Being on the inside, I have seen a lot of good things over the last decade of engineering. One thing, bought music libraries are often lossless. Hard disk space is now cheap enough that lossless makes sense. Of course, if your playing pop, hip-hip, or a lot of similar genres that makes use of samples, what are you really getting with lossless? Program paths are often PCM, particularly in larger markets. Smaller markets are often running analog STL shots, so little has changed there. Processing has gotten better too. Most of the new flagship processors incorporate declippers, and all you have to do is wait for a Red Hot Chili Peppers song to come on and see if its being used

I read something recently about the loudness war being over, as many online streaming entities are pushing the volume of hypercompressed content down, seemingly rewarding material with dynamics by playing them louder. I hope the trend continues, and I have seen some indy stuff lately that looks less like a box on the waveform view of any DAW. Unfortunately the music distribution system is moving farther and farther from a lossless world, and often the only version you can get of a song is MP3. At least bandcamp would let you have a PCM version of a song if you wanted, but again there is no way to tell what the original source was.
As long as people are content to listening to music on the subway with tin can ears, the music industry will pump out content geared toward the lowest common denominator.
+-RH