Wikimedia Foundation elections/2021/Post mortem

From Meta, a Wikimedia project coordination wiki
Jump to navigation Jump to search

This is the post mortem to the 2021 board election. Input is very much appreciated, and can be posted on the talk page or directly to this page. Sign with four tildes as usual.

General thoughts[edit]

  • There seems to be a certain fatigue in the pool of suitable candidates. For instance, we could and should find more women willing to stand as candidates. I suppose the recent rushed bylaws changes and the board expansion didn't help. Nemo 14:52, 1 September 2021 (UTC)
  • I agree to this. In addition, there are sometimes way too many things going at the same time. For example when this board election promotion process was going on, at the same time Grants Relaunch (and call for GRC), Strategy movement charter, ArbCom election and possibly some other things were going on. Possibly there is an "Request for comment/involvement" overload sometimes, which affects participation. Regards. -- টিটো দত্ত (কথা) 14:18, 5 September 2021 (UTC)
  • Hi Nemo, please explain how bylaws changes and board expansion would be reducing the number of women candidates, thank you. heather walls (talk) 02:41, 6 September 2021 (UTC)
  • Hi heather walls, I can imagine one way this might happen, because already privileged individuals will be better positioned to jump on a unique opportunity like 4 Board seats going up for election. But working against that bias, I would also expect the high number of open seats to improve diversity, especially when using STV through which several minority blocs can elect their choice of candidate. Apologies for not having a citation for these guesses—I wonder if Nemo_bis has something else in mind. Adamw (talk) 20:56, 6 September 2021 (UTC)
  • Hi Adamw! Which privilege do you mean? Unfortunately, "I would also expect the high number of open seats to improve diversity, especially when using STV through which several minority blocs can elect their choice of candidate." seeing the results I'd say that it's not particularly true. heather walls (talk) 21:41, 7 September 2021 (UTC)
  • I agree that a disproportionately male candidate pool is a sign that we have a problem, and want to emphasize "we could and should find more women" and additional genders but I can't make the math work out: if less than 10% of Wikipedia editors are female and non-binary,[1] then what kind of outreach is going to give us a balanced gender profile? This leads me back to the point Nemo_bis made elsewhere, that we should consider having some hard quotas for the WMF Board, for example no more than 50% can be of any one gender, country of citizenship, income quartile, and so on. Adamw (talk) 20:56, 6 September 2021 (UTC)
  • There could also be more and stronger support of changes (perhaps cultural) that would alter the representation in Wikipedia editors, as some Wikipedia editors (often women and non-binary) have been campaigning for, for years. We could also take research like that mentioned by Nemo about self-selection and think and reflect more deeply about why that might be and what culture supports and awards those behaviors. heather walls (talk) 21:41, 7 September 2021 (UTC)
  • Titodutta and Adamw make good points, but to clarify what I mean is that the recent bylaws changes were not particularly popular in the community (to use an euphemism), but the board decided to approve them anyway. This sends a signal that the board doesn't care, or (even worse) cares but isn't able to change the outcome. Such an impression is likely to result in a feeling that participation in elections has high costs and low rewards, which tends to exclude people for whom participation is harder. This is typical in the self-selection bias of candidates for job offerings too, on which there is an ample literature. Males on average have a (misplaced) higher confidence in themselves and are more likely to stand as candidates even in the face of negative odds. Nemo 19:14, 7 September 2021 (UTC)
  • How would the board not be able to change the outcome of its own rule changes? heather walls (talk) 21:41, 7 September 2021 (UTC)

References

Questions to candidates[edit]

  • Some of the questions would have benefited from a better copyediting, for clarity or precision. Two example:
    • in question 10 some project names were capitalized incorrectly (WikiSpecies, WikiNews): a very small thing, but annoying for a wikipedian!
    • question 9 was extremely vague ("What do you think could be done to improve combating bias?": what kind of bias?). - Laurentius (talk) 09:17, 5 September 2021 (UTC)
    • ITYM "annoying for a Wikipedian" ;-) Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 14:09, 5 September 2021 (UTC)
  • The election committee selected a limited number of questions (11), based on the feedback from previous candidates. One of the goal was to limit the workload for candidates. This makes sense, but some community members felt that it was too short. As a result, candidates were invited to answer all questions initially proposed (61), and at the end seven of us did. There are a few issues with this:
    • The goal of limiting the workload of candidates was mostly lost (although answers to these 61 questions were briefer and therefore faster to write).
    • These questions were not copyedited, and therefore there were duplicates, unclear questions, biased questions, etc. - Laurentius (talk) 09:17, 5 September 2021 (UTC)
      • Agreed - it would have been better to have a 'core' list of questions (the 11), then a curated list of 'extra' questions that could optionally be answered. Thanks. Mike Peel (talk) 21:19, 5 September 2021 (UTC)
        I agree. I was annoyed at some questions being somewhat redundant. The selection process could improve, maybe advertising it better and seeking more engagement in the preparation stage. Rely on the community to choose the bulk of questions, but edit them afterwards, clarifying, removing redundancies, etc. While I appreciated the effort of some candidates answering the unfiltered list, I don't think it's worth it. Reading answers to 61 questions from various candidates is way too much (and I'm sure answering all of them was even worse). MarioGom (talk) 22:12, 5 September 2021 (UTC)
  • The lack of relevant questions (and answers) discouraged me from thinking too hard about voting. Seemed rather pointless to me. Wbm1058 (talk) 21:45, 6 September 2021 (UTC)
  • A limited number of questions might have been OK but there was no transparency that I knew of about how/why questions were chosen. And so I had very little information about the issue I cared most about despite having my own question endorsed by a couple people and other similar questions proposed. Frankly I felt the questions chosen were written to be as innocuous as possible and that also denied me data on candidates as I wish, as a personality trait, to see how they respond to challenge. A virtue of selecting a number of questions should be the chance to provide a range of question types/styles without overwhelming the candidates. Best, Barkeep49 (talk) 15:27, 9 September 2021 (UTC)
  • The order of answers would ordinarily be randomized to avoid biasing, and the template is built to do that, but nobody remembered to add the qnum parameter this time, so randomizing didn't happen, and the answers were consistently in the same ordering. --Yair rand (talk) 05:27, 10 September 2021 (UTC)
  • There are at least four areas worth considering with the questions, but three jump out to me as warranting specific feedback: which questions were selected, how many questions were selected, and the transparency of the process.
  1. Question selection was immediately disputed upon the list being published. Two specific concerns arose: That 3 of the 4 most community-endorsed topics had not made the cut and that some "tough, but crucial" (my phrasing) questions were missed out while certain easy and/or vague questions made it in
  2. The question number was also felt to be too low. Candidates had the better part of two months if they started considering queries as soon as they were added to the list and many weeks even if they waited until all were added. Even if the list had to be thinned (beyond that of just removing duplicates), 20-25 would have been a perfectly viable number for any candidate who would have sufficient time to be a BOT member
  3. The Community was never notified that the questions would be culled in this fashion. Instead, the phrasing "collating" was used, which on a removal aspect, indicates no more than removing duplicates. This glaring omission in transparency is made more problematic as the issue arose in 2017 as well. Had the Community known that truncation of this sort would be utilised there would been vastly more discussion by us about which questions should make the cut. Although whether it would have helped is unclear - the second area of transparency concern is that we still are not aware of the system used for selecting questions by ElectCom. There must have been a method, as they didn't just pick the most endorsed questions, but we've not been told it. We need to know the system utilised plus the notes on each question, so we can discuss the pros and cons of it, and any changes for the next election. In a section below, I raise the distinct issue of non-communicativeness to these areas Nosebagbear (talk) 22:42, 13 September 2021 (UTC)

Statements and video statements[edit]

  • On the main election candidates page, candidates were listed not with text or table, like in the past, but with an image gallery. It looks nice, but I'd rather have text: we should not cast our votes based on pictures (and being influenced by the pictures is unavoidable). - Laurentius (talk) 10:02, 5 September 2021 (UTC)
  • Also on the candidates photo gallery: the order in which the candidates are presented may create a bias. For these reason, the picture order has been (pseudo)randomized. - Laurentius (talk) 10:02, 5 September 2021 (UTC)
  • Also on the candidate photo gallery: the size of the pictures may create a bias (bigger pictures gets more attention, therefore more clicks, and at the and more votes). As we didn't know of this gallery at the beginning, the size ratio of the pictures varies a lot, and at some point some of them appeared four times bigger than the others (the height is the same for all pictures, while the width changes). - Laurentius (talk) 10:02, 5 September 2021 (UTC)
  • We were asked by the facilitators to make video statements, something that was never done before and we didn't expect. This was welcomed by a few candidates, but others (including me) had concerns, especially considering that videos can easily reinforce biases (based on candidates looks like, their English proficiency, and even the quality of their webcam). The facilitators consequently decided to withdraw their request, and making it optional, but it was too late: if something is optional, every candidate (including me) will still try to do it, if possibile... - Laurentius (talk) 10:03, 5 September 2021 (UTC)
    Was recording the video in the mothertongue of the candidate offered as an option? It is probably relatively easy to subtitle it.--Ymblanter (talk) 17:18, 5 September 2021 (UTC)
    Yes, that was an option (but does not tackle any of my concerns). - Laurentius (talk) 18:20, 5 September 2021 (UTC)
  • Many parts of this election pushed candidates to show up in photos, video recordings, or live video meetings. While I understand that this has some benefits for some voters, I think we should not push people to do that. It is worth noting that one of the current board members never appears in video or in photo and does not want any picture of her around (out of security concerns), so this is not only a theorical concern. - Laurentius (talk) 10:03, 5 September 2021 (UTC)
    • +1, pushing photos is problematic. --Yair rand (talk) 05:29, 10 September 2021 (UTC)
  • +1 all the points above, I was uncomfortable with photos and video statements, and unsuccessfully tried to organize a poll of candidates to take the images down—in hindsight I went about this in the wrong way, I tried to amplify another candidate's question but this attempt fell flat. To reduce implicit bias, I would also prefer to avoid legal names as well, these often encode gender and nationality, it seems better to stick to usernames if possible. Adamw (talk) 17:53, 5 September 2021 (UTC)
  • Agree with photos. Videos could have more clearly been an 'extra' - unfortunately I wasn't able to record one due to health issues, but I'm not sure how much difference they made in the end anyway. Feedback from voters about whether they found the video recordings that some candidates did useful or not, would be good! Thanks. Mike Peel (talk) 21:21, 5 September 2021 (UTC)
  • I still don't get the focus on live and video meetings, when we are a community primarily focused on asynchronous textual communication. MarioGom (talk) 22:14, 5 September 2021 (UTC)
  • I completely agree with the above statements. These issues should be addressed in the next election. Gladamas (talk) 06:40, 8 September 2021 (UTC)

Meetings[edit]

  • In this elections live video meetings were introduced. The first ones were quite confusing and we had many issues, but the organization got better with time, and the last ones were much better both for candidates and for participants. If these meetings will be organized again next time, it is important to remember what we have learned this time. - Laurentius (talk) 12:24, 5 September 2021 (UTC)
  • Candidates presentations tends to take a lot of time. If you have 20 candidates, and each one speaks for 4 minutes, you can easily go over one hour and a half - which is definitely a lot of time. It is better not to have presentations, or at least give each candidate a short time - no more than one minute, possibly less. - Laurentius (talk) 12:22, 5 September 2021 (UTC)
  • Limit the time that each candidate has to answer questions. This can be either a maximum time for each question, or an overall allowance of speaking time that candidates can use as they see fit across all questions. - Laurentius (talk) 12:22, 5 September 2021 (UTC)
  • Facilitators should clearly state the time limits and enforce them. It may seem harsh, but even from the point of view of a candidate, knowing that you have, for instance, 2 minutes is much better than being told to generically be short and be considerate of the timings - because you know what you can do and you don't have to worry about what's fair. If candidates exceed the available time, they should be stopped. - Laurentius (talk) 12:22, 5 September 2021 (UTC)
  • Candidates should have an easy way to know how much time they have. One way to do this is to show a timer on screen; another is to periodically tell everyone how much time they have left. - Laurentius (talk) 12:22, 5 September 2021 (UTC)
  • As a candidate, I spent a lot of time on the video meetings and have the feeling that our discussions reached a very small number of people. The recordings were not always available to the public, and can make for exhausting viewing, live or recorded. Although the meetings were fun and taught me a lot about the other candidates, I hope we can find a more accessible format in the future. Adamw (talk) 17:53, 5 September 2021 (UTC)
    I had a similar impression. We spent a lot of time on video meetings compared to written interaction, although written text is more accessible for the voters (it may be more boring, but it is also more accessible). The meetings had a limited audience; some more people have watched the recordings, but they are long and watching the recording of a discussion is very boring... - Laurentius (talk) 18:26, 5 September 2021 (UTC)
    On the other hand, live meetings complement text-only conversations. Emotions were warm for example, which is hard to achieve in text, and I felt solidarity with other candidates' opinions, including many of their ideas about how to accomplish our shared goals. I liked that meetings alternated timezone and language, although English was heavily privileged. Adamw (talk) 20:28, 6 September 2021 (UTC)
  • I still have no idea whether the meetings were useful to voters or not (any voters want to comment on this?). They did only seem to reach a limited number of people, as others have already commented, and took up a lot of time. Having live translations at the meetings was really good, though, and this is something that should be kept (or effort redirected into other translation work around the election, as needed). Thanks. Mike Peel (talk) 21:23, 5 September 2021 (UTC)
  • What live video meetings? There were "meetings"? There were candidate debates?? Wbm1058 (talk) 21:23, 6 September 2021 (UTC)

Translation and multilingualism[edit]

  • Translations in Italian were better than in some past years, I saw some review going on. I've not checked comprehensiveness of translations. It must be appreciated that WMF invested in communication in various languages. The most important goal is always to have wide and good quality translation of all the most important pages, particularly the vote page and the emails to voters. Nemo 14:52, 1 September 2021 (UTC)
  • Coordination for translating pages was weak. I didn't know where I should go to translate the voting page. I knew that this could be done on vote.wikimedia.org, but my account was blocked there (why the heck accounts get blocked there?) When I asked to be unblocked, I was refused, but shown the landing page for translations. Someone else was quicker than me, but the quality of their translations was not so good. They even made some mistakes in translating variables or constants (those $ thingies). I was asked to fix/improve them, but they had already been imported to the vote.wikimedia.org. I fixed/improved all the translations, but my translations never made it to the voting page as far as I know. Besides, I never saw the voting page in Farsi, even when I went to the voting pages from Farsi projects! I simply couldn't see any Farsi text there. See Talk:Wikimedia_Foundation_elections/2021#Translate_voting_page for more details. 4nn1l2 (talk) 18:53, 1 September 2021 (UTC)
  • Two emails were sent to voters, on 20th and 27th of August. The second one got a lot more translations - something like 4 times more - because in that case it was asked to translate it (it was widely discussed on the telegram group, and probably elsewhere). As this is an email that goes to a very large number of people, having many translations is particularly good. - Laurentius (talk) 09:22, 5 September 2021 (UTC)

Diversity[edit]

  • The pool of candidates is less diverse than one could have hoped. After all, it reflects the decisions we've made in the past decade: since the November 2013 decision to further centralise power in the WMF's hands by reducing local democratic entities, we've basically stopped investing in local organising and democratic skills. As a logical consequence, most candidates come from a previous era where we were still investing in people. Nemo 14:52, 1 September 2021 (UTC)
    • I would suggest that we increase board diversity by further increasing diversity in the movement (and electorate) and by taking advantage of the appointed slots to help the board round out experiences, including with diversity. Best, Barkeep49 (talk) 15:33, 9 September 2021 (UTC)

Quotas[edit]

  • That said, I've always been and I remain a supporter of quotas. The reason is, as recent legislation about the board of directors of listed companies has amply proven, that increasing the chances of being (s)elected facilitates an increased participation (see also the semi-related Gender quotas and the crisis of the mediocre man: Theory and evidence from Sweden, there are many better ones). There could be a quota in the bylaws against having a majority of members with a single nationality (USA-centrism has always been the big problem of WMF, making it a most ineffective organisation) or gender. There can also be other milder methods such as preventing ballots from having consecutive votes for the same gender (so you couldn't put 5 men in the first 5 positions, for instance, but you'd need to choose 4 women to intersperse in the rank), but I'm not sure it has ever been tested with ballots allowing so many preferences. Nemo —Preceding undated comment added 14:52, 1 September 2021 (UTC).
    • In general it feels like this is best sorted with the appointed seats, which can fill in gaps not covered by the community elections after the elections have completed. And also generally persuading more diverse community people to stand in the elections (I know I didn't help with this! ;-) ). Thanks. Mike Peel (talk) 21:35, 5 September 2021 (UTC)
      • But how to persuade more people to stand as candidates? That's a problem whether there's a vote or not. Merely cherry-picking candidates to fill the gaps after an election reinforces the selection bias, while going around and fishing for candidates to appoint is hardly a more robust method as it's subject to all sorts of biases. Quotas are imperfect but they work. :) Nemo 21:09, 7 September 2021 (UTC)

Electoral system[edit]

  • The STV explanations were clear and the interface should have reduced confusion compared to the previous times. The dropdowns could have proved tedious, I'm not sure of the impact on the number of votes. Improvements are possible for the next round. Nemo 14:52, 1 September 2021 (UTC)
  • Electoral system is good. SecurePoll is not good. The design of the voting page can be misleading. The voter is encouraged to rank all the candidates in this particular design, but this is not the best strategy to employ. One should be able to hand-pick some candidates from the candidate pool and then rank them in his ballot. There shouldn't be any placeholders, so that the candidate is not encouraged to select any number of candidates. 4nn1l2 (talk) 19:06, 1 September 2021 (UTC)
  • Using STV is a big improvement over the previous system. I'm happy about this decision (finally!). - Laurentius (talk) 09:29, 5 September 2021 (UTC)
  • The instructions on the voting page where wrong. In particular, «if you add orange to your vote: You have more of a chance to end up wearing an orange shirt» is wrong. One thing should be made clear: when you vote, the likelihood of electing a high-ranked candidate is not influenced by lower-ranked or unranked candidates. This is at the core of this system: you shouldn't worry about strategic voting, and you should just rank your favourite candidate first, your second-favourite candidate second, etc. - Laurentius (talk) 09:29, 5 September 2021 (UTC)
    For what it's worth, single transferable vote is not immune to strategic voting. * Pppery * it has begun 01:01, 7 September 2021 (UTC)
    No electoral system is immune to strategic voting. However, STV makes it quite hard (unless you already know what other people vote - which is not usually the case). - Laurentius (talk) 06:16, 7 September 2021 (UTC)
    Yes, the "Voting Example and Best Practices" section said: Do not rank candidates you do not wish to seat on the Board. Following this advice is ineffectual at best and harmful at worst. See here for discussion. Tim Smith (talk) 23:50, 8 September 2021 (UTC)
  • Despite having participated in all Board elections since I think 2008 or so I had difficulties understanding what I should do exactly. In the end, I ranked the candidates I was prepared to support on top, the candidates I did not want to see on Board at the bottom, and other candidates (I have no opinion about) randomly in the middle. I do not think it was intended.--Ymblanter (talk) 17:26, 5 September 2021 (UTC)
    Actually, I think this is a reasonable way of ranking the candidates. You have recorded your preference, and filled the voids randomly. Not perfect, but it's ok, IMHO. - Laurentius (talk) 18:29, 5 September 2021 (UTC)
  • The specific STV implementation did not allow to give multiple candidates the same rank (ex-aequo). There has been some discussion about this choice in the telegram group and there are different opinions, but personally my opinion is that it would be better to allow for ex-aequos, because it sometimes matches how voters think their votes. For instance, in the comment above Ymblanter writes that he has randomly ranked the middle candidates - which I think is a reasonable approach with the current system, but it would have been better if he could have put all of them at the same level. That would reflect more accurately his opinion. I see a number of use cases for ex-aequos; the main drawback I can think of is that the user interface could become more complex (but I'm not a designer, maybe this is not a a real problem). - Laurentius (talk) 18:37, 5 September 2021 (UTC)
    Agreed, I also wanted to be able to enter same-rank choices. Like others, I ended up randomizing within a preference tier, which is a clumsy workaround. Tim Smith (talk) 00:18, 9 September 2021 (UTC)
    I'd like to further +1 this. STV is not a good system so long as I'm forced to just randomly drop in those I'm either neutral or uninformed on, into the middle, just to ensure that those candidates I'm opposed to don't rise up the list. It's fine as a method if, and only if equal-ranking is brought in by the next STV election. Nosebagbear (talk) 22:07, 13 September 2021 (UTC)
  • UX was far from ideal. A list of numbered dropdowns is really not the best type of control here. I had a couple of issues selecting repeated candidates, and then re-ordering candidates was cumbersome. I agree with Laurentius on same-rank choices. I just had 4 tiers: favorites, liked, neutral, disliked. Everything else within each tier was mostly random. MarioGom (talk) 22:19, 5 September 2021 (UTC)
  • Judging from the results, it would seem that STV was successful in increasing the diversity of voting and the truthfulness of the preferences. I base this not on the specific outcome (who was selected) but just on the number of votes for the top candidates. In 2017 and 2015, the most voted candidates were voted by some 40 % of the (valid) voters, vs. 30 % in 2011 and ~10 % now, despite the fact that in 2011 all top three candidates were incumbents (and in 2015 the field was quite broad). If the voting shares were a fair representation of the voters' awareness or preferences, we'd have to conclude that incumbency is no factor at all, and/or that the winners of the 2015 election run an extraordinarily effective campaign. The far easier conclusion is that people "overvoted" for supposed strategic reasons. In this election the most voted person (with some 20 % of the vote) received half of the preferences during redistribution, which presumably means that half of the persons which would previously have expressed a preference for them were able to distribute that preference first to others. Nemo 20:16, 7 September 2021 (UTC)

Results determination[edit]

  • Results have been published in a reasonable time. I also like the table of the computation very much: I think it's extremely helpful in explaining how people voted and how STV worked. If I remember correctly, moderator elections for StackExchange communities present the results in a similar format (produced by OpaVote, I think), so there is at least some experience in using it. Nemo 19:11, 7 September 2021 (UTC)
  • The format of the result table is very useful to see how close the result was. That in turn helps convince people that their vote might make a difference in future elections (assuming the board of trustees actually follows the result of the election, of course). See also the wikimedia-l discussion on how contended the election was or wasn't. (On the other hand I wish less superlatives were used by WMF. I have no idea in what sense this election would be the "most contended" ever: we have no comparable past elections, but the 2011 elections saw a difference of only 20 votes between the 3rd and 4th candidate in relation to the 1st, for instance.) Nemo 21:04, 7 September 2021 (UTC)

Information for voters[edit]

Voter participation[edit]

  • Compared to 2013 it seems we still didn't manage to get back to 2011 levels. This remains mostly the job of the WMF's board, to convince people that this kind of participation actually matters, and after so many years of accumulated distrust it's not going to be easy. However there might still be pockets of users here and there who could have voted anyway even in the current situation and didn't, for reasons to be identified. Nemo 14:52, 1 September 2021 (UTC)
    At en:WP:ACERFC2021#Voting system - STV, one can see some commenters saying that they chose not to vote for trustees because having to rank the candidates was too burdensome. While that wasn't my experience, I wonder how widespread a problem it is. —2d37 (talk) 09:30, 2 September 2021 (UTC)
    Ranking all candidates may indeed be difficult (for me it was - even though, as a candidate myself, I have a lot of information about other candidates!), but you don't need to. I think a large part of these concerns could be solved by clearly explaining that you do not need to rank all candidates (and giving correct explainations of the consequences), and possibly by allowing to give the same ranking to multiple candidates (in this way, you could even give a support/oppose-like vote). - Laurentius (talk) 09:41, 5 September 2021 (UTC)
    How can I possibly rank candidates that I'm not hardly familiar with at all? Wbm1058 (talk) 21:30, 6 September 2021 (UTC)
    @Laurentius: - I asked back when they were working on STV capability why they hadn't gone for using an STV mechanism that allowed ties. As it is, you can rank those you like and dislike, but everyone you don't know enough to evaluate has to just get plunked in the middle in random order to let you do that.
    The response I got was that they weren't aware that was an option, and by that point they were too committed. If we can get a guarantee that there will be a change to a vote-tie method by the next BOT election that would certainly be a positive. Nosebagbear (talk) 19:40, 7 September 2021 (UTC)
  • The WMF stated goal was to double participation, from approxiately 10% to 20%, and (as far as I know) for the first time some staff was allocated specifically to this effort. There has been indeed some growth in the participation, but by a mere 1%: one tenth of the stated goal. - Laurentius (talk) 12:34, 5 September 2021 (UTC)

Local engagement[edit]

Central engagement[edit]

Campaign by individual candidates[edit]

  • Nobody asked me to vote them. I still find that weird. :) I suppose in some communities it might have happened? Nemo 14:52, 1 September 2021 (UTC)
    I haven't participated in past trustee elections so I don't know how they normally work, but yes, when I saw "7 July – 3 August: Candidates campaign period" in Wikimedia Foundation elections/2021#Timeline, I rather thought candidates would be coming stumping at individual projects. Maybe some did somewhere, but I thought that the biggest city would get at least a whistle stop. On one hand, I suppose it's not needed: at least in that city, an election volunteer posted a notice of the election at a suitable page, and links lead from there to where one could ask the candidates questions and read their opinions. On another hand, editors might feel more comfortable asking questions in their own home wikis than here at Meta-Wiki. Ultimately, I lean towards thinking that actually 'campaigning' (and answering questions across numerous wikis) would be too burdensome for candidates. (On a fourth hand, I suppose it would allow a naturally-arising reward for more multilingual candidates.) —2d37 (talk) 09:00, 2 September 2021 (UTC)
  • As I translated those words I hesitated a wee while, as I don't think such a Candidates campaign has ever happened, it's after all not really a political campaign or such. There were notices about the election in most wikis, and some articles probably in the Kurier/Signpost/Kroeg/whatever suits your home wiki, but that's it. And that's what it should be. It should definitely not be anything in the big wikis, let alone the completely overrated enWP, first, if anything, rather in those wikis with not such good communication structures. Grüße vom Sänger ♫(Reden) 09:13, 2 September 2021 (UTC)
  • Having to do a 'campaign' puts a huge burden on candidates, which I'm not sure is good. Yes, being on the Board is a time-consuming job. But that doesn't mean that running for a seat should take up a lot of time - which if not elected, is simply a waste of time that could have been spent elsewhere. It could potentially also reduce the number of candidates willing to stand, if they know they'll have to put in a lot of time that could be wasted. I think it's much better to keep advertisements of the election centrally organised, so they can be done neutrally (fairly to all candidates) and efficiently. Thanks. Mike Peel (talk) 21:27, 5 September 2021 (UTC)

Email invites[edit]

  • Almost 50k invites were sent in less than 30 minutes. It would be nice to have some information on how many were actually delivered: we can't have exact numbers, but a range might be possible (subtracting permanent errors etc.). For the future I recommend consulting some delivery experts (such as our friends of PHPlist fame) well ahead of time, to work on a strategy to maximise delivery success. Nemo 14:52, 1 September 2021 (UTC)
  • I received an email as Wikidata user, although I'm a sysop on several wikis and not Wikidata. The edit count on Wikidata is easily inflated, so we might have ended up sending messages with "Wikidata" in the subject to people who barely remember what Wikidata even is. It would be interesting to see whether the participation varied in this group compared to others. Nemo 14:52, 1 September 2021 (UTC)
    I got invited with Commons as my home wiki (makes sense), but my bots got invited from Meta. Neither are exactly productive contributors here: 1 2. AntiCompositeNumber (talk) 21:40, 7 September 2021 (UTC)

Security[edit]

  • Information to voters wasn't very clear over who exactly has the keys to decrypt the ballots and how it was ensured that WMF staff does not have access to do so. Nemo 14:52, 1 September 2021 (UTC)
  • In the past we had issues with cookies and login, according to previous post-mortems. That might be important for relatively inactive users, and we had many changes in the login system since the last elections, so it would be worthwhile to collect some statistics from the logs and application metrics on how many login failures we had which were potentially related to attempts to vote. Nemo 15:30, 1 September 2021 (UTC)
  • After voting was complete, SecurePoll offered me a huge string of garbage characters. That's very much not useful for me personally, I think the only value is that we can save this text somewhere and it protects us from the WMF changing our vote without our knowledge. But these precautions are only good if we can actually check this later. I would have appreciated a human-readable printout of how I ranked the candidates, and a QR code for the robot language, which could be scanned by some future vote-checking tool. This would make it much easier to go back and change some of my rankings, for example. Adamw (talk) 17:53, 5 September 2021 (UTC)
  • An email confirmation of the vote, giving both the votes themselves and the hash, would be much more useful than displaying the hash on the page after voting. Thanks. Mike Peel (talk)
    Showing the votes in any way would go against the secrecy of the vote. Sending the hash by email should be ok, but it still needs to be shown on the page after voting, because not all users have an email address. - Laurentius (talk) 07:14, 6 September 2021 (UTC)
    @Laurentius: To clarify, I only meant sending an email to the voter themselves saying 'this is how your vote was recorded' - not to anyone else. Otherwise, all you can do as a voter is to take a screenshot of the form before you send it - not particularly user friendly! Thanks. Mike Peel (talk) 11:30, 6 September 2021 (UTC)
    @Mike Peel: The vote should not be sent even to the voter. It should be treated like a password; once it has been set, never show it in plaintext againg. - Laurentius (talk) 18:08, 6 September 2021 (UTC)
    @Laurentius: Why? Personally, I would want to see that my vote has been recorded correctly, and also be able to later check my vote against who was ultimately elected. I can't think of a reason why a record of the vote shouldn't be sent to the voter. Thanks. Mike Peel (talk) 18:54, 6 September 2021 (UTC)
    @Mike Peel: If you want to record your vote, you should write it down yourself: like a password, of course you want it to be recorded correctly, but (in a well-designed software) you will never get your password by email. If an email were sent, then the plaintext vote would pass through multiple mail servers, and stored in multiple places; threatening the secrecy of the vote. Moreover, it would provide you a (weak) proof of what you voted, which could be exibited to third parties - and that should not be possible. The cryptographic certificate, on the other hand, proves that your vote has not been tampered with, without leaking your actual preference. - Laurentius (talk) 20:06, 6 September 2021 (UTC)

Ballot dump[edit]

  • Will the ballot dump be released, as per custom, so that others can verify the result? Nemo 14:52, 1 September 2021 (UTC)

Eligibility[edit]

  • It is a bit weird that bot are (rightly) not allowed to vote according to the rules, but technically they can, and they received the email voting reminder. Of course, they should not vote, and if they do, those votes will be struck; but it would be better if they couldn't vote to begin with. - Laurentius (talk) 09:35, 5 September 2021 (UTC)
    I am not sure for this election any filters has been used for checking whether the account is a bot account or not before voting. I think it can be easily implemented. Since according to the Bot policy bot accounts will have the name "bot" (as well as the bot flag).- ❙❚❚❙❙ GnOeee ❚❙❚❙❙ 14:35, 5 September 2021 (UTC)
  • There was a point raised on Telegram about affiliate volunteers not being able to vote, if they hadn't made enough edits to the projects, even if they put a lot of time into Wikimedia offline. This seems important, and should be fixed for the next elections. Thanks. Mike Peel (talk) 21:30, 5 September 2021 (UTC)
  • All three of my bots got notices, and they voted for the same candidates I did (none of the above ;) Wbm1058 (talk) 21:34, 6 September 2021 (UTC)

Semi-private channels[edit]

  • The Telegram channel was problematic, in my opinion. There were some good questions raised, but a small number of voices dominated the discussion and the topics covered were jumbled together. Most of all, I was hoping to have an entirely public record of campaigning but instead we were trapped within a "walled garden" which could only be accessed by group members. Telegram tried to upload all my contacts to its servers when I joined, which is too high a price to pay for involvement (the workaround is to join from the desktop app and then link your mobile device, btw). It's fine to have private chats of course, but as a candidate I eventually decided to leave the channel. I would have preferred if we had publicly logged what happened there, and if we could have split out different rooms for each topic. Adamw (talk) 17:53, 5 September 2021 (UTC)
    • Generally agreed - the use of Telegram is bad. There are better free alternatives if we have to have off-wiki chat, e.g., Zulip (Q23680052). Thanks. Mike Peel (talk) 21:31, 5 September 2021 (UTC)
      This is the first time I hear about that Telegram group. MarioGom (talk) 22:20, 5 September 2021 (UTC)
  • Generally agreed. An unrefactorable push channel is not great for this purpose. –SJ talk  22:10, 6 September 2021 (UTC)
  • How many persons actually posted to the Telegram group? I see it says "289 members" right now. Is it possible to ask them to release their text under cc-by-sa, so it can be archived on the wiki? Nemo 20:38, 7 September 2021 (UTC)

WMF staff facilitation[edit]

  • This was something new, and I think this worked really well. The communications with candidates were generally good, and it felt like it was helpful and unbiased throughout the whole election. I think this is something that should be continued in future years. Thanks. Mike Peel (talk) 21:40, 5 September 2021 (UTC)

Election committee[edit]

  • There were several cases throughout the election where the election committee were not responsive / seemingly inactive. I know they do a lot of good work after the vote has closed, but it felt like they should be meeting more and responding to questions more during the election itself. For example, the question of campaigning during the election itself seemed to only get one 'I think it's OK' response from one committee member, rather than a proper response from the whole committee. Thanks. Mike Peel (talk) 21:40, 5 September 2021 (UTC)
  • [Duplicated by 2 requests from Wikimedia Foundation elections/2021/Post Analysis - pings converted to nopings to avoid dups] Immediately after the question selection, TomDotGov, plus numerous others, raised concerns about it, asking ElectCom to explain their choices, as well as discuss amending them etc. As far as I can see, none responded, even to confirm awareness of the issues.

2 weeks later, on the 16th July, I pinged each member of ElectCom and two of their permanent WMF advisors (as distinct from JKoerner, who was a specific BOT-election facilitator, and communicative throughout, within the areas they felt it appropriate for WMF staff to comment on). On the 17th July, another editor dropped a note on ElectCom's talk page.

Further communication attempts were made, including another request on their talk page to respond.

None of these would gather a response. Discussion with JKoerner indicated she could understand our irritation with non-responsiveness, and beyond the methods utilised by myself and others, couldn't offer a logical other communication attempt to try prior to conclusion of the election I unfortunately do not have a new solution for you. The only suggestions I have for reaching out are ways you have already reached out

Ultimately, on the 21st August I tried reaching out through the Community Affairs and Board Governance Board subcommittees. With this I garnered an impressively quick initial response by Shani (WMF), and some well-informed discussion, but was neither able to resolve the issues nor, as of this post, confirm reasoning for the non-responsiveness.

Given the 9 week timescale, I'm not sure what the lack of response, from some well-experienced editors on Electcom, was due to. Obviously in the close-run period there would be on-going demands on their time, but not full-time when the concerns were first raised and especially to the tune of no member even noting the presence of major concerns. This, to me, is significantly beyond what is reasonable. Nosebagbear (talk) 19:31, 7 September 2021 (UTC)

  • As the editor who attempted to bring the issue to ElectCom's attention on their talk page, I endorse this message. The idea that a sitting board member is more responsive to this concern than any of a group of (wonderful) volunteers is both great (how lucky we are to have Shani on the board) and concerning (a community body should be more accessible to the community than the board in theory). Best, Barkeep49 (talk) 15:44, 9 September 2021 (UTC)
  • Hi @Qgil-WMF and JKoerner (WMF):. Per @Shani (WMF):'s most recent comment on the BGC talk page it was suggested I should ping the two of you. I know that Jackie is aware of the above (with the possible exception of Shani's most recent comment at Talk:BGC, which I haven't added to my initial post), although Quim may not be. I would like to stress that this issue is now constituting two main facets: why did (and could) this happen...and what do we do to avoid the issues in the future? Any washup report produced that only handles one, would be a significant failure.
We can't come up with a good method to resolve issues without knowing the full reasoning of the failure, but even if we could, either a egregious communication failure by ElectCom occurred, or there was a good reason/mitigation that has not been provided to the Community. That should not be gone past in a "only look forwards" methodology. Likewise, a set of reasons might make us feel both better and more informed, but must be accompanied by concrete proposals to avoid reoccurrence. I, Proc, the WMF, or the majority of experienced editors who raised concerns no doubt could all propose potential solutions, but without confirmation of what caused this failure, are likely to be either overblown or insufficient. Nosebagbear (talk) 16:29, 13 September 2021 (UTC)
Hi @Nosebagbear, yes, we are working on a proper response this week together with the Elections Committee. The task of producing these questions was distributed, with some bits done by the facilitation team and some by the EC and this is why producing the answers needs the collaboration between both teams. I'm sorry that this is taking so long, and I am happy to take the blame. We were very busy working on other aspects of the election and the EC alone wasn't in a good position to respond because, as said, my team has been involved as well. We all share your interest in clarifying what happened and also what should happen in the next Board election in relation to questions to the candidates. Qgil-WMF (talk) 10:24, 14 September 2021 (UTC)

Voting page design[edit]

Currently, SecurePoll shows a long list of drop-down boxes. Skipping a preference or voting for one candidate twice is possible, but eventually results in an error condition. It would be better to have one list in which the voter chooses the candidates, and deletes those that are completely unwanted. This would look much nicer, and would prevent the errors. --Amir E. Aharoni (talk) 08:10, 8 September 2021 (UTC)

Voting internationalization[edit]

Internationalization is quite flaky. The SecurePoll extension was completely translated to Hebrew before the vote began, but I couldn't actually use the site in Hebrew, even though I came from the Hebrew Wikipedia. I saw some Hebrew text for the election description, but the rest of the user interface was in English. Right-to-left directionality was broken, too. To avoid this, internationalization must be fully complete and tested at least two weeks before the vote begins. so that updates in extension translations and bug fixes can be made on time. (Of course, it's better to give more than two weeks.) --Amir E. Aharoni (talk) 08:10, 8 September 2021 (UTC)

General internationalization issues for translators[edit]

The system for submitting the election description texts is extremely non-robust and non-scalable. It allows only a tiny number people to submit translations to the production site. This could, in theory, be justified by the need for quality control and vandalism prevention, but I suspect that the people who copy the translations from Meta to the SecurePoll site don't actually check every line and just copy everything. It's better to build a more direct system for submitting translations. It's a wiki! :) --Amir E. Aharoni (talk) 08:10, 8 September 2021 (UTC)

A complete list of materials to translate[edit]

I'd love to give the people who speak my language an opportunity to experience the whole voting process in that language, without having to fall back to English at any point, and I'm willing to volunteer my time for that. However, it was quite uncomfortable to find all the texts that need translation. As the time went by, I kept finding more and more texts that need translation. Here are the things I know about:

... And maybe there are some more that I forgot. I compiled this list myself. I'd expect to find such a list on a translation coördination page, and I'd really expect it to be complete :) --Amir E. Aharoni (talk) 08:10, 8 September 2021 (UTC)

Emails language[edit]

How is the language for the emails chosen? I was told that it's chosen according to the wiki where I have the most edits, and it's a reasonable default, but perhaps more options should be given:

  • Add a preference somewhere to allow people to choose the language of such global emails
  • Add links to translated versions to the email body --Amir E. Aharoni (talk) 08:10, 8 September 2021 (UTC)

Home wiki[edit]

In the list of voters, my username appears with the en.wikipedia.org domain. This possibly happens because en.wikipedia.org is my "home wiki" according to the CentralAuth database. However, this is a pretty internal and technical thing: While I read and edit the English Wikipedia quite a lot and even have the sysop rights there, if I have to choose my home wiki, it would be the Hebrew Wikipedia, where I have more edits. I also clicked the banner in the Hebrew Wikipedia before I got to the voting page. So I'd expect the Hebrew Wikipedia to appear in the voters list.

This may be important, for example, for statistics about voter turnout by wiki. Will I be counted as a voter from the English Wikipedia, the Hebrew Wikipedia, or both? It's actually fine if it's both, but if I have to choose one, I'd prefer to be counted as a voter from Hebrew. --Amir E. Aharoni (talk) 08:10, 8 September 2021 (UTC)

If it helps, I'm counted in the Tagalog Wiktionary's stats as that's considered my "home" wiki (and in elections I'm the only editor who'd be counted in those stats), even though that's not the wiki I'm most active in by any stretch of the imagination. That said, I think it would be great if we can get numbers as to who voted from which wiki, or who got to the voting page from which wiki. I imagine these are numbers that we can pull up, right? This would be more accurate in terms of determining which projects turned out which voters, rather than relying on CentralAuth given the immutability of one's "home" wiki with the implementation of SUL. --Sky Harbor (talk) 12:51, 8 September 2021 (UTC)
It is the wiki where the account was first registered. Ad Huikeshoven (talk) 07:44, 10 September 2021 (UTC)

Preview image[edit]

Pages about elections sometimes include photos of several candidates. When such pages are shared on Facebook, Telegram, etc., the link preview may show the photo of a particular candidate. This is misleading because the user may think that it's a page about a particular candidate. This also gives this candidate more weight and promotion.

The preview image for such pages must be neutral. --Amir E. Aharoni (talk) 08:10, 8 September 2021 (UTC)

Unfortunately, Facebook generates a preview image basing on the first share of a page. Pundit (talk) 14:18, 8 September 2021 (UTC)
A workaround would be to include, say, the Wikimedia logo (or other neutral image) in the page as well - so when you share it, the logo is shown in the preview rather than a candidate. Thanks. Mike Peel (talk) 09:13, 9 September 2021 (UTC)

Photo sequence[edit]

By default, when several photos of candidates appear on a page, their sequence should probably be randomized. It may also be nice to add a button to snow them in alphabetical order. --Amir E. Aharoni (talk) 08:10, 8 September 2021 (UTC)

@Amire80: They should already be randomised (they weren't at the start of the election process, but were during the elections). Thanks. Mike Peel (talk) 11:41, 8 September 2021 (UTC)

Export/import[edit]

When I order the candidates in some order, I'd like to be able to export that order and then be able to import it after some time. Then I'd like to be able to keep editing the order.

Thanks to that:

  • I would be able to use a saved order after relogin.
  • I would be able to use a saved order when voting another time (overwriting my old vote).

Grillofrances (talk) 00:27, 13 September 2021 (UTC)

Drag & Drop[edit]

I'd like to have a Drag & Drop instead of selects/dropdowns so I could easily move a given candidate up or down. Grillofrances (talk) 00:35, 13 September 2021 (UTC)