An answer to Martin Colloms’s question.
(Stereophile, May 1998, p.21)

Martin has successfully considered his 701st amplifier.
It was instructive.
Unfortunately, improperly designed amplifiers are always allergic to a negative feedback (like any amplifiers with transformers). I discovered for myself this phenomena almost 28 years ago by hearing a whistle from my first "hi-fi" system.
It became clear later: a generic amplifier has a nonlinear phase response, moreover, the latter varies with an amplitude of input signal. Such a system is hard to control with negative feedback, as prof. Kalman says. The main problem of a real closed loop system is stability.

However, any amplifier stage built with an active element (tube or transistor) has already got some chains of negative feedback, - intrinsic and parasitic besides a circuit for operating point stabilization. A resistor introduced in series with cathode, emitter or FET’s source causes a negative feedback.
A cathode/emitter follower is an amplifier with 100% negative feedback. I suppose Mr.Colloms is informed of these details.

Thereby, all kinds of amplifiers have negative feedback.

Britannica says that a feedback is a basis of the nature.
Why must designers avoid feedback usage?
Since the last two centuries from Watt (1768, governor) and Maxwell (1868, feedback formulating) to present days, a negative feedback has been working for mankind.

I am sure that 25 years ago a tube "old man" (45 years of experience!) sounded better than any transistor "novice". But Robert Widlar had already created his legendary LM101! Soon appeared fast power silicon transistors. The time of solid amplifiers has come.

However, the Compact Disk has returned tubes from dusty shelves. It was much needed to revive unnatural CD sound with a tube "exciter". A problem of "jitter" has eclipsed real problems of sound reproduction for a long time.
But Vinyl has survived! And designers continue to think about the distortion free amplifier. I made one in 1984 (theory and circuit diagram published in 1987).

All things are measurable. Even light pressure, which was measured by Petr N. Lebedev early in the century.
With a pink noise ("harmonically rich") as an input signal and using DSP-based modern statistical methods it’s quite practicable to measure any spectrum modifications in amplifier’s output signal.

Martin is right in the case of a "thermal compression" phenomena. Somebody needs a lot of dB.
However, I have observed a thermal intermodulation in low voltage (only 10-15V supplied) stages at a few milliampers current levels. In general, thermal effects are the main difficulty in precision analog/mixed circuit designing.
Why good do loudspeakers have sensitivities below 90 dB?

Any amplifier stage is a low-pass filter (described with gain, time constant T, and output resistance). In an ideal case, the order of that filter is 1st (T1 only). Having been closed with non-delay negative feedback, the amplifier is stable without a load. Since a real amplifier stage isn’t the 1st order filter (see above) and has a complex load, it’s clear that negative feedback application is a difficult problem.
Therefore "designers ... seeking a safe route toward designs with minimal or no negative feedback" (M.C., 1998).

Thus, a sound coloration mentioned in Martin’s article is a product of negative feedback application to a relative stable amplifier. A typical negative feedback circuit consists of two carbon or metal-film resistors and it can’t be nonlinear. Dynamic nonlinearity is caused by transient feedback breaking under signal conditions in improperly designed amplifiers. Since any nonlinearity causes intermodulation distortions, a coloration appears.   

There is no mysticism in amplifiers design, but the serious science.

Dr. Andrey A. Danilov