This conversation isn't about file compression, it's about dynamic range compression. Very different.
https://www.izotope.com/en/learn/what-is-audio-compression.h...
All we do is listen. Often the goal is to get the record to feel as loud as it can be, but the moment our choices start to degrade the feel of the record we back off. We're doing very little in the way of actual limiting for loudness. We have ways of making the record feel louder, and they don't always involve dynamics processing.
Quite often no limiting is necessary because the mixes that we receive are mixed hot, which leads me to my next point:
Where I'll push back a little - for the sake of conversation - is the assertion that the loudness war died with streaming services. I consider their automatic level compensation features to be a big plus for audio fidelity. For artists and mixers who are sensitive to this issue it relieves them of the pressure of feeling that they need to deliver super hot mixes. The record can simply be what it needs to be and not conform to a certain loudness standard. It's a positive step.
Where my objection comes in is the reality that artists and mixers (who are well aware of the dilemma) nonetheless deliver us mixes that have been crafted for maximum perceived loudness. And they're really good at it. Even world class producers and mixers are sending us records that are squashed to hell, but very often they still sound good! The craft of mixing is evolving and people have become very adept at making an objectively loud record (-3 LUFS, etc) still work. HOW they do that is far too big of a topic to broach here.
I think most modern EDM-pop now is at about -7 or so. These are world-class engineers doing the mixing, they could absolutely push it to -5 or -3 if they wanted, there's just no need thanks to LUFS normalization.
Like when people think about the real victims of the loudness war, it's albums like Californication or What's the Story (Morning Glory), which really didn't need to be that loud except for sounding better (in the louder = better sense) on the radio. Something like Death Magnetic you could argue is both intentionally loud + a victim of the loudness war, but it's also just really poorly mixed/mastered in general.
I'm not sure if the high-res audio version on Apple Music has corrected this sin of the original CDs.
Secondly, more compressed (less dynamic range) doesn't mean worse, or better. It means more compressed. In some genres, tons of compression is desirable; in others, less so. Is an orchestral Hans Zimmer piece that maxes at -10 momentary LUFS better than a Skrillex song that hits -4 during a drop? You could argue either song is better, but I don't think your argument would have anything to do with its LUFS measurement.
Compression is similar to distortion. Too much distortion is generally seen as bad. But if distortion were always bad, people would use acoustic guitars and clean electrics and never run things through a distortion pedal. Same idea with compression. Too much compression will kill a mix, but not enough will reduce the impact the song has on the listener.
In other words, trying to say that LUFS is an "objective metric of quality" is complete nonsense. To me that's like saying that lines of code written is an objective metric of quality of a software developer.