What are the criteria for determining which wikis get Jade? Assuming we eventually (as the amount of time approaches infinity) deploy to all wikis, how quickly do we do this? Who gets to be first? Second?
Description
Status | Subtype | Assigned | Task | ||
---|---|---|---|---|---|
Declined | None | T212435 Review real-world query plans and performance for Jade | |||
Declined | None | T238877 Write Huggle labels to Jade | |||
Declined | calbon | T183381 Deploy pilot of Jade to a small set of wikis. | |||
Declined | None | T199520 Determine which wikis will get Jade and when |
Event Timeline
We're in a position to think about pilot wikis, again. Per IRC, it would be great to have buy-in from interested wiki communities, so that we might be able to deploy in late November.
I put together a spreadsheet comparing various wikis according to what extensions were installed and what third-party services were available to the users of those wikis. Spreadsheet is available here: https://docs.google.com/spreadsheets/d/1WH-srlQulQMT_5BHq4BTfiYoQ1k4mZqTarjFe08V4zw/edit#gid=0. Let me know if there's something that should be explained.
My logic is roughly as follows:
(1) I want to prioritize the wikis that have the ORES extension installed. One of the major motivations for JADE is to allow counter-narratives to ORES, so it makes sense to prioritize the wikis that are actively using ORES. (In principle a wiki does not have to use ORES to make use of JADE, but it's more useful if the two are used together.)
(2) Further, I want to prioritize wikis that make use of counter-vandalism tools such as Huggle, Twinkle, and the edit patrol flag. This suggests a community that is large enough to make use of curation tools.
(3) I am using "revert rate" as a proxy for the general existence of conflict and/or vandalism activity. A wiki where 10%+ of edits are being reverted is one that engages in frequent micro-conflicts that could be described/articulated using JADE.
(4) I am using monthly active editors and admins as a general heuristic of relative project activity.
(5) With the help of @awight I got some data about usage of the major anti-vandalism tools Huggle and Twinkle to get a sense of how much usage a given tool has in that community.
(6) English Wikipedia, German Wikipedia, French Wikipedia, and Wikidata are not being considered at this time due to scalability concerns.
(7) Finally, I want a mix of small projects (but not too small) and medium size projects, with some diversity.
Based on these factors, I am exploring these wikis as the initial places where we launch JADE: Arabic Wikipedia, Catalan Wikipedia, Persian Wikipedia, and Finnish Wikipedia. Additionally, if we can work with the Huggle developers to include JADE views, I would like to additionally explore Portuguese Wikipedia and Russian Wikipedia as pilots. All of this depends on those communities approving the pilot.