I have opened up a new thread as this is about the Ashkenazy Chopin Walzes (and pretty much everything else). Not the Etudes. But for reference I link back as there seems to be a relationship:
There are no less than 10 editions of an album with the Etudes cover in the album identification wizard. The problem is if you look at the editions they are all completely different, all even with different album titles. I have the Walzes but in there are also:
These are all completely different albums although they were repackaged, re-bundled and re-boxed many times. Confusingly what this means in my particular case is that I get the Waltzes at a composition level correctly identified but the album is identified as the Etudes. So this:
I am away from home and I am not able to check if my other Ashkenazy’s are similarly messed up. I also have the Nocturnes for example and it may be the case that hey are also identified as the etudes.
We have looked into what you’ve reported here and have confirmed that this is an incorrect equivalence issue. We have also confirmed that this is resolved in the upcoming improvements to our equivalence that is currently in testing. When this is released you should see this resolved.
Thank you for the report, and apologies for the inconvenience.
Thanks for the prompt response @dylan. I am curious though. Are you able to elaborate what the incorrect equivalence is? That is, what exactly is being defined as equivalent to each other that shouldn’t be? Or will I just be more confused?
Tony, the problem in this case (and with some other bad album equivalences) is that the current equivalencing code does not take into account structurally incorrect contributions to Musicbrainz.
You can see the problem via the following link, which will also answer your question directly about what has been mixed up in this case (a lot, which makes an edit potentially very difficult):
For the Musicbrainz database to be self-consistent from a data perspective, a) every release listed should represent the same thing*, b) individual discs of multi-disc sets should be entered as discs within a release, and c) individual releases which are part of a “series” should be entered as distinct release groups.
That this does not always happen is down to the ignorance of contributors, to moderators (who do not understand the implications of allowing this laxness), to lack of enough experienced moderators, and to the submission “style guidelines” which are incomplete, ambiguous if not contradictory, and/or not adhered to or enforced.
Our new code removes Musicbrainz Release Group membership from its current pedestal as automatically trustworthy: unfortunately, whilst the Musicbrainz database structure is the best one I’ve seen, the data are not. Which in itself is also a lesson about the difficulties of crowd sourcing.
* As an aside, the 200-disc Complete Mozart Edition (Mozart 225) is an — ironically well entered — example of this badness: lovely metadata, but five groups of 40 different discs which are de facto purporting to be alternative releases of the same thing. It’s possible that we can do something to sort out such box sets on a case-by-case basis.
Wow Joel. I see what you are up against. Musicbrainz mods really need to enforce their own model. I thought the difference between a release group and a series on their style guide was clear enough.
It’s interesting to hear the background. I actually thought the splitting of the Mozart 225 into 5 sets was intentional. I’ve used that to create 5 smaller boxes I just roll-up with a “Mozart 225” tag. So silver linings . . .
I investigated this on the Musicbrainz forums a while ago and the only thing I could find was effectively “we’re a bit nervous about having so many discs in a release and we’re not sure if the system will cope”
You really should pin this thread somehow or release a blog posting as information for all of the forum posters who seem to think roon is the meta data problem rather than the source…
