Talk:Contributors/Strategy

The POV of a random editor

The article is tagged as a draft but since it brings up the issue of why Wikipedia is seeing a steady decline of active editors, let me give you a down-to-earth opinion based on my experience here on mediawiki.org.

The WMF sites at large are all very protective of their own interests. Chiefly this includes having high-use, very visible templates protected against vandalism in order to avoid unnecessary wear & tear on their servers due to the way MediaWiki works. This protective and completely reasonable attitude, however, doesn't apply to editor contributions, particularly translations, even when this kind of contribution seems so much wanted with even very draft-level and barely usable articles (example) being marked for translation.

Let me give you a hint: spending any time here translating an article (and to a lesser degree, creating or updating one) feels like a waste of a time, and here's why:

• Anyone can modify your translation because it isn't considered worthy of any protection and well, getting everyone and their dog on board and editing is a pillar of Wikipedia and by extension this site as well.

• You may spend minutes, hours or days translating an article, but vandalising a translation is far easier and takes much less time. Active rollbackers such as Clump and others do what they can, but there's no way they can can watch everything. In the end, it will be up to the original author or other bonafide translators to amend any foul play.

• Extension:Translate is an ingenious piece of software but leaves a lot to be desired when it comes to usability and the way it integrates into the MediaWiki workflow. Admittedly, the translation process itself is currently pretty straightforward, however undoing vandalism (or just bad quality edits) is a pain and requires far too much work. In addition, it pollutes articles with so many tags that updating the base article becomes a chore and puts you off completely unless you are paid for this work.

• The VisualEditor, while a genuinely useful tool to keep things simple and accessible for users new to wikitext (and those who are simply too lazy to learn the syntax), is also responsible for making articles a total mess. Today, editing the average sized Wikipedia article *requires* VE because doing this at source level is downright impossible with such a mess of cite web and other templates intermixed in the text. "Spaghetti code" is a term that totally fits Wikipedia articles these days.

I could add that adding AI to the mix as is currently the trend with everything will only make things worse, but who am I to stop progress.

Despite all these flaws I'm still around here basically because I appreciate the work the MediaWiki developers do and try to contribute what I can because it's the least I can do since I benefit from their work as a non-WMF MW administrator. I do even appreciate the contributions of *some* WMF members, namely those contributing code, documentation or end-user support. I could however live without those pushing their political views or personal agendas, and those who leave actually useful resources to die (see the WikiApiary fiasco) but sadly Wikipedia and the WMF movement are full of it. Tactica (talk) 13:23, 26 June 2025 (UTC)

Hello Tactica

Thank you for sharing your thoughts. I hope I'm in the right bucket regarding your assessment of WMF employees, so that you'll have a pleasant conversation with me. :)

If I understood correctly the key points of your message, you ask to have a way to protect translations from being freely edited by anyone, as you notice some vandalism. I sometimes miss not being able to protect translations. I also really appreciate the work of all Mediawiki.prg admins (and translation admins who review all the translation markup). You also expect to have better patrolling tools so that vandalism could be avoided on mediawikiwiki's translations. There is a potential solution with the Moderator tools, that use a language agnostic model to filter edits that are likely to be vandalism. I don't know if it would be possible to apply this model to mediawikiwiki (it is not possible to use it on Wikidata).

I'll be honest: the strategy we publish and you comment on may not be the best place to find a solution to your concerns. It guides the work of four teams committed to improving the editing experience, primarily for Wikipedia. Each of these teams keeps the other wikis in mind so that they can offer their tools there as soon as possible. The most appropriate place is the global annual plan, which influences how Mediawiki is built and developed. Unfortunately, the call to participate in the 2025-2026 plan has just ended. :/

If you have any suggestions for improving staff contributions on mediawikiwiki (documentation, translations, etc.), please let me know so that I can try to improve things, taking into account the limitations of the ecosystem we all live in.

Best, Trizek_(WMF) (talk) 17:25, 26 June 2025 (UTC)

Positive feedback

I'm responding to the request of User:SPerry-WMF at a village pump.

The draft correctly identifies positive feedback as important. It also says the volunteers like "being able to track their impact" and I suppose that's because it's a strong positive feedback. Here are some observations and experiences I have had which may help in this.

1. On the English Wikipedia, I wrote a template called "annual readership" for use on the talk pages of articles. This displayed a graph of the daily readership of the article over the last year. This seemed useful in various ways by showing trends, periodicity, spikes and the overall level of attention that the article was attracting. This was enlightening to many editors who were not aware of other ways of finding these stats and so the template was added to thousands of articles. But, unfortunately, this was broken when the infrastructure for displaying graphs was destroyed and the replacement for this still does not support the needed functionality.

2. So, what's badly needed are more and better ways of telling editors that people are reading and appreciating their contributions. For example, I occasionally write reviews of restaurants and other places on Google Maps. Google now regularly strokes me to encourage me to keep doing this. For example, it emails me a monthly report telling me that I'm now a level 6 guide, nearly an expert photographer and that thousands of people have read my reviews and so on. The Special:Impact feature on Wikipedia provides similar feedback but that feature seems too hidden from editors like me -- I only discovered a way of adding it to my user page recently.

I could say a lot more about editor motivation and how it does and doesn't work on Wikipedia. That's because I have been editing Wikipedia for nigh on 20 years and have created many articles which have now accumulated over 120 million views. This is the best way to understand the issues -- try editing yourself. See Eating your own dog food!

Andrew Davidson (talk) 23:07, 6 December 2025 (UTC)

Hi @Andrew Davidson, thanks for the message. I'm responding because I work with @SPerry-WMF in the Contributors space, and my team (the Connection team) focuses on creating a more meaningful experience for editors, particularly through making it easier for editors to collaborate, receive recognition, and understand their impact. So, I really appreciate you sharing these thoughts, as well as the background on annual readership. We agree that editors need more accessible and in-depth data points on editor impact. Like you mentioned, there is the Impact Module, but more can probably be done. You shared a great example of the insights provided to Google Maps contributors (which we'll think about some more). At this stage, we're still collecting feedback and learning from people about potential next steps to take under the Contributors Strategy. So, comments like these are helpful in sharing examples of what motivates you or others as editors. Finally, as a side note: Many of us in the Contributors space genuinely enjoy editingǃ I completely agree that it is one of the best ways to learn about the wikis (as well as talking to many people, so we're not too swayed by our individual biases). Thank you againǃ IFried (WMF) (talk) 22:35, 10 December 2025 (UTC)

Here's another fresh example of demotivating decay. The article Abraham Lincoln is featured today at the English Wikipedia. Looking at its talk page, I notice a banner which reports that "This article has been viewed enough times to make it onto the all-time Top 100 list. It has had 90 million views since December 2007." The template which creates this message – All time pageviews – is now threatened with deletion too.

The problem there is that the statistical series hasn't been maintained. The new pageviews tool only reports about 62 million views because it only goes back to 2015 and so only covers 40% of Wikipedia's 25-year history.

This sort of thing is corrosive because it sends a message that great achievements on Wikipedia are actually ephemeral because the records will not be maintained and respected. See Ozymandias...

Andrew Davidson (talk) 09:18, 16 February 2026 (UTC)

Another fresh example of a different sort occurs to me. Yesterday was the 20th anniversary of the creation of my Wikipedia account. I had planned to celebrate it at the monthly Oxford meetup but that was cancelled and I got no messages of any sort about the anniversary on the day.

I'm fairly thick-skinned and so not especially bothered about such ceremonies myself. But people notice such courtesies as a sign of community health.

Andrew Davidson (talk) 09:36, 16 February 2026 (UTC)

Agree with what you said. Re 1 it would be great if a replacement was built & added to talk pages – most readers & users don't know of the pageviews stats and how to get to them. Is there an issue or wish about this? Would be even better if one could add labels to spikes and display totals for eg 30 days & 1 year.

For example […] Google now regularly strokes me to encourage me to keep doing this. For example, it emails me a monthly report telling me that I'm now a level 6 guide, nearly an expert photographer and that thousands of people have read my reviews and so on. good points. One could also send users summaries like that and create more meaningful achievements/badges than the meaningdevoid edit-counts which can work via distinguishing between different contribution types and then providing metrics/stats for these where you can find some ideas here (especially on the 1 page in that category; e.g. the Commons app already shows the percentage of your files that are in use in a project; that's not what I mean but helps illustrate). reddit & StackExchange for example also have motivating & fun/feedback badges for achievements and genuine contribution levels.

Special:Impact is near-useless as far as I can see (in the vast majority of cases at least) – a missing piece toward it being interesting, meaningful & motivating is to distinguish edits: it should separate out articles one has created (including redirects turned to new articles and if possible also separating out articles that are just translations) and articles with comprehensive edits (not just a typo fixed but at least a whole paragraph if not section added) and then provide separate stats (charts & list) for each of these.

One nice motivating and/or interesting feedback users get is a notification when one of the articles they created has been linked from another article; one could also add notifications when files they uploaded get used (phab:T77154). One further area of things that could be done is not mentioned here are edit campaigns – e.g. see here and here – where there could be things like campaign statistics & tables that make the whole thing more fun and engaging in part via the direct and largely indirect feedback (such as seeing yourself in a leaderboard or seeing some stats of a group effort compete with that of others or seeing the overall campaign impact stats one has helped with etc). Prototyperspective (talk) 18:52, 29 January 2026 (UTC)

Thanks for the comments. On your last point about campaign stats and leaderboard, note that there's already an Outreach dashboard tool which is used by event organisers to record the activity at editathons and similar events. In my experience, it's just used with a short timescale of a day or two so I agree that there ought to be a version more focussed on campaigns which last for a week or more. Andrew Davidson (talk) 20:34, 9 June 2026 (UTC)

Follow-ups on "LLMs in Wikipedia: Investigating How LLMs Impact Participation in Knowledge Communities"

Hello @SPerry-WMF: the Wikimedia Research Showcase for Feb 2026 had a presentation with the topic "LLMs in Wikipedia: Investigating How LLMs Impact Participation in Knowledge Communities". The concept of using artificial intelligence in a tool that would provide suggestions and scaffolding to contributors sounds very promising. Is your team doing any follow ups that are linked with this research project? (CC: KGordon (WMF)). ↠Pine (✉) 03:52, 1 March 2026 (UTC)

Hi @KStoller-WMF: I haven't heard back from SPerry-WMF. Are you able to respond to the above? (CC: KGordon (WMF)). Thanks, ↠Pine (✉) 21:22, 20 March 2026 (UTC)

@Pine Thanks for the ping! I think @SPerry-WMF is best positioned to speak to this from a broader Contributor perspective, but I'll share what I know about this.

From the Growth team’s side, we have collaborated closely with the Machine Learning team and continue to do so to support newcomer experiences. This includes developing Structured Tasks such as Add a link, Add an image, and Revise tone, which provide guidance and scaffolding to help newcomers get started editing.

These features rely on task-specific machine learning models and heuristics that are designed to support well-defined editing workflows. While they incorporate elements of automation and prediction, they are more lightweight and targeted than the large language models discussed in the research presentation.

In parallel, the Editing team has been developing Edit Check, which provides real-time feedback during the editing process. It uses similar underlying approaches to identify potential issues and guide contributors before they publish potentially problematic edits.

The Moderator Tools team also explored this space through work on automated moderation support. For example, AutoModerator uses the 'revert risk' machine learning models developed by the Wikimedia Foundation Research team. As with the other systems I mentioned, this approach focuses on task-specific predictions rather than general-purpose LLMs.

While this work is not a direct follow-up to the specific research project you mentioned, it is closely aligned in spirit. Across teams, there is growing interest in using automation to streamline repetitive tasks, reduce friction, and make contributing more efficient and rewarding for volunteers. KStoller-WMF (talk) 05:14, 27 March 2026 (UTC)

Hi @Pine, I'm sorry for not getting back to you earlier, I've had it on my to do to rewatch the research session and I'm glad @KStoller-WMF jumped in so quickly to make up for my delayed response. Kirsten already gave a great overview of projects we are investing in that use machine learning. Just to add to her points and coming back to the take aways from the research presentation, I see a lot of alignment between how we're approaching AI at the Foundation and how Moyan Zhou is framing the opportunity, specifically when she's talking about "AI for Process" and "Show My Work". With regards to "AI for Process", our focus at the Foundation is on using AI to enhance the work of editors. You can find more details on that in the AI strategy we published last year. And with regards to "Show My Work", a good example is how we're adding tags to some of the edit checks Kirsten mentioned above, for example by adding the editcheck-tone-shown tag for Tone Check. I'm particularly interested in this topic, because I'd like to find ways to better equip new editors to be successful while also making sure that by doing so we're not inflating backlogs and we're providing experienced editors with the tools they need to handle backlogs and trust newcomers with these edits. SPerry-WMF (talk) 18:16, 27 March 2026 (UTC)

• Hi @KStoller-WMF and SPerry-WMF: thanks, I may re-engage with this topic a few weeks from now. @KGordon (WMF): does WMF Research have an open line of communication with Moyan? I may want to follow up with Moyan also, and it might be helpful if someone can introduce the two of us via email, or a talk page ping if Moyan has a Wikimedia account that's already publicly identified. ↠Pine (✉) 20:32, 28 March 2026 (UTC)