Problem: Article Alerts is an automated subscription-based news delivery system designed to notify WikiProjects and Taskforces when articles tagged by their banners or placed in their categories enter various formal workflows (such as Articles for Deletion, Requests for Comments, Peer Review, and many more). See WikiProject Physics/Article alerts for example. This is currently an English-Wikipedia exclusive, and is maintained by one user (en:User:Hellknowz), meaning the bus factor leaves the whole project in a fragile state.
Who would benefit: Every edition of Wikipedia. For scale, on the English Wikipedia, 1467 WikiProjects and Taskforces subscribed to the Article Alerts system. Virtually every active project is subscribed, and the system is one of the best lines of defense against improper deletions and one of the best ways to advertise ongoing high-level discussions to communities of interest.
Proposed solution: Have the WMF / larger Wikimedia community create a more solid and scalable framework for Article Alerts that can also be deployed to all languages. Maybe it's not possible to have something as customizable as the English Wikipedia's implementation, but there are loads of things that could be ported and deployed. (Edit: See en:Wikipedia:Article alerts/Roadmap for a tentative roadmap.)
I rank this as one of the most useful watchlist tools on en.wikipedia, and the bus factor worries me, although the code could presumably be shared even as is. However the main barrier to overall global implementation is probably on places like simple.wikipedia where there are no WikiProjects and the category’s are generally incomplete. Therefore it will probably be of more use on Wikipedia language editions with a larger userbase like de.wikipedia. A Den Jentyl Ettien Avel Dysklyver (talk) 15:08, 9 November 2017 (UTC)
To clarify, there is redundancy in code access (two people have it), the problem is that there's only one coder. This makes new features / bug fixes /general maintenance dependent on the will, time, and capabilities of one person. Running the bot on the toolforge would also be great and make the bot more reliable during holidays if there's a power failure and such. The bot clearly is more useful for the larger Wikipedias (certainly all 10 Wikipedias feature on the main landing page would greatly benefit from this), but if a framework can be designed and ported, it wouldn't be a lot of work for dedicated subcommunities on smaller wikis to benefit from this as well even if it's easier to follow every discussion on a smaller Wikipedia. Headbomb (talk) 18:33, 9 November 2017 (UTC)
Not even all larger Wikipedias use WikiProjects, though. It's not only a size thing, but also history (how have we organized) and editing culture. It's probably good for editors to be keenly aware of the fact that if they don't have WikiProjects, the infrastructure for something like this isn't there on their home wiki, and it won't work for them. /Julle (talk) 12:29, 4 December 2017 (UTC)
@Julle: To a point. You don't have to be organized by Wikiproject, you can be organized by categories for instance, or organize via any other grouping of topics. For instance, you could have alerts based on the transclusion of an infobox, such as en:Template:Infobox biography. So even if you don't have a Wikiproject structure in your own wiki, there are still ways this could be used. That's part of what the proposal is. Have the WMF take over, and have a team that's dedicated to making it work for the specific needs of other (non-en:wiki) Wikipedias. Headbomb (talk) 20:20, 4 December 2017 (UTC)
I love Article Alerts! It is a great tool that can be used for informing large groups of editors about issues that interest them without annoying others. It is also wonderful for those interested in collecting wiki-statistics. Wikiprojects that have been set up to look after a group of articles can add Article Alerts to the wikiproject with minimal effort, and have the Article Alerts available to anyone who knows where to look for them. There was a good Signpost article about Article Alerts back in 2009.
The en.wiki has thousands of Wikiprojects (the Signpost has many articles about individual wikiprojects). Wikiprojects are a great way for editors to find other editors interested in a certain topic, which can be a great motivator for many. Tools such as Article Alerts help shift the burden of running wikiprojects from human editors to BOTs. This in turn helps retain editors since much of this task is repetitive/tedious and tends to burnout humans.
I would love WMF to make this as a wiki-compatible configurable tool (at least to English WP for start) rather than an external bot. The benefits are too numerous for me to list, but multiple languages is certainly one of them. This would be much too big of a project for me to do to convert it into an extension or tool labs bot or something of the sort. It's been quite a few years since coding AAB and real life certainly means that one volunteer coder for the project is unfeasible for any serious expansion into other languages or complex workflows and on-wiki features. The code is also pretty atrocious and team-unfriendly now that I can look at it with some 10 years of experience. It was more of a "wow, this sounds like a cool bot I can code" rather than a plan for a community-driven open source tool. The things that would be different/new/incompatible for a tool that directly works with database and/or MediaWiki is again too numerous to list. I can fix and keep the bot running when issues occur, but getting to new features and even bugs is a hurdle that could be abruptly terminated by a bus. — HELLKNOWZ ▎TALK ▎enWiki 23:26, 1 December 2017 (UTC)
Don’t underestimate yourself User: Hellknowz (and you too User:Headbomb). The code you wrote ten years ago may not be up to your standards of today, but it is robust and has served Wikipedia’s community well. I worry that the code the WMF staff replaces it with will be inferior in terms of features.
Also I wanted to add a comment about statistics, which are not harvested and put to good use. For example, Alerts of AFD discussions show the number of participants in each deletion discussion when they are displayed on the project(s) page(s). However, no one is capturing this information for statistical purposes as far as I know. Wouldn’t it be nce to know how many articles are deleted overall based on the votes/consensus of one or two people? Ottawahitech (talk) 19:39, 2 December 2017 (UTC) Please ping me
@Ottawahitech: The bot has served the community well, we're well aware of that. I gave a restropective at Wikimania this summer, where WMF devs expressed interest in taking over [assuming it's endorsed on the community survey]. To be frank, the code's status has gotten to a point that whatever new feature we wanted to have, we can't have either because it would blow up the whole bot because of en:spaghetti code, or take too much of Hellknowz' time to implement. Very little has been done on en:WP:AALERTS in over two years. WMF taking over would mean much better integration (e.g. possibly even integrate this with en:WP:Notifications), have much better/reliable performance [we need manual restarts on our personal machines if the bot crashes, if we're busy, that can take a few days before we get to it], and allow other languages to benefit from it. It might have less features at first, but over time they would likely do more, and do it better. Headbomb (talk) 19:51, 2 December 2017 (UTC)
Support: while I only use a lot for the enwiki Ireland project, I find it great for reviewing proposed deletions that allowed me to save two articles this week alone. I sure other language wikis will find it useful too. Ww2censor (talk) 12:38, 2 December 2017 (UTC)