"Piotr Wyrostek" wrote in message
news:7241c.443481$I06.4948340@attbi_s01...
*snip* quoted text

Yes, and the same problem exists when you listen to A or B in a DBT. That
is,
you assign a "label" to the A sound and then to B sound in the same way
(i.e that A is "rolled off and that B is "bright") as you described above.
There is no difference if you know that the amp is "Harman" or "A".

So when you listen to X, you try to hear if it is "bright" or "rolled of".
Since this "sound labels" where constructed in your mind, you have to
obtain
random results from this test, because neither A nor B nor X is
bright/dull.

If A & B do, in fact, sound different, I should be able to correctly assign
"sound labels" with the repective amps and to then identify X as being
either A or B. What is your point?

The problem is that the results are random and they are random
INDEPENDENTLY
FROM ANY REAL DIFFERENCES between A and B, if any exists.

The results are only random if there aren't any audible differences between
the amps? What are you trying to say?

This invalidates the whole idea od the blind testing.
That is, the (correct) premise, that what we hear is influenced by not the
sound
alone, invalidates the DBT testing method.

The purpose of blind testing is to remove the variables other than sound.
If you don't know what amp is playing, it's impossible to prejudge it based
on anything other than its sound, and if its sound is audibly different than
another amp, you should be able to hear the differences and correctly
identify X.