looking at the Fresh Releases page on ListenBrainz has reminded me about this topic, implementing new release feeds from labels.
I believe the way to best implement this would be a new release status and a report of all releases with this new status. any such ingested releases would get this new status, that way it might show up on the artist page (provided they’re properly linked, and depending on what’s decided by the community), but it’s still clear that the data might need fixing before it’s properly usable. it could be a new release data quality, now that I think about it, though we might want to make it editable in the main release edit page if we go this route
releases with this status (or data quality) should maybe be auto-editable, either within a certain timespan (i.e. 24 hours after first non-bot edit, within a month of ingest, something like that) or until it’s given a different release status/data quality. that way we wouldn’t have to wait a week to use the corrected data once the release is found by an editor.
this wouldn’t need to be limited to labels either, as we could consider ingests from places like Bandcamp, SoundCloud, BeatPort, YouTube, and other places independent artists upload music. although I admit, this might mean more release group/release merges, which aren’t really fun, lol
that said, I’m surprised how many new releases are on the Fresh Releases page. I count 28 for today alone, which I suppose isn’t many, but it’s more than I expected to find…
I wasn’t actually aware that was previously a thing… that of course doesn’t necessarily mean we can’t use data quality, but certainly something to keep in mind
I still believe in this case it would be a good use of a perhaps underused feature tho~
People would abuse this “New Release” status to bump their own edits through faster without checks.
I personally I have made errors on some new releases I have added and had a week or three of waiting for edits to complete so that typos can be removed. It is due to those 7 day waits that my typos were spotted and could be corrected. (Often by me 25 hours later… )
Usually a friendly Auto Editor can help push that timescale along, depends on the error.
I see no problem with waiting. A bit of patience for perfection.
that’s the thing (which I guess I didn’t mention), the status would only be used by the import bots. regular users would not be able to set this status/data quality
I only really mentioned the auto-edit functionality as it seems to have been a concern in the previous thread. it’d be nice, but not essential, I suppose
An import bot is a user. A user can modify the bot. Still seems open to abuse. And even weirder to allow a bot a right a user can’t have.
Also who would then be allowed to update it? Only the bot? Or any user?
Nothing personal, just trying to poke holes in things to make sure they are bullet proof. Especially if this was a bot as they would not reply to messages (hard enough to get some users to reply as it is… )
If a release is entered with pre-release data, and then this data is different on release day, just bribe a friendly Auto Editor to help out. Much safer to have humans involved. (I think most of the AEs are human… but some I am not sure )
if there’s much worry about this, it could be a select subset of bots, maybe approved by admins or whatnot? I’d also consider it more of a curse than a right, lol
it would be open to any user to edit, merge, delete, etc. just like any other release. the first set of edits could be an auto-edit (save for any merge or deletion, of course)
for clarity, the sequence of events for such an imported release could be as follows:
release gets added by one of these bots, with basic data; release date, tracklist, label, catalog number, album artwork, etc. at this time it is also marked as an import (release status, data quality, or [third option]) note that this is the only thing the bot would do
after this, any one of 3 things happen
a user finds this release in the report, fixes any errors like artist credits, capitalization and punctuation, higher resolution artwork, etc
a user gets notified, as they’re subscribed to the artist (or the bot), finds the imported release is the same as the one they added last week, and put in a merge edit
the release remains in import limbo, unedited. this isn’t bad in my eyes, as it would be data that places like ListenBrainz could still likely use.
Still think it is an odd idea. You are making a hierarchy of different powers that are not required. Handing a small set of people who can create bots the ability to have special privileges seems against the normal way MB operates.
And I don’t know who you think these “admins” are - We are all equal here. Some are more equal that others (AEs) but otherwise we are all equal. Let everyone vote and edit equally.
This assumes that it will all be fixed in that one edit? What happens if someone updates this while it is waiting… now you break the “First is an auto-edit” special case.
I see no advantage to making a special release channel. It would require extra developer time to allow a small selection of users to feed the database. It just seems overly complex.
Why can’t you do the normal wait 7 days? This also allows more eyes onto the edits. And something popular is voted through quicker anyway. A normal release added in the normal way will get more eyes on than something only available to a be seen to an import limbo report.
I’ve added plenty of releases a month or two early. Info based on what was available. On day of release corrections are made. Anything major an Auto Editor can often help out. Just like any release that is usually added.
(I’ll stop waffling now… )
haha - this is funny:
" a user gets notified, as they’re subscribed to the artist (or the bot), finds the imported release is the same as the one they added last week, and put in a merge edit"
That sounds like you are asking permission for a bot to make a mess of the database with dups and not doing searches? This is a prime reason why I say “bot bad, human good” To me this modern “AI” is “Automated Incompetence”. A bot needs to know how to make a search first to check for a potential dup.
Short version:
Current MB system is efficient and allows more eyes on.
A bot adding stuff can follow the same rules as a human.
Bribe your friendly Auto Editors to help anything that needs a poke.
I can see why this is asked for - you and I may be happy noodling about with older releases, but there’s an immense amount of new releases coming out that MB doesn’t have, and probably never will, which is makes the service unappealing to a different set of users.
When adding a digital release from a digital source, it’s only humans introducing mistakes
If there are mistakes in the digital source, e.g. barcode or track titles, then we would be mirroring that ‘correctly’, except for in edge cases, like the wrong artist being linked in the store, which editors would have to fix.
However, dupes would be an issue. I would imagine that with something like this editors would eventually get used to the idea of not entering new digital releases (from certain stores) at all, in the knowledge that the bot will add them and then they can do edits as required.
(I think this is unlikely to get implemented unless someone just goes ahead and writes the bot/s… but it’s interesting to discuss nonetheless)
that’s fair, lol… there could probably be a basic search for matching releases, but I don’t know how available stuff like UPCs and catalog numbers are for unreleased music, so we might not be able to tell if it’s the same version… it would theoretically be possible to not make new release groups if one already exists though, provided the artist hasn’t done any shenanigans with their release titling (looking at you, Weezer and especially Brouillard)
this is exactly the reason, yes. it’s understandably difficult to find how many releases come out every week, but I’m sure there’s at least a thousand, certainly many more than the roughly 200* we’re getting already. having this data on new releases could give some people a reason to use MusicBrainz who wouldn’t otherwise
*based on yesterday’s releases multiplied by 7, lol
I will note again, the whole conditional auto-edit thing is entirely optional, I was just addressing some editors concerns from the previous thread
the core of my idea is how to denote such imported releases and integrate them into the database
I don’t understand bot coding, but could it do a lookup? Check for a digital release with that name and track list? Then flag this for the bot owner to manually check. This leads to a lot less work for the bot owner to do - automatic data entry for most releases, with a small set to check manually.
Don’t want to put off humans from adding new releases of their favourite artists. You need that human to check the data, so don’t scare them away. Especially as you’ll need the human to merge recordings.
I have no specific negatives against a bot. As long as the comments on the edits are checked and responded to. Yes, automate, but respond to the feedback. If it is all clearly coming from one user with “bot” in the name then we know what we are dealing with.
Using on new releases also means it would avoid that old error of digital releases from 1996 It would also be much more likely to have valid artist links, etc.
It would help if the bot does the basics like Guess Case. Otherwise it may cause more work than it saves.
The main thing I would worry about is a bot from a big brand like Spotify who thinks they don’t need to bother checking the data being output. Especially if they got used to a little team of MB editors always cleaning up for them for free. This is why the feedback would be important. Some kind of mechanism may be needed that allows MB to pause a bot due to an issue. We’ve all made that coding error
Slightly tangential to the direction the discussion is taking, but I want to bring up the fact that MusicBrainz used to have something similar already in the 2000s. It was extremely helpful for monitoring both genuine news and wrong data. It was left behind when migrating to NGS, but we have a long neglected ticket supporting its reintroduction.
From the ticket description:
dashboard included: edits with most votes, needing love, and expired; recently updated artists/releases/labels
music news included: upcoming and recent releases, recently deceased/broken up artists
I just wanted to bring up that apparently BookBrainz has a similar (upcoming?) feature which is very similar to what I’m talking about here (the automatic data import, that is)
based on my limited knowledge of this feature (i.e. what’s been talked about in the #/bookbrainz IRC), it looks like data is automatically imported into BookBrainz and added to a listing of recent imports (first screenshot). here they could be looked over by editors and checked for errors
additionally, on the entity pages (second screenshot), it’s noted that the entity has been automatically imported (including source and import date), with buttons to approve, edit+approve, or discard the entity.
I still think some form of automatic data ingest could be a great thing for MusicBrainz and everyone that uses our data, if handled carefully
there’s now a blog post about how BookBrainz is implementing their big importer project
I still think that an automatic import feature could be a great asset for MusicBrainz too, to at least get very basic data for a ton of music which might not get added otherwise
I don’t disagree in theory, but the data source needs to allow doing this. For books, things like the Library of Congress and OpenLibrary clearly allow it. For music, the places where most people want to import (iTunes, Spotify and the like) often explicitly disallow it.
)
)
)
It would also be much more likely to have valid artist links, etc.
