From an epistemological point of view, the whole field of audio reproduction is fraught with a myriad of problems. Actually, it's a miracle that we can agree about anything! From the recording venue, the position of the microphones in the hall, the brand, model, number and mix of microphones, the different arrays, the cables, the recording medium, the engineer, all the way on through to the manufacturing of the CD that you're going to feed into your machine, there are just so many variables involved!
So what to do? Obviously, as many people remark, returning to the concert hall to refresh one's auditory memory as to what live sound actually sounds like (to you, on that date, in that hall--more subjectivity and variability) is important, as it is also to establish a kind of benchmark for what one hears at home.
But I think consensus is important too. Bertrand Russell might have you wondering if that coffee table sitting there on the carpet in front of you is "really" real, if it's really even there at all, but we get around this epistemological stumbling block by agreeing to accept that it probably is there, and therefore also we tend to walk around it (even if it isn't "really" there) so we don't get knocked on the shins.
If three or four like-minded friends agree that your system sounds good, then it probably does. (Reluctantly, I had to put that "like-minded" in because there is such subjectivity and variability out there too as to what "good" sounds like. Different ears; different criteria.) But here we have a minimum of consensus.
Likewise, if six or seven recordings of small classical chamber groups sound good on a given system, and one other recording doesn't, then I'm prepared to believe that there's some kind of problem with the recording somewhere in the recording chain, and that it's not the system's fault. (I'm talking about a level playing field here: for the purposes of argument let's stick with 17C sonatas with continuo and not drag Shostakovich into the mix...)
How does that, er, sound?
