Amplifier specs, does they matter?

Psaq, In normal test for THD amplifier is fed with pure sinewave ant tested at the output for anything else present. Not only that this test says nothing about THD when driving your speaker but also won't show any high order harmonics manufactured by deep negative feedback in Transient Intermodulation.

40dB of negative feedback means that amp was designed with 100x higher gain that is reduced back to normal by feeding output signal to the input in opposite phase (cancelling 99% of the input signal). This scheme is wonderful and improves everything 100 fold, but when rapidly changing signal (music) is applied amplifier for a moment has higher gain (because of input/output delay) and output overshoots (signal fed back to the input is too late). It can be easily shown when testing amplifier with square wave. This overshoot translates into unpleasant high order odd harmonics. Such amp will have fantastic THD, IMD, DF etc. specifications but it will sound horrible - guaranteed. Even power ratings can be very misleading. Because of all that I think that specifications are pretty much garbage. I would use them only, as Al stated, to exclude certain amps that, for instance, cannot drive my speakers etc.

^^That.

I would say the most important spec is weight within a given amp type. The amp with the best power supply and running the most in class A always sounds better and controls the speakers better imho. This means larger transformers, filter caps, and heat sinks.

Kijanki, you appear to have taken a stronger position against use of NF than in the past where you argued results were very implementation specific.

Or are you just pointing out pros and cons, which all approaches always have?

Based on a stated 100X improvement with NF, negative effects would have to be quite significant to eliminate that completely.

The "its all implementation specific" argument still seems to ring true to me based on what I see, read and hear.

At least one company claims that their really-fast negative feedback loops eliminate the problems traditionally associated with negative feedback. Maybe the situation has improved to the point that it can potentially do more good than harm.