More punch anyone?
I figure if you set the attack long enough you should get a BIGGER dynamic range because of the increased audible ratio between the start of the note and where the compressor kicks in and squashes.
I also compress the final mix if I am looking for a ducking-like effect, where either the vocals or bass drum squash the rest of the mix. It's kinda nice to bus out a track, EQ it, and run it back into the sidechain so the compressor is working on only certain frequencies. It's fun to play with... Listening to each kick drum, vocal, or cymbal splash eat up the whole mix.
Sometimes its necessary to compress when there are passages where one or all of the instrumentalists freaks out and increases the dynamic range beyond the point of reasonable listening. Have you ever listened to a classical station, only to find yourself turning it up... and up... and up.... and then BAM quick, turn it down!
The noise floor makes it unreasonable to let the dynamic range sit at such a high level. Perhaps some classical stations assume that the listeners want the music untouched, or that they are sitting in a perfectly quiet room with their $1,000,000 stereo system.