I think we can all agree on the semantics: "fidelity" means
"true(r) to the original" (Viggen) or "accurate reproduction (of the original)" (Marco-Jax2). The original being what's on the RECORDED medium -- NOT the actual event that was recorded.
AS this is all about the listening enjoyment, at home, of an actual past event (musical or other) the ASSUMPTION is: the more accurate REPRODUCTION at home is, the better.
Many have already noted that that, ENJOYMENT, rather than ACCURACY or true to the original, is a better goal. Of course we'd like both -- but most systems fall short of this. What many are saying is that, (I'm putting it in other words) 'as long as I can "control/choose & match" the system's distortion (i.e. departure from the original) in a way that the OVERALL result is enjoyable (i.e. makes my ears happy) -- that's what I want/choose'. I "trust my ears", despite the possible compromises.
Quite rightly so: this implies that our ears choose homogeneity/acceptability of the overall sonic result OVER certain details, etc, that may be contained in the recorded medium but are slightly masked OR where the system introduces sounds that are NOT contained in the "original" OR where the systems slightly alters detail. As we have lots of experience with live performances, our ears are led to choose the sound that is more reminiscent of the "real thing ", than otherwise.
As to reproducing the actual event, esp. the dynamics -- FORGET it. To do this we'd need: signal, energy & an electrical-to-acoustic power converter commonly known as "speaker".
Assume we want the speaker to be capable of matching our ears' dynamic range, i.e. what we can hear: 120db SPL, with minimal distortion (I'm dreaming, but, hey, it's an example).
Think about this. Take a speaker rated 90db SPL/1W/8ohm (2,84V) offering a pure resistive load (how nice!). The extra 30db, (i.e. 90+30=120) in SPL terms, mean a bit over 30 times LOUDER. As we're talking about SPL (force/are), the relation between intensity (watts/area) & force/area (pascal) is I=Pressure^2/p (air impedance constant) which means when spl doubles, watts quadruple. So, we're talking about a minimum of 900 watts here. Which mean an electrical potential 84V and a current of 10,6 Amps. Not bad. Not to mention the "hi-fi" PS required for such amplification: surely over 2x the output rating, +4x if it's class A (i.e. ~4kW /+45amp trannies).
Now, let's find a midrange capable of sustaining that -- let alone not distorting. In fact, I doubt any driver will have the time to distort: it'll just go poof.
Even we opt for the highly efficient, ~100db/1w, cheapo Lowther / AER / Supravox etc drivers of this world ($1k each); +20db SPL means 10 times more, which means a good 100W... I don't know of any such driver capable of sustaining this power level (despite some manufacturers' claims).
OH, and I forgot: all these SPL's are taken at 1m away from the source! At each doubling of distance, we lose 6db (anechoic) -- but add a few db because of the two near-identical sources + some room help, say we lose 6db spl at 4 metres... so we really need 126db. Whew....
Let's just listen to music:) Cheers