A LONG post, but essentially simple concept.
You record a band, putting some squash on the room mics (when the compressor responds, they go up and down in level, right?) A little compression on the snare, to make it 'sit' and be more stable (it, too, goes up and down, with the leaked cymbals and room tones on that mic mirroring the snare's level changes).
You record the vocal and bass with dynamic control. Nothing drastic, just some nice smooth LA-2 or RNC, depending on budget. (They go up and down in a limited fashion, nicely controlled.)
You mix, putting about half the tracks through some sort of dynamic auto-control: compressors, limiters, distorters. (Up, down). Your drums have the "bus it out and compress it, feeding it back under the kit" mix trick (Up and down, plus whatever the individual tracks are doing: snare up and down, kick up and down, toms up and down, cymbals and room smashed once again, just a little bit.)
So, in your mix, look at any one thing: such as your "snare sound"; it's coming from the close mic, the overheads, the leakage, the sub-compressed track, the added verb or ambience. ALL these components that make up your "snare" are moving in different volume "directions" due to their levels and compression ratios....up and down. Maybe they all move downward together on a hit, but at different rates and distances.
So the overall result may be your snare is rushing down for a few milliseconds, then up for a few milliseconds, then down for a long while, then up for a while, then down... One single track can be a dynamic nightmare of contrasting changing levels and tones.
Look at the whole mix process, then 2-bus compression (see other thread), then basic mastering, then even radio limiting. You can have music that is SO hard for your mind to focus on. I think it's one of the big secrets of "old records" - 1 or 2 compressors in the whole building. Tracks almost HAD to have natural dynamics and EQ shapes. This is MUCH easier for your ears to listen to... and the SOUND is better.
Hmmm?