Wikimedia Foundation Annual Plan/Quarterly check-ins/Audiences3 Notes October 2017

From Meta, a Wikimedia project coordination wiki

Present (in the office): Toby N, Sarah R., Amanda B, Anne G., Trevor B, Niharika K, James F., Heather W

participating remotely: Danny H, Ryan K, Katherine M, David B., David M., Caroline Sinders, Sydney Poore, Johan Jönsson, Jack


Community tech update

  • Top 10 wishes for the year, 5 completed, 2 on-going, will likely get 8 for the year.
  • [Santa dogs for the outcomes.]
  • Overall happy and getting happier
  • Global preferences is to have one place to set preferences across all wikis - set preference on one wiki and have it available across all; Example interface language.
  • Goal was to figure out a technical plan, based on Kunal's extension.
  • Held a community consultation to figure out how to handle local exceptions, and if we need them (we do). In the new quarter we'll figure that out for version two.
  • Edit summaries for non-latin languages.
  • non-latin characters use more unicode characters so their edit summaries were forced to be shorter. This will allow them to have the same length comments as latin characters
  • The grant metrics tool was one of the projects taken on to support smaller communities and programmes wishlist.
  • Helps Foundation grantees analyse the impact of their editathon/etc.
  • Working in liaison with Community Engagement department colleagues.
  • Will work on this next quarter.
  • ACTRIAL research
  • Landing page to encourage new editors to create articles in draft namespace.
  • Working with English Wikipedia new page patrollers. Analysing impact.
  • This quarter we'll be writing up findings, published by end of quarter, and having discussions about what to do based on those findings.
  • [TN] Long-standing problem. Goal was to make sure the trial was run in a way that analysis was valid and we seem to have reached that goal.
  • Big part is building trust and co-operation.
  • New wishlist in three weeks' time. 6–19 November for proposals. Voting 27 November-10 December.
  • New top-10 wishlist entirely based on community ideas and prioritised by community votes.


Anti-harassment tools

  • First goal was to measure admin confidence for first FY. Using admins as proxy for community leaders.
  • Over 100 responses in twelve days
  • 97.4% of admins feel confident with their skills to mitigate vandalism
  • 39.3% of admins feel confident with their skills to mitigate harassment
  • Used wiki hounding as one of the statistics, talked about what admins were feeling unsure about, what different problems are
  • [TN] Really telling – strong community confidence in working on vandalism, but not harassment. Big piece of work.
  • Yes, not afraid of these numbers, lots of room to improve, lots of directions we can go it, important to know where we're lacking.
  • Partnering with SuSA to better understand harassment, with a focused survey on forms of harassment and how people respond to it.
  • Methodology round how to measure expectations around disputes; how to create better policies etc.
  • Also made some Q1 work on improving existing tools, which continue into Q2. (Slides to follow.)
  • AbuseFilter runs for all edits for all users, lets the community craft rules for different situations and make decisions about allowing the edit through, and/or tagging them.
  • Mostly used for vandalism and spam.
  • Hypothesis that allowing more filters will let us help them craft filters for some forms of harassment.
  • Wanted to understand performance. Now measured, want to integrate data so filter authors can make decisisons based on it.
  • Interaction Timeline tool will help people understand timeline of user conduct.
  • Inspired by existing tool (Editor interaction timeline). To understand how editors interact with each other on similar pages. Used to analyse wikihounding. Visualize how forms of harassment look like, put all information on one place and is more visual. Prototypes tested with one editor. Will be testing with more admins.
  • Not designed to work with all forms of harassment (e.g. beyond three editors), but for certain kinds of harassment it works.
  • Estimating tickets this week.
  • Also working on more narrowly-targetted blocks, not from the whole site but just a page etc., scapel not a chainsaw.
  • Allow users to mute notifications from some users, and e-mails.
  • Worked with Global Collaboration folks.
  • Q2: Lots are continuation from Q1
  • Want a version of the interaction timeline out by the end of the calendar year, hope to have it out in November.
  • Big research project will be on blocking functionality, broader than page-specific blocking.
  • Built the ability to block individual user from email mute. Use case - A being harassed by B, B creates sockpuppets to send harassing emails to A. Something like B needs 10 edits before sending emails.
  • Continuing performance work on AbuseFilters.
  • ProxyBot - blocks open proxies. Anti-vandalism technique. Will decide whether to invest in it.
  • Working on internal team communication processes so we do our due dilligence.
  • [TN] Also having discussions on how we can talk outside the Wikimedia community more publicly about the work we're doing here.


New Readers

  • Improving access, offline readership etc.
  • Pageviews seem to be lagging or going down.
  • [TN] Maps to our experience of mobile users browsing less compared to desktop users.
  • Baseline understanding of awareness.
  • Video campagins in Iraq and Nigeria, strong collaboration between the Foundation and the communities. So important, crucial to make it successful.
  • Particular thanks to Dan Foy for the before and after phone surveys, crucial to understanding.
  • 27% of Nigerian users have heard of WP. Launched a video marketing campaign. For Nigerians, by Nigerians. Hired local celebs. 9.2 mn views in 3 weeks.
  • The Iraqi Wikipedians approached the Foundation asking for us to help increase awareness. Saw opportunity in collaboration with the community in Iraq and the mobile operator AsiaCell.
  • SMS blasts, PR campaign, videos. 5.1 mn views on those videos.
  • In Iraq, a 55% increase in awareness; metrics were set out in advance and evaluated.
  • Increase in mobile pageviews following SMS blasts. Not sure which activity contributed to the shift. Looking to get help from an analyst to understand the data and get a full picture.
  • Lot of great learning.
  • The need to build additional time into working on community collaboration is key – important to do, but takes time.
  • Using different types of metrics with different granularity to understand things in the right level of detail; adds complexity but helps understand, will probably need an analyst.
  • Looking forward, excited for collaboration between Community Resources and Global reach.
  • Inspire campaign- begins Jan. Low cost projects for local communities. Rapid grants. Special thanks to Kacie.
  • [TN] So much to like here, so much to learn from, thank you.
  • Access as one umbrella, rolling up Offline and Affordability. Will focus on the former.
  • Partially offline (often individuals) and fully offline (often institutions)
  • Partial offline space - Web and Android.
  • Web - allow readers to download mobile-formatted PDFs.
  • Delayed due to dependencies on OCG infrastructure.
  • Android - Reading lists and Enabling reading ZIM files in the app. Wanted to push it farther and let users download, not just view.
  • Got feedback that content packs already existing are not appropriate for this format.
  • Invested in research. Stepping back to make sure we address the issues brought up.
  • Paying down debt important but slows us down.
  • Having a dedicated PM for Android would have helped; looking forward to that happening next month.
  • [TN] Was talking to someone yesterday about MVP and this shows that we can deliver a product(?)
  • Offline for no Internet, generally an institutional use case
  • Long discovery phase for this.
  • Trying to understand the needs. Very complex. [Diagram of interdependencies between stakeholders]
  • Planning to develop and run strategic tests this quarter and next to help us decide.
  • Four current options we're looking at. May narrow as we get more understanding.
  • Learnings - One takeway is to narrow scope earlier and run tests against that.


Structured Data on Commons

  • Making Commons more useful by making it machine-readable.
  • Outcomes -
  • Making metadata easier, making search more effective, GLAM, License compliant reuse, support tool development
  • Building up scaffolding for program through creating a series of documents as an investment to create alignment around program.
  • Document on how stakeholders relate to changes we want to create and how we can help.
  • Communications plan - based on groups and stakeholder matrix. How to target each group. How to reach them.
  • Roadmap.
  • Risk tracking doc - breaks risk, cause, effect, and strategy for mitigating that. Has a lot of moving parts. External forces. Been really useful for realizing problems and letting us deal with that.
  • [TN] Historically we haven't put in work upfront like this and hopefully we'll do that more.
  • Goal for building structure data on Commons infrastructure
  • MCR goal was a stretch, we didn't hit it, will get there this quarter.
  • Making editing metadata easier. Jon K. and Nirzar created straw designs. Multimedia working on core piece of infra - MediaInfo extension to get Wikibase and Commons to talk to each other.
  • Really wonderful design discussion at Wikimania, Jon K. and Nirzar presented designs, lots of good constructive feedback.
  • People were enthusiastic about building the right thing rather than something that looks like what's been done before
  • JMo interviewing commons users to dig deep into their workflows.
  • Making search more effective - need to measure baseline metrics to understand improvements. Good for reporting in the grant and helping the funder know how to judge us.
  • One contributor lost 5000 files because they missed a square bracket while uploading. :(
  • GLAM contact doing great.
  • 10% of all Wikimania conference attendees came to one of the Structured Data on Commons sessions
  • Initial grant report coming soon.
  • [TN] Thanks everyone.