While I agree with the sentiment, there are few exceptions to be found. Notably the rather expensive
Marantz SACD player. It has a 100+ SINAD, and also has a respectable multi-tone. But that is just at low frequency and high sample rate. At CD rate, THD rises sharply beyond 1 kHz and even goes over 1% at 7kHz, or is just ~70 across the board with the other filter. Both are evil.
A single number is never great to represent dynamic systems, but it seems reasonable to center on a more meaningful measurement. For example:
WORST_CASE_SINAD ÷ (HARMONIC_ORDER_1 × HARMONIC_DB_1 + ... HARMONIC_ORDER_n × HARMONIC_DB_n)
...would adjust the value taking into account psychoacoustic value. But then you'd have enormous ranges, so maybe take the log...
Anyway, it's imperfect, but that's what graphs are for. We use them to evaluate frequency response, no reason they can't be standard to evaluate distortion. (As we're already presented with, at least here on this site.)
One thing that's missed, however, is the cumulative nature of these distortions. As each component hits the next, the structure of the signal is changed nonlinearly. A couple "good enough" components could interact such that the result is rather imperfect.
Aince this is a psychoacoustic evaluation, now we're discussing the difference between MP3 and AAC, for example. So one could say the "320Kbps MP3" equipment chain is "good enough", but matters are made worse when you rely on a clean signal. Say you want to EQ. If you EQ an MP3, it's not going to be as good as a 24-bit FLAC, for example. Same with these signal chains: several "good enough" pieces on the way to the EQ now change that challenge.
So what does this matter? Well, is SINAD "good enough"? Well, no. It never was. It's a single number to describe a dynamic system. Is it "good enough" in the context of a purchase decision? Odds say yes, but if you get an unlucky chain that exacerbates distortion... then no.
What we'd need is a standard for the raw measurements at varying rates and frequencies, and the ability to chain them together to observe/simulate the cumulative effect in a given combination.
Is that ever going to happen? No. It's too much work. And the interest? Probably not high enough.
But what's interesting is, who would benefit most from this sort of evaluation? Ironically, not really us. It would be the subjective audio listeners.
Why?
Well, at some point, someone is going to realize that they can make a lot of money by *simulating* various popular, colorized equipment. (FIRs, analog simulation, etc.) And when you do that, the absolute cleanest, most transparent system is to your benefit. (Avoid garbage in, garbage out.)
So, maybe one day that'll become the popular way to consume audio among subjectivists. "Use our Vintage Audio Domain Processor and get the greatest systems at your fingertips at the push of a button!"
And, if so, that would be to the benefit of everyone, objectivists included. Photography has already gone down this road, for capture, at least. (Which is all they really need, since most displays now are good enough to reflect that quality.) I know the same exists in audio capture, but to realize it on the playback is a whole other potential...
Anyway, I'm wandering way off topic. Point is, single numbers are never "good enough" to measure dynamic systems. But it's what we've got, and it's better than what we had.