(Or perhaps, why is The Loudness War unsustainable?)
My LUFS Soapbox is for a reason - because The Loudness War is destroying what we call music.
The goal of recording before we had Digital technology was to get the peak levels as hot as possible without going into the nasty distortion zone. This was fine with analog tape because we had to make it loud enough to get heard over the inherent tape noise and as a bonus we could also 'push' levels over that 0dB line on the meter and get some of that 'warm' tape compression that could 'work its magic' on a track or the whole song.
When digital recording became ubiquitous, we were suddenly blessed with a much lower noise floor so we could now actually use the full range of dynamics on an instrument without having to level it up (or use potentially destructive Noise Gating or Reduction Systems) get it heard above the tape hiss. But we were also blessed (cursed?) with a limit of how loud a signal could be: 0dBFS (Zero Decibel Full Scale).
For some frame of reference, a typical magnetic tape recorder's 0dB meter reading is equal to about -14dB (that's negative 14dB) on today's digital recording systems. Analog tape could be pushed into 'the red' (peak levels) by about 6dB. which is the equivelent of digitally recording with peak levels of -8 to -6dB today. Also you need to keep in mind that the best tape machines recorded with about 13 bits of dynamic range, while digital systems start at 16 bits and capable of many, many more.
Back to our analogy - we could push tape a little harder and get some nice distortion that might make things sound a bit more pleasing to our ears, but when Digital became the standard we couldn't go above that 0dB limit without getting digital distortion. Digital Distortion is nothing like analog distortion - it's a very nasty sound; grating and unlistenable.
So The Loudness War proponents keep pushing that 'louder is better' - but with that digital cap always there, you can't make it louder without terrible artifacts. So how do you make it louder?
Easy-peasy. Compress the source so the overall sound has limited volume range. This gives the illusion of 'louder' but at a cost - the loss of the dynamics that were played or programmed. We could also use a 'Maximizer' (a fancy term for automatic Gain Leveling and Brick-Wall Limiting) on the final output to bring the overall levels as close to 0dbFS as we want. Let's take a look at simple 'Maximizing':
For some people Overcompressing and HyperMaximized levels sound okay. They'll concede that some of the dynamic range is lost, but extoll that the mix sounds 'louder' and it's 'bigger' or 'punchier' than the other mixes out there. Mission Accomplished.
As I said in a previous post, it doesn't take long for the competition to catch up when you're a soldier in the Loudness War. In our Digital World® new tools are just some coding away, so the competition will soon have better (or one could say more devastating) tools and plugins than you had on your last release, and they will compress and hype each and every track and then the whole song even more to get 'that big punchy sound'. Before we know it, any semblance of dynamics in music is gone as the contestants leapfrog each other into insanity. Metallica's 'Death Magnetic' was hoisted as the big red flag on this front, but even the Red Hot Chili Peppers 'I'm With You' was a come up to that. And this process of mixing and mastering is still being done today, and it needs to stop.
(If you were looking for the TL;DR part, here ya go):
If we mix a song using only Peak Metering then all we're aiming for is how close we can get to the Zero mark without making it a distorted mess. This is a leftover from analog recording that we think is the key to getting good mixes, but in truth those Peak Meters are lying to you. Our ears don't work on peaks. They respond to average (referred to as RMS, or Root Mean Squared) levels. The LUFS 'Standard' and it's accompanying Meters are similar to RMS and designed to measure the average (or Integrated) level of a complete work. Because we're now measuring the entire piece from start to finish we can get accurate, translatable levels from using 'The LUFS Standard' and its metering.
So let's think about this: if we have a standard to accurately represent recorded audio levels that matches to how we actually actually hear them, and we can define a set of values that allow us set a maximum average level for playing back that recorded audio in defined listening environments, then we're not fighting each other to find out who's the loudest anymore - we're now free to make music however we want because everything will play back at the same overall average level.
It can be as dynamic as you want, or as compressed as you wish, but when loudness doesn't matter anymore creativity will be unleashed again.
And do we need us some creativity in here.