br85,
Are you implying, by any chance, that BB's will have their greatest effect on amps without GNF circuits? Or that the greatest effect will be on low output Z amps (mostly tube amps), GNF or not?
Well... that requires sort of a complex answer, but Ill do my best to make it simple to understand. First of all, amplifier output impedance is basically the function of two issues. The dominant factor is the "channel" impedance of the output stage.
As an example, in order for an amp to be able to output higher levels of current so that it can more easily drive lower impedance speakers, many manufacturers will use 2 or more output devices (transistors or tubes) in a "parallel-wired" arrangement. That means the final output current being delivered to the speaker is (hopefully) shared equally between each of the individual devices. By doing this the "on" or channel resistance of an individual device is divided by the number of devices being used. As the number of devices being used increases, the total output stage resistance (impedance) decreases. The reason the term "impedance" is used is because the values of this resistance can (and virtually always does) vary with frequency as a function of design. Of course the amplifier's power supply must also be able to provide the current being called for at any given moment as well, but this is another subject.
The above is the basic reason most larger high-power amplifiers also have a lower rated Output Impedance. This fact applies whether or not any Local or Global Feedback is applied or not. Now, regardless of how many output devices are being used, once Negative Feedback is applied it has the effect of forcing the output stage to "turn on" more (harder) in order to squelch (attenuate) any signal that is not supposed to be present (distortion) at the output. This then has the effect of lowering the "Effective" Output Impedance of the amp. We must remember that
ANY alteration of the output signal from that being being commanded for by the over-all gain structure of the amplifier... is considered a form of distortion. This means that a simple voltage dividing process which is always present is considered distortion too.
VL = ZL* IL and IL = IT wherein IT = VT/(ZA + ZL).
VL = Voltage across Load (speaker)
IL = Current through Load (speaker)
VT = Total Output Voltage from Amplifier
IT = Total Output Current from Amplifier
ZA = Amplifier Output Impedance
ZL = Load (speaker) Impedance
As we can see from the above simple equations, as the amplifier's Output Impedance increases, the voltage developed across the loudspeaker decreases. Conversely, if the amp's Output Impedance remains fixed, as the Speaker Impedance decreases the voltage across it also decreases.
Well, if the amplifier uses no Negative Feedback (be it either Local around the Output Stage or Global) it can do nothing to correct for the voltage drop occurring across the loudspeaker resulting from the voltage that drops across its own
internal (Output) impedance.
That means the voltage across the loudspeaker will "sag" to some value less than that which would otherwise be programmed by the amplifier's internal gain. Such an amplifier's ultimate performance then is extremely vulnerable to the effects of the loudspeaker load impedance. Seeing that speaker impedance is truly a complex vector sum of pure resistance, reactive capacitance and reactive inductance, (we're talking imaginary numbers math here, +j, -j, etc.) any amplifier using no significant feedback in any form will suffer from a widely varying frequency response at the speaker output, depending on the variations of the
loudspeaker's impedance. Oh, and loudspeaker impedance can (and usually does) vary to significant extremes.
Alright-y then. Conversely... the application of Negative Feedback to the design will cause the amplifier to "see" any un-programmed voltage drop at its output terminals as an error (form of distortion) and then adjust its output signal to correct for the error. This is good. The upshot is that the amp's Effective Output Impedance will decrease as the Load Impedance decreases and increase as the Load Impedance increases. Doing so thereby corrects for the variations in Load Impedance and results in a flat (hopefully) frequency response at the amplifier's output terminals.
OK, so now somebody tell me why you still want to use an amplifier that doesn't use any negative feedback? You say you have speakers that have perfectly flat impedance? Well, OK then... but if not... don't waste your breath.
Now, we must realize that this process is happening on a microsecond-by-microsecond basis and is actually far more complex than I've just described, but I think you get the idea. The key point is to remember that while Negative Feedback generally does far more good than bad, its usually not a perfect process and is highly dependant on the over-all design. When using feedback, the area of worst offense is usually at high frequencies. If you can imagine both complex musical signals being generated... clear up to 20,000 potential variations per second... AND a complex louspeaker impedance - both coming into play at the same time, it's not hard to imagine how the circuitry could make an occasional "mistake" trying to correct for it all.

Back in the early days of amplifier design many engineers applied massive amounts of GNF without consideration of numerous important variables, believing it to be a sort of cure-all, particularly in Solid-State designs. That led to the idea that there was something intrinsically "wrong" with the "sound" of SS and drove many audiophiles back to tubes. It is a shame that stigma is still at work to a great degree these days, and in many cases falsely so as SS design has progressed light-years from what it once was.
But... in most cases amplifiers aren't perfect yet and feedback loops can still generate small errors that were not part of the recorded material. This is true whether the amplifier is of a Linear SS, Tube or Switching design, as the majority all use some form of NF for the reasons clearly seen above. The upshot is that in most cases then, the
Black Boxes offer the potential for some significant improvement in system performance. Again, even in those few cases where the amplifier uses no Global Loop but the Output Impedance above 1KHz is around 2-Ohms or so (less is better), they still offer a potential for improvement in performance - GNF or not.
Now... go measure your amp's Output Impedance from 1KHz on up and you'll have a "sure-fire" set of facts to go by to know whether or not to give the BBs a try!

Take care,
-Bob