Rules of thumb for the clean-up of releases with multiple DiscIDs

Tags: #<Tag:0x00007f7d018dc040> #<Tag:0x00007f7d018e3ed0> #<Tag:0x00007f7d018e3d40>

Some releases in MB are burdened with an excessive number of attached DiscIDs, e.g.:

As I understand it, large collections of DiscIDs are partly the legacy of the old database schema and partly the accumulation of DiscIDs attached to a non-matching edition. It is very well possible DiscIDs are shared among different releases. However, it should be less common to see any single release host multiple DiscIDs. Exceptions to this rule exist, e.g. discs containing data tracks can have two valid DiscIDs because old versions of Picard once included some bug that resulted in different TOCs.

Since we expect only one or a few valid DiscIDs per release, it is clearly desirable to detach superfluous DiscIDs. Although this topic has come up here and there before, we are still lacking guidelines on how to best clean up these releases. Here is my suggestion:

You can safely remove DiscIDs different from the one you can verify for a given release when:

  1. You made sure that your copy matches the MB release, as confirmed by close comparison of catalog number, barcode, track list and cover art.
  2. There is no annotation stating these are valid DiscIDs for this release
  3. There is no evidence in the editing history that the release was hijacked at some point
  4. The unverified DiscIDs are already linked to a different release within the release group
  5. You are not dealing with a CD with data tracks

Feedback, corrections and additions appreciated.

9 Likes

I appreciate removal even if the release was hijacked (if there is still another release with these DiscIDs). In most cases images were added later, so you can never be sure. And a related Discogs release might have been changed too.

1 Like

When I’ve done a really detailed cleanup I’ve asked people in past edits what releases their IDs were for. Then I eventually remove all of the ones that have no reply/are from too far back to have a note.

Bit of a scorched earth approach tbh.

Then they can be applied to the correct one in future. No complaints yet but maybe someone disagrees?

4 Likes

I personally don’t think we should remove disc IDs that are not assigned to a separate release. It disables lookup by this disc ID, and keeping it does no real harm.

4 Likes

If we don’t want to remove some Disc IDs, for some reasons, we can simply create a new release for our edition and leave the multi Disc ID release as it is, out of our collection.

3 Likes

My view is that the harm is that they will never be added to the correct release. If the lookup finds a release why would someone go into mb to submit it/remove it from the others? It permanently retains vague data.

2 Likes

I guess that depends on the particular scenario. For example, if a release was started by someone adding his/her DiscID and hijacked later on, then it seems a bit rude to remove it.

Anyway, the rules of thumb are deliberately on the conservative side. That is, following them should not incur the risk of one getting oneself into hot water, and hijacking a release is frowned upon. That does not mean that you or me or anyone else are barred from taking more rigorous measures when we feel justified to do so.

2 Likes

Boy, you are zealous. Kudos to you! :smile:

Same here as in my answer to ernstlx, I guess. I agree that some bold editor may have good reasons to submit complete DiscID removal edits and if those pass the voting procedure that is fine. However, I do not feel comfortable endorsing this procedure in the rules of thumb.

My hope is that even having this very conservative procedure will encourage editors to clean up some of the excessively burdened releases.

5 Likes

When I was a noobie using Picard if I was offered a list of CDs that told me I had a “German” or “Dutch” release I would not believe it. So head to the site, find the original GB release and attach a discID. Only a few months later did I start to learn about different pressings from different countries.

I can see the similar problem happening a lot with noobies who don’t know the MB rules. Especially re-issues, and re-issues that don’t have dates visible in the database.

I agree that discIDs that are attached to multiple releases can be detached and moved, but I would not delete a unique discID. It would be really useful if discIDs could get a number attached like AcoustIDs so we can see how many times they had been attached to that Release.

Personally I tend to leave them alone - following the “if in doubt” rule. I use it as an excuse to do what @jesus2099 suggests and make a new unique release with fresh art, pressing details, etc. Though sometimes this isn’t possible if the release in question is already well detailed

6 Likes

That’s a nice idea. Still, I suspect that as long as the list of versions is not exhaustive a lot of DiscIDs will bleed into inappropriate releases. The act of attaching them to an existing release is just too darn easy in comparison with properly adding a new release.

Yep, same here. I tend to add fresh releases with cover art scans, CD hub info, DiscIDs, the works! and then watch them like a hawk to prevent them from being hijacked :grin:. Nevertheless, if MB is to be taken seriously as a music encyclopedia, it won’t do to have releases with dozens of DiscIDs. That is obviously incorrect information.

7 Likes

Well it is only attached once.

But we used to have kind of that, I don’t remember the name of it, but the number of times the Disc ID was looked up, probably.
Then it stopped being updated, then it was abandoned. :slight_smile:

Actually, when I came to MB, it was already no longer updated, if it ever worked before… :thinking:

Woops no, I confused with something else. There was a number for artist aliases, how many times each alias was searched and matched.

1 Like

This post was flagged by the community and is temporarily hidden.