We’ve been offered feeds. That won’t work until we have one of two things: much better quality of data provided by the labels (unlikely) or a clear way to distinguish imported, not verified data from data verified by a user. IIRC BookBrainz already has a concept for the second option - if it works, we could copy it.
In any case, our experience of label feeds is basically that the community would kill us if we accepted them untouched, because they’re often terrible.
I didn’t know you had received an offer of some, okay I agree some verification is required but that is fine verification is still quicker than adding the data, would be great if the imported idea was to happen. Dont know how the feeds work but I would think just a feed of new stuff would be priority as older releases already well covered.
You could also probably also autofilter out particularly bad entries in the label feed so they wee discarded at the start.
Thats a good example of bad data !, but my own checks have shown the data to be okay so its by no means all bad. Clearly MusicBrainz is much better than Jaxsta, but there is still the problem that there is a quite a lag in getting new releases added. I also find that many singles never seem to get added, only the albums get added.
I wonder if this is the direct result of how many editors have interest in the singles that you mention. I add Jazz and Classical, and especially the later one is quite current with new releases. There are a number of editors who are very quick to add new releases, user scripts help with the more tedious tasks.
I think the biggest asset of mbnz is it’s schema and curated entries, this in my mind needs to be guarded.
I know I’m repeating others (and myself too Have you ever heard of Jaxsta? - #5 by Skeebadoo), but Jaxsta made big promises about getting their data cleaner and yet it still strikes me as a middleman business relying on automated data gathering and not much beyond that. Don’t get me wrong, I love the idea (like any such database), I like the presentation, but execution? Not so much.
Honestly, I basically just typed the artist I’m listening to right now and this is the result
They may be wrong but there are alot of people who would trust data coming from an official source much more than community curated sources, its the same argument for not liking wikipedia.
Anyway my point wasn’t about the quality of the Jaxsta, (although I thought quality was adequate for matching albums it doesn’t really matter that much if they have multiple versions of the artist if the details of the attached release are correct), my point was because they are getting feeds from the labels they are getting new releases into their database much quicker than they get into MusicBrainz. Also they know what new releases a label is releasing whereas MusicBrainz doesn’t, so nobody really know what it is missing.
I can offer the opposite point of view. For the music I listen to (classical and opera) the data coming from the official sources is sometimes OK, sometimes really sloppy. MusicBrainz data on those releases is much more reliable — in part, because I edit it until it is up to my minimal standards.
There have been studies showing that Wikipedia has broader and more accurate coverage of topics than e.g. Brittanica, an encyclopedia edited by experts. There are weakness with Wikipedia, and weaknesses with MusicBrainz, but I trust both precisely because they have good editorial standards and a community which does a pretty good job of curating.
I saw this statement on another topic… What is an album?
In the digital world: An album is a bunch of individually downloaded songs. A single is a song that is individually downloadable.
It’s not like the ‘old days’ where the single is a two song 45 and the album is the ten song 33.
Today, you download the songs you want.
Basically, today’s albums are 20 singles. Do we really want those listed separately?
Jim I agree with your entirely, it isn’t my point of view that offical sources are better. I’m just saying this is the view of many people, and in their eyes if MusicBrainz said it had some feeds direct (but still checked) from official sources that would give it a level of credibility it doesnt currently have amongst some people.
Can you be more specific who are those people who believe that direct feeds from Commercial Businesses that clearly are not interested in consistency would give a Community Driven project that is driven by data quality more credibility ? Another Commercial Business ? Academia ? Some newsgroup ?
We’re all ‘hardcore’ users (in one way or another) if we’re in the forums, but surely we can agree that there are people (and orgs/programs/sites) out there who look at MB, see that it’s got less on it than Discogs, and leave and don’t ever look back.
Having low quality data is better than nothing, but duplicates and all that crap that’s harder to tidy isn’t.
I wonder if MB could charge a fee to bigger labels for editors to clean their data and add it… if we’re confident having a release in MB provides value. p.s.
Last time I remember that we had a huge batch of automatically added releases was in 2014: Tag “likedis auto” - MusicBrainz
It was not really that bad, it is even maybe positive as it added many small artists.
It consisted of importing releases from artist.cdjournal.com which is a rather good source.
But automatic import has its flaws that we are still fixing 7 yeas later, like singles with missing karaoke tracks, tracklist not as printed, bad release group type, etc.
I am not suggesting just dumping in a load of new releases directly into MusicBrainz without verificaction, I envisaged them being either loaded into MusicBrainz but marked with some flag so they that are not visible by default each one would have to be converted into a release once checked by an editor, or loading them into a MusicBrainz Import database and then users could check them and import them.
But the reason I think this would be a good idea, is firstly that the labels do know better than MusicBrainz editors when they are about to release new albums, and it would give much better coverage of new releases. Secondly even if the data has errors it is surely easier to modify an existing release that is nearly right than to add all that data manually in the first place.
And after all we have plenty of scripts that are used to seed releases already from various websites, but this still requires you to find the release on another website, check release not alrealy in MusicBrainz, and dont add in the performer credits for you so you end up with the only the basic information unless you are prepared to do further manually editing.
I understand your wish for more up to date entries, there aren’t many who wouldn’t like that. But I do not share your opinion that bad entry is better than none because it is easier to correct later.
Corrections always (for regular editors, not the super ones) go thru the voting process which can take 6 or 7 days. So when I tag a new release and it is badly incorrect, I spend time correcting it and then I have to wait for a week before I can actually use it.
In the other scenario, a release I want is not in the database, I uses various users scripts to add it, make an effort that editing is correct and I can use it immediately.
It takes me 15 minutes or so to add a new simple release OR I need to correct an existing release and wait 7 days before it’s effective.
I’m all for a volume, off page, API style new releases adds - but on MBNZ terms.
After all Discogs already exists (I don’t think they have API feed from labels either) and one can use Discogs as they have more releases and get them quicker that MBNZ. I tried and had to say thank you, but not good enough.
It’s indeed more difficult to check and fix stuff, it’s boring too.
It would be difficult to find editors willing to review tons of label provided data, rather than working, as usual, on their collection of favourite artists.
You are obviously an experienced editor and this would not prevent you continuing you adding your releases as you currently do, but for casual editors its much more difficult and time consuming. I occasionally add releases to MusicBrainz and have done over many years but I can never remember the details of the various scripts. Users who are not particularly interested in MusicBrainz but are interested in good metadata (e.g roon users) are always finding MusicBrainz entry unfathomable. For my part I add releases with release seeding but not performer credits because takes too long, you may consider that a bad entry but it is better than not having the release at all, and if there was a way to preseed the credits I would use it.
And in your example it doesnt take 7 days to enter the information, just 7 days to wait for the data, very different. Im not recommending adding super bad entries but tigerman said
so someone could be selective about feeds and just take good ones.
Regarding Discogs, they tend to concentrate on physical releases so not so good for the shift to digital releases.
Well if noone is interested in then they don’t get done, but that is no different to releases not being there currently because no interested enough to have done them, but I am sure there are many editors who have favourite labels not just favourite artists.