I've read this article a few times and there's a lot to process, after each reading I come away with a slightly different appreciation for the details.
It is generally agreed that overuse of compression is a problem in music. Compression, even if used a bit excessively, can occasionally lead to stellar results. The music packs an aggressive punch (i.e. sounds good when played very loud) at the expense of almost all dynamics and the unpleasant, hard to define feeling of ear fatigue. For years (before I I knew there was something unique but also exhausting about listening to albums like Oasis' "What's the Story Morning Glory?" and Verve's "A Northern Soul". They convey feelings of excitement and invincibility that comes across completely differently than music that is mastered with lesser amounts of compression. The evolution of music production has taken it to places it never intended to go, and now it's like an arms race, where the problem keeps compounding and a wannabe hit song must be mastered loudly if it has any hope of competing with everything else on the radio.
Apple might be the only company big enough and ubiquitous enough to force this kind of change. At this point, it has to be an industry driven (and enforced) change. Most consumers aren't even aware of the issue. The average listener won't be turned off of buying music because the amount of compression isn't to their liking. I doubt they'd even notice the changes if iTunes Radio standards become widespread.
They mention "Wrecking Ball" as an example of a horrifically overcompressed song -- if you haven't listened to it closely, pay attention to the change in the volume of her voice during the verses and then during the intro to the chorus (before the full force of the backing music comes crashing in). She's screaming the chorus but on the recording, her voice sounds much quieter than in the verses! It's an unavoidable consequence of having to squash everything in the chorus to make it sound as loud as possible.
I wouldn't have guessed that the loudness range of hit songs hasn't changed in the past fifty years, according to the study cited in the article. This means, if I understand it correctly, that even the more "properly" compressed songs that are dynamic in *parts*, are on average getting louder and more squashed. So rather than Nirvana going quiet/loud/quiet/loud, there would have been a few seconds of quiet, followed by a uniformly loud/semiloud verse/chorus/verse sequence with Kurt Cobain's voice getting lost in the muddle somewhere.