Not everyone agrees that the goal of a playback system is realism —> some look for reproducing everything on the recording.
I believe them to be one and the same.
Yes, I think in principle that would be the case.
In practice the variability of recording quality, the various distortions added by choices of micing/mixing etc, and then any possible colorations in speakers (even "neutral" speakers) tend to break that game of "telephone."
So, as it happens, changing the signal from what is on the recording can very often result in greater realism/naturalism. Even something as simple as a slight dip in EQ to tame exaggerated sibilance in a recording. Much of the job of our dialog editors and mixers in film post production is doing precisely that: Altering the characteristics of the original recording (e.g. the production track made on set) to sound more natural - EQ dips and boosts, de-essing, cutting out extraneous sounds or tones from the original track, adding tones, you name it. The original tracks are messed with in all sorts of ways. So one is distorting the original recording, but with the result being a more natural, realistic character to the sound.
Unless your talking about distorting the sound on the recording to add
something to your taste.
See above. If we are talking about trying to produce more realistic/natural sound, altering the original sound (which is thus distorting it) doesn't automatically
mean it's being altered away
from realism "to someone's taste
." It can in fact be altered TOWARDS greater realism.
And once we realize that, it's perfectly reasonable to ponder how other ways of distorting a track can not just "be altering to taste" but "altering towards greater natural/real sound."
Along those lines: Unless we can appeal directly to very well constructed live vs reproduced tests with different types of recording, different types of distortion etc, I don't think certain claims can be settled as to "X distortion enhances or detracts from the realism." It's more argument from inference, personal experience suggesting things, etc.
Since I have no such well constructed tests, I can't offer anything to help settle such questions. I just have my personal experience and inferences. So for instance, as I've said I'm very often comparing the sound coming from speakers to real sounds. If I hear a vocal that strikes me as sounding fairly "real" I will often take a moment to listen to a real voice (for instance my wife) and then "oh yeah, real voices have a different character than the reproduced voices."
But my little Spendor LS3/5s can sound amazingly close to capturing the essence of a human voice. They survive this comparison better than maybe any speaker I've owned. Except...when I replaced my CJ tube amps with a Bryston solid state amp at one point! Voices that seemed regularly rounded, organic and human via the CJs sounded a bit squeezed, harder, brittle, "recorded" on the Bryston. Slightly more accurate to how that recording sounds? Likely. More realistic sounding? Not to my ears. It didn't seem to survive those little "real vs reproduced" tests I'd do as well as when I drove them with the CJs.
This isn't meant to convince anyone else - but if the CJs are changing the sound they seem to distort it with those speakers in a way that doesn't seem just "to my taste" but actually "a bit more real."
Which is one reason I keep the CJs. (Though if it were easy enough, it would be interesting to do a live vs Cj/Bryston blind test. But, alas, not so easy...)