Wikimedia Futures Lab/Dashboard/Experiment Tracker/Centralised Experiments Log
Possible model for the log:
| Experiment | brief description | date of experiment | status | Category | Outcome (anticipated or unexpected) | Link to data | Language/project | Creator(s)/Maintainer(s) |
|---|---|---|---|---|---|---|---|---|
| A Media Face for Commons | Swapping Wikimedia Common's frontpage with a more engaging and media oriented homepage, while paying attention to SEO to improve referral on "free image" searches. | tbd | in preparation (detailed requirements drafted) | Multimedia content | increased page views on Commons | link | Commons | TheDJ and Fuzheado |
| Access to AI tools for developers | Provide Wikimedia volunteer developers with free access to AI tools for “vibe coding” to improve their productivity and encourage experimentation | tbd | ideation | Tools for developers | increased experimentation on-Wiki | All | User: Netha Hussain | |
| AI generated video summaries for articles | Use AI and VideoWiki to make video summaries of Wikipedia articles, by generating a script and identifying Commons images | tbd | ongoing | Multimedia content - Video | See prompt for Claude AI and examples here: Wikimedia Futures Lab/Dashboard/Experiment Tracker/AI generated video summaries for articles | All WP / VideoWiki | Olubusola Afolabi (User:Ifedohlapo)
Andrew Lih (User:Fuzheado) Tochiprecious (talk) | |
| Audio versions of Wikipedia articles at scale | Procedures for creation of article narration audios that can scale to be used for millions of Wikipedia articles. | tbd | phase 1 completed | Multimedia content - Audio | increased page views on Wikipedia | w:Wikipedia:WikiProject Wikipedia spoken by AI voice | ENWP | User: Prototyperspective |
| Can you make Wiki fall in love with AI chatbots? | A Grand Prize Challenge to the best prototype of a stand-alone AI chatbot that uses Wikipedia (or Wikimedia) data that pays back the most useful data to Wikipedia. Low hanging fruit includes reported mistakes in Wikipedia articles, unexpected knowledge gaps in Wikipedia articles. | tbd | ideation | Ethical AI | elevate ethical AI to the top of the field and increased brand visibility | Wikimedia Futures Lab/Dashboard/Experiment Tracker/Can you make Wiki fall in love with AI chatbots? | / | Effeietsanders, Tanveer |
| Centralised Experiments Log | Log of experiments carried out involving Wikimedia movement by any individual or body r | Start 1 February 2026, revisit at 1 July 2026 | ongoing | Brilliance | Centralised Experiments Log | All | Eugene233, Carla (Soylacarli), Risker, Mayur-WMF, Eva Martin | |
| Coalition to build health metrics of the Commons | Bring together a group of aligned partners at Wikimania in Paris and other regional conferences in order to research, define, and draft health metrics for the digital commons. | tbd | ideation | Coalition of the Open | increased collective power to better sustain openess | / | Jay Prakash (Tech), Bisi, Claudia (WMAT/WMEU), Jens (WMDE), Ben Vershbow (GLAM-E Lab) | |
| Community perceptions about AI | Research to study how experienced Wikimedians across multiple language editions perceive, use, and govern AI-generated content, including tools, workflows, risks, and policy implications | 2025 - 2027 | ongoing | Research | better understanding of community perceptions about AI | Research:Between Prompt and Publish: Community Perceptions and Practices Related to AI-Generated Wikipedia Content | English, Italian, Bangla, Malayalam Wikipedia | User:Netha Hussain; User:tigrotta_ac |
| Creating editing assistant tools utilizing AI | utilize the AI to create the wiki-friendly gadgets to assisting the editors | reduced time spent on repetitive tasks | Research:The Impact of WikiVault on Korean Wikipedia | Korean Wikipedia | User:Youngjin (WMKR) | |||
| Innovation wikis | We want several wikis to agree to be "early adopter wikis", in which they elect to have experimental features deployed in A/B tests without the usual discussion of each feature. This will allow Wikimedia Foundation product teams to rapidly discover which ideas work and which don’t, and then move on. | We are aiming for four communities to agree to be early adopter wikis by April 2026 | Phase 1 complete | Ecosystem for experimentation | quicker experimentation of new features on-Wiki | Innovation wikis | Bengali, Czech, French, Turkish; more soon | MMiller (WMF), SGrabarczuk (WMF) |
| Future Labs Circle | Creating an “ecosystem for experimentation” in our movement will greatly increase the chance of experimentation being done at all and being done effectively. Be present with a contact/booth at every Wiki gathering in the next 12 months. | never stop experimenting | ideation | Ecosystem for experimentation | larger amount of success of experiments and quicker adoption of successful experiments | link | All | Jan-Bart,Lucy Crompton-Reid, Nazneen Nawaz |
| Improving admin recruitment for small wikis | Identify wikis ready for new administrators and share information (via village pumps, talk pages, or banners) encouraging them to select admins, including guidance on the process and available mentorship. | ideation | UWER recruitment | increased number of admins on small wikis | Small wikis | User:DerHexer User:HakanIST User:PMG | ||
| MainPage refocus | If we reform the main page to focus on recruiting new users we will break the trends of decreasing account creations | determined by individual project | increased account creation | individual projects | ♥Ainali | |||
| New user experience for readers | Create prototypes for a new product with a different user experience to attract users consuming our content who are currently not visiting our products anymore. Do user research to understand users journeys, pain points and needs better. Based on that, do ideation sessions on product ideas with students, community and staff members on a variety of different solution ideas. | Prototypes ready by the end of 2026 | ideation | Readers experience | new product that attracts more readers than current Wikimedia projects | New project | User:Raja Amelung (WMDE)
User:Lena Meintrup (WMDE) | |
| Reimagining non-Wikipedia Wiki interface | reimagined the format of Wiki projects through the existing system to provide tailored alternative to "usual" Mediawiki interface | EmpAhmadK (talk) | ||||||
| Simplifying new account creation | Make account creation zero-click (upgrading temp to full accounts) and make "Create account" a button with a simple one-click experience that makes clear what accounts are for. ("track your edits!" "join us!") | Start 1 February 2026 | increased account creation and editor retention | User:Sj/Design chats/Create account | Any | User:Sj | ||
| Social integration | Helping casual editors integrate into the community: If we invite more new editors / casual contributors to online + offline events, introducing them to (more clearly defined) roles and show more appreciation for all kinds of participation. | ongoing | increased account creation, editor retention, editor activity | DerHexer, Chinmayee Mishra, Pinky sl, Hakan | ||||
| Temporary Admin Role for Medium WIkis | Introducing an experiment of temporary admins as a way to (a) meet urgent needs, and (b) find potential candidates for long-term. | ideation | UWER recruitment | increased number of admins on medium wikis | Medium wikis | Renvoy, SanBonne | ||
| Topics for readers | "Give them what they care about" → special interest topic page |
|
|
| ||||
| Twinkle AI: AI-supported feedback to newcomers | Develop an AI tool for newcomers that can generate a nice, instant contextualized and personalized message explaining admins actions, like deletions and rollbacks, and post it to talk pages. | ideation | Contributor retention | increased newcomers retention and improved quality of edits from newcomers | Discussion notes: Wikimedia Futures Lab/Dashboard/Experiment Tracker/Twinkle AI: AI-supported feedback to newcomers | All | Albertoleoncio,
Laurentius | |
| Update Agent | A (customizable) AI agent capable of analyzing incoming data (open access journal data, newspapers, other unstructured data) that tries to find discrepancies to existing information in Wikipedia and Wikidata that are opportunities for updating, maintaining or improving information on Wikipedia | tbd | improved accuracy | Mathias Schindler | ||||
| Wikipedia Brand Visibility for Millennials | Increase Wikipedia visibility to bring more readers/contributors through marketing approaches targeted to millennials using digital media platforms and in-person activations. | 1.) Marketing template created for the movement by March 30th,
2) create example social media videos for AB testing in 5 languages - April 1-15, 3) launch campaign - April 15th, end of campaign - June 15th 4) evaluation and documentation - June 16th- June 30th. |
in preparation | Brand visibility | increased brand visibility for millennials | Wikipedias in three regions: Germany, Arab region, USA. | User:Pacita (WikiNYC), User:Alli Khalil, Julia Gebert (WMDE), User:Friederike vF(WMDE) | |
| WikiVoice | Community-based language digital archive for underresourced language communities to publish oral history. | started in 2025 | ongoing | Oral culture | better representation of oral culture in the Wikimedia projects | Wiki Voice Visuals | WikiVoice | User:Biyanto (ESEAP Hub)User:Psubhashish, User:Tochiprecious |
Experiment background
[edit]Owners: Eugene233, Carla (Soylacarli), Risker, Mayur-WMF, Eva Martin (WMDE)
Hypothesis:
[edit]We believe that if we create a central log of experiments with outcomes, lessons, etc then:
- We reduce risks of duplication
- Better sharing of lessons
- Faster evolution.
Brief explanation:
[edit]We propose a centralized experiment log where people can document their experiments so the community can track what worked, what didn’t, and access any supporting documentation they want to share. Entries will be organized by categories (e.g., education, community, technical, etc.)
What's in it for experimenters?
[edit]- Share your experiments with the wider community - wider profile for your work
- Find new collaborators
- Log your experiment so that others can carry it on
- Star trek jokes
Next tasks
[edit]- Confirm definition of an experiment. We propose it to be inclusive (e.g., edit filters, software, policy modification, activities funded by grants, etc, etc). This will increase awareness of how many experiments we actually do and how they do not have to be "big" to count.
- Identify existing repositories of information. Determine who will populate the archive with existing information. Is there a way to automate some of these processes?
- Who will be responsible for ongoing maintenance? Discuss with wikidata community what information they would be willing to include in available fields. Also discuss the possibility to work with metabase (wikibase instance).
- what to do with experiments that are not publicly viewable (e.g, we will soon have edit filters that are visible only to oversighters as they will flag certain information covered by the privacy policy (ANPDP). Some anti-vandalism experiments are not publicly visible in order to make them harder to "defeat". Possibly some kind of exclusion statement?
