Talk:Wikimedia Foundation Board of Trustees/Call for feedback: Community Board seats/Ranked voting system

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

Single transferable vote[edit]

Can anyone tell me why this isn't a slam-dunk that the Board should have just proposed at the same time as the first round of changes to the Bylaws? According to the Board's own proposal page, STV "is designed to resemble proportional representation" and "should [be] a good fit for the Wikimedia Movement". On top of this, it's a legitimate voting method used in real-world elections such as the Australian Senate and the Irish delegation to the European Parliament. It certainly meets the demand for a real voting system from community members angry about the Board's shenanigans! BethNaught (talk) 22:09, 1 February 2021 (UTC)

I oppose using proportional representation while the Board is structured as it currently is. With about half the Board held by "Board-selected" (which in practice look to be heavily influenced by staff), there are risks of tilting the balance away from community control. If the WMF's continual explosive growth doesn't slow down soon, the staff will make up more than 1/6 of the votes within one election, effectively allowing the addition of a member which could be opposed by the entire community if proportional representation were used. (WMF staff are guaranteed a vote, regardless of edits. Relevant thread from 2015.) The lack of firm community control over the Board makes for quite a mess, and limits safe options for voting systems. --Yair rand (talk) 01:54, 5 February 2021 (UTC)
In practice I don't think that WMF staff all vote, or that where they vote, they all vote the same way. So I wouldn't be that worried about the circumstances where staff form a block-vote to ensure a WMF favoured candidate gets elected. I do think that where the WMF staff get to vote, affiliate staff should also be able to vote on the same basis. Chris Keating (The Land) (talk) 13:54, 10 February 2021 (UTC)
Also; the number of votes cast is something like 6,000 in the 2017 and 2015 elections. I think WMF + Affiliates is some distance off having 1,000 staff, let alone 1,000 staff who actually vote in elections! Chris Keating (The Land) (talk) 16:10, 10 February 2021 (UTC)
@The Land: The community also doesn't all vote the same way, and shouldn't have to just to prevent loss of the WMF. The 2017 and 2015 elections had 5120 and 5167 valid votes, respectively. The WMF hired 102 full-time staff and 37 contractors last year, and the org has been growing at a dangerous rate for its entire existence, fast enough that getting to 1/6 within an election cycle (three years) seems well within the range of possibility. (If you include affiliate staff, which I'm not as concerned about, it would likely get there even faster.) If we come to a point where the interests of the staff sharply differ from the interests of the movement, we shouldn't have a giant vulnerability of the system in place. --Yair rand (talk) 02:02, 11 February 2021 (UTC)
Hi there, Yair rand, I hear what you said about concerns about Wikimedia Foundation staff getting to vote and those votes influencing the election in a way that would not align with the community. I know some staff members are volunteer contributors too. Do you have examples of where staff voting has been an issue? I am trying to better understand so I can better capture this feedback. Is there a solution to this? Best, JKoerner (WMF) (talk) 23:17, 11 February 2021 (UTC)
JKoerner (WMF) I would say there has long been a generalized concern that staff are often badly out of alignment with the community, offset by the community-value toward inclusive participation. However there probably aren't any concrete election-result issues to cite because staff numbers were so much smaller and because we never used Single Transferable Vote (STV). STV is a problem for two reasons. First, given certain vote patterns, STV can fail badly. Each round of STV can knock out the best available candidate, leaving the final round selecting of the second-worst out of all candidates. In such a scenario it is very possible the community would cast blame on staff for the pathological result. The other problem is that STV candidates represent factions (Condorcet candidates represent everyone). STV latches on to fringe groups and elects candidates who are polarized opponents. STV is essentially designed to organize staff votes into a block electing a candidate actively hostile to the community and simultaneously organize blocks of votes electing candidates actively opposed to staff. Now... from a craven-partisan-tactical perspective STV is arguably advantageous to the community, it creates a polarized majority-faction actively fighting and winning any conflict with the isolated staff representative. In contrast Condorcet systems elect moderate/centrist candidates all of whom have the broadest possible acceptability to all voters. Staff wouldn't be able to elect any candidate who was actively hostile to the community, but they would have the power to reject hostile or unreasonable or uncompromising candidates. I would hope we all want a movement (and a board) dedicated to collaborative common-interest, not one composed of warring ideological factions. Alsee (talk) 22:28, 13 February 2021 (UTC)
Thank you all, I have captured this feedback for the upcoming weekly report. Best, JKoerner (WMF) (talk) 19:53, 15 February 2021 (UTC)

Opinion[edit]

I think this is a good idea. Chris Keating (The Land) (talk) 20:43, 3 February 2021 (UTC)

Hi Chris Keating (The Land), Glad to see you sharing feedback here. Do you want to share more about why you like it? Best, JKoerner (WMF) (talk) 22:52, 4 February 2021 (UTC)
In short, because I agree with the rationale that's stated. ;) STV or a similar system does a better job of balancing different constituencies or kinds of opinion, particularly with more candidates to be elected at once. The current system also puts a very high weight on "oppose" votes - one "oppose" is worth 3 or 4 "supports" which really penalises candidates who are in any way different or controversial. Chris Keating (The Land) (talk) 14:30, 5 February 2021 (UTC)
The only further thing is that there would have to be very clear communications about how the voting system worked, and ideally a demonstration and clear instructions would be built into the voting interface (and translated). While much of the world is familiar with this system, not everyone is and in the ASBS election there were problems with affiliates apparently misunderstanding. This is not an inherent problem. The ASBS process was rushed, confused, and under-resourced, and with better planning and communication these problems could be avoided. Chris Keating (The Land) (talk) 17:45, 6 February 2021 (UTC)

Seems fine to me, based on the feedback from BethNaught above. It's not a system that I know well, but it's use in real-world elections from real countries makes it seem reasonable. TomDotGov (talk) (hold the election) 00:57, 5 February 2021 (UTC)

I'm generally a fan of STV and would support introducing it. However if it was adopted, I think staggering the trustee terms would be a good idea so that you're not electing all eight trustees at once. The more seats are open and the fewer votes that are cast, the more sensitive STV becomes to small variations. If anyone wants to see the nitty gritty details, Antony Green's blog has everything you ever wanted to know and more. --RaiderAspect (talk) 12:17, 5 February 2021 (UTC)

I strongly support this system. It already works at ASBS elections and has shown good results. It has an advantage of favouring more consensual candidates that might not necessarily be your first priority but who satisfy a significant number of electors. ASBS elections have shown that ranked voting system results in many affiliates voting for their closest (regional, linguistic, thematic etc.) candidate as a first choice, first choices with only narrow support get quickly eliminated, and second or third choices become decisive to select really qualified candidates among those having a wide enough support. It should give good results for an online community election — NickK (talk) 11:09, 6 February 2021 (UTC)

Thanks all for the feedback. I included this in the first weekly report. Best, JKoerner (WMF) (talk) 23:22, 11 February 2021 (UTC)

Issues[edit]

While I am not categorically against STV, as I am against, say, a selection panel, there are some major issues, some coming from Wikimedia related aspects and others from general STV elections, that make me currently a significant opposer. They include:

STV can risk getting non-objectionable candidates in over generally positive. It also makes the election process more complicated, not to mention that if, say, I had 4 candidates I really didn't want in a 20 candidate election, to maximise the chanced to them not getting in I'd have to cast intermediate ranking votes on those I didn't know, without a clear ability to decide (say) who warranted "7th" and who warranted "16th".

  1. Complexity - voter numbers are already fairly low in BoT elections, and considerations of tweaking en-wiki's arbcom elections methodology has previously run into resistance on grounds of complexity. Even a 10% loss in votes cast (as opposed to voters) would be too egregious a negative. Even explanation and so-on, as used in actual elections that shift to STV by no means resolve the issue
  2. Voter knowledge - STV requires knowledge of everyone if you want to be both fair and give functional oppose votes. For example, If had 4 candidates I really didn't want in a 20 candidate election, to maximise the chances of them not getting in I'd have to cast intermediate ranking votes on those I didn't know, without a clear ability to decide (say) who warranted "7th" and who warranted "16th". Whereas in a normal set-up I could support the ones I wanted, oppose the really problematic 4, and go neutral both for those I was neutral on, but also the candidates who I felt insufficiently aware of (as a note, while I read everyone's statement, to me, that would be wildly insufficient research for an election as important as BoT).
  3. WMF voting share - this issue probably can be resolved (but remains one until a viable solution is offered), but its mentioned in the first comment above, so I note it here as well
  4. Objectionable candidates - say we got a lot of unqualified candidates, which I can think of several circumstances that might cause it. STV can make it easier for one to achieve position than the current approach
  5. Non-objectionable candidates - likely to be a more common issue, this is where generally positive candidates who get lots of high ranks but very few mid-rank votes, can lose out to non-objectionable (neither heavily positive or negative) candidates.

There absolutely are positives from STV, but in my experience, they only work fairly when you can trust in the vast majority of the voters knowing significant amounts about the vast majority of candidates. Nosebagbear (talk) 10:20, 10 February 2021 (UTC)

Just my 2p, on the second point: you can put the candidates from 7th to 16th at the same level, this will make sure you give no specific preference to any of them and at the same time you prefer all of them to the ones you want to vote against — NickK (talk) 12:31, 10 February 2021 (UTC)
Hi NickK, Thanks for your continued feedback! I'm asking some questions to see if you have ideas for possible solutions. I numbered them to go along with your points:
  1. Is there a way you see to increase participation in the Board of Trustee elections?
  2. Is there a way you see to improve learning about candidates before elections?
  3. I asked the same question above, because it was noted as a concern, is there a situation where the Wikimedia Foundation staff voting has become a problem? Or is this something that you see could become a problem in the future?
  4. How might a solution to provide qualified candidates be developed? Does it align with the idea Call for types of skills and expertise or is it something different?
  5. Maybe I need to study this a bit more, but could you clarify this for me so I can better document your thoughts about this?

Thanks for your time on this! Best, JKoerner (WMF) (talk) 23:35, 11 February 2021 (UTC)

@JKoerner (WMF): Along with my points? I think you wanted to ping rather @Nosebagbear:, I really put in just my 2p. (If these questions are still relevant to me, I can still answer them) — NickK (talk) 01:06, 12 February 2021 (UTC)
I suspect @JKoerner (WMF): did mean to ping me. So my proposal for improving participation in terms of candidates, is for the WMF to go to the communities where it would like more candidates from, much further in advance (probably at least 6 weeks) of formal nominations opening, and ask why not run, what issues they have and see if they can be mitigated. In terms of getting voters, a more "multi-spectrum" approach is needed: the general banner, dropped messages at as many of the local communities primary watering holes as possible, a better method to allow individuals to ask Trustees who speak different languages questions, a request to chapters to notify their members, and so on and so forth. Some research into why people don't vote, might also be worthwhile.
As well as the multi-lingual questions bit, I'll give some thought on learning about candidates.
So long as WMF staff remain under 5% or so of the electorate, I don't particularly have a concern - I included it as others have noted concerns (not many, but more than on this page). It's probably the least of my issues there. It could be a concern in the future. Not because I expect the WMF to start canvassing its employees or anything, but there is a benefit to staff votes to back a trustee who would aid their projects (etc), which is not necessarily the same as trustees representing the community Nosebagbear (talk) 10:27, 12 February 2021 (UTC)
Ack! Yes, my apologies, @NickK:! I should not work so late in the day. That means mistakes!
Thanks, Nosebagbear, for elaborating on this. This helps. Best, JKoerner (WMF) (talk) 12:42, 12 February 2021 (UTC)
@JKoerner (WMF): Well, as Nosebagbear answered something difficult, I can add one argument to your questions.
Regarding increasing participation, in Ukrainian Wikipedia we have managed to significantly increased participation by two measures:
  • Translation of the main candidate information (brief bios and programmes, 1 paragraph each) into Ukrainian with a discussion on the village pump.
  • Personal voting invitations in Ukrainian, explaining what these elections mean and giving a link to candidate presentations.
This worked, as roughly 25% of eligible participants actually voted — NickK (talk) 21:26, 12 February 2021 (UTC)
Thank you both for this. This is great feedback. I wonder if other communities have experiences like this, NickK. @Nosebagbear: and @NickK:, could there be an idea you want to suggest in this? I sense there might be something emerging here - something like a way to design election planning to encourage participation (voting and candidates). Best, JKoerner (WMF) (talk) 21:46, 12 February 2021 (UTC)
@JKoerner (WMF): The key word is 'election'. All of this makes sense if there is an election (and there is also a good amount of ASBS discussions as well). If there is no democratic election, there is no point in recruiting community members — NickK (talk) 16:21, 13 February 2021 (UTC)
Got it. I hear your feedback on that. Is there a sort of best practices you might be interested in drafting to encourage broader participation? It sounds like you and Nosebagbear have a lot of practical experience with this. Best, JKoerner (WMF) (talk) 13:17, 15 February 2021 (UTC)

Support ranked choice, preferably Schulze or other Condorcet method[edit]

I definitely support ranked choice, preferably returning to en:Schulze method which we used to use, or other en:Condorcet method. (Schulze is a member of Condorcet method family). I could write an entire essay on the subject of elections systems. Mathematicians have extensively studied elections systems and how they can fail. When there are more than two candidates you really need Ranked-choice to actually record the voter's preferences. The proposal here suggests Single Transferable Vote. While Single Transferable Vote is indeed better than a single-candidate-ballot (and better than the Support/Neutral/Oppose system we currently use), Single Transferable Vote can potentially to select the second-worst candidate. If electing multiple candidates, Single Transferable Vote could select the second worst, third worst, and fourth worst candidates.

If there is a candidate who could beat every other candidate in a one-on-one contest, Condorcet systems are the only systems guaranteed to elect that candidate. In rare cases voters may create a en:Rock-paper-scissors loop between candidates, and different Condorcet systems merely offer different ways to break that tie.

In short, any time any election system disagrees with Condorcet, it is mathematically provable that the other election system is clearly wrong and the Condorcet system is clearly right. If you put those two candidates in a head to head matchup the winner will always be the one selected by the Condorcet system.

The benefit Single Transferable Vote might have over a Condorcet system is that some Single Transferable Vote methods are easier to explain to the public. However no one uses those simple versions because they are clearly inferior. Once you start using sophisticated system for transferring votes, it gets complicated. At that point there's no reason not to use a Condorcet system. Alsee (talk) 08:21, 13 February 2021 (UTC)

We're here for the Sum of all knowledge, so simplicity must not be any criterion, that would override correctness. The worst would be some "First past the pole" systems, were bigger minorities could rule over the clear majorities, like in Britain in the Thatcher years, or with presidents without the popular vote. Grüße vom Sänger ♫(Reden) 10:59, 13 February 2021 (UTC)
I have to disagree, because it is not a dichtomy - "simplicity or correctness". All other wikipedia governance is using some degree of simplification from what the ultimate ideal would be - because otherwise you get a net-loss of correctness unless you can guarantee that every voter will have a flawless understanding, first time, of whatever more complicated method is selected. Nosebagbear (talk) 16:52, 13 February 2021 (UTC)
Nosebagbear I accept your point against a radical disregard for simplicity. I would suggest the primary practical decision here is whether to use a Ranked Choice system. Non-ranked Choice systems can fail rather badly on "correctness", and I would hope we can avoid them. I acknowledge that not everyone is familiar with Ranked Choice, and that there can be confusion mainly regarding unranked candidates. Those issues can be largely addressed with a decent voting interface. Perhaps give each candidate a slider with clearly indicated directions for "good" and "bad", and start the slider in the middle. Then there are no unranked candidates and, to the extent people are still unsure, they would be unsure about things that are literally irrelevant. Casting votes of +10 +9 -10 is the same as voting +10 0 -10 is the same as +10 -9 -10 is the same as +1 0 -1 is the same as +3 +2 +1. Or have an interface to make a ranked list, maybe dragging candidates up or down.
If we do go with Ranked Choice then there aren't really any good arguments for not using a Condorcet system (the most mathematically correct option). The Ranking interface is the same no matter which back end system is used, and the various back end systems wind up being roughly comparable in complexity. Alsee (talk) 00:33, 14 February 2021 (UTC)
Thanks all for this discussion. I have noted the feedback for this upcoming weekly report. Best, JKoerner (WMF) (talk) 20:33, 15 February 2021 (UTC)
I fully agree with Alsee and support adoption of a Condorcet method. Anthere (talk)
I also support adopting a Condorcet method, with two caveats:
  • The Board should not be deciding the method used for their own election.
  • The problems with the voting UI that existed during its prior use must be fixed before readopting the method. The use of plain text boxes (which defaulted to blank, which was interpreted as "lowest value" by the system) was very problematic, as many voters misinterpreted the system. Either use a number input defaulting to 0, or a fixed-step range input with a spectrum going from "oppose" to "support", or a drag-and-drop list ordering system like the Mediawiki logo vote.
--Yair rand (talk) 06:05, 18 February 2021 (UTC)
Hello Yair rand, I am compiling the main points for the RVS and want to ask you: if the Board shouldn't be the one deciding the voting system, who specifically should be involved in the decision for any 2021 elections? Thanks --Oscar . (WMF) (talk) 14:50, 23 February 2021 (UTC)
The Wikimedia Foundation elections committee is the org, that runs elections (and that hasn't done their job up to now, and changes could done by RfCs. The Board is just a servant and trustee of the communities. The current board is not legitimately constructed, it has not been elected in a proper election, and only a proper election validates the Board. Grüße vom Sänger ♫(Reden) 21:43, 23 February 2021 (UTC)
^ That about sums it up. --Yair rand (talk) 23:49, 23 February 2021 (UTC)

Collected Feedback about Ranked voting system from the first and second weekly report![edit]

Hi folks! Here is the summary of the main views and opinions of Ranked voting system from last week and this week Weekly's report, these reports summarize ideas from all over the movement. We think it might be helpful to share the feedback pertaining to each idea on each idea's talk page in order to provide coherence and move the discussion further.

Ideas from the first weekly report[edit]

  • This should be used as it is similar to other real world examples in governments
  • This should not be used because the Wikimedia Foundation staff are allowed to vote and their votes could potentially sway the elections
  • Education content on how this voting system works is critical to its success
  • Feedback during a Russian WikiCommunity conversation had most attendees favoring a ranked voting system

Ideas from the second weekly report[edit]

Please reach out with questions or comments, cheers --Oscar . (WMF) (talk) 00:04, 19 February 2021 (UTC)

Common sense reform[edit]

I believe this would be a common sense reform, that should not be controversial. It should help in opening up the election in a transparent way, that would be in the best interests of everyone.--Pharos (talk) 05:10, 23 February 2021 (UTC)

Collected Feedback about Ranked voting system from the third weekly report[edit]

Here is the feedback from the third weekly report covering February 15 - 21 of the Call for Feedback.

10 users from 5 different home wikis have participated in the conversation on this idea's talk page so far.
  • Attendees at a meeting of the Kurdish community said a ranked voting system would bring some trustees who don’t have a clear idea of what they need.
  • The Condorcet method is getting more support. One person said the Board should not be deciding the method used in their own election and the interface must be fixed.
  • Jon of Wikimedia Norge, former member of many election committees, mentioned STV has been discussed since 2008 and it went well in the affiliate seat selection process.
  • The 2019 ASBS facilitators and other volunteers disagree with the idea of the Board deciding a voting system for community elections and say this decision belongs to the community.
    • “ASBS elections have shown that ranked voting system results in many affiliates voting for their closest (regional, linguistic, thematic etc.) candidate as a first choice, first choices with only narrow support get quickly eliminated, and second or third choices become decisive to select really qualified candidates among those having a wide enough support.”

Do reach out with any questions or feedback. Cheers, --Oscar . (WMF) (talk) 20:45, 25 February 2021 (UTC)