As to the latter, I know someone around here has a free tutorial and reference info on compression. Now who was that again?....
G.
Reeeeeported.
Compression is a tricky thing, because sometimes it is necessary, sometimes it makes the overall sound of a track better, and sometimes it can ruin things completely.
The instances above are going examples of when you need compression - when the dynamics are, for one reason or another, out of control.
Times where it sounds good are completely subjective. One instance is in a vocal recording I was running tests using last night. I just wanted to try out some positioning in my room and such, so I was singing the first verse and chorus of the Del Shannon tune "Runaway". The beginning is a moderate volume, but by the chorus, the notes get high enough that I have to start pushing some air. The difference isn't so major that I was good at starting volume and clipping at the end, but there was too much differential among the high and low peaks, as the track progressed. Slapped the compressor on, messed around a bit so the Threshold was only hitting the heavier stuff, and had to add make-up gain of about 3dB when all was said and done.
This made the vocal sound fantastic - it wasn't set to music, so I didn't have to worry about how it would sit in a mix, of course, but the dynamics were still there, only they now sat in a range tolerable coming out of my monitors with the volume left untouched through the track.
Of course, this is one of just many scenarios where it helps. It helps give drums a more up-front sound, since it "amplifies" what would otherwise be diminishing overtones coming from the drums. It is mentioned a lot in regard to distorted guitars since usually they want all the volume they can get, so compressing the signal means that there is more high-gain output than there is lost overtones, etc.
Situations where it is overused and/or bad? Turn on the radio. The Avril Lavigne song "Girlfriend" came on as I drove to work this morning, and it about damn near blew my ears off (I was swiching from CD where I was listening to a Bob Dylan tune from his latest album). While those two songs represent entirely different genres of music, the Dylan tune had a very nice jazz-tone to the lead guitar playing, and the player was using volume as well as tone (and the notes, of course), to make his musical point.
The Lavigne song was a constant volume, and that was loud as humanly tolerable. Again, I realise the point of the modern rock stuff isn't to be overly emotionally expressive - its more geared toward toward upbeat dancing in clubs and crap - but they also don't need to compress every last instrument flat and then crank the gain to -2dB.
I rest my case!