It seems that The Loudness wars are being discussed more than anything. With the drive for maximum volume has come the collateral damage of minimum dynamics.
Compression, Limiting, Brick Wall Limiting and Normalizing can all achieve increased loudness in many circumstances. There are restrictions and inherent problems with each. Normalising, for example, can only take the loudest points in a track to the selected maximum volume the range of volume between the loudest and quietest parts is preserved - the dynamics are retained. If you've mixed a track with an aberrant, super loud cymbal hit in one spot the amount of volume or level increase will be minimal if any at all.
Compression, (my 1st experience was with TV ads where everything is as loud as everything else), reduces peaks that will then allow the general volume to be "normalized" in relation to those reduced volume peaks.
Compression with "catch up" or "make up" gain will reduce the peaks, (by a ratio as set my the user and with a bundle of other variables as well), and automatically lift everything else up toward those new peaks.
One of the biggest issues regarding the use of these soft or hardware items are "pumping" where the effect is obvious and the response of the unit/program can be heard starting and ending on the caught peaks. "Colourization" is when the process alters the sound by more than its expected effect due to errors or planning in design and manufacturing(EQ, phase etc). Often we perceive these colourations as advantageous which is why so many old analogue units are being replicated in the digital world.
LOUDER is a competitive thing that became a fad and is now, thankfully, receiving a backlash. Will my track sound loud enough beside the previous track on the radio? Well, my songs won't get played there so, from that point of view the issue is moot. The fact that people don't like to adjust the volume control on the phone/MP3 Player/I pod/DAP has a lot to do with it too.
Take a track and run it through a sonic maximiser or Aural Enhancer. 1st listen will probably be impressive but, like LOUDness war tracks, repeated or sustain listening will probably render the listener and her/his ear fatigued. Sometimes, when used subtly, the effect is wonderful and not deleterious to the song and listening experience.
Mastering? I've experimented with Ozone & the ilk - even LANDR (they offer a free MP3 jobbie and I wanted to do a Mix - LANDR Master- ME master comparison for the sake of thoroughness). The LANDR job was unsympathetic and what I expected for an algorithm based, (upload, select genre, run, listen to result, select variation on that result, give up) attempt.
Managing dynamics, by whatever means, requires subtlety and, most crucially, GOOD EARS. I don't have much of the former and almost nothing of the latter so I get a mix as good as I can, (with reference to ears here and other forums), then send it to an M.E. who's work I've sussed out, done before & after comparisons etc. I currently use 2 M.E.s, (more correctly production master engineers), because one does a certain type of my stuff in a way I prefer and the other covers different stuff appropriately.

Cheers
rayc
"What's so funny about peace, love & understanding?" - N.Lowe