But wait, there's more. It's actually really, really easy to avoid the damage of extreme loudness processing - Just Don't Do It! There you go! All sorted? Nope, not quite because there's another factor waiting to bite you in the butt; The Fletcher-Munson Wars (I made that name up).
We get so used to dealing with levels of signals in both digital and analogue domains that we forget that at some point we need to move air. I can max-out my levels in my DAW, but I can still make it move more air by turning-up the amplifier that it is attached to (actually I reduce the attenuation before the amp, but let's run with it). All the signal processing is done, I turn-up the knob and it moves more air; it gets louder and it sounds different. Er, hang on a mo. Simply by hitting my ears with more extreme pressure levels I make the sound change, because my ears, and yours, don't react linearly to SPL. As you turn-up the level, the ears' mechanisms respond more efficiently to different parts of the audio spectrum, and the apparent eq balance of your mix changes. And this time you can't simply "Not Do It" because final listening level is completely and utterly outside of your control. What to do?
Many years ago we had a family stereo that had a "loudness" switch - when you played music quietly you switched this in and it all sounded "better", it re-eqd the amplifier to attempt to correct for the effects of the Fletcher-Munson curves and whisper it quietly but it switch actually worked rather well. Ultimately audio puristry got in the way and I refused to use it, or indeed to upgrade to any component that had anything similar fitted.
I suppose that in an ideal world we could issue each track we mix or master in different versions, one with "Play Loud" on the tag, one for "Normal Home Use" and one for "Shush, The Baby's Asleep". Until such an enlightened time, we need to at least be consistent.
I recently mixed a gentle track featuring female vocals and piano. The supplied audio was fairly quiet, and I mixed it quietly. Having a balance I liked I exported a couple of versions of it and put it to one side to master it. A couple of days after I finished the mix I created a project for the master and imported the stereo tracks. Running through with the monitors muted I checked the levels throughout the track and turned them up by about 11 dB - no compression, no limiting and no normalisation (quiet jazz really shouldn't be that loud) and then listened. Absolutely horrible, strident, harsh and boomy mess it was. I hit the stop button and checked that I hadn't left a limiter set to "kill" in line. I hadn't, I'd just made the signal bigger. Turning-down the monitor controller made it better; turning-down the gain in the DAW and turning-up the monitor controller made it bad again. The reason that this particular track hit me so hard was that the sparse arrangement had little to mask the higher frequency ranges. What to do? I revisited the mix and re-balanced it with the monitors turned-up. Problem solved.
Bob Katz recommends mastering with a calibrated monitor controller such that a -20dBFS pink noise calibration signal produces 83 dB SPL per monitor at the listening position. I tried this and it was FAR TOO LOUD - CAN YOU HEAR ME? WHAT? However, by setting my system to a level where I'll comfortably listen to music at what I consider to be a "robust" level and a "quiet" level and checking the SPL at my normal listening position (there are some good enough SPL meters available as freeware for IOS and Android devices if you don't have a meter or even just do it by ear) and marking my monitor controller I know that programme material with a RMS value of -14dBFS on a loud passage will be at a consistent enough SPL when I have the knob set to either of my marked positions. I can still mix as loudly or quietly as I want to, but I have a reproducible level that I can check against, and that's easy to set if I'm working in someone else's studio.
I'd be interested to know that levels other folks mix and/or master at - perhaps you could stick your readings in a response below.