Relationship Type "Is The Earliest Release Of" for Tracks
Link Phrases
track is the earliest release of track
track is a later release of track
Description
This advanced relationship type indicates that two tracks are absolutely identical. No remastering has been applied and the track length is the same: they've just been released on different albums, for example on a compilation album. If the track has been remastered, you should use the RemasterRelationshipType instead.
The MusicBrainz database contains separate "track" entries for every time a track appears on the album. Technically, as the database is structured now, this isn't necessary. The same track could be used by several different albums, and this advanced relationship type would be unnecessary. However, the user interface doesn't currently support this, and it would be a lot of work to figure out how to present this to the user without confusing them. In the meantime, this advanced relationship should be used. If the system is ever changed to avoid the need for duplicate tracks, then this advanced relationship data will be used to figure out which tracks are duplicates that can be deleted.
In line with the DontMakeRelationshipClusters principle, all duplicate tracks should be linked to their earliest release. In any group of identical tracks, only one should ever have more than one "is the earliest release of" relationship.
There is a usability problem with the way this relationship currently appears on the website. When making the relationship, you end up with something that says "Track A is the earliest release of Track A". Because the tracks are identically named, you can't tell which one is really the first one. You have to click on the links to figure out which one came from which album.
This AdvancedRelationshipType is part of the AlternativeVersionRelationshipClass.
Relationship Attributes
None are appropriate
Examples
![[WWW]](http://wiki.musicbrainz.org/-/musicbrainz/img/moin-www.png){kind=small}
Where the Wild Roses Grow was originally released on the Nick Cave and the Bad seeds album Murder Ballads. It is the same track as the one that appeared on the Kylie Minogue album Ultimate Kylie. Note that although Ultimate Kylie is probably more well known than Murder Ballads, all versions of Where the Wild Roses Grow should link to the Murder Ballads version.
Discussion
Now that all references to 'album' have been changed to 'release', this AR is a little confusion. Under our current defintions, a track on an album can't be the earliest release of something. Perhaps this AR needs to be rephrased? "is the earliest appearence of" or something like that? --Gecks
Well tracks still get released on releases. The term "release" for the objects in the database surely creates more problems than benefit though.
--Shepard
The MusicDNS designers claim that a PUID will be the same for tracks that are "identical" according to this SameTrackRelationshipType definition. "The Open Fingerprint™ acoustic fingerprint ... [achieves] rigorous, consistent identification of identical master recordings." ( OFA Whitepaper, p. 3.) If that claim is true, and if every track had a PUID, then in theory you could go through the database looking for tracks with identical PUIDs, and they should be identical. In practice? Well, in theory, theory and practice are the same. -- JimDeLaHunt 2007-12-29
In theory, they're identical to within +-5 seconds of the original. So theoretically, you could have three identical AAD-conversions from the same master and mix, one coming in at 1:55 (slightly time-sped up, one at 2:00 (time corrected), and one at 2:05 (slightly time-slowed) and they would have identical PUIDs. On the reality side, also, it is theoretically and practically possible to have two different tracks, albeit very similar, with the same PUID... just very very uncommon. (I have found one example of this, a "with vox" mix and "without vox" mix of the same work, audibly very distinguishable, but both resolving to the same PUID.) Even given both, however, real cases of each would in all practicality be rather rare. The much better reason I'd be hesitant to use PUIDs as definitive of identicalness is that puids are rather invisible to the submitter, and easy to submit, coming in as they do without any voting and without any user-prompting. If we could guarantee that all puids were correctly assigned, in theory this could work, so long as the two aforementioned issues were taken into consideration. In practice, however, I think there's a good chance perhaps 5 to 10% of the puids in the system are attached to the wrong something of a track, be it the wrong version, mix, live puids attached to studio (or vica versa), the wrong performer, whatever. Better than we ever had with TRMs, but still the same caveats I think must be considered before treating them as more than an indication of potential identicalness - and never do I think we should be using them as the end all be all guarantor of identicalness. -- BrianSchweitzer 2007-12-31 16:43:42