Understanding Disc IDs

I have this release: Release “Songs From Lonely Avenue” by The Brian Setzer Orchestra - MusicBrainz

The MB release has 2 disc IDs attached, one of which matches mine. There are 2 releases in this RG, one marked as US release, the other GB. The editing history on the release I have reveals that one of the disc IDs was added from the editor’s “UK” release. I then found that both releases in the RG have both disc IDs attached.

I’ve seen many releases that have several disc IDs attached, which seems wrong in my feeble understanding, but at least in this case, there appears to be clear evidence that something is wrong. I’m inclined to remove the UK disc ID from the US release, and vice versa. Based on what the editor said, and the US physical copy that I have, it’s clear which disc ID is which. But would it be correct to do that?

I found this thread, which has a lot of good information, but it hasn’t really helped me come to a conclusion.

3 Likes

These are pre-NGS releases.

Back then, all editions with same tracklist were together in a single release, with as many release events as editions.
So all disc IDs were attached together.

When NGS came, it split these to one release per edition but it was impossible to decide which edition was which disc ID.

So now all these pre-NGS releases are sharing the same disc IDs.

So yes, when you confirm a disc ID for a specific edition, you should remove the other(s), IMO.

8 Likes

Ok, that makes sense. Thanks.

You were lucky to find a nice clear one like that. :slight_smile: Sometimes you can get 20 DiscIDs attached to a dozen Releases. :exploding_head:

The main thing is when you do a removal like that Explain Loudly in the edit notes.

The pre-NGS CD issue is one reason I often add my known DiscIDs to the Annotations along with Matrix details. Aiming to help out anyone else doing this kind of thinning out.

This is also why when adding a DiscID good notes are useful too.

2 Likes

I find this can be hard to do as I have seen some releases with a batch of almost identical ToCs on those DiscIDs. Makes me more wary of removing them. Especially if some are unique and only attached to that release.

Yes or another way is to add a new release linked to dale recordings, very precisely described including matrix, manufacturer and everything, and you only attach your disc ID on it.

But the case here was quite simple as the original Edit #11335198 - Add release (pre‐NGS) shows 2 releases and now the 2 releases have the same 2 disc IDs.

So it’s quite safe to remove the wrong disc ID from the release possessed and confirmed by @Beckfield. Nothing more. :grin:

5 Likes

Trust me, I won’t be removing disc IDs unless it is just this unambiguous. :sweat_smile:

Edit #87448875
Edit #87448976

2 Likes

Yep, seen ‘em. Aiiiiin’t touchin’ 'em. :sweat_smile:

2 Likes

It’s not too scary, you can scroll through the edit history. Get rid of any that are too old to have edit notes/are pre-NGS (probably all of them), and add yours :stuck_out_tongue:

The others you can ask the user if it was really that release.

I think personally think we’re reaching the point where we could just programatically remove all pre-NGS discIDs (or maybe just any attached to multiple copies). Unless we don’t care about having release-specific confidence in DiscIDs - I don’t use DiscID’s much so maybe we don’t, I dunno!

2 Likes

So which copy do you decide is the correct copy? And which do you delete? :thinking: :crazy_face:

It is a minefield. And I am running around it a lot as I am starting to geek out a lot on Manufacturing. :nerd_face: In most cases the USA CD can be different to the Euro one. In other cases it is being pressed from the same master and is identical. There is not enough of a pattern to be sure. I am trying to follow a pattern of the different companies - but it is not clear. It can be common for the exact same DiscID to stay around for decades and be associated with a dozen releases. And then that EMI UDEN CD will also pop-up in South Africa and used as a master for their manufacturing.

What would be real useful is if we could see how many people had a specific DiscID. That would make rogue DiscIDs stand out like with AcoustID, but we don’t have this data. So they all look “equal”.

Our current best option is to get people to add DiscIDs again, or note them in the annotation. Or note them somewhere in the history for when a trawl happens. It would be real nice to see a “verified” flag to stick next to a DiscID. Something that can have a “me too” option to get a popularity vote to it. I don’t really like deleting a DiscID that is attached to only one release - but maybe it was badly calculated by only one person?

Trying to sort out something like The Wall takes ages due to there being 20+ DiscIDs attached to many release. And that one is “easier” due to the disk layout changing. BUT we have the trouble that we still don’t know which ones are correctly attached to a Release. Too often a random disc has been chosen to set track times from, and rarely the correct disk.

I am building up charts of the Floyd CDs. Learning those pattern changes and manufacturing changes. And now and then applying the napalm when I am confident of a layout pattern. :fire: :fire: :fire: At least dropping out the clearly wrong option. Waiting and watching for new DiscIDs being added to prove theories.

It is a long task. No simple answer. :exploding_head:

And then after you have spent months sorting a band out, neatly having single DiscIDs to many Releases, an editor appears and adds DiscIDs without notes and doesn’t reply to questions… :rofl:

4 Likes

There’s no way to know which is the correct copy. Hence delete all that can’t be confirmed*

*assuming that we want release-specific confidence in DiscID’s.

Agreed! And first removing them is really the only way to make this happen, in my experience.

If you are putting that much work into Pink Floyd releases I think you can absolutely remove pre-NGS ‘unknown’ DiscID’s, + those from unresponsive editors if they don’t fit.

1 Like

I tend to see that as too brutal as people still use that data for matching. Maybe I lean to much on the “If in doubt, don’t change it” side of things too much. Once deleted it can’t be put back. This is why I’d love a “data quality” flag for disc IDs.

These discIDs have been there for a couple of decades. I don’t need to rush a fix through. I also find patience is good as one of the discIDs will suddenly pop-up and unlock a chunk of the puzzle. :nerd_face:Those original Japanese pressings were very different layouts, but would be the first glass masters used in Europe and the US and first releases followed them closely. A few years later a different layout then springs up. Had one appear last month which helped unlock a number of these releases…

And just because and editor does not reply it does not mean it is bad data. Not everyone reads their MB emails every day, and many editors have moved on to other things.

At the heart of my tweaks is the fact I know many people who use this database don’t really care. They just want to tag something. This is why so many discIDs get added without data. I know I was guilty of that when I first used Picard. Due to not understanding I would add my UK bought CDs to wrong editions because I wanted to link to the earliest release. Now I understand the details and differences I know the errors I made in those days.

3 Likes

It only makes sense when you have one of the editions in hand and you are cleaning up the matching MB release.

If you can’t find exactly your edition, you create a new MB release with only your disc ID.

If you find an MB release that has no conflicting data with your edition:
You clean it up by removing all other disc IDs.
You can also remove your disc ID from all the other MB releases that were added in the same (example with 2 releases) Edit #11335198 - Add release (pre‐NGS).

2 Likes

You can not do that. DiscIDs are not that unique. At this moment I am adding an Enigma release for my CD in hand where many Releases legitimately share the same DiscID. EMI UDEN and EMI SWINDON were pressing the same CD for many years across Europe and changes happened to the CD packaging cause multiple Releases. But the CD stays the same.

In that example I expect the USA release can be removed as it is a scrappy entry, but the UK and Euro editions are verified to be using the same CD. (Okay - you made me brave enough to drop the link to the USA Release as the USA release is scrappy)

Same happens with PDO\PMDC\etc pressings. And other plants. The same plant presses the same DiscID for many years, but different Releases are required due to packaging\company changes. A focus is needed on the factory before removing DiscIDs.

(I am spending way too many hours digging into this subject :nerd_face: :crazy_face:)

Edit: Between phone calls I was digging more. This same DiscID is shown as attached to an Australian release. And edit notes confirm that too. It is a good example of how one CD Master can travel far and wide…

3 Likes

You totally can, and imo should (if they are pre-NGS IDs).

There’s little value to a user in your addition of a DiscID to a correct release if it’s on a bunch of others as well, with no way to know which is correct. The user is just getting a ‘fuzzy’ match to a random release in the release group, maybe right, maybe wrong.

As long as there is still that discID in the release group there is no downside for the tagger, even if they have a different CD.

If you remove the others they are getting a 100% match to a CD with that DiscID. That is different to “they are getting matched to the only CD with that DiscID that exists” because, as you say, multiple CDs can have the same DiscID. But it is still a substantial improvement over “who knows if this CD has this DiscID”.

And they have no way of knowing if they are getting the right release.

Each one of those people could be adding their ID to the correct release, if the DB hasn’t already peppered their DiscID everywhere, making it pointless.

If you don’t remove discID’s they, and future users, can literally never assume they will get the right release from a DiscID lookup :sob:

If you don’t know if something is true, it’s not great data.

If you know that a lot of it is wrong I’m happy to call it bad data (doesn’t = bad editors, just talking about the dataset)

4 Likes

Please look into the example I added about Enigma. I added an EMI Swindon pressing today. If I had deleted all other references to that DiscID I would have caused real issues and damaged database quality. There are many other Releases sharing that ID - most of the EMI UDEN presses and an Australian press. Edit Histories clearly show this. I’ve done too much reading of those edit notes today. :smiley:

You must be careful removing DiscIDs blindly. Yes, if a release has next to no actual user edits and it is just pre-NGS, then it is suspect. But in this example many of the Releases attached to that DiscID had been clearly validated as legit in the edit histories. You must look at the factories involved. DiscIDs are not a one to one mapping with Releases. You cannot assume your Release is the only one with that DiscID.

I do get your points that most pre-NGS data is unreliable confusion, but you can’t just make assumptions about all old data.

This is also why I keep adding Release links to the manufacturers to make clean ups like this easier to happen.

4 Likes

Then the data is good and there is no need to remove it. @jesus2099 and I have only been talking about pre-NGS or otherwise suspect discIDs :+1:

4 Likes

As @aerozol said I was telling this for pre-NGS releases (all seen in the same add release edit), where you see a bunch of editions in the same edit and you have the same bunch of disc ID in each post-NGS split releases.

Some disc ID edits will tell for which versions they were made so you can keep them there.
But otherwise, when you have confirmed an edition, remove the bogus disc ID from there and if there is no confirmation for your disc ID on another edition, remove it from there (it will still be attached to your edition).

2 Likes

That irked me too, which is why I suggested some rules of thumb for safely removing superfluous DiscIDs.

1 Like

Wow, yes. That happened to me recently. A brand-new editor appears out of nowhere and as it its very first edit attaches a DiscID to a release that has two already. So I welcome the newcomer to MB and ask to check whether the cover art matches their copy. Later I find three cancelled edits to remove the DiscID and the editor transformed to “Deleted Editor #xxxxx”. Oops. Was it something I said? :frowning:

Of course, the DiscID is uniquely attached to this release. Remove? Leave it?

3 Likes