FORUMS
Not sure where to put this, but I think that both "sides" (if one can in fact break any of them into sides rather than a free-for-all) ought to consider something about devices like SET's.
The primary characteristic of an SET is that distortion grows with level, and quite rapidly as it approaches its maximum output. What's more, the distortion is quite unsymmetric as well.
First, the implications of the asymmetry: That means that you are getting even order harmonic terms as well as odd-order terms (symmetric == odd order, a particular kind of unsymmetric == even order, not all assymetric distortions are purely even order).
Now, on to how the ear works:
The ear has, inside the cochlea, a mechanical filterbank that literally sorts out the signal from the eardrum/etc across frequency, with bands about 60 to 80Hz wide at low frequencies and about 1/4 octave or so at higher frequencies, with a crossover between the two kinds of frequency scale in the 600-800 Hz range.
Why does this matter?
Well, inside of one of these 'filter bands' (it should be understood that there are more than 30 or so bands, one can calculate a filter response about any given point on the organ of corti, so there are many, many overlapping filters with a bandwidth as above), if you double the signal energy, the loudness will go up by about 2^(1/3.5), which is about a factor of 1.22.
If you put this additional energy somewhere else where signal was not previously present, the loudness will double.
N.B. I guess I need to define two terms:
Loudness: The perceived sensation level of an acoustic signal.
Intensity: The measured energy of an acoustic signal
Anyhow, how does this relate to SET's?
Well, distortion necessarily widens the bandwidth of any signal, and one inescapable conclusion is that as distortion rises, the intensity goes up more slowly than the loudness of the signal.
The point? SET's have less measured dynamic range, but may sound like they have more, due to their distortion mechanisms.
This can be simulated in software, has been simulated in software, and does in fact have the same effect.
Just FYI.
If you go to www.aes.org/sections/pnw/ppt.htm and pick up the "loudness tutorial" you will find a graph showing the loudness of a signal of the same energy, as spread over 1 to many bands.
The change in loudness due to spreading of energy can be equivelent to a change in intensity (for an undistorted signal) of at least two orders of magnitude, which is quite a bit, to say the least.
Seriously. In terms of loudness, perhaps more. In terms of electrical measurements, definitely less. Does the added distortion count as distortion or added loudness. 
