|The real problem may be that there isn't any sound card out there that can provide the performance necessary to elucidate the differences between a PPA and a META42, so RMAA dutifully reports that there is none - not something you'd really want to share with the masses, despite the fact that the lack of any difference might be entirely due to the sound card.
Not true at all. You can very easily see differences in noise performance, dynamic range performance and especially channel crosstalk. I am not talking about PPA specifically as I haven't built it yet but I did measure different amps, sources and variations of the same design and you can always pick up differences. Distortion - unless to a headphone - is usually low enough that your point may have some merit but otherwise... Maybe what you need is a better soundcard. I bought a Revo just for the sole purpose of measurement as it has the second best ADC chip from AKM. And indeed it has 104dB + noise performance. Audigy is much worse and indeed if sound card is worse than what you're measuring then the results are meaningless.
Also in my experience, test results are very repeateable and don't really vary from run to run. You can of course debate merits and demerits of the methods they use. For example the real value of channel crosstalk is masked by the noise floor, i.e. if it's better than noise floor then the value of noise floor will be shown. You could probably use auto-convolution over a long period of time to get the real value but of what use it would be is questionable. You can also debate if certain flaws are audible or harmful to the sound or even if the whole thing matters subjectively. But RMAA is very useful to pick up potential trouble with a particular amp.