The point of an ABX test is to remove any influence in your capability to "tell the difference".
Basically creating a double-blind test.
which it fails at in my opinion, by
A) streaming over the internet,
B) choosing some shitty acoustic guitar thing, and
C) having no clue what the source file's original quality was....
have a friend rip the cd and play it, there's a difference, only reason i care is because i think the music industry as a whole lost a lot with the transition to mp3, that it may never get back,
i agree the diff can be subtle, and without a comparison i have trouble myself guessing mp3, or lossless