Vladimir said:1W = 1W = 1W regardless of output stage biasing, that was my point. How you percieve it doesn't change the fact that an amplifier will begin clipping after it excedes it's power output capabilities.
Let's say you want to prop up a desk with a couple of books. What matters is how thicks they are, not if they are poetry or phonebooks.
There is a threshold in amplifiers where further qualitative improvements in performance stop being audible. Same is for DACs and CDPs. Same for cables. Fortunately components that pass this sonic quality barier are cheap and plentiful. You can pay premium for added features, better build quality, aesthetics and larger quantity, but you won't get improved sound quality out of it. You can imagine it if it pleases you, but it's not real.
I agree that there is a threshold. Though we do need to keep in mind that even that is somewhat relative. While I don't believe in "golden ears" or any such, I do believe that trained listeners notice things that the untrained don't notice. So even Harman trains listeners, so that at least know what to look for during tests.
Years ago I took an online distortion test, that was being discussed on another forum. The first time I took the test, my threshold for hearing distortion was pretty average, but as I practiced, I found I got better scores. The same thing was noted by other persons who tried the test. It's not absolute proof of anything, but it certainly suggests that training and just knowing what to look for can have an affect on hearing subtle differences.
Now, as I will always say: Whether you're willing to pay an extra $200 or $20,000 for subtle improvements is a very personal decision.