This is not bad idea in principle however often an artist is well represented either in Discogs or MusicBrainz but not both, just look at the first page (artists are ordered by total number of releases discogs + mb) and you’ll see that in many cases there is only one release in either mb or discogs.
Go down to page 20 and the total number of releases mb + discogs is only seven
So the chances of getting three matches is quite low
I could do this if it was required to get agreement
In my opinion, the criteria that ijabz has established are strict enough to combine the arithmetic of dicogs and Mb.
You should not be too fussy with such an update, otherwise nothing will ever come out.
Given the specifics of this import, I am absolutely fine with this.
Ideally the bot will be created and used only for this specific import, to make these edits easy to find. The bot’s bio could explain the matching used.
Edit notes could perhaps specifically list the criteria for that match (“same name (X) and unique in both + share these exact release names: A, B, C”, but them in nice English and layout). (Including the artist name is important, in case artists get renamed later).
Remember to contact us before running it in the main database, so we can check it and give it the bot flag etc as needed. You can go wild testing in http://test.musicbrainz.org/ if you want though! (that’d actually give us a good idea of the kind of edits it’ll do, the edit notes it will leave, etc.)
I’ve looked at the report, and I understand the premise. I think there is still potential for error, and I’d like to dig into the list and look for examples, but it is difficult to do so because the list cannot be filtered, sorted or searched as it is presented, so I’d have to wade through 65 pages looking for any artist I am able to comment on. Is it possible to make it easier to work with?
I don’t know how many artists you feel able to comment on are actually going to be in this list, as most are pretty obscure. It would seem more efficient to just pick some artists at random and check to see if the match is valid but if you really want I could generate either a csv file containing the full list that you could download and search with your favourite text editor.
Okay, rather than create a bot I have created two greasemonkey scripts
The first one i install on the report albunack that clicks on each link with a 5 second delay between each. (I have used a modifed version of this to only run against a few links)
// ==UserScript==
// @name AlbunackSubmitArtistLinksOnPage
// @version 1
// @grant none
// @include http://reports.albunack.net/mbartist_discogsartist_report2*.html
// ==/UserScript==
function delayedOpenLink(link)
{
link.click();
}
function check()
{
var links = document.getElementsByName("link");
var i=1;
for(link of links)
{
setTimeout(delayedOpenLink, 5000 * i, link);
i++;
}
}
setTimeout(check, 5000);
The second is applied to on the artist edit page when editing artist url to automatically click on the submit button
I’m unclear if this constitutes MusicBrainz definition of a bot as it is a bit different to the bots that already exist, and is just a way to automate the clicking of a few buttons. But Im happy to submit these with an ijabzbot account if desired.
Seems the key thing to me is not these scripts but if data in the reports is correct, this is explained on the report page.
These matches are very likely to be correct because a match has been found based on the artist name, the artist names are unique within the MusicBrainz and Discogs databases and both artists have at least one release with the same name.
and all testing done over the years have so far shown the results to be correct.
It’s a bot if it doesn’t require manual input/verification of data. As far as I can tell your userscripts just run through and seed and submit data without you manually checking, which would constitute bot behaviour.
Read some of the notes from a few years ago above, and now added the musicbrainz artist name and dicogs artist name to the edit note.
FWIW the difference between this and usual bots, is the other bots go off and do their thing and there is no way for other mb editors to check the results before they have been submitted to MusicBrainz. Whereas in this case the results are already in the report and can already be checked by anyone before they are submitted to MusicBrainz, so there would be nothing to stop somebody checking every submission now if they wanted.
So I thought the key thing about designating something as a bot was to make it easy for others to check the submissions, rather than requiring the user doing the submission manually checking before submissions to avoid it being a bot.
The main difference with bot edits* is rather that it is queuing its edits, massively, even if it starts to act wrongly.
Cannot happen when editing manually.
In this case, the edits are all created in advance so it cant really start to go wrong unless there is a technical issue with the MusicBrainz website, it would be useful to expand that bot page a bit I think.
Fair point, because Im concentrating on a particular data item (e.g mb artist links to disocgs) and dont use subscriptions I forget that the majority of editors focus on particular artists.
