Wikipedia talk:Requests for comment/Arbitration Committee Elections December 2019

Electoral commissioners

• Make up of the commission (follow up from prior year 3 vs more than 3 active seats discussion). — xaosflux ^Talk 16:05, 21 October 2018 (UTC)

  Should probably remain at three, or at least at an odd number, so the Commission can make a 'ruling' by majority if necessary.  Swarm  talk  00:08, 28 November 2018 (UTC)

• Remove the requirement for commissioners to sign the ANIP. We receive absolutely no non-public data, it's positively silly to require a signature. SQL^{Query me!} 07:28, 18 November 2018 (UTC)

  Or go the opposite route of handing temporary oversight, and maybe checker, to commissioners for ACE related matters only, given the recent situation on Fred's questions page, to be revoked at the end of the ACE. —CYBERPOWER (Chat) 13:22, 18 November 2018 (UTC)

  @SQL: I believe it is so that if there are issues that the scrutineers need to raise regarding private information they may discuss them with the commissioners. This is not well codified but could be (i.e. commissioners may receive and evaluate information from checkusers related to the election). — xaosflux ^Talk 18:49, 18 November 2018 (UTC)

  @Xaosflux: One of the, probably the original reason would had been because of SecurePoll. If you're an admin for the election, then you see CU like info for all votes. Commissioners are no longer made (or rather no longer retain before the election begin) admin for the election. -- KTC (talk) 19:05, 18 November 2018 (UTC)

  @SQL: According to the scrutineer instructions, Commissioners, like Scrutineers, are temporarily made Securepoll admins during the election, which "enables access to CheckUser-like data on each vote cast". Is this information incorrect? I'm not even sure who's in charge of granting Securepoll adminship.  Swarm  talk  23:49, 27 November 2018 (UTC)

  Swarm, It is incorrect, as of this year. SQL^{Query me!} 23:51, 27 November 2018 (UTC)

  @SQL: Do we have any idea when or why it was changed? Was it just a fluke, or was it intentional?  Swarm  talk  00:12, 28 November 2018 (UTC)

  @Swarm: Intentional, and this is not the first vote-wiki SecurePoll election where that's the policy. Underlying reason being security and minimising the number of people who has access to non-public data, the scrutineers being the only people mandated to decide on the validity of a vote, the coordinators thus have no need to have access to the CU-like data that comes with being admin of an election. -- KTC (talk) 11:16, 28 November 2018 (UTC)

  Swarm, To answer the other half of your question - I'm not 100% sure, but my access was initially set up by JSutherland (WMF). We were notified via email by Jalexander-WMF on Nov. 17th that we would be removed as election admins before the election goes live this year. I'm not sure how much of the email I'm allowed to quote or paraphrase. SQL^{Query me!} 00:21, 28 November 2018 (UTC)

  Okay, thanks both. I went ahead and got an on-Wiki clarification for the community to reference, so I'm thinking it should be fairly uncontentious to delete the confidentiality requirement going forward.  Swarm  talk  23:23, 29 November 2018 (UTC)

• Void the restrictions on the Electoral Commission's mandate. The 2013 decision to shackle the Electoral Commission, to prevent them from taking "active and ongoing role in the day to day operation of the election" was made by three users and there is no particularly obvious reason maintain these restrictions. These are highly trusted users appointed by the community, and they're already expected to take a leading role in running the election, and they can and should be trusted to oversee and moderate the election at their discretion.  Swarm  talk  00:08, 28 November 2018 (UTC)
  • As a side note, the original intent of the Electoral Commission was to "deal with unforeseen problems, adjudicate disputes", due to unusual events occurring during the previous election in 2011, after the annual RfC had closed. As far as I recall, the intent was to provide a way to resolve open questions in a timely manner during the election when a community consultation is infeasible without unduly disrupting the election process. Of course, the community could decide for the next election that they want the commission to take a leading role; I don't believe, though, that the commissioners are currently vetted by the community with this in mind. isaacl (talk) 04:46, 28 November 2018 (UTC)

  @Isaacl: Yeah, the Commission isn't mandated to actively run the election, and I'm not suggesting we change that. However, the Commission's responsibilities to deal with "unforeseen problems" inherently make them the party ultimately responsible for the proper progression of the election. In theory, this means we have a strong team of volunteers handling everything, and the Commission can sit back and do nothing until the unlikely event they're needed, but in practice, it means they have a de facto dual role as election volunteers, due to a notable lack in volunteers.  Swarm  talk  23:23, 29 November 2018 (UTC)

Voters

• Voter suffrage - was identified last year that for users with multiple accounts, including bots, may all be 'eligible' so long as their controller does not vote multiple times. — xaosflux ^Talk 20:56, 26 October 2018 (UTC)
  • IIRC last year or the year before one user voted with multiple eligible accounts voted with account A and then several days later with account B. The determination was that the vote from account A be discarded and the vote from account B be allowed to stand as if it was an amended vote from account A. Thryduulf (talk) 16:40, 20 November 2018 (UTC)
• New voter requirement suggested: Must have 1 live edit within the past 2 years. We are consistently maintaining voter rolls of people that have long left the project. — xaosflux ^Talk 21:32, 26 October 2018 (UTC)

• Support. If anything that seems a bit lenient, but I think it would be a good idea. – John M Wolfson (talk • contribs) 22:59, 11 August 2019 (UTC)

• I'd like to see Check User run on all voters by policy. One person controlling 20 accounts can tip an election. Carrite (talk) 01:04, 28 August 2019 (UTC)

• @Carrite and John M Wolfson: the way secure poll works this kinda already happens. Stewards have access to the CU data during the scrutiny period, and refer cases that raise concerns to local CheckUsers for review. Local CUs are allowed to use their tools to assist stewards in this process by checking the regular editing data, but do not have access to the poll data beyond what is shared by stewards. In cases where there is a concern, stewards have held off on certifying the results pending further assistance by local CUs. TonyBallioni (talk) 02:12, 28 August 2019 (UTC)

• Thanks for the information. That seems reasonable enough as is. – John M Wolfson (talk • contribs) 02:15, 28 August 2019 (UTC)
• Also, the stewards are granted local CU for the election and can perform their own investigation. — xaosflux ^Talk 11:22, 28 August 2019 (UTC)

Something weird happened here. User:John M Wolfson's 11 August comment was originally under User:Xaosflux's (Special:Diff/910413780). Then User:Carrite inserted their comment between the two (Special:Diff/912810030). And then User:TonyBallioni replied to John M Wolfson as if their comment had been in reply to Carrite rather than Xaosflux (Special:Diff/912817163). I moved Carrite's comment to after John M Wolfson's, but didn't try to reorder anything else. Anomie⚔ 11:32, 28 August 2019 (UTC)

• I would like reconciliation with other elections if it is possible. The requirements to vote in the steward elections are that an account
  • not primarily use the account for automated (bot) tasks;
  • have made at least 600 edits globally (on all Wikimedia wikis) before one year before voting day
  • have made at least 50 edits globally (on all Wikimedia wikis) between one year before voting day and voting day

I would like for voting rights to be more obvious. I know that ArbCom seems English Wikipedia focused and Stewards have a rather different role but there are some similarities. More elections are appearing and I wish we had some common or expected criteria. Blue Rasberry (talk) 19:25, 4 September 2019 (UTC)

@Bluerasberry: feel free to add a RfC section, I've already opened one on voter suffrage here that may be a good place to add in: Wikipedia:Requests_for_comment/Arbitration_Committee_Elections_December_2019#Voter_activity_requirements. — xaosflux ^Talk 19:49, 4 September 2019 (UTC)

Mass-messaging

• For mass-messaging, codify not to message users with blocks that extend beyond the election. — xaosflux ^Talk 21:31, 26 October 2018 (UTC)
  • Codify not to message bot flagged accounts, and accounts in Category:All Wikipedia bots, Category:Wikipedia alternative accounts, and Category:Wikipedia doppelganger accounts. -- KTC (talk) 13:34, 12 November 2018 (UTC)
  • And Category:Deceased Wikipedians. Thryduulf (talk) 01:16, 20 November 2018 (UTC)
  • And globally locked accounts. -- KTC (talk) 16:31, 20 November 2018 (UTC)

Candidate requirements

• What is meant by "good standing" in terms of candidate requirements? Options include:
  • Not blocked or banned - at the time nominations close or at some other time?
  • Not subject to active sanctions - at the time nominations close or at some other time? - any sanctions or just some types?
  • Have the trust of the community - determined how? determined by whom?
  • Not been recently desysopped - what counts as "recent"?
  • Whatever the consensus of the commissioners is - does no consensus mean good standing or not?
  • Some combination of the above. Thryduulf (talk) 11:26, 12 November 2018 (UTC)

• This topic has been rendered moot. Per WT:ACE2018#Official statement on Fred Bauder's standing as candidate, the "good standing" verbiage has been removed outright as it was discovered that it was wrongly added unilaterally and without explanation, and it was never formally intended to be implemented by the community, and it remained over the years erroneously and without the community mandate. Theoretically, if we want 'good standing' to be a formal requirement, it can be proposed, but such a proposal would likely not be reasonable. A unilateral admin action such as a block should be able to influence an election.  Swarm  talk  00:21, 28 November 2018 (UTC)

• Whether any of these should be requirements for candidates are questions I would still like to put to the community next year, only the words "good standing" are moot. Thryduulf (talk) 11:36, 28 November 2018 (UTC)

Committee structure, term lengths

• Has the smaller committee size helped, hindered, made no difference or is it too soon to tell? Thryduulf (talk) 11:26, 12 November 2018 (UTC)
• Number of vacancies - one year or two year term for any seats coming up for election that was as a result of a successful candidate last year receiving 50%-60% support? -- KTC (talk) 13:41, 12 November 2018 (UTC)

Candidate lists, question pages

• Are candidates allowed to remove and/or move questions asked of them to the talk page? Other editors? Only the commissioners? Do the usual exceptions apply? Thryduulf (talk) 11:26, 12 November 2018 (UTC)

• Should candidates be permitted to RevDel problematic edits on their own page, or should they seek an independent administrator or a commissioner to handle it instead? This immediately came to mind. OhKayeSierra (talk) 21:21, 5 December 2018 (UTC)

• I appreciate where you're coming from on this — normally I'd say no, get someone uninvolved to take care of it – but that candidate has been the constant target of a vile and ongoing attack for a while now, so this seems perfectly fine to me. ~ Amory (u • t • c) 22:45, 5 December 2018 (UTC)

• @Amorymeltzer: There’s no doubt in my mind that his integrity is beyond reproach considering the unwarranted vitriol he’s faced this election, but I still believe it’s a question worth considering, in light of that and everything else involving this year’s election. OhKayeSierra (talk) 01:30, 6 December 2018 (UTC)

• I'm wondering about the number of questions. On one candidate's page it looks like there are 35 questions, but one is actually 9 question and another 7, etc. I'd estimate there are at least 70 questions although I think this example is an outlier. Now I wouldn't suggest limiting the number of people who can ask questions, but perhaps the number of questions they can ask. Doug Weller talk 13:04, 16 November 2018 (UTC)
• Should questions to candidates be required to be phrased in a neutral manner (i.e. avoiding leading questions)? Even if a non-neutral position is accompanied by evidence? Thryduulf (talk) 16:11, 17 November 2018 (UTC)
• Are withdrawn and/or disqualified candidates listed on the candidates page? -- KTC (talk) 13:44, 12 November 2018 (UTC)
• Does the community want the Commission to take a more active role in curating the Questions page? -- KTC (talk) 09:50, 15 November 2018 (UTC)

Voting system

• I think we should consider replacing the current voting system with some kind of ranked-choice system that allows users to fine-tune their votes, e.g. to denote which candidates they would like to see on the Committee if their preferred choices don't gather enough support and/or which candidates they definitely don't want to serve. Regards SoWhy 08:12, 20 November 2018 (UTC)

  @SoWhy: If it's not a voting system that SecurePoll supports, then you would need to propose this well in advance of the election, gain consensus support, and have someone volunteer to code it up and reviewed. Alternatively, you would need to gain consensus to move away from SecurePoll altogether as part of the proposal. -- KTC (talk) 12:46, 20 November 2018 (UTC)

  @KTC: As far as I can tell, ranked-choice was (supposed to be) added during mw:SecurePoll/Redesign. Maybe Anomie, Jorm or RobLa who worked on the redesign according to MediaWiki-wiki can confirm this? I don't see a way to test SecurePoll myself. Regards SoWhy 13:46, 20 November 2018 (UTC)

  The project you refer to did not include adding any voting methods. But, looking at the extension, I see it includes the Schulze method (without any tie breaker), which appears to be the only method using ranked voting currently available. The extension also supports Approval voting, Plurality voting, and two variations of Range voting (sum and average). As far as I can tell SecurePoll doesn't support the S/(S+O) method directly, so I guess it must be done by people taking the histogram output by the averaging Range voting method and manually calculating S/(S+O). It looks like there was a tallier for Instant-runoff voting (where the wikilink ranked-choice redirects) at one point, but it was removed in 2015 per T107400.

  If you want to test SecurePoll, probably the easiest way is to request adminship on testwiki. Anomie⚔ 15:25, 20 November 2018 (UTC)

  @Anomie: that might be broken right now? I tried making some and it always fails with "Our servers are currently under maintenance or experiencing a technical problem.". — xaosflux ^Talk 16:11, 20 November 2018 (UTC)

  Looks like that's being tracked as phab:T209804. Anomie⚔ 19:01, 20 November 2018 (UTC)

  If it were to be necessary for a new voting system to be implemented, I suggest considering Ranked pairs, which is also a Condorcet method and is pretty straightforward to implement. It satisfies all the voting system criteria met by Schulze according to its Wikipedia article. Of course, there are always going to be scenarios where one method is worse than another; I can't recall at the moment if those that are worse with Schulze are considered by experts to be more likely that those that are worse with Ranked Pairs. isaacl (talk) 23:19, 20 November 2018 (UTC)

  (+1) ∯WBG^converse 12:37, 21 November 2018 (UTC)

  @Isaacl and Winged Blades of Godric: if you would like this option in the 2019 RFC then you will need to ensure either that it is supported by SecurePoll before the RFC begins or get consensus to use some other system that does support it. The same goes for any other proposed voting system of course. Thryduulf (talk) 10:09, 22 November 2018 (UTC)

  It's not a decision solely for me; I think interest should be gauged far ahead of time before a request is made to add it to SecurePoll. Otherwise it would be a waste of effort to add an unused feature to SecurePoll. isaacl (talk) 18:25, 22 November 2018 (UTC)

  I propose Meek STV which is arguably the fairest multi-winner voting system. Please also see Phab:T117127. Furthermore, we are holding our elections using Schulze method and SecurePoll at Persian Wikipedia: 2016, 2017, and 2018. 4nn1l2 (talk) 07:35, 22 November 2018 (UTC)

  Yes, SecurePoll already supports Schulze but Meek STV or Ranked pairs (or even Schulze STV) has to be implemented yet. Point is, at the 2019 RFC we should discuss a) whether to switch to some kind of ranked choice at all and then b) which system to use (from those that were implemented in SecurePoll by that time). Considering that T117127 is open since 2015 with no takers, we might have no choice but to use the system that is currently implemented. Regards SoWhy 10:21, 22 November 2018 (UTC)

  Given the lead time required to support a new voting system, I suggest a discussion on it should be started earlier rather than later (though I would wait until after the current elections are over). I agree that it's quite possible a decision to change to a new system may have to be suspended until support is implemented. isaacl (talk) 18:30, 22 November 2018 (UTC)

  I think the best thing to do is, work on getting one or more different vote systems supported by SecurePoll as soon as possible, then when they are added have a discussion about picking one or two (at most) of those systems to run off against the current system in next year's main RFC. I suggest you make that first discussion a centralised discussion, notifying those who've commented on voting systems here and on the election talk page, those listed in the below asking to be notified and anyone else you know to be interested. Thryduulf (talk) 23:06, 22 November 2018 (UTC)

  The trick would be in getting people to do the work of implementing, reviewing, and merging code for multiple voting systems, on the chance that the community might later decide to use one of them. Anomie⚔ 02:10, 23 November 2018 (UTC)

  I agree with Anomie; I think it will be challenging to get multiple systems implemented on spec. As Schulze is already in use, my guess is that trying to get a consensus for that will be the (relatively) easier approach. Not sure if there is really enough desire to switch, though. isaacl (talk) 03:58, 23 November 2018 (UTC)

  According to T117127, the change would only require adding a different tallying mechanism since the other parts could be copied from Schulze, so it might be easier than writing the whole thing from scratch. Still, you need someone to actually add a different tallying mechanism. Regards SoWhy 10:14, 23 November 2018 (UTC)

  Even better, SecurePoll separates the "ballot" and "tallier" so you wouldn't even need to copy the code for the ballot. Anomie⚔ 19:21, 23 November 2018 (UTC)

  The ballot can be reused for methods where the voter specifies a simple ranked order. For methods where the candidates are given scores, a new ballot would be needed. (I am dubious about whether a method requiring scoring could gain consensus agreement, given the additional work it would require of voters, but maybe I'm overestimating the effort of using say a simple 5-star scale.) isaacl (talk) 19:35, 23 November 2018 (UTC)

  FYI, that ballot already exists too. It's the ballot we use for the current ArbCom elections, with the three values labeled "Oppose", "Neutral", and "Support" instead of "-1", "0", and "1". Anomie⚔ 20:21, 23 November 2018 (UTC)

  Your best bet might be to see if someone would be willing to commit to implementing one new voting method for enwiki ArbCom elections, with the specific method to be determined by the result of an RFC. That could avoid both the "developers do a bunch of work that won't be used" and "we hold an RFC that can't be implemented" problems. You might try asking volunteer User:Huji who is currently maintaining the extension, or WMF's Community Tech team (since enwiki ArbCom elections are related to "administrative layers of the Wikimedia projects"). Anomie⚔ 19:22, 23 November 2018 (UTC)

  There's a little bit of chicken-and-egg problem: personally I'd like to first see a bit more interest within the community as a whole (versus a few voting system fanpersons :). I don't have a good feel yet if there's even a small chance of a change gaining consensus. isaacl (talk) 19:40, 23 November 2018 (UTC)

• Unsure if it has been discussed here in the past or not (it is likely that User:4nn1l2 has mentioned it before, though probably not on this page), but to make sure everyone is aware: for the last three years (2016-2018), we have been using Schulze and a ranked-choice system for the fawiki Arb. Com.-equivalent elections. It has been successful and, despite some initial nay-sayers, it has been well received. I would like to repeat 4nn1l2's point that Meek is the better solution but implementing it requires tens of hours of work, which neither myself nor other SecurePoll maintainers have not been able to set aside. huji—TALK 22:21, 23 November 2018 (UTC)

  As the person who implemented Schulze in BoardVote (predecessor to SecurePoll) for the 2008 WMF Board Election, I would like to point out that Schulze is technically designed for a single winner election even if that's not how Wikimedia have used it. -- KTC (talk) 22:58, 23 November 2018 (UTC)

  @KTC: Schulze method can also be used in multi-winner elections because it generates a ranked list of candidates. In fact, any single-winner voting system can be used for multi-winner elections. The only advantage of a genuine multi-winner voting system such as STV is guaranteeing proportional representation (i.e. fairness). Having said that, the current voting system of the English Wikipedia ArbCom and Board elections is a simplified version of score voting which is also a single-winner voting system. 4nn1l2 (talk) 10:22, 24 November 2018 (UTC)

proportional representation should be a goal of this election. It is generally not a good idea to use a single-winner method to select multiple candidates for a committee, since 51% of the voters can completely determine the outcome, leaving 49% completely unrepresented. I think basing the form of the ballot off of Approval voting rather than requiring voters to rank choices would be easier, considering how many voters don't know as much about the candidates or have a nuanced a set of comparisons of them as in other election situations. It would seem worthwhile to adapt a method like Proportional approval voting for this election. I have written open-source software (https://github.com/nealmcb/pr_voting_methods) to calculate the optimal committee to represent as many voters as possible, and represent them as fully as possible, as outlined here. I would think it could take into account the 50% requirement for ArbCom by simply dropping candidates who received less than that level of support from the pool from which the optimal set of winners is identified.e ★NealMcB★ (talk) 01:33, 25 November 2018 (UTC)

@Nealmcb: As outlined above, not only does the community have to decide to switch voting systems, the chosen system needs to have a software implementation. So unless you are willing to implement PAV into SecurePoll, this won't be even an option. Regards SoWhy 12:31, 25 November 2018 (UTC)

@SoWhy: Yes indeed. I've inquired on the SecurePoll Talk page about whether the necessary data is available. I think it best to do some investigation of how it might work (there?), along with community conversation around the desired goals (here) in order to proceed. ★NealMcB★ (talk) 01:45, 26 November 2018 (UTC)

Nealmcb, were you able to get an answer about this? StudiesWorld (talk) 11:02, 1 July 2019 (UTC)

• Perhaps Meek is best for single transferable vote methods. Personally I prefer methods that take into account the ranked preferences of all voters, not just some of them (those who voted for candidates that exceeded quota, or for candidates that are getting excluded for finishing last). But this can be discussed further at a later time. isaacl (talk) 23:02, 23 November 2018 (UTC)

• "If it ain't broke, don't fix it." Current voting system works fine. Carrite (talk) 13:55, 28 August 2019 (UTC)

General/Other

• Review Wikipedia:Arbitration Committee Elections December 2018/Feedback. — xaosflux ^Talk 14:09, 20 November 2018 (UTC)
• Do not start election on the weekend in case of technical problems. -- KTC (talk) 14:17, 20 November 2018 (UTC)