運動憲章/起草委員會/經驗傳承

點此投票

運動憲章

補充文件

專案

社群諮詢

起草委員會

批准投票

常見問題

本報告總結了關於組成運動憲章起草委員會過程的「反饋、經驗傳承，以及未決問題」。這些經驗傳承旨在作為維基媒體運動未來治理流程應記取的建議和考慮因素。

概要

經驗傳承：

• 所有利害關係者都同意，在這一流程中以及由此產生的候選人名單中，需要更強大、更廣泛、更深入的多樣性、公平性和包容性。
• 需要清晰度、透明度和時間來促進利害關係者的公平參與。
• 一個選舉委員會或中立的單位將促進未來的流程。
• 每個利害相關者群體在如何內部選擇候選人上有透明度為佳。
• 使用的技術必須完善。

出現的問題：

• 如何處理過程中的變更？
• 每個利害關係者群體（項目、自治體、基金會）是否應該使用相同的流程來選擇/提名/任命候選人？不同的利害關係者應該有不同的流程嗎？
• 誰負責確保流程的公平性？

Introduction

The Movement Charter Drafting Committee, a group trusted and tasked with drafting a major governance document proposed in the Movement Strategy Recommendations, was formed on 1 November 2021. The process of forming the Committee involved lengthy discussion and planning. Eventually, the 15 seats of the Committee were identified in a mixed process that included Projects elections, Affiliates' selection, and a Wikimedia Foundation appointment. This process provides a potential model for future activities that seek to improve equity in participation and the way we govern our Movement.

This report summarizes feedback about this process. The feedback comes from a feedback page on Meta, a video conference discussion with the Community, and internal reflection sessions in the Foundation's Movement Strategy & Governance team - altogether engaging roughly 60 community members and staff.

These Lessons Learned are meant as tips and considerations to keep in mind for future governance processes in the Wikimedia movement, and they are based on concrete recommendations collected in the feedback. There are also some questions arising that require further thinking - careful and pluralistic consideration of those questions will help future processes learn from and improve upon this one.

Lessons Learned: Considerations for Integration in Future Processes

• All stakeholders agree that there is a need for stronger, wider, deeper diversity, equity, and inclusion in the process and in the resulting slate(s) of candidates
• Clarity about the process is crucial
  • Clear communication about the plan and changes to it is essential
  • Ensuring clarity and transparency about the process and in communications requires time
• Clear mandates for various stakeholders involved in the process are necessary
  • A neutral body - such as an Election Committee - mandated to make decisions about the process of Selection & Election was suggested to help resolve tensions, disagreements, real and/or perceived trust gaps, and other frictions
• Transparency on how each stakeholder group internally selects candidates is desired (e.g., WMF, Affiliates)
  • The appointment of Affiliate Seats through a regionally-representative committee of selectors was accepted in principle; although it may further staff support to implement
    • Risk: Wikimedia Foundation staff support to Affiliates may be (mis)construed as meddling
• Stakeholders agree on areas for improvements to the selection and presentation of candidates
  • Clear and transparent eligibility and expertise requirements are communicated in advance
  • Presentation of candidates is clear, concise and harmonized to help voters more equitably access the statements and compare candidates to each other more easily
• The mechanics of the Election (including the translation, compass, STV methodology and SecurePoll) must be secured and smoothed
  • Resources (hard and soft) must be allocated to each of those components to ensure a successful Election process

Questions Arising for Further Reflection / Clarification

• Course correction during a process can be disruptive, even if/when that change aims to correct the imperfection in the process itself. Is it better to stick to a plan while we are in it, even if we discover it could be improved? Or is it better to make changes to improve the plan, even if it means this may cause disruption or confusion?
• Should each stakeholder group (Projects, Affiliates, Foundation) use the same process for selecting/nominating/appointing candidates? Should different stakeholders have different processes? Would different processes emphasize an arbitrary separation between the overlapping groups within the Movement?
• Would safeguarding the commitment to equity fall under the mandate of the proposed neutral body (election committee)? Who is responsible for establishing expectations, standards, and exigencies for achieving equity?
• How might the lessons learned from this process inform other processes such as UCoC ratification, establishment of the (Interim) Global Council, Charter adoption/ratification, Board of Trustees election processes, collaborations between the Global Council and the Board in the future, etc.? Will one approach work for different kinds of processes?
• How can a process apply more candidacy criteria (e.g. minimum edits or account history) without creating barriers to participation that are not equitable?
• How can data be used to provide an accessible assessment of candidates (e.g. visualized) without  misrepresenting qualitative expertise?

Preparation

Planning Process

After the January Global Conversations, there were multiple proposals to help move forward with the new governance structures for the Wikimedia movement. These proposals got “stuck in limbo” due to “lack of facilitation”. Because the design of the process was “dragged on for such a long time”, there was also a “steep fallout” in participation. To move things forward, the Wikimedia Foundation presented a compromise proposal that was discussed during online events on 27 June 2021 and, subsequently, revised per feedback.

The proposal turned out to be successful. It did not directly acknowledge the many community proposals that preceded it, though, nor did it reach a “level of dialogue” thus frustrating some of those who led such proposals. There was also no explicit or clear mention of how the proposal skipped the previous idea of creating an Interim Global Council.

During the elections and selection, several aspects of the process changed (mentioned in their relevant sections). The majority of these changes were met with objections and “raised some serious questions about the validity of the process”. These issues, though, may have been avoided if there was a neutral group making decisions in the process. There was, overall, a lack of clarity for who had the “mandate of making decisions”. While some liked the “multiple processes for stakeholders”, others thought it furthered the arbitrary “division between Affiliates and communities”.

Recommendation

• In reference to “best practices from previous movement-wide elections”, appoint an “election committee made up of community members” to run the process.
• Process changes need to be considered carefully on a case-by-case basis. Clear communication about any changes is imperative.

Call for Candidates

The call for candidates was sufficiently “clear” and “low bar” to attract a very diverse range of nominees. Due to a number of reasons, probably including the wide outreach, two week extension, compensation and the lack of almost any candidacy criteria, the number reached a record of 72 eligible candidates. This “was not necessarily a good thing”, due to the difficulties of voting and selection from an “overwhelming number of candidates” on the slate.

While there was opposition to extending the call for nominations by two weeks, it is common in Wikimedia elections for nominations to peak just before the deadline, thus making the effect of this extension less certain. It was also perceived that the lack of diversity in candidates at first may have encouraged the huge flux towards the end.

Additionally, the process of screening candidates for Trust & Safety was not thorough, as some of the criteria (i.e. project blocking) cannot be fully-checked automatically. This led to a situation where a candidate was initially informed to be eligible, but in fact was not. Because this candidate was selected “through a democratic process” to represent an Affiliate, they were substituted in an “extraordinary [procedure]”.

Recommendations

• Evaluate the extent to which voters were able to use the STV system with a large number of candidates.
• Visualize the table or data of candidates to make it more easy to “digest and interpret”, such as the statements’ visualization that was created by a volunteer candidate.
• When any candidate criteria cannot be fully-verified by the organizers, the responsibility should be delegated to candidates themselves to “declare any/all compliance with the guidelines of the election”.

Skills Assessment

Candidates were asked, before the elections, to self-assess themselves by selecting their top three skills from a skills and expertise matrix. The way that this matrix was applied, though, may not have had much use for neither the elections nor the selection processes. As multiple voters and selectors expressed, “ticking a box” for a certain area of expertise does not give sufficient information.

Expertise greatly varies not only in its length in years, but also in the individual and personal experience associated with it, which can only be assessed qualitatively. Because diversity, on the other hand, is more quantifiable, it was often easier to evaluate candidates by their diversity instead of expertise, thus leading to “tokenism and underappreciation of experience”.

Recommendations

• A greater emphasis is needed on the competence sought in candidates: Even when diversity is a priority factor in selection, it should not be considered in isolation from expertise, or it would become pure tokenism.
• When nominating themselves, candidates should either be asked to already provide expertise within their statements, or should be clearly alerted of any upcoming requests for information.
• Assess skills with more detailed qualitative and quantitative criteria. For example, candidates can be asked about the number of years of experience in each skill, or can be given the chance to elaborate on their specific strengths and achievements in each area of expertise.

Wikimedia Foundation-Appointed Seats

There were several complaints about the Wikimedia Foundation’s decision to appoint two members of the MCDC before the official elections, something that constituted a change to the originally proposed process. This was objected to as a part of “constant changes and variations” in the set up. The planned process, however, would have led to the appointed Foundation’s staff participating in the community elections and Affiliate appointment processes, thus coming into a “direct competition” with volunteers, which would not have been helpful to the nominated WMF staff members nor to volunteers. Some community members also asked why the Foundation granted its two seats to staff members, instead of using them for added diversity.

Recommendations

• The Foundation should clarify its process ahead of time. For example, there should be clear expectations regarding not only who appoints seats, but to whom should the appointer (in this case, the Foundation) grant their seats.
• Delay the Foundation’s appointment process until after the elections, so that it may contribute to appointing two people who "balance" the diversity aspects, similarly to the affiliate-selection process.
• Transparently disclose who in the Wikimedia Foundation is making the appointment (as in the selection process of the Affiliates), and based on what criteria are they making this appointment.

Affiliates-Selected Seats

Regional Approach

The Affiliate selection track followed the relatively-unique process that was used when convening the Movement Strategy Design Group, which planned the “transition phase” between the publication of the Recommendations and implementation. In that process, Affiliates were distributed into a number of regions that each appointed a “representative”. In this case, however, the chosen “representatives” were a group who would meet together and reach a common decision, based on certain criteria, to appoint 6 seats of the Movement Charter Drafting Committee. This was an unprecedented approach.

Geographically-focused Affiliates were allocated into eight major regions following mostly the existing collaboration structures (like CEE, ESEAP, etc.). Thematic Affiliates were allocated into their own “thematic region”. Affiliates could switch regions if preferred. There was still some dissatisfaction about the distribution, especially in the cases of: The Caribbean and Central Asia (two unique regions grouped with Latin America and East & Central Europe, respectively) and all the Thematic Affiliates, who had never worked as a group before.

The process of convening Affiliates in each region was perceived as difficult and “confusing”. Multiple Affiliates felt they lacked the “mandate”, because of the short notice about the design, which was a part of the agreed proposal for establishing the Drafting Committee. Many also mentioned late outreach, ineffective contact methods, and being “late to the party”.

Additionally, the selection process needed more guidance and support from the Movement Strategy & Governance Team, in order to avoid placing “a lot of burden” on the Affiliates. However, in spite of its many difficulties, the process did encourage regions to “make a collective decision” in a way that has never been done before, and, more importantly, to “move forward the idea of a ‘hub’”, which may involve much more of this regional collaboration and decision-making in the future.

Recommendations

• Create and widely advertise a long-term regional distribution list where Affiliates can choose their regions across all upcoming Movement Strategy processes.
• Consider creating a specific-mailing list for announcements that require Affiliate contacts engagement or action.
• Announce the process at least 6 weeks in advance and allow Affiliates to provide feedback, and provide them with a more clearly-defined selection process (e.g. with steps A, B, C).
• Offer immediate support by the Movement Strategy Team to reduce the workload on Affiliates.

Selection Sequence

In the original Wikimedia Foundation proposal, it was stated that: “This [Affiliate] ranking is done while the election runs, and is published together with the election results”. During the elections, there was a renewed discussion on Meta about whether it should take place before or after the elections results are announced. Eventually, the selection period was extended until after the results of the elections were unofficially announced.

There were many community objections that this delay took place on a discussion held “with selectors and not others”, that this was a “MASSIVE” and a “drastic change to the system”, that it “upweight the opinions of the Affiliate selectors” and that it would have affected the decision-making of many voters. The counter-argument was that: without knowing the election results, selectors would not be able to “do their job properly “ in ensuring diversity. The Affiliates seats ended up selecting committee members from five geographic regions (ESEAP, MENA, Sub-Saharan Africa, South Asia, and Latin America), of which four were missing in the election outcomes. According to a volunteer, “this is the fourth election” where all of these four regions are missed; therefore, “it is a pattern”.

Recommendation

• Explore “best practices for fair sector representation with regards to geographical, language, cultural or stakeholder group”.

Selection Process

A group of nine selectors, representing different geographic and thematic regions, met and discussed for three weeks to agree on six of the Drafting Committee seats. The selectors referred to the predefined diversity and expertise criteria and screened the candidates against it. The formal expertise criteria, however, may have not been very helpful, and the selection committee did come up with more detailed aspects of expertise to evaluate, such as the candidates’ experience on major Wikimedia committees, in outside organizations, or in sister Wikimedia projects.

Overall, assessing expertise was difficult, due to its qualitative nature. Many of the selection committee members took initiative to consult their regions’ Affiliates about the selection criteria, but there was no consistency across the different regions in prioritizing diversity over expertise or vice versa. Eventually, the selectors group was able to make their decisions by common consent.

Recommendations

• Offer more support to regional representatives in communicating and convening with their regions to consider their recommended selection criteria.
• Seek wider agreement about how to apply diversity and expertise criteria in selection processes.

Elections

Translations & Outreach

Candidate statements amounted to over 30,000 words, the equivalent of about 120 pages in average print. In only a few weeks, this “huge amount [of text”] was fully translated into a dozen languages: a major success. Some translations, though, have been under-utilized or unused. Also, the massive translation workload affected the Movement Strategy facilitators' ability to communicate with communities as they were focusing on translation first before outreach.

The timelines and set-up process for the elections of the Board of Trustees and for the Movement Charter Drafting Committee were linked. The main reason was logistical, because one team was organizing both elections, and another was to offer non-elected Board candidates the opportunity to run for the Movement Charter Drafting Committee instead. This caused several timing issues, including both frustration from the Board election results and the feeling of exhaustion and tightness from communities. Several communities were unmotivated to participate in yet another election.

There was mixed feedback about the outreach. Some perceived that it “lacked reminders” and was “less enthusiastic” than Board elections, some appreciated the Movement Strategy & Governance messaging, and some communities complained of excessive announcements. It was also unclear whether “candidates could do a campaign”.

Overall, the participation in the elections met the Movement Strategy & Governance Team’s projected goal of 1,000 voters.

As a side topic, the Anywhere on Earth (AoE) time zone was a source of confusion. It was criticized for “not being well known”, having a “Wikipedia article only in four languages” and being “more complicated than one might think”.

Recommendations

• Follow a clear facilitation plan in the translations and outreach process and “stick to it”. There should be a specific period of time dedicated for translations, so that they do not interfere with outreach.
• Start translations earlier, during the call for nominations, to save time.
• Assess the “value for money” of translations into minor languages and whether translation is “[needed] for all languages”. Future elections should have more clarity about which languages are eligible for translation funding.
• Better communicate the impact and value of these elections for the future of the Wikimedia Movement, and give them their due space in the movement calendar.
• Use only the UTC standard time zone instead of the Anywhere on Earth (AoE).

Elections Compass

The elections included an unprecedented experiment with the election compass (also called “voting advice application”), a questionnaire that helps identify voters with the candidates who share their views. To create this, the community was invited to propose statements towards which candidates would position themselves. The community suggested 111 statements, and were then asked to “upvote” (adding their signature) to select the top twenty. However, not all voting community members may have entirely understood the upvoting process, as many seemed to vote for statements they were approving, instead of upvoting statements that would have been helpful to learn more about the candidates.

A separate homepage was set-up on Toolforge, adapting the open source tool “OpenElectionCompass”. Besides the English and German version of the interface, staff members as well as volunteers provided additional interface translations. In the end, the tool was available in English, German, Spanish, Portuguese, Italian, Indonesian, Polish, French, Turkish, Hungarian, and Arabic.

The final result was “praised wildly”, “drew voters in”, and even “kickstarted interest even beyond the tool” in creating a spin-off website. It did come a little late, though, creating an unexpected load on the candidates. Also, there has not been a recognition that many communities may not have the “assumed knowledge” about the tool’s concept, which is not known everywhere.

Recommendations

• Provide more time to create statements for such a tool, including stronger facilitation and better explanation about how such statements should be written.
• Announce the election compass together with the call for candidates.
• Dedicate specific outreach and/or support to help introduce the concept of the “election compass” to communities.
• Keep collecting the compass statements through an open on-wiki process; it was appreciated and helpful.

Elections Method

This was the second major election, in the history of Wikimedia, to use the single transferable vote (STV) method, in which voters are asked to rank candidates according to preference. Although the methodology was useful in adding diversity to the election results, such as a majority in elected women, it was quite complex for voters to understand and use in ranking roughly 70 candidates. Several community members criticized the lack of a “negative vote” and a “neutral vote” equivalent in STV, which cannot be made up for by “randomly ranking neutral candidates”. Eliminating negative votes was seen to be a “huge problem “that “turns a strong ‘no’ into a negligible ‘yes’”.

The rules included a limit that does not allow electing more than two candidates from a single Wikimedia project. The rule did not end up affecting the results. While this was agreed to be a “fair rule” in theory; in practice, it has flaws. If implemented, it would have eliminated a candidate from South Asia, and it may have affected candidates from other lingua franca-dependent underrepresented regions, like Africa or the Caribbean. Alternatively, there were several suggestions for diversity based on the “geographical region”.

A number of stewards were asked to be scrutineers during the elections following a suggestion from the community.

Recommendations

• Prepare more material and communications (e.g. infographics) to explain how the STV method works and how voters should interact with it.
• Consider a two-step process where the community can pre-filter or negatively cast votes for candidates they deem unsuitable for STV ranking.
• Group the candidates into "slates" according to their background: Voters can be encouraged to focus on one candidate from each slate, for example.
• Consider election rules (e.g. electorates) that are focused on geographical regions, instead of languages, in order to ensure diversity.
• Place a threshold for the number of candidates above which we have to move from STV to a different method.
• Ensure and arrange scrutinizing of elections.

SecurePoll & Results

The voting software in the elections was SecurePoll, the “tool of choice for major Wikimedia community elections” since 2009. The tool offers great advantages including community trust, wiki account log-in, eligibility checks, and data privacy. For multiple reasons, though, including the implementation of STV in these elections, SecurePoll “proved a very difficult tool to use”. Developing this tool would be a “VERY significant investment in the free and open source software tools that facilitate collaboration”.

There were multiple issues in both the interface and the script. For example, the interface and user experience were both challenging, ranking the 70 candidates proved to be difficult, and the technical setup would not have allowed implementing some elections rules including both the two candidates/project rule and the alternate election (see: Ranking task for developpers). As a community member mentioned: “Before we try to effect change, we need to build the tools that enable it”.

Recommendations

• Ensure the availability of technical and “engineering support” before elections on WMF’s side. It is “not possible to run an election this large without this support”.
• Consider alternative tools for future elections, especially if they are based on the STV method. If any alternative is to be selected, it should be publicly suggested and tested beforehand to avoid affecting “people's trust”.
• Create a public task checklist (e.g. on Phabricator). If we share the details of the elections beforehand, developers can have clear expectations about the potential technical needs.
• If SecurePoll is to be reused, some of the technical issues are: inability to rank all candidates, lack of flexibility in applying specific election rules, and the impossibility of editing election rules halfway through.
• If SecurePoll is to be reused, there are several user interface issues to fix including: visual interface options (e.g. drag and drop), saving voter choices and editing them, and a confirmation message before submission.

Sources

• Report of the live feedback meeting (13 November).
• Meta pages, including: official feedback page, elections talk page and candidate nominations talk page.
• Internal Movement Strategy and Governance Team reflections, with four sessions (each dedicated for a specific topic): Set up & nominations, affiliate selection, elections setup and elections communications.

Notes & references