@frogman: You entirely missed my point regarding "inappropriate": Of course "the use of the sound of unamplified instruments in a real space" is "the best way to judge accuracy (naturalness) in the sound of a component." I didn’t say otherwise (reread what I wrote if you wish). That is the basis of hi-fi, after all!
What I DID say was that metric can not be used when the source material used in the evaluation of components is not a recording made with the intention of capturing "the sound of unamplified instruments in a real space", but is instead a recording made to merely sound "good". What does "good" mean? In the world of Pop music recording, if you think good means "the sound of unamplified instruments in a real space", well then you obviously haven’t spent much time in a recording studio.
Beside the obvious fact that most instruments in Pop music recordings are not unamplified, they are very rarely recorded in "a real space", and certainly not one that resembles what you hear in most studio recordings. Electric guitars, for instance, are often recorded with the amplifier placed in a small isolation booth. To create the illusion of the sound being produced in a larger space, electronic reverb is then added to the signal. Electric basses are often recorded plugged straight into the board, not an amplifier. Same with electronic keyboards.
Another example: The most commonly-used microphone for recording snare drums in studios in the Shure SM-57. Why the 57? Because the mic has a built-in presence peak; the mic was designed as a vocal mic for use in PA systems. The presence peak helps vocals to "cut through" the sound of loud instruments on stage. In the studio, engineers use it to create a snare sound that "pops" in the mix.
Another example: The better studio engineers apply compression to the overhead mics employed to capture the sound of drumset cymbals. The compression makes the drumstick tip hitting the cymbal have a more pronounced percussive "click" sound. The first time I recorded at Flora Recording & Playback Studio in Portland Oregon (where Bill Frisell has recorded a lot---a great studio), I played on the house set---a nice DW kit with Zildjian A cymbals. After the first take we listened to the playback, and I was startled by how different was the in person sound of the cymbals compared to how they sounded on tape. I instantly recognized the sound of compression, and verified it’s use with the engineer. When mixing all the tracks, he then added electronic reverb and considerable amounts of equalization. While the results sounded "good", you can’t really consider it "the sound of unamplified instruments in a real space", now can you?
As for the use of the term "sound effects": the quotation marks were intended to convey the subtle employment (apparently too subtle ;-) of humour. I’ve seen the term used in hi-fi critique before, by writers better than myself (perhaps even by Art Dudley. Oops, there I go again ;). I guess I should have used the much less subtle ;-) to convey my intent. But the idea does raise a valid concern: TAS reviewers quantified the ability of a component to reveal low-level detail as an indicator of the components’ degree of resolution. Okay, fine. But what if that ability comes at the expensive of other, more musically relevant characteristics? I mean, I want to hear the sound of Tony Rice’s fingers sliding across the strings of his Martin D-28 acoustic guitar, but what if that sound is being exaggerated by a component? How does a listener know how "much" of that sound is contained in the recording, versus what the component is doing to the signal?