environments, etc. are not going to be tractably modelled.
Measurable sometimes for sure, but not symbolically characterizable or calculatable.