Talk:Board elections/2011/Post mortem/Lessons

From Meta, a Wikimedia project coordination wiki

Lessons Learned from 2011 Election[edit]

2011 -vs- future, A 'diff' of ways we can do better next time.

Some[edit]

(by Alecmconroy)

Explanatory philosophy:

  • You can never have too many candidates. Candidates are people stepping up to volunteer to do a huge amount of work for us. Many good candidates may not be known to the WM community, indeed, good candidates may not even be editors of our projects. If we get enough candidates that we have lots and lots of obviously bad candidates, we can have run offs, we can have multi-round voting, or we can just have increased voter participation.
  • You can never have too much civil, good faith discussion.
  • You can never have too many voters.
  • You can never have too much help. Election coordinators, assistants, clerks, and translators are essential to a global election.
  • You can never have too much work left to do
  • Voting is evil. Discussions have their downsides, but straight voting without discussion reaches the worst decisions of all methods. Sometimes we do need a vote, this is one of them-- but we should make it as discussion-like as possible.
  • There is no deadline. Indeed, make sure there aren't any hard deadlines too nearby. Given the complexity of a truly global election, unforeseen delays are, in practice, quite predictable.

Before everything, diagnose what went right and what didn't in the 2011 elections.[edit]

fundamental or systemic issues
  • Disinterest and apathy toward the board/foundation, despite great loyalty to their local project.
  • Foundation as mere "non-profit ISP" to a project.
  • No pre-existing coherent massively-global community, no coherent traditions of how to run a massively-global election.
  • No substantial discussion at global-community level
probably specific to this case
  • not enough coordinators, and there can never be enough translators
  • translation delays, no pauses or extensions
  • notification emails were sent late
  • candidate statements requirement too short
  • question translation may not have taken place

Pre-pre election: starting right now[edit]

  • Designate a central location for election process discussion. (have per-language subpages)
  • Designate a central location for talking about the candidates in general. (again, per-language subpages)
  • Designate a central location for discussing individual candidates-- split first by candidate, then by language.
  • Use a private mailing list so that people can email the entirety of the election committee easily.
  • Designate a central on-wiki location for communicating with the elections committee, with per-language subpages as needed.

Building the global community--- Community- and Public- relations[edit]

  • Absolutely must convey a strong sense of who the board is and what the board does.
  • During the year we need chances to "meet" the current elected board members as individuals as much as possible (and the unelected ones too, of course). I know many want to our board to be like a traditional non-profit board-- one that presents a "unified front" to the community. But right now, there's such uniformity that the board is positively reflective and thus, invisible to the global community. Board members, especially community- & chapter- chosen members, need to draw contrasts between each other during the year. We do this some, but should do it more.
  • During the year, perhaps the board should periodically 'refer/defer' some judgments to vote or consensus of the global community-- let each board member that wants to summarize their thoughts.
  • Except for elections, the global community has no power. Because it never has had any power, the global community doesn't really exist, as a coherent body. And because it doesn't exist, it continues to have no power. Autonomy, however limited, is what gets people engaged. Slowly give the global community more and more to decide, and more and more people will show up to do the deciding.
  • Vote every June on something. If there are no seats to fill, find something else to vote on.
  • Consider giving the global community some sort of "advise and consent" role for the unelected board members. A sufficiently strong consensus of the community can 'veto' an appointment. This margin can be arbitrarily large-- 2/3rds perhaps to veto. Or 4/5ths. The point is to engage the global community and get them to invest in the foundation's future the way they are invested in their projects future. Existing board members should feel free to advocate for the approval of the community to a given expert. It doesn't matter if they ever veto an appointment, the point is that it'd be good to ask the community to get involved in all aspects of its leadership.
  • Election banners and text should be accompanied by a "countdown" to let readers know how long the current phase of election is scheduled to take.

Envision Future Recruiting[edit]

  • EnWP's arbcom election had approximately 30 people listed as helping in some capacity. In the 2011 board elections, our candidates outnumbered our coordinators.
  • In the future, we will need a much larger group of people to help coordinate an election across such diverse populations.
  • Each project should appoint its own "local election coordinators" to help ensure an orderly process.
  • The movement as a whole needs a large staff of election coordinators to interface between the english hub and the 750 projects.
  • Many phases of the election will require large amount of translation resources, requiring us to pre-recruit a network of translators capable of fast, high-volume, high-quality translation.

Tackle the language barriers[edit]

Many of the difficulties encountered in this election stem from how many different languages we now contain in our community.

Think this is a challenge? This is nothing. Some day, once technology spreads to the entire world, more of our 'viewers' won't be able to read, write, or type. There will come a day when we'll have to do text-to-speech and vice-versa for every language in our movement. In the event that we invent Star Trek's Universal Translator or find a Babelfish before the next election, disregard this section. :)
I'm also intentionally not investigating the full extent of our current translation endeavors. It's easier for a mind to list all ideas, rather than all ideas not already in use. Stifles creativity. Brainstorm independently first.
Investigate our current language skills
  • We need to determine what words, if any, have become "global" to the point that anyone who participates in a WM project can reasonably be expected to understand them. "Wikimedia" for example, might be such a word. I've 'heard' the word "OK" was, at some point, the most widely understood 'global word'. We need to develop a "Wikimedian Lexicon" of words that can be used within the projects with a reasonable expectation that it somewhere can be understood.
  • Translate the Wikimedian lexicon into as many languages as possible, have it prominently displayed in all projects so that roughly all admins (or similar 'serious editors') would have a chance to learn it.
  • Lexicon need not consist only of English Words, and there would be some huge advantages if it turns out some of our most widely understood words were not English.



Muster an army-- "Translators for World Peace"
  • We should take steps to actively recruit a standing peace-army of translators dedicated to promoting intercommunication among the projects. I know we already have a lot of people helping out in this way-- they just did a LOT of work for us. But we need infinitely more. (additionally, such a translators-for-world-peace internet group would be useful to other internet organizations besides just ours). I have no idea how you do outreach to "people willing to be translators" but we need them.
Improve translation-related UIs.
  • Improve UI for "Request Translation" Requesting a translation needs to be well advertised and a true "UI improvement focus". Everyone on everyone single project should instantly know how to get prompt help with a translation any time. Do we have a bot or utility that take a chunk of text and parses it up in a reasonable way, preparing it for translation? It looks like we're doing the translation set up manually with the help of experts, which requires templating skills again. (not a problem for most of us reading this, but for a translator in the global south-- learning templates is a huge barrier to them focusing on what they do best-- translating.
  • If we could make the translation UI nicer, especially for elections-related translation, that could help-- right know, you have to know wikicode, templates, and the relevant two languages-- that's a lot to ask for in a single person. IF we can make the UI very very simple, so that anyone who knows basic internet navigation and the languages can instantly contribute.
Screen where a user can save what languages they speak, to which fluency.
Henceforth, user can automatically see translations they can help in, priority and topic ranked.
Clicking on a translation shows very very simple form: Source text, field for translation-- no template, no wikitext if possible, just source words.

An all-inclusive roadmap for translation[edit]

We need to create a detailed plan for how to BEST communicate from any one project to any other.

Instant but dumb
a machine translation network.
  • Which languages don't have access to machine translation from English to their language. Machine translation is the 'lowest quality, but also instant and easy". Which languages don't have access to a machine language translation of a "core language"?
  • What percentage of the community doesn't have instant machine translation access in ANY language they speak? That number and demographics need to be identified so we can make a
  • For those who can get a machine translation, what is the quality of the translation of the 2011 election Q&A? That has our most nuanced discussion?
  • Make sure instant and once-click-accessable translations of all english pages are available. Page length was a problem this year, but maybe we can work out a way to have our pages fully translated during the election somehow.
  • Pre-planning should make sure each language decide on or or more 'best' machine-translation strategies-- directly from English, Semi-direct from a core language, or indirectly by translating a non-core translation. Incorporate those recommendations into official election documents for a language.
  • Analysis needs to reveal "isolated islands' of languagess that will have trouble getting translation. Election coordinators & foundation staff should help these languages find a viable machine translation solution if it exists.
Perfect but slow-- building an infrastructure for "The Synaptic Switchboard".
  • Wikimedia is the first organization I can think of that needs near-real-time translation for every single language actively spoken on the planet.
  • This special need could be met by creating an informal network of human translators who, combined with machine translation, could allow a member of any project to send a message to every other member of ever other project. That is, we must build a system of human and software that is capable of letting the global community 'talk to itself' in near-real time.
  • We need to actively assess the language fluency of each and every editor. If we could seriously ask optionally at some point after registration which languages each editor speaks, that'd be very helpful. We need to get a sense of how interconnected different language-communities are. This is how to get the language data. Multilingual editors will edit in multiple languages and thus will have global accounts on projects in multiple languages. Analyze project interconnectedness.

Question for the nerds:

  • We have 86,912 active users on Wikimedia projects. Of those, I am interested in identifying the "demonstrably multilingual" users.
  • Let's define a "demonstrably multilingual" user to be one who is simultaneously active on at least two projects of different languages. Meta and Common perhaps should not count, as they are inherently multilingual. I want to take 86,000 users and filter that set down to users down to just "demonstrably multilingual" users.
  • Now, each such user has a set of projects they are active on, and thus a set of languages we can infer they are able to speak. We use this date to answer the following question:
  • For each non-english-language project, how many of their active users are ALSO active on an english-language project? (the answer is should be a single whole number for each project)
  • For any two projects, how many users are there who are active on both? (answer is a triangular matrix)
  • For any two languages, how many users appear to speak both languages? (answer is a triangular matrix)
  • How do we build a tool so that users with translation needs can automatically connect with online users who can meet those needs? (the answer isn't math <grin> ).
  • Based on that, choose a 'hub' language-- the 'working' but not 'official' language of Wikimedia. Currently our 'hub language' is English, and I suspect it will continue to be upon analyzing the interconnections. But can't hurt to actually check.
  • Pick a set of "core languages" that must be translated before moving to a next step. If needed, pause the election until translation occurs. These should be the 'election-committee-provided' 'high-bandwidth-traffic' languages, languages guaranteed by the election committee to be available at each step in the process. Core translations should be best quality translations-- (ideally candidate-translated to the extent the candidates can).
  • It is the responsibility of the community to actually find a connection from the "Core Languages" to a language they speak. It's the responsibility of everybody to make sure it's easy for them to find that connection, and we all need to make sure the whole 'system' works.
  • Analyze project interconnectedness to screen for "island communities"-- those with the weakest connection to the hub language, where many steps are required to connect it to a "core language". Election committee should focus on making sure those projects have extra help.
  • Create a basic translation roadmap for each language-- what's the best way to for them to get information from En to their language? optimize roadmaps for "fast but narrow-band translation" and a roadmap for "high-volume translations" that will need lots of translators working in parallel. In pre-planning, use global account edit data to find people with requisite language skills.
  • Test the system beforehand. Have a 'trial run' where non-election-related texts are translated back and forth from projects. See if everyone can actually talk to each other with any coherence. That is, for an obscure language, can its members rapidly find a path to ask a question, get an answer, ask a followup, get a reply, and have the conversation be both timely and meaningful?


  • Each project needs community members responsible for making sure their project members are aware of election, understand the process, not having problems, able to get their translation and discussion needs met, etc.

Pre-Election[edit]

  • Recruit large numbers of coordinators, including many of those involved in the Dec 2010 EnWiki Arbcom elections, which had good turnout.
  • Assemble the translation network, test it.
  • Announce schedule well in advance. Give everyone an idea of what's coming and when.
  • Plan that if things go wrong, we'll pause. In particular, if translation to core languages is slow, we'll call a time out and get all core languages translated before proceeding to the next phase.
  • Make sure local projects have coordinators in place to alert us if their project isn't getting needed translation.
  • Consider having "civility coaches" to do things to encourage civil discussion, avoid personal attacks. Don't outright remove merely uncivil remarks of course, that would just make it worse, but do ask people to zealously focus on the positives will coolly and regrettably sharing the negatives. Either on-wiki or off-wiki, some people will ignore these encouragements, as it should be-- but most people will heed a clear, logical, candidate-neutral call for how to have a 'civil election discourse'. Those who are unnecessarily uncivil will merely serve as a cue that their words should be discounted.

Sparking Candidacies & Discussion[edit]

  • Release document "About being a board member" discussing in clear, concrete terms what the duties of a board member are. Let readers be able to realistically see what life as board member is like, such that they can try to 'see themselves' in the job. Use document to actively recruit potential candidates from within the community.
  • Simultaneously, release a document "About the Wikimedia Foundation", discussing in clear, concrete terms what WMF is, what its staff do. Talk about what the board does." Use this document to recruit community members into the election process.
  • Potentially release a document "Questions for the global community" where the uninvolved-board , foundation, and individual community members ask questions of the global community. Use this to spark discussion about WMF and its future direction. What are some things where the board could use input from the community.
  • Consider putting a few "issue" questions on the ballot just to drive up turn-out. Encourage rationales on the answers.
  • Encourage a 'nomination' phase where you actively ask the global community who it looks to as a leader / trustworthy? Enlist the community to recruit candidates. Begin discussion about who SHOULD run, who'd be a good candidate, and what the next chapter of WMF is like.

Questions[edit]

  • Collect a "question base" of optional questions candidates can optionally choose to answer immediately upon deciding to become a candidate. Questions for the question base could carry over from year to year, be submitted at any time. Candidates must be under no pressure try to complete them all, so include lots of especially-skippable questions to make sure candidates don't feel the need to treat the whole thing as a giaaaaant questionairre. The point is for candidates who have time to discuss to start discussing. This will generate, in turn, people discussing during the question gathering process.
  • Each election, solicit Formal Questions to be answered by all candidates. Have all candidates answer individually and separately, privately submitting their answers for translation, translations in core languages to be published on the first day of formal discussion.
  • during formal discussion and voting, allow informal questions. Present questions in the order they were answered, so readers can see who got to see what answers prior to their writing. A "Rapid-response Translations teams" needs to get these informal questions and answers out to the projects. Members of all projects should have the chance to ask a question and have us do our best to give them answers in their language.
  • Allow wide variety in question type, include optional questions, 'fun' questions, 'for the rest of the community' questions.
  • Allow candidate-specific questions, aimed at a particular candidate, but upon which any other candidate may comment.
  • Question UI should be able to filter by question, as it did this year.
  • Questions UI should ALSO be able to filter question answers by candidate There comes points in discussions where it's helpful to see only two candidates responses for pairwise comparisons, or see only a handful of candidates.

Candidate Presentations[edit]

  • Designate a "suggested length" we tell candidates. Beyond that, 'suggested length' is non-binding.
  • Create a "guaranteed translation length"-- some length that the we've decided is "definitely enough" for a candidate presentation. Increase it to beyond 1200 chars-- get this year's candidates to suggest how much they thought would be "enough" and then double it to get "definitely enough". One 2011 candidate suggest 2400 char would be enough, making 4800 characters our "definitely enough" point, if based on that single candidate's opinion.
  • Remove word limit on candidate presentation / Massively increase the word limit to "good faith limit". Candidates should understand that statements exceeding the "guaranteed translation length" cannot be guaranteed to be translated beyond that length. -- candidates should have lots of latitude at this point. If they want to write a 300 page manifesto as their candidate statement, let 'em-- it'll be informative. Don't guarantee to translate it, but allow it.  :)
  • Allow and encourage links
  • Consider including videos of candidates.

Voting[edit]

  • Use a better UI for the voting method, so that 'ranking' is automatic via dragging.
  • add the ability to 'flag' unqualified candidates on ballot.
  • Consider moving away from a culture of secret voting to a system of optionally secret ballots, where a portion or entirety of a ballot can be hidden if desired, or published if desired. Encourage rationales and discussion for the top few candidates.
  • Add rationales for top candidates to ballot.
  • Add 'saved' ballots so people can update ballots more easily. (if I've voted and try to vote later, I should see my last ballot as it was)

Post-voting[edit]

  • Reflection. Ask candidates, coordinators, and community users to reflect on how the election process went and how it could be improved. Write down the lessons before you forget them.
  • Endorse the unannounced winners. If you have confidence in the voting process, say so now, before the winners are announced.

Announcement[edit]

  • Release a video / animation showing and explaining how the schulze method picked the winners, in a way that is clear and easy to comprehend. (create and translate this earlier, with the results added in only at the last moment). The Schulze method offers us huge advantages over traditional voting, and just using it and telling people about it is an opportunity to spread important knowledge.

Post-Announcement[edit]

  • Encourage posts of the form: "I supported $WinningCandidate because of X, Y, and Z". Let the global community give each winning candidate a 'mandate' explaining its vote.
  • Conduct a study into the active editors who chose not to vote. Why didn't they vote?

Some more[edit]

  • The 1200 character limit on presentations is very strict. You can hardly squeeze in a decent introduction of yourself, your achievements and your beliefs.
  • Maybe force a self-presentation / achievements / worldview split on the candidate presentations.
    I like this idea. A good achievements section should have more links directly to projects - it could be short descriptive clauses, easier to translate, and not limited in the same way. A few more elements of 'worldview' could be included into the header and also not character-limited.
  • Send out the voter invitation mail on the first day of the elections.
    Yes!
  • Voter turnout will remain low as long as the board remains unknown and irrelevant. People just won't care. The board should publish some kind of report of their own and present it directly to editors in various projects, they need to communicate what it is they do and how their decisions matter.
    Yes. I will publish something like this before Wikimania.
  • I feel that 19 candidates for 3 open slots are too many. It's difficult to review and evaluate and rank so many candidates. Maybe we need better barriers for candidates, as some candidates are almost farcical. Maybe requiring to show some initial support (50 signatures?) will help reduce this inflated number of candidates.
    This is probably the most important bit. If we have more than, say, 8 candidates, we should require either a show of support or a primary election to narrow things down to no more than 8. 6 might be better. Then there could be a proper debate / public discussion.
  • Eventually, people like to know what others think of the candidates. Voter guides are one decent way of doing it, as long as they're based on fact and not on defamation.
  • It seems few people read the Q&A section, certainly with up to 19 answers to every question it's more than any person can handle. Also, I find that some of the questions were not very good - they either handled esoteric topics or repeated other questions or were provocative.
    We need clerks to organize and present the Q&As. And a group (the kurier or signpost or wikinews could take this on) to start the process off with a set of good, well considered questions based on the popular discussions of the past year(s).
  • So as an alternative, maybe we need a kind of forum (inasmuch as wikis provide a forum) for voters to discuss candidates and freely talk to candidates. While this was never strictly forbidden, it was never encouraged either.
    Interesting thought... SJ talk | translate   10:13, 18 June 2011 (UTC)[reply]

I'll add more when I think of some other lessons learned. Harel 20:21, 12 June 2011 (UTC)[reply]

Even more[edit]

Please add your own lessons