To illustrate the specification issue I'm talking about, let's take the example of THD. Once, THD was given simply as a a single metric. A percentage number, that was presented as though it was all of the relvant information needed to assess the subjective impact of distortion on the sound of an given component. Later, we started seeing 2-dimensional graphs and FFT plots of distortion versus frequency, and more importantly, the harmonic order of the distortion. Also relevant to the sound, however is the distortion versus signal or power level. But how best to present that information. Logically, a 3-dimensional waterfall plot makes sense. But, wait, there's yet another aspect to harmonic distortion (without even addressing intermod. distortion here) which is, of course, distortion versus impedance. This parameter is often presented as it's own 2-dimensional plot, but it interrelates with the other three aspects of harmonic distortion. So, perhaps, a 4-dimensional presentation is required to properly show the complex interrelationships? How would that information be presented so as to be usefully interpreted by humans? What complex 4D shape would indicate superior subjective performamce over some other variation in shape? And this is only for the parameter of harmonic distortion, what about for all the other system parameters?
The logical engineering approach to solving this problem is to design components with objectively inaudible distortion for ALL conditions. Very high amounts a negative feedback is, of course, the current solution - to varying degrees of success. But this solution brings us to the conclusion that we should then all own nothing better mass market receivers or integrated amps, or listen only to CD. Better, has no practical meaning, once a pramater is beyond human perception. Except, our subjective experience leads us to often conclude otherwise. Perhaps, more guilding of the distortion Lilly, or that of some other parameter, will yet bring uniform approval of component subjective sound quality.