I'll add one thing here: for the past 56 years I've build, bought and designed amplifiers, both valve and solid state. The same applies to speakers and spend a fortune over all those years - expensive and cheap all included.
After a mega expensive solid state amplifier decided during one night to blow its output and taking the speakers with it that were unrepairable (since the drivers were no longer obtainable) I went back to valves.
I've had a number of well selected (bought) valve amplifiers, inclusive monoblocks with 4 KT88 ea. driving some LS5/12A yet was not happy. So started to make my own. It took me ten years but being invalid I did have the time on my hands. The last one took me 18 months from start of desing to completion (including testing).
It is my belief that global negative feedback is a two edged sword that often introduces instabilty depending on the complexity of the load which never is a pure resistance. My last amplifier exceeded all expectations, only local cathode feedback in the output stage, less than 1% distortion at 5 watts, less than 0.2% at 1 Watt and at 100mW it is 0.02%. Square wave perfect up to at least 50Khz (at which frequency my signal generator starts showing its limitations) and frequency response flat to 100Khz. Introducing global negative feedback created instability and frequency rabge had to be severly limited to around 30Khz. Square wave started to have rining, overshoot. The feedback loop is affected by the speaker reactances, and hence stability goes out the window again.
Yes it has around 3% at 8 watts but it is the first watt that is important. At night I often listen in the double digit mW region, during the day in the 200Mw to 300Mw with some ProAc Tablette 10's. Now replaced with some Falcon Q7 complete@home that need a tad more power. But listening at 500mW has me quickly turning the volume down - too loud for an extended period in my room at my distance away from the speakers.
This was the first amplifier that I made without UL or GNFB and using cathode feedback and I could not be happier. I now firmly belief that (too much?) GNFB robs an amplifier from "life".
Tube amplifiers can be a mix of current or voltage drive to the speakers and the damping factor has little to do with the speakers yet everything to do how much the voltage rises with a different impedance (current load).
As
@Tony L stated earlier: speakers are far from a perfect non-inductive load and hence will react differently to the combination voltage/current change depending on the load.
Just my view, YMMV.
