There are not too many examples of great stereo gear that have not gone through the crucible of extensive measuring.
I was going to have a
massive ramble here, but I'm trimming it down to just a
large ramble instead...
I think we need to look a bit further than the measurements we have, since they're just being abused in studios by people who don't understand what the outputs mean (ie. think the spectrum analyser should show equal power across the entire spectrum at all times) and abused by manufacturers who want to pull the wool over our eyes with meaningless data (although there are some exceptions here).
By way of an example, say an amplifier is rated 35W into 8 Ohms over 5Hz-25kHz with no more than 0.1% THD...
35W is assuming an 8 Ohm resistive load, which a speaker isn't - at some frequencies, the amplifier may be supplying considerably less (or more) power, which in turn may affect the distortion performance
and usable frequency response.
If the speaker can reproduce frequencies outside the rated response of the amplifier, then there is no data at all to allow even an educated guess as to what might happen. There could, for instance, be a sudden peak of noise at high frequencies which was overlooked during the design owing to the designers already "knowing" the response of the amp. Supertweeters and older MOSFET amps come to mind, here...
Finally, the THD measurement. This is one of my pet hates. Trying to explain to someone why an amplifier with 10% even-order harmonic distortion can sound nicer than one with 1% odd-order is very difficult. Manufacturers would rather have it that way since adding masses of negative feedback is an easy way to increase frequency response and decrease the total distortion, so making the amp look "better" on paper. Sadly for our ears, NFB and many other distortion-reducing technologies are better at lower frequencies with even-order harmonics, and so tend to cancel the "nice" even-order distortion and leave some of the higher odd-order instead, or even increase it in some circuits. Graphs or tables of the relative levels of each harmonic would be an excellent way to compare the actual THD performance of two pieces of equipment, but this is virtually unheard of...
To sum all that up, I'd say we need a better picture of what equipment can do by using *all* the currently available testing methods and showing *all* the results from these tests. This means the manufacturers can't hide behind "useful" figures, we get a better picture of what we're buying is actually capable of and we don't need to introduce any fancy
new tests ... Yet!
The problem here is then twofold: Cost (to manufacturers, and thus to us) and ease of understanding (providing that much data in an easy-to-understand manner for "Joe Public" is virtually impossible).