So, I’ve seen in the documentation that the official policy is that a remastered recording is NOT a separate Recording from the original, in MusicBrainz. Why is that, though? If people want a way to distinguish which remastering is found on which releases (in cases where a remastering creates a clear, audible difference), why shouldn’t MusicBrainz have a way of tagging that? Marking releases as remasters of each other only helps so much—often, songs appear on compilations or soundtracks, and it would be useful to be able to indicate whether the copy found on such a compilation is the remastered recording or the original. Documenting stuff like that is one of the things I love about how thorough MusicBrainz’s relationships can be, and this just seems like a weird deviation from that, that I never understood.
It’s often hard to figure out exactly what remaster we have, so that would lead to a lot of unmerged recordings by necessity. How do you determine properly whether a given compilation has the original or a remaster? In most cases, it won’t say. Additionally, the main benefit of sharing recordings (only having to enter relationships once and have them shared everywhere) is generally seen as more useful than the benefits of distinguishing remasters.
I know what you mean about wanting recording relationships to carry over, but that’s a problem that still exists with edits & instrumental versions and so on—it might be good to have some system in place where when a recording is linked as based on another, they carry over automatically until the ones that are no longer relevant are discarded?
That’s a fair point about not knowing which recording to use, although maybe there could be a generic ‘unknown’ recording for every Work, or for every Work & Artist combination…
Actually come to think of it, that kind of ties in to something I’ve been thinking would be handy on MusicBrainz in general—a ‘confidence’ rating for every piece of information that gets entered. Like if you’ve researched something carefully, you can tag your edit as ‘high confidence’, but if it’s something where the liner notes give a name, and there’s two violinists by that name and you can’t determine which, you could tag it as ‘low confidence’ so that at least it’s still entered as a starting point, but people will know not to assume that it’s a researched piece of information… I know it sounds nicer in theory to just not include low-confidence information, but in the case of something like a recording, something HAS to be provided—and so the compromise that’s been settled on is to make the information incapable of being higher-quality, so no-one has to specify that theirs is lower? Whether as a confidence rating, or just a generic entry that isn’t linked to other entries until the person linking is certain, it might be a way to approach things like this…
Anyway, I know there’s bigger fish to fry, but I thought it was interesting to consider 
This seems like an unfortunate compromise made because of a lack of a way to specify information occasionally not being known, which deprives it from ever being stated even when it is known.
OK, so I just found out about this when I added new recordings for the songs on a remastered CD and when adding the “remaster of” relation I got the “deprecated” message.
https://musicbrainz.org/edit/95210558
How can I revert that automatically applied change?
There are three, clearly distinct, versions. The original 1984 mastering is very soft, the 1995 mastering is too loud and has an indexing error in track 2, and the 2011 mastering has a lot of audio compression. It would be great to show that distinction here.
Shouldn’t each remaster be a separate release within the same release group? It will have a different release date and artwork,
The question is about remasters of individual recordings, i.e. which (easily distinguishable) remaster of a recording appears on different releases like compilations.
Correct; but not the focus of this thread 
I was considering compilation release such as greatest hits by an artist where some tracks are remastered vs the previous release of the entire greatest hits compilation…. Those should be separate releases. I agree with you that it would be helpful to indicate which tracks are remastered.
Especially when trying to get all versions of a certain recording, knowing which version of a recording is on a certain compilation is very useful. Of course, this requires the person to enter the details to know the difference between the recordings. I like the “unknown” tag suggestion to recordings but I can imagine that will be hard to implement.
On some greatest hits compilations - Bob Seger Ultimate Hits (Ultimate Hits: Rock and Roll Never Forgets - Wikipedia) for example, it is described on the back of the CD case or on the liner notes. Some might believe that remastering is a record company ploy to get customers to buy a new version. It’s also true that many tracks were not very well mastered on the original releases. Regardless, it still needs to be entered.
I guess sometimes it is, but for a while I’ve become really interested in the differences, because I want to keep the best-sounding master and get rid of the rest. (It started when I learnt about the loudness war.) Sometimes just a repress has a different mastering in the sense that there are different output levels.
I first added annotations on this release group:
The three CD releases have clearly different mastering. The first is a bit soft, the second is too loud with even clipping (and an indexing error so should be different versions anyway), the third is marketed as a remaster and has a lot of audio compression (loudness war). I want to make that clear in the recordings as well.
What we buy, though, is releases.
So the annotation telling bad remaster is enough on the release.
If we would separate all remasters again, we would have as many recordings as tracks, because it would be usually very difficult to know the remasters or same, without actually listening to everything, with some kind of deep analysis: it would never happen.
There are long old debates on this. Short version, right or wrong, but MusicBrainz database there is no way to note the mastering of a Recording.
It is all about the source Recording. And if nothing was added or removed then every one of the remastered versions of the recordings released over 50 years get merged into the same recording. No matter how many times an engineer may have returned to the original tapes to make a new version.
Trying to find the “best sounding master” is not something Musicbrainz is designed for.
The best you can do is add annotation notes to specific releases and\or release groups about the different remasters.
The only time a recording can be different is when something has literally been cut out or spliced into that tape. A change of volume of vocals, quieter drums, or awful loudness war clipping is not enough for a new Recording.
An indexing change IS something that causes a new set of recordings. When you create those it is worth making an disambiguation note in that recording so it is easier to select
I have dozens of copies of Dark Side of the Moon and can spot the differences. But then I am also a bit mad. So I am with you on how useful this could be if there was a way of noting it.
I’ve spent far too many hours untangling different versions of The Wall where the CD Mastering sliced and diced that album up in many different ways over the decades. That caused differences in the physically cutting of the audio and therefore separated recordings.
I’ve also merged tons of Dark Side of the Moon recordings that are from 50 years of different releases, but must all link to the same Recordings in MB language as they all came from that same studio tape.
Also on the other side of this I have a 1984 CD of Wish You Were here, and a 1994 “remastered” version which ARE literally identical. Whack them into Audacity and you can see it. Look at the fractions of a second of a DiscID and you can see that the con happens. Sometimes remasters are not what they claim which causes a level of scepticism on it mainly being a marketing trick to resell that same album again. (Pink Floyd are awful skilled at this art). Most of the time “remastered” release is mainly about renewing copyrights.
But for each release group we know a lot about, there are 10,000 release groups where we don’t know that much.
The list of recordings would become a list of tracks, where the similitude would be unverifiable.
I feel it would not be good, but maybe I’m wrong. 
I agree with your thoughts. So few people would look at recording differences in Audacity, etc as most will not have two copies to compare. And anything someone “hears” is purely a personal opinion. There is also a large difference between listening on some iPhone headphones or £5000 speakers.
It would be close to impossible to really create anything of use in the database and yes, we would likely be left with as many tracks as recordings due to not knowing. (I can imagine the complexity of trying to merge Various Artist collections 
)
I usually prefer the older copies of non-remastered albums as this is the sound I grew up with. It is personal opinion. We are all different.
I often help people setup HiFi systems. Was working with someone the other day whose speakers were the size of a washing machine. Listening to Dark Side of the Moon on those was like I had never heard the album before. We all hear something different.
woah, don’t say never - it might happen just not right now, but maybe one day 
Less so in the days of digital downloads where you can buy tracks from a cd.
There is no suggestion to separate all the existing ones, just let the known remasters be marked.
Then remixes and remasters should really be a new entity, I guess.
I understand why, but the fact that there’s a different recording if someone during mastering makes a mistake in the track index, we do get a new recording/version but not at a remix or remaster, feels off. I mean, there is really is the same recording, but a different track index, which is at mastering level. Again, I understand why, since the link between the recording and track duration, but it just feels off.
I understand. Now I come to think of it (as a software developer and architect), adding a new entity like remix (could even be from different recorded unused tapes) or remaster could make it possible (though not necessarily easy…). Of course, we need to ask ourselves, to what end? In my particular case (but I guess also applies to @IvanDobsky when he talks about ‘‘The Wall’’ or ‘‘Dark Side’’) it is very interesting to see what remix or remaster a certain release contains. But we will probably be part of a very small group.
I think in a perfect world we would have them separate, but as it stands we can barely manage (usually don’t tbh) to get recordings merged even discounting mastering.
Happily (for us) MB errs towards the more complex and fine-grained, which is like a delicious drug to your general MB editor, but unfortunately at some point we just hit a wall of person-power and being able to get new users/have an understandable and usable (if we’re being optimistic, lol) system.
On paper, yeah separate them, in practice… 