> I didn't see any misunderstandings of evolution in the book
The greatest feature of evolution that it makes teleology unnecessary. It can operate as a mindless/goalless algorithm. Of course evolution may be operated by someone/something with specific goals instead of pure chances, but thats just an unnecessary assumption until it proved - so we may employ Occam razor here until someone proves it. The possibility of design without a designer is one of the main point and contribution of the theory and it should be fully understood - even if if later turns out (which have not happened yet, and no sign for that happening beside metaphysical opinions...) that there was a designer.
> of actually existing computers is also extremely complex, and not obviously any less complex than the mapping from these states to physical states of rocks
You can't be serious:
First map a Computer chess program states to human interpretable visualization to play chess against it
And then do this with your rock.
First is mainly one-to-one mapping of digital signs to pixels by using a static table, the second is..., I'm not even sure how to inject my move into the states of the rock, but no idea either how to extract its response. You know why? There is simply no one to one mapping there. The states of a rock is a large bundle of everything without interpretation, the states of the chess program is CLEARLY more constrained - the proof is empirical: it CAN be implemented.
Like if you say that the set of all positive integers contains the solution of factorizing any large number. Of course it does: you just have to point to those numbers. ;) But how? All states are not the solution - selecting the proper states is the computation, and if those selection mechanism is in the mapping process then the mapping IS the computer.