I will also say that I prefer amps with zero feedback, but @atmasphere and I will diverge there, and that is just fine.
@donsachs I wasn't a fan of feedback for a long time. Here are two things for you to consider: the first being that in most amps with feedback, tube or solid state, the feedback node (which might be the cathode of the input tube) distorts the feedback prior to its being able to do its job. In the case of a tube, the tube isn't linear. So using the cathode as an input for the feedback means the feedback signal gets distorted by the tube as it mixes with the incoming audio. So IMD, harmonics and inharmonic noise is generated, literally causing the noise floor of the amp to be composed of that rather than actual noise.
Norman Crowhurst wrote about this problem in the 1950s but didn't suggest a solution.
However the solution is simple: wrap the feedback around the amp in such a way that it can be mixed with the incoming signal using a resistive divider network at the input of the amp (in the case of a tube amp, at the grid of the input tube rather than using the cathode) so that the feedback signal isn't distorted in the mixing process. IOW, much the same way you see opamps do feedback.
One interesting tube preamplifier design that used this idea was made by Leak, called the Point One. Properly refurbished and given a good power supply its quite musical.
Of course, to do this effectively you usually have to have more gain that can be used for the feedback. That is why Leak used pentodes but I've seen Fisher circuits that used this idea too using 12AX7s.
The second problem is if the amp lacks gain bandwidth product, if you try to use too much feedback, at some frequency in the audio band distortion will rise since the feedback is being decreased on a slope of 20dB/decade. Bruno Putzeys wrote a fine paper about this problem. I'm convinced this causes brightness and harshness. This is a very common problem in amplifier design, although not nearly so now as it was up to about 20 years ago.