Maybe his point was that the thread was so off-topic, that he made an off-topic post in yet another direction.
One paper that I think was mentioned here before but didn't get much discussion is
http://www.essex.ac.uk/ese/research/...%20testing.pdf
It's Hawksford proposition of a distortion metric that seems significantly better than THD/IMD (though there isn't a related human study to see how well his metric correlates with perception).
So, anyone have any comments on that? It would be interesting to try to run one's own amplifiers through such a testing procedure.