Commission des affaires communautaires de la fondation Wikimedia/2022-02-23 Conversation avec les administrateurs de la fondation.

Recording of the Conversation with the Trustees on 23 February

Le comité des affaires communautaires - un comité du conseil d'administration de la fondation Wikimédia - organisera la prochaine conversation avec les administrateurs le 23 février de 19:00 à 20:30 UTC. Cette conversation est l'occasion pour les membres de la communauté de s'entretenir directement avec les administrateurs au sujet de leur travail. Le conseil d'administration est un organe bénévole composé de dirigeants du mouvement et d'experts externes chargés de guider la Fondation Wikimédia et d'assurer sa responsabilité.

Comment participer ?

Cette conversation aura lieu via Zoom avec un streaming live sur YouTube. La durée de l'appel sera de 90 minutes et comportera des mises à jour ainsi que des questions et réponses ouvertes et une conversation.

The meeting was recorded and is available via the same YouTube link. It will also be uploaded to Commons and posted here in the days following the event.

Soumettre vos questions

Participants asked live questions and submitted them ahead of time to askcacwikimedia.org.

Interprétation de langue simultanée

Il y aura également une interprétation pour d'autres langues s'il y a cinq membres de la communauté ou plus qui sont intéressés. Pour solliciter une interprétation, adressez un courriel à $1. Vous pouvez le faire jusqu'à 5 jours avant la réunion pour nous permettre de prendre les dispositions nécessaires.

Agenda

• Board updates: celebrations, projects, milestones
• Knowledge as a Service under this year's Annual Plan and its continuation and evolution in next year's Annual Plan
• Leadership Development Working Group

Notes

Board updates

Work on projects the Board is overseeing

• Universal Code of Conduct:
  • The UCoC reached an important milestone with the vote on the Enforcement Guidelines this month.
  • UCoC was created as a baseline of acceptable behavior and an outline of positive, constructive behavior. The Enforcement Guidelines are an explanation of how the code will be enforced.
  • 3097 voters from 146 Wikimedia communities participated in the vote, and 76% voted in favor of the Guidelines.
  • The Board will be reviewing community feedback from the process, then make a decision about ratification during our Board meeting
• Fundraising on English Wikipedia:
  • Banner fundraising campaign on English Wikipedia in December: following RfC, the fundraising team started a process to write banner language together with community members.
  • Unfortunately, during first few days banners resulted in 70% less revenue than on the corresponding days in the prior year. The performance of the banners during this campaign led to the Foundation raising $24.7M, less than the original $30M goal.
  • That campaign ended up with a shortfall of just over $5 million below target revenue.
• Affiliates strategy
  • Working with Affcom and others to develop our first affiliation strategy. Will help us be more strategic about the work the Foundation does in collaboration with affiliates.
• Sister projects task force
  • There has never been an organized process on how to propose a sister project. CAC has this in its charter.
  • Looking at the full lifecycle of how sister projects come to be, are approved and sunset.
  • Process will include volunteers

Foundation’s Annual Plan this year: Knowledge as a Service

• Commitment to knowledge as a service is a big part of this year’s annual plan.
• Small updates on: New Page Patrol work, Commons, Vector 2022, Community Wishlist
• Wikifunctions/Abstract Wikipedia:
  • Wikifunctions will be a new Wikimedia project, goal to provide a compendium of functions. Functions answer a certain form of questions not usually answered by Wikipedia. Knowledge in Wikifunctions can answer complex questions: does the habitat of two given species overlap, does the size of geoshape on Commons match what is stated in Wikidata, create a graph of the population of a city on the fly.
  • Functions are a genuine new Knowledge Format. Functions can be called directly and integrated into third party apps, and thus make it easier to embed knowledge in many applications out there.
  • The Wikifunctions Beta was launched last year, and you can try it out online. You will see that there are still a few rough edges, but the main functionality is there.
  • One main goal of Wikifunctions is to provide functions to generate natural language text (Abstract Wikipedia). So we can take the basic data from Wikidata about a city and use that to create an article about the city in many different languages. This way we want to fill in some of the gaps that we have across the different language editions of Wikipedia, and allow the communities to really focus on the topics they care about most.
  • Also, it is really important for us to make sure that everyone can contribute to Abstract Wikipedia in their own language. The goal is to work on a common baseline of knowledge, that is created once, maintained once, but available across all of our languages.

Q&A

Is Wikifunctions a ChatGPT simulation but for Wikimedia?

• Similar in that it answers questions, but that’s about it
• Wikifunctions is predictable and editable, contributable by communities
  • Parameters of questions are narrower than ChatGPT, but with reliable and trustworthy answers

How are Wikifunctions different from info tables?

• Building a function in wikifunctions means you can benefit from it in all our projects. It is not limited to a particular wiki. A central place where code is being maintained, doesn’t need to be copy/pasted from wiki to wiki.
• Example: Wikidata has a birthday of a person. You want to display age of person, probably you’ve copied a function from another wiki and no one is really maintaining the original function. Easier to maintain original function and use across projects.

Is there a more detailed timeline for Abstract Wikipedia than what is on Meta?

• More updated timeline is coming soon and will be available on the timeline page on Meta. There is also a newsletter people can follow.

Knowledge as a Service under Next Year’s Annual Plan

• Initial draft plan on Meta for review and feedback.
• The Foundation's annual plan this year will be led by the needs of product & technology
• These are built around three buckets of work. Idea is that responsibility and focus areas are really clear, so that things happening in one bucket won’t block the work from another. Three buckets:
  • Wiki Experiences
  • Signals and Services
  • Future Audiences
• Throughout April Maryana and Selena will be visiting community calls and likely hosting a few of our own to get your thoughts on this draft.

Q&A

Do you already have anything about maps in the bucket for Wiki Experiences?

• In the past, Foundation would take mostly completed roadmaps and offer to community for feedback. Instead now we’re sharing our highest level priorities as early as possible. Then we’ll move to talking about objectives and key results, which is a level above specific project plans.
• Creating key results is challenging and this is the first time we’re doing that across Product and Tech together.
• We are trying to release as early as possible at every step. Mid to late March will likely be a point where we get more project specific and key results specific.

Will there be drafts from departments other than Product and Tech for this year’s Annual Plan?

• We are starting from the P&T framework. Other departments are building on top of that. What you will see is additive. Starting with P&T focus and add layers to that.
• This is the first year that any department has released info and asked for a conversation this early in a long time. This is the first step, the foundation to build on. Each department will have its own process.

Leadership Development Working Group

• Group of 15 community members from different backgrounds, representing each region of the Wikimedia movement. Our group members hold different roles, from campaign organizers and skills trainers to on-wiki functionaries and affiliate leaders.
• Group formed in June 2022 as a response to Movement Strategy’s “Invest in Skills and Leadership Development” recommendation
• Set out to accomplish 3 goals in our 1-year term:
  • 1 – Draft a shared definition of leadership.
  • 2 – Create a Leadership Development Plan.
  • 3 – Guide the implementation of the Leadership Development Plan.
• First goal of the year has been accomplished
• Right now, we are busy working on the leadership development plan, which will be a resource kit for people who want to develop leadership in themselves or others
• Four focus areas:
  • Understanding leadership roles and skills
  • How to build a leadership development program
  • Specific aspects of leadership (e.g. burnout, diversity, passing on knowledge)
  • Showcasing leadership initiatives
• Hope to have a draft for review in the coming months.
• To follow the group’s work:
  • Meta page where you can find the leadership definition as well as monthly progress updates.
  • MS Forum
  • Courriel : /leadershipworkinggroupwikimedia.org

Q&A

What kind of leadership are you referring to? Affiliate leadership or unaffiliated leadership? Is there a regional focus?

• Work highlights the collective with respect to individual efforts. Conduct surveys and calls to get input from different regions.
• Affiliates are a critical part of the movement, but always room for individual leadership for new volunteers for example. We look at those two types with even further, different subtypes. We aim to include everyone. We will gladly take feedback if there is any form of leadership we have overlooked.

How do you plan to share your outcomes with all communities?

• Work will continue to be published on Meta. We also share on regional lists and on the MS Forum. Visit community calls and conferences.

Open Q&A

Have you started reviewing the content of the Form 990 and when do you plan to publish the Form 990 for the last fiscal year?

• Form 990 is the statutory filing the Foundation does to the IRS of the United States as a charitable org within the US. Required in order to stay in compliance as a charitable org. Finance team is working on it right now. It involves a lot of US legal compliance. Intention is to release as early as we can.
• The 990 has generated a lot of community questions. Knowing that community is interested, the Board has requested more community engagement around the form this year. We want the community to understand it better and how the Foundation is operating, be as transparent as we can.
• Additional information, added by the Finance Department after the call:
  • The form 990 is based on the Foundation's audited financial statements, which are typically finalized in October.
  • After that, it requires data and collaborative work across the Wikimedia Foundation, including teams in Legal, Advancement, payroll, finance, and grants administration to compile the remaining information.
  • It then goes through several reviews in multiple departments for accuracy and completeness. Given that the form 990 typically runs around 50 pages, this process takes considerable time.
  • The form is made public each year by the required submission date to the IRS in mid May. The Finance Department works to improve and expedite the process as much as possible every year, but because its publication is reliant on a number of moving parts, it is often hard to publish very far in advance.

Now that the UCoC Enforcement Guidelines have seen approval from the movement, what can the trustees say to help give minoritised editors and editors in minoritised content areas confidence that we have the Trustees' support in countering systemic bias, that our contributions are valued and recognised?

• Board has overseen UCoC process since its inception. Establishing an equitable UCoC is how we show we as a movement value retaining minoritized editors
• The Enforcement Guidelines received approval from the movement but have not yet been ratified
• In March Board will review the all the feedback from the vote and make recommendations about next steps. Once EGs are ratified, Board will work with project team on training modules to get people aligned on what is and isn’t acceptable behavior on the platform and a manual for how to respond.
• UCoC and the Board’s involvement stemmed from an older resolution that the Board took to have a healthier space on all wikis. UCoC is one aspect of having safer communities. Also resource allocation toward safety measures, trust and safety work that happens in the background, work with community functionaries. The vision is wide and the Board recognizes that a lot needs to happen in order for the movement to see a cultural change.

Is the machine learning content creation in the proposed annual plan in reference to ChatGPT?

• Languages team has been doing work for a long time around translation. Certain communities like Indic Language Wikipedia communities have been wanting to work with the Foundation toward that end. This is in reference to models to find topic areas for coverage by machine translation. As communities request support for content generation experiments, this is about supporting those requests. Toward the end of March we’ll have more specific project proposals around this.

Why don’t Foundation fundraising campaigns keep running until the goal is met? That’s what my local radio station is doing now.

• The more time the banners are up, the less impact they have. People don’t interact with them as much. There are diminishing returns, there is a short time limit to work with to make sure the banners are as effective as they can be.
• This time we took a few more days because of how the holidays fell on the calendar.
• Banners are not the only revenue stream, the team is diversifying the way we are fundraising.
• Additional information provided by the Fundraising Team after the call:
  • The fundraising campaign did run for a longer period of time and at higher banner levels in 2022 to recuperate revenue from the decline in banner performance.
  • The 2022 campaign ran for 6 days longer and readers were shown nearly 50% more banner impressions than in 2021, but we raised $10M less than in 2021. While increasing the length of the campaign did help recuperate some revenue, it was not able to make up for the full reduction in revenue from running lower performing messages.
  • We've found that typically the first few banners a reader sees are the most important. The donation rate declines by showing readers more and more banners while the disruption to the reading experience increases.
  • The team balances raising revenue with the impact of campaign length on our community or readers, donors, and volunteers. Over the years, the team has worked to limit the disruption to readers and volunteers by making the campaigns more efficient, reducing the length of banner campaigns and limiting the number of banners readers see.
  • The 2022 campaign was a change from that trend. In trying to balance all of these factors, the team used the final end of year fundraising push to end the campaign.

Is the Foundation planning any layoffs in light of what is happening in the tech industry generally?

• Wikimedia Foundation has tech elements but is not a tech company
• We will have more information by April on our future financial projections. There is a revenue gap in the current financial year. We anticipate we’ll have a reduced budget.
• Through annual planning we’ll gain more insight into what we’ll be working on and how many staff are allotted to which workstreams.

For Jimmy: What are your early thoughts and questions about how ChatGPT intersects with Wikimedia?

• Any Wikipedian who has played around with it realizes it’s not a threat to Wikipedia in its current form. It gets things wrong and makes up sources. Bad actors creating hoax entries may create things that sound more plausible now, but our editors are pretty good at spotting these.
• What might be more interesting are the possibilities for us, like the tool that can describe any random image of the internet in great detail. This could be a great bot someone in our community could run. Longer term possibilities=giving AI 10,000 articles on related topics and find contradictions. That might eventually be possible with this technology.
• New external trend we’re exploring this year in the process of generating our annual plan: how machine learning and ai are impacting how people search for content and the type of content that’s available online. We will be hosting a call to get input on that.