Topic on Talk:Growth/Personalized first day/Newcomer tasks

Hi WMF Growth folks! An idea for a possible newcomer task occurred to me, namely helping create redirects, and I wanted to run it by you all.

The appeal of it is, first, that there's a lot of need: the search functionality isn't great, we currently have no good way to tell which popular searches are candidates for redirects, and the sheer number of possible redirects means that there's a lot more of them to add.

The second part of the appeal is that they're extremely easy to make and to prompt. I imagine that the workflow would go something like this: First, a reader searches for a term, say "Window washer", that does not have a redirect. They are brought to the search page like normal, and they click on Window cleaner. At this point, the software recognizes that this is the 10th person who's searched for that term recently and every one of them has gone to Window cleaner, so it generates a pop-up box asking if "window washer" should be redirected to "window cleaner". The user is taken to the page afterwards whether they click yes or no, but if they click yes a redirect request is generated in their name (or, if this works well enough, the page could be generated directly). The thing I like about this is that, since it only requires a single click, it's unobtrusive enough that we could use it even for readers who have never expressed interest in editing before and thus get them over the hurdle of making their first edit.

The main concerns that come to mind are (1) privacy, since this involves use of search data, and (2) that creating new pages, including redirects, is limited to autopatrolled users, so we'd either have to carve out an exception to that, or established editors would have to handle the queue. I think both of those challenges are surmountable, though. Feel free to let me know what you think! Courtesy pinging @MMiller (WMF).

I am quite opposed to this idea, precisely because it is so easy to create a redirect. I recently had to threaten to block a user on en-wiki for creating innumerable redirects for every single person's name ever credited with making any minor contribution to anime films they were interested in, and they made over 80 redirects for alternative spellings to one single article (see Example.) The fact that a redirect for a trivial name has been created can seriously hinder searching for that name across multiple articles. Having a redirect does not improve search functionality - it weakens it if not done sensibly.

Thinks of an example wherein a minor session musician contributed to backing music for a notable band. Their name might be listed in the infobox or main article text, but they might have worked in a lesser way with twenty other bands. Ten of these bands or albums have articles about them, and this session musician is mentioned in three of them. Without any redirect present, a user searching for that name will encounter search results offering all those different articles, and so they can go an look at each one in turn to find the information they seek. But if a well-meaning new user and music-lover decides to accept the invitation to create a redirect to the first article, their 'helpfulness' may actually hinder many other users finding the information they seek.

Now, the example of the troublesome user I've given above was extreme, but I could envisage us being flooded out with inappropriate and unnecessary redirects made by uninformed editors, every one of which will need a deletion discussion to get rid of. So, no, please let's not consider offering this facility.

The troublesome scenario you're describing sounds like what could happen if we just let people loose and told them to make redirects however they wanted. That would indeed be bad, but what I have in mind here is something a lot more controlled. It certainly wouldn't allow something to create 80 redirects to the same page, as it'd only be triggered as I described above. For the example of a musician, if they were part of several bands, different people searching for them would click on several different pages and the system would not suggest a redirect. Is their affiliation with one particular brand is really so strong that all 10 searchers chose that page, then the redirect would likely be appropriate as a related topic.

Regarding "every one of which will need a deletion discussion", not if this is done via w:WP:Requested redirects.

Redirects are indeed a tricky topic. Some communities allow plenty of them, some other have very strict rules. Since we are looking for universal workflows, I don't think working on redirects would be easy.

This said, the search issue you raise Sdkb is a concern. I don't know much about Search, but I though that aliases listed on a given Wikidata item are taken into consideration on search. But maybe I'm totally wrong and, if so, you should suggest it to the Wikimedia Search Platform team.

@Sdkb -- thank you for coming to us with this fresh idea! I haven't thought that much about redirects, and I would need to learn more. I definitely like that Community Wishlist idea you pointed out, as such a list could be useful for both redirects and new article creation. I also understand the kind of burdensome scenarios that could come about as @Nick Moyes describes. If I'm thinking ahead to how such a task might work, it might be something like a feed of suggested redirects, surfaced from an algorithm that looks at search data, and the user would be asked, "If someone searches window washer, should that take them to the window cleaner article?" Then the user would be constrained to confirming suggestions, as opposed to being able to create any redirect that they think of on their own.

I'm going to add this idea to my team's list of possible structured tasks, just so we don't forget about it. I'll also ask the WMF Research team if they know of any research into redirects and patterns with their creation.

@Nick Moyes -- if you have some time, it would be really helpful to us if you could check out the "add an image" project that the Growth team is thinking about, and add any thoughts to the discussion page. Your input was very valuable when we were first talking about "add a link", which we're building now -- but this "add an image" project is more ambitious with more pitfalls that we want to avoid.