Suggested tasks based on contributions history (user interests) also for experienced editors (Community Wishlist/W316)

Description

There are three main problems addressed by this proposal:

How to keep contributors engaged (many stop contributing after a while) and increase their constructive contributions
Structured tasks/Suggested edits are usually boring
Many editors have way too many tasks with no way to offload some of them to other contributors looking for tasks

It is proposed that the already-developed Suggested Edits are extended to be more engaging and interesting by having it suggest tasks for topics the user in interested (and likely somewhat knowledgable) in which would also turn Suggested Edits into something also relevant for more experienced users rather than only new users. For this, both selected interests and the user's contributions would be used.

According to Isaac, there are three WMF-supported recommender systems:

Suggested Edits: tasks on the iOS/Android apps that are oriented towards surfacing quick tasks -- i.e. constrained to mobile.

Newcomer Tasks: tasks via the Newcomer Homepage that are oriented towards aiding new editors in discovering simple edits to make to learn/improve as an editor -- i.e. mobile or desktop.

Content Translation: high-impact translation tasks that can be on mobile but are still largely desktop-oriented.

He also noted that Content Translation is already personalized based on past edit history. So this could be used for this and improved. Categories, WikiProjects set on the talk page, analysis of wikilinks (e.g. those in the lead and the WikiNav most clicked ones), which templates link to the page, which lists link to the page, and AI topic generation (similar to summarization) using the lead text, etc could all be used. There already is a tool that shows similar articles based on an input article and its results are already quite good: the list-building tool (I have suggested renaming it to something clearer like "Related articles finder"): https://list-building.toolforge.org/. It could be improved as described here and used for what's proposed here. Moreover, topics and selected interests are used Newcomer Tasks and these could be used more or exclusively until the contributor has made some contributions (of course the contributions history can't be used until there is a substantial history to base things on and even then it probably shouldn't be based only on that). When the contributions history is used, I think articles edited often or with large byte changes (that are not reverts and weren't reverted) should be weighed higher.

This may be less about building some large new thing than about making something already built far more useful. Consider how much of this has already been implemented and that it would also improve things for newcomers (more interesting/fitting suggestions) while there is a clear scarcity of contributors (& contributions) problem in Wikipedia which this may help mitigate.

I think these tasks are best shown by default on the Home page dashboard when logged in (in place of the 'Main page') and there already is a Newcomer Homepage dashboard which could possibly be reused or adapted for this. It could also be a separate page that is one option when clicking on the user dropdown in the top right.

Example tasks

A few examples of suggested task types that could be integrated into this collective intelligence system (each showing only tasks likely of interest/relevant to the user):

Connecting experts (of broad & narrow subjects) with articles that need expert attention (explicitly via hatnote or implicitly) – e.g. these (usually academics) may only edit a few articles of their niche subject and then not edit anymore
Connecting people who can create illustrations (problem: way too few) with articles missing illustrations, data graphics & other scientific images like these
AI suggestions of citations for phrases that have been identified as needing any or better citations … see study/tool "Improving Wikipedia verifiability with AI"
Suggesting categories for articles where the user if experienced only needs to confirm to add them or showing articles with identified too-few categories that match the contributor's interest so the user can add some categories to it
Some UI that shows media in an article's linked Wikimedia Commons category if the article has no media files (or only few despite having high pageviews) while there are either lots of media files in the WMC category or some files with indicators of good quality (such as having been featured as Media of the Day) and let the contributor select some good media file(s) from there. See related proposal Suggest media set in Wikidata items for their Wikipedia articles (I think these are best done partly automatically but not all useful media files are set on Wikidata items and the task type is about these instead of files set on WD items)
ToDos of WikiProjects (example)
1. Especially stub expansion of articles the user is interested in. Like with task 11 and 7, it's not recorded how many stubs a user has expanded substantially and changing that alongside this would make it far more impactful (especially if there was a chart on the user-page and things like badges such as 'top 1% of stub expanders of the week', 'expanded 20 stubs', etc).
A better requested articles system that matches people likely able to or interested in writing an article to actually learn about the request
A better media request system matching potential request-fulfillers with requests (related to #2)
Open talk page issues – see related wish Do not fully archive unsolved issues on Talk pages (not just any threads on relevant articles but those the contributor is relatively likely to able to / interested in solving)
1. Especially reviewing Add-a-Fact proposals and a subset of talk page requests that propose specified changes like these – this could speed up things a lot and keep things more up-to-date
Suggesting checking and improving the generated article & section summaries (and displaying such summaries of articles may enable people to quickly understand an article or issues with its contents to then edit it – e.g. the summary may be flawed because the article is flawed)
Suggesting adding missing sections (partly based on the see also section or other language version articles) of articles
Integrating tools that are meant to be used by more experienced contributors – e.g. instead of suggesting adding individual wikilinks (one at a time), integrate FindLink for articles identified as being underlinked. More task tools like this and #3 could be developed but would need some task system also used by experienced users to get integrated in to be useful.
Enabling contributors to add todos that other contributors can also pick up instead of keeping these to themselves – e.g. related to #4 see c:Commons:Categorization requests and this tasks system could be extended also to Wikimedia Commons
Relevant protected page edit requests, requests for comment, etc
Reviewing article drafts (to speed up article reviews and connect relevant contributors with drafts relevant to them)
Checking and/or checking of flagged phrases in a machine translated article of an article the user recently edited / edited a lot / is interested in – as described here if that project is made real
Integration of various scattered hard-to-find underused tools (gadgets, scripts, ...) and enabling the community to add new task types. Current approaches (see below) are about getting newcomers become familiar with Wikipedia (or are indiscriminate backlogs) while this system would be about getting things done (efficiently) and keeping contributors engaged & productive.
Helping out with new articles, drafts and stubs about topics one is interested in (related wish W132: Make it easier for users to find others to help in collaborating on writing articles)

Current approaches

See also Wikipedia:Task Center for relevant resources. Other sites keep people engaged by showing things interesting to them in their feeds every once in a while – this isn't the same but the particular task type (or getting lots of important things done via these tasks) may not matter more than keeping people who currently short of things to do on Wikipedia for whatever reason to still have fun contributing to the site and to keep contributing.

New users can already select interests
SuggestBot can already show personalized suggestions but it's not as comprehensive as what is suggested here, is used little, one can only use it when requested to get weekly fresh suggestions instead of all users seeing private suggestions and being able to scroll through many of them whenever interested
…it does not have a good UI, or much potential (it's just a boring list of articles), and is only based on hatnotes; maybe parts of its algorithms or ideas could be used here
Wikipedia:Community portal/Open tasks is not adjusted to personal interests and skills or experience level

Critique

I don't know how many would use this and think it could be that most editors are also overflowing with things to do instead of being out of ideas & motivation of what to do. Nevertheless, there's probably many who would greatly benefit from this, e.g. to keep being engaged when for some reason shortly running out of things to do or looking for easier or novel/uncommon tasks. One can always go back to such tasks and tackle some and this kind of thing itself is kind of fun because it's more interactive, has more feedback, etc. I think there are downsides to tasks getting editors who would not be involved with an article to be involved with them or that some contributors may turn from doing slow complex tasks to short quickly-rewarding microtasks but the benefits outweigh, especially if this is designed well. This is a complex subject and this proposal is under development.

Assigned focus area

Make it easier for patrollers and other editors to prioritize tasks

Type of wish

Feature request