Its a matter of degree though. You could say that the record is made of individual molecules but they fit together perfectly and occur at a scale that makes it insignificant. Similar with digital. It all depends on sampling frequency, sample size, (and accuracy of the device that creates the samples). Gets back to Nyquist Theorem or similar models assuming the minority opinion perhaps that Nyquist does not cut it as teh basis for CD format. I think it is an extremely close call in theory especially for younger better ears (although older ears are better trained perhaps even if not able to hear above 12-14Khz or so in general) but a good one in practice. Plus, as time goes on and technology improves and becomes more affordable, teh bar can be raised further if needed until it finally becomes clearly insignificant, like those molecules.
Digital is clearly improving all the time. Vinyl format stopped getting better probably almost 50 years ago now. The conclusion down the road seems inevitable if not already the case.
We had a $72,000 digital system at a recent show. I had heard an earlier version several years earlier and back then it was easily the best I had heard, and this new version was even better. The designer was in the room with us, has an LP system (a good sign), and upon hearing our analog system in the room, turned to me and said 'digital has such a long way to go' and sighed...
Molecules are no comparison to bits, its really not an acceptable analogy.
Analog also continues to improve :) It did not stop at some sort of roadblock 50 years ago. I guess you could say its my opinion that analog is much more accurate. Also, I try to be careful not to assume that one example of the medium, whatever it is, is representative of the whole, just as one playback of such is not either.