The problem with evaluating 2 amps in the same system without making any other changes is that it also has its pitfalls in really being able to make a determination about the ABSOLUTE quality of the amp. Example:
- You have a speaker with a wicked drop in impedance, say 0.5 ohm at 60hz-75hz...and your current wonderful amp ("Amp A") has a great frequency response into a standard 8ohm and even a 4ohm load...but buckles slightly (more than the target 0.5% variance) with that 60hz-75hz frequency range during peaks. So you end up backing the speaker 2 inches closer to the wall to provide some reinforcement. Fair enough. No system is perfect.
Then along comes Amp B...measures about the same as Amp A at standard 8ohm and 4ohm loads...but is designed to maintain much better performance down to 0.5 ohms.
What happens when you plug Amp B into the system? At 60-75hz, it probably maintains the recording's original signal better...but with the speaker's backed up 2 inches closer to the wall, now the 60hz-75hz bass is slightly too much. Is Amp B over-hyped in the bass? Or is it that the original setup was optimized for Amp A which measures great to 4ohms but happens to struggle at 1ohm to a variance of more than the target 0.5% and therefore bass at this range decreased by 1-2db and lacked oomf.
Put another way, if I am looking for the best amp, which do I prefer...Amp A that sounds great with my current speaker location...or Amp B that sounds inferior with the current speaker location...but significantly better if I had to slightly adjust the speaker a few inches? For me, its B.
I agree with the high level principle that you have to eliminate variables and change only one thing at a time to do evaluations. That is basic scientific method. However, in complex systems, it is very rare that you get to make 1 simple change and ALL observations thereafter have NOTHING to do with the intricate interactions between parts of a complex system and are ALL simply about the one element that was changed.
In the above case, the 'over-ripe bass' with Amp B in the system was actually because the original speaker placement had suited Amp A which did not handle 0.5ohm loads during peaks quite as robustly.