Block adding Wikipedia relationships to anchors (STYLE-835)

Tags: #<Tag:0x00007fe316aa9028>

As I wrote on the STYLE-835 ticket:

An entity should only be linked to a Wikipedia (or Wikidata) page that is indeed for the entity in question. This follows from these two guideline segments: “The URL should still be specific to the entity being linked” (although this is mostly meant as “do not link to too wide pages”) and “Points to the Wikipedia page for this artist”.

This suggests that we should block linking to anchors in Wikipedia pages.

Generally, this is being used to point to things like a section for a particular member of a group, or a particular album, on the page for the group as a whole. This not only goes against the previous idea of a 1:1 matching between page and entity but also causes other problems, like MBS-9463 (the abstract extracted is of course for the page, not the section linked, which does not itself have an abstract). Additionally, since sections of a page do not get Wikidata pages, anchors will in any case be lost once we finish a migration from Wikipedia to Wikidata.

I am opening this to see if there’s any big problems I’m missing regarding this change, although unless a solution is offered with regard to the Wikidata migration dropping anchors anyway, it might very well be a moot point to be honest.

5 Likes

Well, the anchor (technically, “fragment”) is part of the URL, so it’s technically following the first guideline. But yeah, it doesn’t really make sense with Wikipedia, so I don’t see any problem blocking it there.

As long as we’re only talking about blocking it for Wikipedia.

Some sites do weird things like displaying entirely different pages via client-side JavaScript based on the fragment, and we should still be able to link to those for sites that aren’t Wikipedia.

BTW: It won’t be a complete fix, as sometimes Wikipedia editors merge multiple small pages together, and then leave redirects to sections of the larger page. E.g., https://en.wikipedia.org/wiki/Bleeding_Gums_Murphy

3 Likes

Definitely :slight_smile: I have nothing against anchors all the time, and as you said some sites depend on them.

Heh. Well this guy does have a Wikidata page - so we would have something to link to, just no English article to display.

I think we should just move ahead and block Wikipedia altogether and finish the migration to only use Wikidata. But eh. :stuck_out_tongue:

2 Likes

I agree this is what should eventually happen, but first we need to write some code to let people paste a Wikipedia link and automatically add the Wikidata equivalent instead :slight_smile:

3 Likes

My $0.02
I think if John Smith has a WP page, we can link to WP. But we shouldn’t be linking to Band X. Nor should we point to, Band X#John Smith.

One of the problems with only linking to articles, rather than subsections of articles is that WP is fluid.
An article today may become a redirect tomorrow.

For example:
Fred Eltringham redirects to his band, The Wallflowers. Not a specific subsection. His article was created as a redirect. It never had content.
And then there is Jesse Money (I created her MB entry), where her WP article redirects to a subsection of her father’s article.
But, Normani Kordei previously had an article. It no longer is an article. It redirects to her band, Fifth Harmony.

Tomorrow, who knows, those three people could have their own article instead of pointing to somewhere else.

It’s just something to think about.

Added a new guideline to Wikipedia relationships:

“Do not link to anchors on a wider page (Page_Name#Page_Section), even if they are specifically about this work. You should only add this relationship if the page itself is about the work.”

Also entered MBS-9515 to actually block people from entering anchor links.

Is this change really meant to remove WP-Links completely, even if there are no artist pages itself?

https://musicbrainz.org/artist/59cd71a7-3f62-46fa-97d2-e2afa8d49eca/edits
https://musicbrainz.org/artist/6b3bf30a-f481-4d72-8196-fe32439a02a8/edits
https://musicbrainz.org/artist/3d7a4772-d825-4c48-bd94-f13cd947784e/edits
https://musicbrainz.org/artist/7379b590-4c35-4e39-a551-c9f8878a220b/edits
https://musicbrainz.org/artist/db09b829-817d-4024-9d32-aff8568cb713/edits

1 Like

I’m looking at the loss of these links as a reduction in quality of the database.

I suspect the 1:1 ideal is just too far from the real-world situation where 1:“many different interpretations” is the general case.

I read that these links will be lost anyway eventually with a move to Wikidata.
I don’t know the cost of not going the Wikidata route but the cost of losing these links and reducing MB towards being a spreadsheet style db is significant IMO.

2 Likes

Yes.

A fair amount of these might be acceptable for pages at Wikidata, even if they don’t qualify for their own German Wikipedia page. So there’s that at least. That said, I feel the wrong info being given by the links is worse than the loss of removing them. It means nobody using our data will accidentally show the info for the band instead of the desired info for a member.

The presentation of extraneous poorly linked data is a negative.
Editors and users not having access to the good information is also a negative.

As an Editor/User I value availiability over aesthetics.

There will inevitably be tensions between people wanting nice clean displays and those wanting whatever data is available

Perhaps someway of quarranting/separating off the dirty links can be found?

I’m coding illiterate but I imagine that the differences between a link to a WP article and to a WP article section could be used to separate.

1 Like

Maybe these WPsections links belong in the Annotation?

If so would a automated bulk dump from Extetnal links to Annotation be possible?

That way they would get clean and I’d get the dirt and we’d both be happy.

On the one hand, you beg for editors note and on the other hand you let WP links be deleted which give a valid source for the informations (just because this informations are anchored in bigger WP articles).
I don’t have to understand it, I just stop adding WP links at all (because the get removed anyway…)

2 Likes

This seems a bit backwards to me. If I understand this correctly, any links that are to a Wikidata page that are not 100% bang on topic will be removed.

So if we have an obscure band and an obscure release which is only mentioned part way down a Wikipedia page that is now not allowed? This seems backwards to me. Some of us don’t know how to add Wikidata pages and don’t have an account with them.

Sometimes it is hard to pull out all the details of an obscure band. When I am doing research on something and then find details part way down a Wikipedia page then I will add that reference as I know it will help the next person. Hard to find data is hard to find! :wink:

Removing that link from MB because Wikidata don’t see the band as important enough for a separate page damages MB data quality.

Another example I have been working on lately are Film Soundtracks. I add a film soundtrack to the database. I then go to Wikipedia and search for the Film Soundtrack. If I find it, I then add a Wikidata link for it.

Sometimes there is nothing special about the Soundtrack and Wikipedia merge the page into the Film page. Now part way down the Wikipedia page are the details on the Soundtrack. So in this case I then add a Wikidata link for the film.

Next I go add an IMDB link for the film to that soundtrack.

My understanding of the above is that you will now delete that Wikidata FILM reference page, even though it holds details on the Soundtrack. And right along side it is an IMDB link that goes to a FILM and not a Soundtrack but that is allowed.

That’s Illogical Captain :smile:

1 Like

IMDB links are defined as being “IMDB link for the film this is a soundtrack for” (we don’t store films, so that’s the only thing they’re likely to ever be, too!). Wikipedia/Wikidata links are defined as “link for this thing here specifically”. We could have a separate relationship, “Wikidata link for the film this soundtrack is for”, but otherwise, adding that link is breaking the expectations of people using the data.

I don’t think anyone is going to complain if you link there in your edit note as a source for the data! :slight_smile: It’s just not the Wikipedia/Wikidata page for this specific entity. I also sometimes find info about a person at their spouse’s/parent’s Wikipedia page, that doesn’t mean it makes sense to claim that’s the page for this person; it’s still their spouse’s or parent’s, and if you link to this Wikidata page, and someone uses that Wikidata page to, say, find all other relevant links for this artist, they’re going to get a very weird, very wrong result.

2 Likes

@reosarevok
These links to WPsections appear quality information that would suit the Annotation section.

Can they be semi/automatically/magically shifted to Annotation?

Guideline I see stops addition of these links.

It does not direct or imply the removal of the information contained in existing links from the database.
The best method for dealing with with these existing links needs to be discussed and agreement reached.

I am not happy with the idea that I might be editing in the dark when needed information has been unnecessecarily deleted.

2 Likes

What about a “Related Article” tick box? Some way of saying “This is very relevant, but not the exact article because nothing exact exits”.

3 Likes