A few relevant ideas:
- An almost universal technique in Popular* music is to compress the dynamic range of finished songs. Tools called "compressors" and "limiters" are used to achieve this, and they basically make the quiet parts of a song louder. Many members of the audio engineering community feel that these tools are grossly overused by the modern music industry (including many of those engineers who are compelled to overuse them in order to keep their job), but that's a completely different rant.
- There's a tremendous divide in the approach sound engineers take to making a Classical recording versus a Popular recording. There's variation within each community, based on genre and sub-genre, performer, engineer, etc., but the Classical Music world typically attempts to capture and reproduce an original performance as accurately as possible, while the Popular Music world uses the studio as an instrument, and all sorts of heavy processing are acceptable and commonplace.
- A whole lot of people listen to music in noisy environments. College campuses and mass transit are full of music fans listening to mp3 players with earbuds. I personally listen to over an hour of music accompanied by significant road noise each workday. In these settings, music with a relatively small dynamic range is great - the listener sets the volume at one comfortable level, and the music never gets too quiet to hear or too loud to bear.
In an ideal listening setting, a large dynamic range is fantastic, and I don't think the old standards should be abandoned, but in the modern world of downloadable music (i.e., where there's very little cost associated with the actual transfer of audio data), it should be easy to provide both compressed and uncompressed versions of a recordings to listeners.
*For this discussion, I'm making a tremendous generalization and dividing EVERYTHING into either Classical or Popular - frankly, though, I have no idea how Jazz fits into this discussion.