I am dead serious. Just look at the frequency response of a decent solid state amplifier with low output impedance: it will be flat within about 0.2 dB across the entire audible spectrum, and often even a bit beyond. Now compare that to some of the best speakers, like my Quad ESL 2805 and my Harbeth P3ESR desktop speaker. They are not flat at all, even if excellent by speaker standards. The explanation is very simple: speakers (like microphones - recordings matter very much as well, but are out of our control) operate at the interface between the electrical and the mechanical, and thus they have mass. Also, they operate in a room that has its own problems of reflections and room modes in particular. Just look at frequency response graphs in a real room compared to in an anechoic chamber or outside in the open air. You could easily have 10 db deviations from the ideal. Hence, when I added a subwoofer to my main system, I quickly decided to get a dsp room eq system as well (an Antimode 8033), which did clean up the sound significantly. But that can also be verified with measurements.
As for inaudible gains: I did once participate in a blind test of (excellent and very reputable) amplifiers. I thought I could distinguish them, but I was wrong. I was no better than random. Some time ago I replaced an old Quad 303 amplifier with a more modern Q606-2. The new one did indeed sound better and more dynamic, but only at higher listening levels. What I was hearing was the beneficial effect of bigger power that was not compressing/clipping the dynamic peaks of symphonic music. Real music and classical music in particular often needs big power (those needs can be measured, and you would be surprised - I was too).
