Regarding the effect of a surface finish on the guitar's sound, I think it'd be fascinating to hear a high-fidelity recording of a guitar that's been assembled 'naked', and then have it taken apart enough to properly finish it, reassembled and then repeat the recording.
That would eliminate any variables such as wood densities, grain differences, etc. Heck, if one was careful you could use the same strings, although using the same kind of strings from a common production lot would probably be just fine.
I don't know that anyone has done such a test, or whether any maker would want the results of such a test publicized but I do think it'd be really neat.
-K