Not sure how much "circumstantial evidence" I should have before editing track lengths / "splitting" recordings?



Found a single “unmatchable” track from an unknown source, which should be:

but it is 4:00 long, not 5:34

I know nothing about that track or the band (I just look up random unmatchable files I come across sometimes)… so I took a look around:

Checked spotify, itunes and the associated ASINs of all releases here:

All sources I could find say “4:00”. There’s a “limited edition” mentioned in the edit history. Its ASIN is the only one that I couldn’t find any track lengths for.

Is that enough “circumstantial evidence” to assume that the “5:34” recording should only be on the “limited edition” (if anywhere at all)? Would it be appropriate to create a new recording + change track length for the other two releases which have ASINs that say “4:00”? Or should I just trust that the people who first entered and associated the ASINS with the wrong lenghts did so on purpose (maybe amazon is to blame) and create a new release with the 4:00 version?


I don’t want to sound rude but let’s first edit what we have and/or know. :wink:
Btw one of those release has a TOC so there is little doubt.



So, that basically means:
Always trust the TOC / TOC > ASIN (or anything other information I can find on the Internet)?
And the ASIN should stay connected to the release no matter if Amazon says it has a different track length than the TOC indicated?
Maybe Amazon lists the length of its digital releases instead of the real one from the CDs sometimes?

If that sounds about right, I’ll just find and add a general digital release from spotify or something in cases like that…


Amazon is sometimes wrong. Guess how much the people doing data entry there earn per hour, and how high their motivation will be…

When a toc exists, we assume that somebody had the actual CD with this track lengths in their drive. While you could fake this data, it’s a bit of work, and casual vandals will probably be too lazy. And it’s hard to get wrong by mistake.

So I very much think there are two versions of this song, and you have the shorter one ( has „Album version“ as an extra title for this track.

Whether all these releases have the right recording linked is up for discussion. You could leave notes on a recent edit of it, or write a recent editor that he/she should double-check.


You could just use

for your file. Uploading a AcoustID for this recording would also help.


That’s the recording from the “digital release” I added (verified length with itunes / spotify).

Oh, I assumed it was possible to attach a toc to the wrong release by accident… or the wrong ASIN to a release that already has a TOC. Is that not how those things work?

Yes, that’s the one that combines the TOC (long version) and the ASIN with the short version though. Which makes it look as thought the ASIN’s of the two CD releases might have gotten switched somehow… And that the one that is “short” according to Amazon might really be the short version?

Or not. shrug

Guess I’ll just post a link to this thread in the notes… sooner or later someone who has the CD is bound to notice that it is recognized as “digital” instead of “CD” release (by Picard or whatever) if the length of that track really matches and whoever that is will probably be in a better position to fix it?


My rule of thumb: never ever change the length specified on a recording, unless you have absolute proof that that specific recording has an incorrect length set. If it’s on a release with a TOC, that trumps your proof (unless you have absolute proof that the TOC is wrong, which is very unlikely). ASINs do not matter at all for this - Amazon’s data, like discogs’, is a starting point for data to enter, nothing more (especially since there’s a lot of “inbreeding” there - you might find the same wrong information in many places, who all copied it from each other).
Basically, it’s better to start off assuming that you have a different version than what MB already has when track lengths differ by more than 5 seconds. Someone with more evidence can always merge later.


Oh, thanks!

Is creating duplicates of stuff (not just recordings; releases, featured artists & everything too) and just trusting that someone else will merge it later if it’s really a duplicate considered good practice / a normal part of the workflow here in cases where there’s even just a little bit of doubt?

That would be a bit counterintuitive, but also it would make sense and some things much easier / safer… so if that’s the case, someone should probably put that piece of advice somewhere where newbies are (more) likely to see it…


I’d say yes, as this is basically the inverse of - it is better to create one MBID too many, than it is to have had an MBID that, as it turns out, was actually referring to two separate things. Ie., it is easier and cleaner to merge two entities into one, than it is to split a single entity into two or more.


I disagree with this a little bit, always check the edit history, TOC’s are extremely often on the wrong release (especially older releases). A lot of people get lazy and just add the TOC to whatever release they can find, instead of adding their own specific version to MB.

Other people will say to play it safe, but personally I think that if there’s little unique data on a release that helps you identify it/ figure out if it’s real, then there’s zero loss if you merge it into one you know is correct. Someone can always add their version again.
But again, always check the edit history, leave notes asking for info from the original editors, and try track it down yourself.
After that… I merge. If I can’t identify it, nobody else can, and it will literally be there forever (as you can’t prove that something doesn’t exist).

But as a rule of thumb, if you’re even slightly unsure, just leave it.
And add whatever release (physical or digital) as well as you can, that’s all you really have to worry about. Even if it possibly duplicates data.

edit: as Freso says, it’s much easier to merge than split, so if you’re unsure, duplicate that info!


Occasionally there can be subtle differences between pressings of cd.
I cannot remember the release but I have seen the same album with 1-2 seconds different between my toc and another toc produced in another country.


Agreed. But in this case, even if the TOC links to the wrong release, it is still evidence that a version of the song exists with this length.