I think what he meant was that the act of mastering popular music often isn't exactly a subtle process. During mastering they try to make the music as loud as possible, reducing dynamic range in the process.

"With the advent of the Compact Disc (CD), music is encoded to a digital format with a clearly defined maximum peak amplitude. Once the maximum amplitude of a CD is reached, loudness can be increased still further through signal processing techniques such as dynamic range compression and equalization. Engineers can apply an increasingly high ratio of compression to a recording until it more frequently peaks at the maximum amplitude. Extreme uses of dynamic range compression can introduce clipping and other audible distortion. Modern albums that use such extreme dynamic range compression therefore sacrifice sound quality to loudness. The competitive escalation of loudness has led music fans and members of the musical press to refer to the affected albums as "victims of the loudness war"."

http://en.wikipedia.org/wiki/Loudness_war

On the contrary, I really love it.

I also love the cognitive dissonance which suggests that even a slight loss of fidelity in the 'inaudible bits' is far worse than a distinct loss of fidelity that occurs with an astonishingly imprecise trip through a multitude of lossy analogue processes.

Or that the addition of audible hiss, wow and surface noise doesn't matter.

You seem to suggest that algorithms like MP3 and AAC work primarily by throwing out the inaudible bits. A bit of that does happen, but most of the compression occurs by removing redundant data. After all, over 50% of the data can be removed before even a single bit has to change -- you can remove a shitload more if you're willing for it to be a very very very very very very close approximation instead. Psychoacoustic modelling is used to ensure that priority is given to sound that matters more over sound that matters less.