Wow, there are a few points to cover here:
1. It's not always about the mastering. A lot of the problems can come from the mixing stage. There are examples where the mastering stage did nothing to the audio, only the usual sorting out the fades, track order building the CD layout etc. I think Metallica's
Death Magnetic is one example.
2. Here's the "but" to the above. It's not always about trying to make the loudest possible digital product. Sometimes, maybe more often than we want to accept, the production is done the way it is to sound a certain way. I suppose there is also a contradiction to that, which would ask how can the engineers determine what it will sound like when, for example, true peaks will be handled differently? Surely better to leave the headroom to maintain consistent playback on any DAC and actually bake in the sound of clipping distortion using plugins or play out of a DAC and re-digitize the analogue clippped sound.
3. It has been shown time and time again that when put through radio processing a cleaner, clearer and less processed version sounds closer to the intended and has less artefacts, whilst still ending up being a dynamically squashed radio-friendly sound. Using the same music, but hyper squashed already, just end up being mangled more by the radio processing.
4. Adding to point 3. I don't think digital releases are done so much for on-air radio play, but more to push little earbuds, cheap little stereos and lousy car audio. Grabbing a bunch of older material with Integrated LUFS around the -18 LUFS ballpark and putting on a phone, in a car etc. all seemed too weak, and like there wasn't enough I could do to make them sound fuller. Bunging the same audio through a few VST plugins including a limiter at the end, allowed me to bring up the loudness to around -12 LUFS which seemed to work better in these noisy scenarios.
When comparing my 're-mastered' version to the original at home, in a quieter room, it was louder and had some artefacts that weren't the most pleasant, and when level matched in this scenario it was clear that the quieter, fully dynamic and properly peaking audio sounded much better.
I guess my point is that in noisy situations and when not really relaxing and listening a lot of audio nasties disappear and we just want more level to get above background noise. Buying well isolated headphones would be a better option instead, but they aren't sold as standard.
5. Old specs for CD samples allowed for something like no more than 3 maximum value samples together. What this actually means in terms of the analogue waveform is something else, but interesting none the less. Quite a lot of boutique audiophile releases have 100% maximum value digital peaks and I wondered why they did this, especially when they had gone through so much effort in sourcing the cleanest tape source etc.
These aren't ISPs / true peaks, these are the sample values in the digital data. Presumably the ISP would be greater than digital 0 in these cases, but likely such fast transients I doubt they would be audible.
6. What about transients in analogue and (pre-)amplifier input sensitivity headroom? Could, say, vinyl with particularly high peak transient amplitude not be amplified with the phono pre-amp and overload the input of the headphone/speaker amplifier? I guess the frequency would likely be high and the RIAA roll off would fix the problem before it even was one, but technically would it be possible?
I doubt that anyone is going to start making some test vinyl for the sake of checking that out though.
7. I thought the loudness war was already over. A few folks have said this. Having a look here :
TT DR Database I see an awful lot of DR5 material for this year. So much for that then.