Wikipedia:Arbitration Committee Elections December 2009/Feedback

This page serves as a forum for editors to provide feedback on the December 2009 Arbitration Committee Elections, and ideas on how the process can be improved for the next election, in 2010. The 2009 elections departed in some ways from those of previous years, most notably in the introduction of the SecurePoll secret ballot.

Please interpolate comments, feedback, and discussion below; if adding issues, use the formatting consistently. Note that this is not the place to discuss candidates or the Committee itself.

Proposed timetable

edit
  • From the announcement of the new arbitrators onwards: seek resolution of the outstanding issues for the 2010 election; liaise with developers if software changes are required
  • 25 October 2010: preparation—call for volunteer coordinators; establish new pages for the election; invite and confirm scrutineers and administrators; confirm timing; liaise with developers and confirm software situation; confirm electoral roll and update eligibility tracker; apply for and implement watchlist notices, for "nominations are open" and "voting is open"
  • 1 November 2010: call for and manage general questions
  • 10–24 November 2010: call for candidates, call for individual questions; manage candidate statement lengths and individual questions to candidates; remind developer of arrangements for starting the vote
  • 1–14 December 2010: voting period; construct draft announcement for scrutineers, plus message to scrutineers
  • 14 December 2010: close of voting – cascade full-protect of all voting pages to ensure clean cut-off
  • 15 By 18 December 2010 (expected): certification and announcement of the results
  • 19 December 2010: solicit feedback from scrutineers and other election officials
  • Total duration: 49 days

Discuss "Proposed timetable"

edit
  • I think this years election shows that having the results announced within a day is unrealistic. I think 3-7 days is more do-able. ~ DC (Talk|Edits) 06:09, 21 December 2009 (UTC)[reply]
  • I agree with DC about this, and would suggest that the whole process be moved to start about 10-14 days sooner, so that announcements can be made around Dec. 15-16 or so. That gives new arbitrators time to identify to the WMF, start navigating through the myriad lists, and everyone to get through the exam/holiday season with a little less stress. Risker (talk) 06:23, 21 December 2009 (UTC)[reply]
    • One thing I thought of was moving the the whole timeframe up a month. I have no evidence to back it up, but I think there's be less activity in early December than early November, because people tend to get busy/spend less time online during the holidays. Moving it up 2 weeks would put it in conflict with Thanksgiving in the US, but a month would avoid it. ~ DC (Talk|Edits) 07:01, 21 December 2009 (UTC)[reply]
    • I agree with both DC and Risker. I know that my task, even when mostly ceremonial in nature, is one that I take seriously. I do my due diligence... I look at the history of the candidates, I read everything over, I ask for feedback, and I listen and read. I search my email archives. I don't know if all this helps - it hasn't ever led to anything useful yet, but I do take it seriously, and I need a touch more time after the results are announced before I can formalize them.--Jimbo Wales (talk) 23:51, 24 December 2009 (UTC)[reply]
      • Having it in the first few weeks of December also puts time for voting and reviewing right at the end of American and Western European (at the least) university semesters, quite a problem for us considering our userbase. I would suggest moving elections to either November or January. NW (Talk) 23:50, 22 December 2009 (UTC)[reply]
        • Very northern-hemisphere-oriented. January is dead in some places. Tony (talk) 02:59, 23 December 2009 (UTC)[reply]
          • Considering that is where a large portion of our userbase resides, that's not such an awful thing. I think both of your proposals below are good, though I would lengthen the voting period a bit longer than you have it (two weeks is fine, no need to shorten it). NW (Talk) 23:38, 24 December 2009 (UTC)[reply]

Proposal 1: earlier but same duration. There have been calls to hold the election earlier. This is the way it would look if moved forward by three weeks, retaining the duration of each activity, except that I've increased the timing of the preparation period and, of course, the scrutineering period. The other durations are the same as those used in ACE2009. All start times assume 00:01 UTC and all end times assuming 23:59 UTC.

  • 1–13 October [14 days]: preparation
  • 14–22 October [9]: call for general questions
  • 23 October – 6 November [15]: call for candidates, individual questions
  • 7–12 November [6]: fallow period (why so long?)
  • 13–27 November [15]: voting period
  • 28 November – 4 December [< 7]: scrutineering, certification and announcement of results
  • Total duration: 49 days

Proposal 2: earlier and shorter. ACE2009 took well in excess of 50 days, of which the wider community were involved in a period of 49 from the first public call for questions to the announcement of the tally. I suggest that to take almost two months of every 12 with the electoral process is unnecessary drama on a wiki. We can lessen the risk of election fatigue by planning for a slightly crisper pace for the public parts: what about a 32-day schedule, not including prep? This would make for a more effective event, holding people's interest with less of the drawn-out drama. This timetable would start one week earlier than the one at the top (equivalent timing and duration to ACE2009) and would finish almost three weeks earlier.

  • 19 October – 1 November [14 days]: preparation
  • 2–6 November [5]: call for general questions
  • 7–13 November [7]: call for candidates, individual questions
  • 14–17 November [3]: complete individual questions
  • 18–27 November [10]: voting period
  • 28 November – 4 December [< 7]: scrutineering, certification and announcement of results
  • Total duration: 32 days

Tony (talk) 11:41, 24 December 2009 (UTC)[reply]

Election personnel

edit

In the 2009 election, most of the on-wiki running of the election was done by self-appointed co-ordinators. Editing the interface and monitoring votes was handled by WMF-identified "election administrators", while "scrutineers" drawn from the ranks of stewards not active on the English Wikipedia monitored the integrity of the election and signed off on the results.

Is the current structure of coordinators, administrators, and scrutineers optimal?

Should the term "administrators" be changed to avoid confusion with site-wide admins?

Should the roles and responsibilities of each position be written out?

Should scrutineers continue to be entirely drawn from the ranks of non-en.WP stewards?

Discuss "Election personnel"

edit

Possible term for (election) administrators might be "officials". Then: coordinators, officials, scrutineers. Tony (talk) 12:15, 11 December 2009 (UTC)[reply]

I think the Election officials → { Election administrators, Election coordinators, Scrutineers } breakdown works. All three roles are essential, and the requirements for the admins and scrutineers are pretty inflexible. Happymelon 12:49, 18 December 2009 (UTC)[reply]

Yes, I think it would be of benefit to the community to write the roles out. If we are to continue with a secret ballot then simple acts like this greatly improve transparency, which itself engenders trust and confidence. "Election co-ordinator" and "Election administrator" work well for me. "Scrutineer" is a good word also but I am disappointed with the limited community involvement in the role.

For the purpose of the check user aspect of the role, I think the current way of selecting scrutineers (and of performing the role) is quite good. But that is a very limited a view on the role of an election "scruitineer" IMHO. For the purpose of garnering community satisfaction in the administration of the election (no offense, Happy-Melon), I think it falls short of the mark. I think there has to be some level of community involvement in scrutinising elections. (Many comparisons have been made that conducting ArbCom elections by secret ballot is more like a "real-life election". In this aspect - i.e. the ability or inability of the community to satisfy itself that the election was carried out fairly - the election process departs most starkly from "real-life" elections carried out by secret ballot, at least those carried out in democracies.)

There are many ways we could do this while maintaining the privacy policy and the secrecy of the ballot. Three such examples, ranging from an option that would maintain least secrecy of the ballot to one that would maintain it the most, are:

  • We could open the ballots completely (minus check-user type information) to the whole community after the scrutineers have performed their duties as present. The list could show votes stricken by the scrutineers marked in some way as having been discounted.
  • In parallel to the current scruitineers work, en.wiki community scruitineers (drawn from across the range of editors) could scrutinise on an anonymised ballot list showing the check-user data visible to the current scrutineer.
  • A selection of en.wiki users with check-user rights could be chosen through an open process in advance of the election to be act as scrutineers (even in addition to the current method of selecting scrutineer).

I'm not proposing these as suggestions (I'd prefer a variation on the above three combined) but rather simply as feedback here. More thought, I believe, should be put into the community confidence aspect of what is meant by scrutinising an election, as opposed to seeing the role in a limited sense of catching double-voting. --rannṗáirtí anaiṫnid (coṁrá) 00:23, 19 December 2009 (UTC)[reply]

I think the failure by the current scrutineers to spot an obvious discrepancy in the count on the part of the election administrators demonstrates the practical benefit of community scrutineers. It mirrors too the benefit of community-based scrutineers in commonly spotting similar error in "real world" elections conducted by secret ballot.
This is no personal slight against either the election administrators nor the scrutineers, it was human error on both parts. However, it is one easily spotted by community scrutineers, who have a greater degree of motivation to spot such trivial errors. Indeed, given even the dearth of opportunity the community had to scrutinise this election, it was a community member that spotted the obvious error despite it having evaded the election administrators and strutineers. --rannṗáirtí anaiṫnid (coṁrá) 09:44, 19 December 2009 (UTC)[reply]
Interesting points. The open-list-minus-checkuser-data is already published: Special:SecurePoll/list/80 shows you exactly the same data as it does me or the scrutineers, only without the checkuser data. You can see the current, old, and struck votes in exactly the same way. Anonymised data will probably not be possible; it's too difficult to anonymise properly, and the privacy policy protects a logged-in user's private data absolutely, not just in connection with them. I think the requirement that the election appoint scrutineers who are identified to the foundation and competent with checkuser, is essentially set in stone. I wouldn't have a problem with them being enwiki checkusers rather than stewards (or perhaps a mixture of both). The "opportunity for the community to scrutinise the election" is unlikely to change: the community already has access to all the data that can legitimately be published, there was just no enthusiasm to investigate it. Happymelon 10:42, 19 December 2009 (UTC)[reply]
Replied on the talk page to save space. --rannṗáirtí anaiṫnid (coṁrá) 12:05, 19 December 2009 (UTC)[reply]

The SecurePoll system

edit

The SecurePoll system was used for the first time in an ArbCom election in December 2009, following a discussion in the weeks leading up to the election. The community needs to decide whether it should be used in 2010.

There are some technical issues that invite attention:

Display previous votes? Should the developers be asked to change the software to display a voter's previous vote so that they do not need to vote in a single sitting, and do not have to start from scratch on returning to their ballot paper to make changes?

How to manage mid-election withdrawals: User:Secret withdrew half-way through, but it was not possible to remove his name from the list, and votes could still accrue for that candidate.

Minimise call on developers: Are there ways of minimising the necessity for developer input/set-up? For example, is it feasible for an election coordinator to upload the names of candidates? What would need to be done to the software, and would it require too much technical skill for non-developers to upload the names?

Are the actual votes going to be deleted after all this is over? (Or erased, purged, or whatever the technical term is.) The process requires that the record of how each person voted has to be kept so that a person can change their vote and so that invalid votes can be thrown out. But after the process is completed, checked, verified and so on, there wouldn't seem to be a need to keep them around. (Modified version of Neutron's question at the election talk page).

Advantages and disadvantages in destroying the data after the election process

Discuss "The SecurePoll system"

edit
  • I'm uncomfortable with the idea of being able to display your previous vote. In a real-world election, that would make it possible to sell one's vote by logging in and showing it to the buyer. It's quite possible that the abstraction provided by interaction over the Internet makes this less feasible, and it's certainly true that vote-buying on Wikipedia is improbable, but we should try to be a model for good elections. rspεεr (talk) 17:59, 7 December 2009 (UTC)[reply]
What, by taking a screenshot of it, emailing it to the potential "customer", promising to vote that way, and here's my bank account number to shovel money into and you can trust me to honour the screenshot? (Other voters cannot access one's personal ballot page.) This is, let us say, a fatally flawed business model! Tony (talk) 00:14, 8 December 2009 (UTC)[reply]
Screenshots and bank accounts are indeed a poor way of making such an exchange. The way you get this kind of privileged information, as developers of shady Web games know, is just to ask the user nicely to type their username and password into your form. In fact, now that you've got me thinking about it, the fact that the entire election takes place on the Internet makes this easier, not harder.
You don't even have to pay real money; many Facebook and Twitter users will "sell" their online identities for a game of Mafia Wars or the ability to upload photos from their phone. Imagine how cheap a single vote is in comparison. Again, it's not socially likely to happen in the current environment of Wikipedia, but it's technically quite straightforward. rspεεr (talk) 06:56, 8 December 2009 (UTC)[reply]
Rspeer, although the thought is chilling, I'd say the previous, low-tech manual "scroll down and vote" system was also vulnerable to vote trading. We can only do our best to avoid dishonesty whatever system is used. Tony (talk) 07:51, 8 December 2009 (UTC)[reply]
Yeah, it's probably not a huge concern, especially as we're coming off a system where you could basically "trade" your vote for an arbitrator liking you more. (I'm not claiming that this happened to any extent that would affect the elections.) rspεεr (talk) 08:34, 8 December 2009 (UTC)[reply]
Rspeer, I don't understand your point at all. Thanks to the public voting log it is already trivially easy to verifiably sell one's vote with a screencast. The concern of vote-selling is not a small concern but it's just unrelated to the proposal at hand. --JayHenry (talk) 20:28, 8 December 2009 (UTC)[reply]
  • On displaying a voter's previous vote: Yes, absolutely this should be done if it is technically feasible. (And we already know that the software is keeping a record of how each person voted; otherwise it would not know which vote to over-write when someone votes for a second time to change their vote.) It should be easier to change one's vote. I also don't think the software should necessarily reveal to the public who has changed their vote, as it did this year. It's really nobody's business. Neutron (talk) 00:06, 15 December 2009 (UTC)[reply]
    • There are disadvantages to a public record of when and how often someone changed their votes, but the advantage is that the person changing their votes sees a visual confirmation in the log that their vote did in fact change and get recorded. If the log only showed the fact that you had voted previously, you might think "did the new vote really get recorded?" Carcharoth (talk) 03:05, 15 December 2009 (UTC)[reply]
    • It should be possible to have a system where you don't have to vote for everyone at once and can easily change your vote without displaying previous votes. It could go something like this: If you haven't voted for a particular candidate before, the ballot would give you four options: Support, Neutral, Oppose, No Vote. If you have, there'd be a checkmark next to the candidate's name and four options: Support, Neutral, Oppose, Retain Previous Vote. Just an idea; might need better wording on that last option. --Chris Johnson (talk) 01:29, 16 December 2009 (UTC)[reply]
    • i'd definitely be in favour of a system where i don't have to do all the voting in one sitting and where i can see who i have/haven't cast a vote on yet. i'd also prefer a system that lets me know a changed vote has been duly recorded without publicizing that information. Sssoul (talk) 17:43, 16 December 2009 (UTC)[reply]
      • I was considering voting, but when I realized I need to do it all in one sitting, or remember all the decisions when I continue it, caused me to change my mind about it. This system may also cause people who usually edit from public computers to save their votes on a subpage (publicly visible to all), which defeats some of the purpose of a secret ballot. I think that the ability to see one's previous votes isn't that relevant, but one should easily be able to see who (s)he actually voted about(for or against), and be able to modify some without needing to remember the others. עוד מישהו Od Mishehu 08:44, 24 December 2009 (UTC)[reply]
  • On withdrawals: Although I don't feel terribly strongly about it, I think that once the voting starts, the ballot should remain stable. It's a secret ballot; nobody really knows how they (or anyone else) are doing, though they might think they know. A candidate may certainly "suspend" his/her "campaign", as is done in real-world elections, without changing the ballot. But I don't think the ballot should be altered once people have started voting. Neutron (talk) 00:06, 15 December 2009 (UTC)[reply]
    • I agree. Withdrawals in a secret ballot system are pointless, and in theory someone could withdraw after voting once in support of themselves, and hence end up with 100% support. Not that this would really work, of course. Carcharoth (talk) 03:07, 15 December 2009 (UTC)[reply]
    • I agree w/ Carch. We can simply adopt a system like most real-world elections: create some buffer prior to the start of the election where candidates may not be added or removed. We don't have the hassle of paper ballots and multiple polling places to necessitate that this time period be very long, but it should be non-zero. After that I don't see the benefit of removing names or the cost of retaining them. In the unlikely event that someone notionally "withdraws" during the election but is still elected, S/he can either resign or refuse the seat. Protonk (talk) 22:07, 15 December 2009 (UTC)[reply]
  • The move to secret ballot, which is used in most 'real-world' elections, seems to have so far significantly reduced drama, potential for distortion etc. kudos to getting it set-up and looks like it would be a good idea to keep on using it and expanding its use where suitable, Tom B (talk) 13:20, 15 December 2009 (UTC)[reply]
    If by drama you mean discussion about the election then it's certainly reduced that. I don't know on what observable information we'd conclude the process has reduced the potential for distortion, certainly when we've not seen the results yet. --JayHenry (talk) 03:00, 18 December 2009 (UTC)[reply]
  • It ought to be possible for the developers to set up the SecurePoll system so that a certain group of users (let's say for the sake of argument there's an ElectionAdmin flag that can do this) that can set up new elections and edit the list of candidates, deleting them if necessary in the case of withdrawals. They could also appoint scrutineers and vote counters if needed, I'm not sure how that part of the extension works. Hersfold (t/a/c) 05:00, 16 December 2009 (UTC)[reply]
  • I just want to go on record here as liking the way the way this was done. We still had conversation, user were still able to publicly state positions if they wished, but in the end you could make whatever decision you thought best without consideration of how you might be viewed by the candidates or others in the future. I'm sure this idea would be shouted down if I actually tried it, but I think this could also be new model for RFA. Beeblebrox (talk) 18:51, 16 December 2009 (UTC)[reply]
  • I was very glad that we switched to a secret ballot. I found the interface reasonable enough, although I seem to remember I wished there were more cross-links between the voting page and the candidates' pages. — Carl (CBM · talk) 12:18, 17 December 2009 (UTC)[reply]
  • Overall good. However would like it to be possible to be able to not have to vote all at once and be able to change your vote without having to revote on all the candidates. I did not want to go back and change my votes when it became clear there would be an extra place being elected as I could not remember all the previous votes I had made and there was no way I was going to spend the time to remake the decisions again. I hope the votes are destroyed after the election process is completed to ensure there is no way candidates will ever find out how each voter voted on them. Davewild (talk) 18:08, 17 December 2009 (UTC)[reply]
Privacy of voting at RfA? <Draws breathe sharply in astonishment>Beeblebrox, I was enjoying thinking I'd thought of that first. Tony (talk) 13:25, 18 December 2009 (UTC)[reply]
  • I'm curious about the order in which the candidates' names appeared on the ballot. I'm sure there's some logic (order of filing to run?) but think that alphabetizing it would make things a bit easier for many voters. (Apologies if this topic was already discussed elsewhere. I couldn't find it.) Rivertorch (talk) 06:47, 19 December 2009 (UTC)[reply]
    It was randomized for each voter, in order to be fair. rspεεr (talk) 06:52, 19 December 2009 (UTC)[reply]
    I'm not sure why an alphabetical arrangement wouldn't be fair. It seems both fair and feasible in many elections off-wiki, anyway. Personally, I found the discrepancy between the order of the pre-voting pages (e.g., candidate statement and questions pages were alphabetical) and the ballot page slightly disconcerting. It lengthened the time it took me to accurately cast my vote from (rough guess) 60–90 seconds to 5 minutes—not a huge hardship, by any means, but considering the hours required to wade through all of those pre-voting pages, an unexpected and unwelcome obstacle. Also worth a second thought next time around: increasing the line spacing between the rows on the ballot. Even with the (faint) border lines, aligning name with radio button was a little tricky, especially going all the way over to the Oppose column. Rivertorch (talk) 16:10, 19 December 2009 (UTC)[reply]
    Scientific study has demonstrated conclusively ([1][2]) that position on the ballot paper has a tangible, measurable, and serious impact on the allocation of votes in real elections. Candidates listed at the top of the ballot paper receive, on average, 2.5% higher support than those listed at the bottom. Perhaps that effect is less pronounced here, it doesn't matter: it is a known source of bias in ballots that can be completely countered by randomisation. It is, as you say, "not a huge hardship". Happymelon 16:20, 19 December 2009 (UTC)[reply]
    Good point. I am aware of such studies. I'm not sure that it doesn't matter: the rate of participation in Wikipedia elections is so low that I'd expect voters here are unusually well informed and have more reasoned opinions on the candidates than typical "real world" voters do. Without conducting a study specific to Wikipedia (a difficult proposition, to say the least), we cannot know whether ballot position makes a difference here. To be on the safe side, randomization is probably the judicious way to go. I'd suggest a brief note on future ballot pages stating that the order is random and why. And increased leading. Rivertorch (talk) 18:43, 19 December 2009 (UTC)[reply]
    Strictly speaking, randomization doesn't counter the effect but distributes it evenly on every candidate so that the net effect on relative rankings is neutral. It still has the observable effect of, on average, bringing support percentages closer to the median. — Coren (talk) 18:52, 19 December 2009 (UTC)[reply]
  • Build in a period of at least a week for the scrutineers to do their job without community pressure to rush out results. Remind the scrutineers to do some basic sum-checks beforehand to make sure everything adds up. And set up some sort of competition or content drive (maybe on articles on election systems) to distract those who are hanging around waiting for the results and engaging in stand-up comedy... :-) Carcharoth (talk) 16:28, 19 December 2009 (UTC)[reply]
  • I'm ambivalent on whether voting data should be destroyed, but if it is, the voters should be notified when they vote that the voting data will be destroyed at a later date (and when), and sent a receipt (encrypted if needed) telling them how their vote was recorded. What should be destroyed is the IP data that is collected during voting. Carcharoth (talk) 16:38, 19 December 2009 (UTC)[reply]
Carcharoth, as an election coordinator I almost felt it my duty to entertain the crowd with tap-dancing, stand-up jokes, etc. This should probably be written into the duties. Tony (talk) 09:24, 24 December 2009 (UTC)[reply]
Such notification can be given before next year's election, but for this year, I think the votes should simply be deleted. I think that when people were told the election would be "secret", that meant that their votes could never be revealed in any way, through any means. As it turns out, at the moment, that is not really the case. The data is sitting on a computer and it could be leaked, accidentally disclosed, or otherwise made known. At the point where there is no longer any need to be able to invalidate votes (which I would define as being anytime after Jimbo announces the appointments), I think that who each person voted for should be made completely "secret", by deleting that data. Neutron (talk) 21:16, 19 December 2009 (UTC)[reply]

Improving instructions to voters and voting rules

edit

Voter eligibility. Is there a need for explicit rules about blocked and banned users? For example:

  1. Voting is not permitted for the duration of a block or a ban.
  2. The use of a multiple account to cast more than one vote or to evade a block or a ban by voting will result in the voiding of all votes by the user in that election.

Voting system used. Voters should be given information about the voting system used.

Discuss "Improving instructions / voting rules"

edit
  • The page Wikipedia:Arbitration_Committee_Elections_December_2009 describes the voting system only with the words "with the traditional options of support/abstain/oppose". It is not clear to me what that means; how will the winners actually be determined? Are we sorting on the difference #support-#oppose, or on the ratio #support/#oppose? If all I want is one particular candidate to succeed and I don't care about all the others, is the strategically correct approach to vote "oppose" for all of them, or can I "abstain"? AxelBoldt (talk) 18:30, 9 December 2009 (UTC)[reply]
The relevant metric is (unchanged from previous years) the support percentage, calculated as supports/(supports+opposes). Abstain/neutral has no effect. So a candidate with 8 supports, 67 neutrals and 2 opposes will have a "score" of 80%. See last years results. Regarding strategy, a rational voter will not use the abstain/neutral option.  Skomorokh  18:35, 9 December 2009 (UTC)[reply]
Thank you, that clarifies it. I think all voters should be given this information; they can't be expected to hunt for old election results in order to deduce the tallying method, nor do they know whether the same method is still being used. Specifying the tallying method ahead of time would also avoid the possibility of nasty disputes after the fact. Cheers, AxelBoldt (talk) 22:15, 10 December 2009 (UTC)[reply]
"Regarding strategy, a rational voter will not use the abstain/neutral option." I don't think that holds. Strategically, I want my opposes to count, as much as I want my supports to count - for entirely rational reasons I'm not going to invest unlimited additional time to form an opinion on candidates who I am 'neutral' about after I've researched them fairly. Were I to randomly vote my neutral candidates 50% support and 50% oppose, then I risk bumping out (or in) a candidate I support (or oppose). ‒ Jaymax✍ 01:55, 19 December 2009 (UTC)[reply]
(I'm going into voting geek mode, not expressing an opinion on our elections.) But you also risk bumping in or out a candidate when you only make your vote for or against them count half (ish) as much against your neutral candidates. Voting neutral rarely improves the expected value of your vote to you (see Range voting#Strategy), it only decreases the variance. rspεεr (talk) 07:15, 19 December 2009 (UTC)[reply]
  • As to the other issue, there is little point in telling blocked or banned users that they can't vote. If they have a sleeper sock so well put together that it fulfills the requirements to vote and has not been detected, telling them not to vote is extremely unlikely to have any effect. Beeblebrox (talk) 18:55, 16 December 2009 (UTC)[reply]

Supplementary voting

edit

In the 2009 election, the franchise was given to accounts rather than editors, forbidding some eligible editors from voting using the SecurePoll system. It was proposed that "postal vote"-style supplementary voting should be enacted to facilitate these voters; ongoing discussion for the 2009 election is here.

Can some hypothetical examples of these disenfranchised editors be given for context, are they editing outside the main namespace, editing under IP addresses(!?), or something else? ‒ Jaymax✍ 01:43, 19 December 2009 (UTC)[reply]

Discuss "Supplementary voting"

edit

Postal voting is an exceptionally poor idea open to incredible amounts of abuse with almost no legitimate use whatsoever. This has not been thought through at all. --JayHenry (talk) 20:40, 8 December 2009 (UTC)[reply]

The question seems to be more about users in good standing who have sufficient votes but spread across multiple accounts. The automated system doesn't recognize them.
Maybe a simple "fix" would be that such users can approach scrutineers and if it's agreed they have franchise, the scrutineers can add their primary (nominated) account to SecurePoll as an exception in the election. This would mean SecurePoll would need a minor upgrade to allow an automatic rule, plus a manual whitelist. But it would solve the problem completely. FT2 (Talk | email) 12:24, 17 December 2009 (UTC)[reply]
You are still describing something with almost no legitimate use whatsoever, although certainly an improvement of the postal voting idea. There are lots of totally legitimate users, interested in ArbCom elections, who do not have a single account with 150 mainspace edits, do have that spread across multiple accounts, and there's no impropriety at all? Come on, you're more familiar with Wikipedia than that. This is opening the door to abuse for no good reason at all. Someone without 150 real edits on a single account can stick to one account and wait until next year. --JayHenry (talk) 03:04, 18 December 2009 (UTC)[reply]
The main concern is active contributors who RTV for a legitimate reason close to the election.
If a supplementary voting system is ever added, the election officials will need to ensure they only accept votes from people who have a legitimate reason for not being able to vote using their old account.
I can agree that this is hardly worth the extra voters, and the cost is a lot of extra effort to prevent abuse.
John Vandenberg (chat) 02:44, 19 December 2009 (UTC)[reply]
I believe the Right To Vanish is the right to leave. Not the right to start over secretly and receive secret suffrage with nobody except a few scrutineers having any clue who you are or why you have suffrage (and being able to evaluate whether your reason is "legitimate"). Misusing "RTV" to cover the latter case unfortunately has trivialized the meaning and deference we ought to show people who leave. In practice, if you are part of the in crowd you can "exercise your right to vanish" as a way to erase all your past negative behavior while maintaining the benefits of your tenure. (I am aware of at least one instance where an editor "RTVed" and was given adminship on their new account; it's hard to think of a more abusive and outrageous use of RTV.) If you are not part of the in crowd, little respect is shown to your desire to extricate yourself from the project. We should weaken rather than strengthen this dynamic. The community must mature beyond that cloak-and-dagger nonsense which is prone to cliquishness and outright abuse. With all due respect to (the incredibly minuscule number of) people who truly are forced to "start over", if this happens shortly before an election, you can wait until next year. --JayHenry (talk) 03:46, 21 December 2009 (UTC)[reply]

Questions to candidates

edit

There has been unresolved debate on measures to bring the size and number of both "general" and "individual" question under control (examples: 1, 2, 3). Should this be done?

Possible measures include:

  1. the requirement of a seconder for each question;
  2. a limit of one question per user (properly observed and enforced);
  3. a word-length limit per user;
  4. an absolute limit on "individual" questions per candidate;
  5. a proscription on the mass pasting of the same "individual" question(s) on more than one candidate's page.
  6. closer editorial control over the questions.

Discuss "Questions to candidates"

edit
  • I have to agree that the number of questions is simply staggering here. I can't say with any certainty what the "best" resolution to this is, although I would suggest that limiting each user to 1 question could improve things. Perhaps if there were a group (not unlike the group that has worked here to co-ordinate things and oversee the results) which looked at the questions and simply picked out 15 or 20 of the best for each candidate it would assist the process. I found some of the questions to be a bit ... ummm ... let's just say perhaps asking the candidates on their talk page might have been a better choice for a few of them. Even Arbs have real lives, and I can sympathize with the sheer multitude of work that went into each and every candidates page. Surely we can find a way to tone it all down a notch for next year. — Ched :  ?  15:05, 8 December 2009 (UTC)[reply]
  • From a voter's standpoint I would say that the huge volume of questions made the question-and-answer process mostly worthless. I tried to wade through all of it, but there was no way I could. I also think that a lot of questions are a little too "inside baseball." Maybe a little less focus on the political and interpersonal disputes that have taken place within Wikipedia would make the election a little more "accessible" to the "rank and file" who are more tuned-in to editing articles than some of the other stuff that goes on around here. Neutron (talk) 00:14, 15 December 2009 (UTC)[reply]
  • Rather than questions, make candidates draft an arbitration case (or pick some other arbitration activity if they don't particularly like the idea of drafting a case). Or some other form of practical test, such as analysing various forms of evidence. More practical stuff and less wordy stuff. Carcharoth (talk) 03:09, 15 December 2009 (UTC)[reply]
  • I think voters must be allowed to ask questions. One method by which things could be somewhat improved, as I suggested last year, is for the "General questions" section to be formatted in such a way that each question is put only once, and the candidates list their respective responses one after the other. That would also have the advantage of allowing voters to compare the answers more readily. Gatoclass (talk) 08:14, 15 December 2009 (UTC)[reply]
  • If the single question rule is to remain the wording on the general questions page desperately needs to be revised. Both this year and last year it said on the general question page This questions page may be used to pose a single question to every candidate in this year's election. Does this mean a single question per candidate? A single question being asked only once, so no duplicates? It is clear that a few users did not understand it actually meant a single question per person, as shown on the discussion pages and by the fact it has been ignored by many two years running. Personally I think such a rule, even if clear and enforced, is unhelpful as there is not limit to the amount of question people can ask candidates directly. I did like the idea of general questions being submitted, being re-organised into question categories with duplicate merged, and then submitted to the candidate. This could also work well with the Board Election like set-up of questions being listed and each candidate putting their individial response underneath each one, as suggested by Gatoclass. Camaron · Christopher · talk 14:42, 15 December 2009 (UTC)[reply]
  • I agree with the idea of restricting users to one question. To be honest I only glanced at the questions and answers because it was just too much to read over and I don't think that I would be the only one, so I think that it is self defeating allowing so many questions and answers.--Literaturegeek | T@1k? 14:45, 15 December 2009 (UTC)[reply]
    Well you don't have to read them all. You can just pick out the questions from users you trust, for example. Gatoclass (talk) 16:05, 15 December 2009 (UTC)[reply]
  • I think questions suffer from the many-to-many networking realities at hand. RfA's are somewhat manageable (with an obvious and notable exception) mostly because the import and scope of the "office" are such that they attract <200 participants (usually), in english, they are "no big deal". By contrast we have made Arbcom into a very big deal (relatively speaking) and we have many more users participating in the election. Enough that I didn't bother reading the questions or answers (general or otherwise). I'd be interested to learn how many users read all the questions, or even a preponderance of them. Somewhat cynically, I suspect we won't throttle or control questions in the next election, so discussion of the merits may be fruitless. I do note one pleasant outcome of the secret voting system: it substantially reduces the ability of a voter to grandstand after seeing his/her pet question go unanswered. Perhaps we will find ourselves at an equilibrium with candidates refusing to answer late or redundant questions. Protonk (talk) 22:02, 15 December 2009 (UTC)[reply]
  • As a candidate, I don't think the number of questions was terribly excessive; I was able to answer the vast majority of mine in the space of a few days. However, I did join in rather late and I believe got fewer questions than many as a result. Hersfold (t/a/c) 05:04, 16 December 2009 (UTC)[reply]
  • The volume and variety of the questions serve as an introduction to the firehose you will face if appointed. Just as there is no way to fully do what is required as an arbitrator, there is no way to comprehensively answer all the questions adequately. It does no harm to get a peek at the over-whelming situation you will face. Fred Talk 19:13, 16 December 2009 (UTC)[reply]
  • Per Fred. But also, a lot of questions were off-topic on what Arbitrators do. If current and ex-arbs described the role better for next year, and an indication of useful and likely less useful areas were given, then questions might be more salient. FT2 (Talk | email) 12:27, 17 December 2009 (UTC)[reply]
I'm not so concerned about the candidates (although the idea of perpetuating the "trial of hell" aspect just because I myself survived it doesn't seem like a good reason). No, the real concern for me is the voters: they want to make their judgements in under 10 hours of reading, thanks. Several people have already told me they didn't vote because it's too hard to wade throught the text. Of course, in previous years we just acted like a herd and voted as we saw our friends did; fortunately, that is no longer possible. Bringing the question process under control so that it serves both voters and candidates better is a prime consideration for next year, in my view. Tony (talk) 12:38, 17 December 2009 (UTC)[reply]
  • I did wade through most of the questions, skimming past questions I was not interested in. It took about 4 hours for me to go through them and make my decisions. I think a limit of one question each is too strict but would support having an enforced limit of some sort as some people did ask a lot of questions (and there were quite a few I paid very little attention to). Perhaps something like each eligable voter could ask 2 questions, with any further questions from that person requiring another 4 people to endorse the question being asked. This would have to apply to both general and individual questions as the individual question part just became an extension of the general questions this year. Davewild (talk) 17:58, 17 December 2009 (UTC)[reply]
I would suggest a two question limit, with parts, subparts, and the word "and" being counted towards the limit. Questions violating the limit to be removed not by the candidate (why should he take the heat?) but by an election official. In addition, perhaps it is time to look at the election guides, some of which seem very idiosynchratic and based on personal preferences having little to do with ArbCom. Perhaps it would be a good idea if there was a standard they had to meet if they wanted their page in the election template Just a thought.--Wehwalt (talk) 18:59, 17 December 2009 (UTC)[reply]
Two question limit? That's a singularly bad idea. Standard to meet for an election guide? Even worse idea. ++Lar: t/c 23:10, 21 December 2009 (UTC)[reply]
  • Non-controversial suggestion: Collapsed (show/hide) answers of all the candidates under each general question in one page to easily compare the answers. Sole Soul (talk) 08:06, 19 December 2009 (UTC)[reply]
    • Easy for voters to compare, but I'd rather each candidate took a fresh, original approach to responding to the questions rather than reading their colleagues' responses just above. Tony (talk) 13:18, 19 December 2009 (UTC)[reply]
      • I don't think that where you put the answers influences the answers. Or it shouldn't. A candidate who wants to crib from everyone else's can open multiple tabs. But putting the answers together makes it easier for voters to compare. Given how much you said you are concerned about how hard it is for voters, I find it surprising that you're opposed to this idea. ++Lar: t/c 23:10, 21 December 2009 (UTC)[reply]
      • If that's the case then it's better to know before they read each other's responses in an ArbCom case. P.S: I noticed a lot of similarities between the answers to Majorly's no. 1c question. Sole Soul (talk) 13:08, 22 December 2009 (UTC)[reply]
  • Way too many questions. There were over 400 question/answers for people to sort through (21 general questions *23 candidates; this doesn't include individual questions). I hate to admit it, but I didn't even try to skim through them, instead I looked at the candidate's statements. In the future, I think we should have a limit of five to ten general questions (or get rid of general questions all together). ~ DC (Talk|Edits) 04:58, 21 December 2009 (UTC)[reply]
  • Perhaps there should be a set period during which asking general questions is permitted, and during this period, collaborative editing to refine off-the-point or similar questions should be encouraged, perhaps with a target of producing five to ten key questions - a kind of compositing utilising the wiki software. Warofdreams talk 04:08, 23 December 2009 (UTC)[reply]
  • There certainly needs to be a cut-down of the questioning. It's all kinds of silly at the moment. Stifle (talk) 10:25, 15 April 2010 (UTC)[reply]

Feedback from the candidates

edit

The 23 candidates in the 2009 elections are invited to give feedback and make suggestions in this section.

  • This was a relatively painless and mostly drama-free election. Arguably, some of that can be attributed to the Committee's performance during 2009 which, by most accounts, was much less controversial than the previous year and thus less likely to arouse passions.

    I'm a bit disappointed by the paucity of direct discussion, but perhaps that is a normal side effect of editors being able to keep their votes to the privacy of the booth; so there can be no badgering or "why did you vote this way, you fool!" effect. — Coren (talk) 00:29, 15 December 2009 (UTC)[reply]

  • Agreed, secret ballot worked well. It might not be a bad idea to shorten the election to a week, voting fell off dramatically after the first few days. I may have some comments on the amount of time it is taking to announce the results, but am content to let the scrutineers do their thing for now.--Wehwalt (talk) 18:54, 17 December 2009 (UTC)[reply]
Wehwalt, one week of that sounds just great from where I was standing (it was a lot of work). But I think the complaint would be that seven days only is two short a window for those who might be on vacation or distracted or otherwise unavailable. Compromise 10 days? Tony (talk) 09:29, 24 December 2009 (UTC)[reply]
  • Dislike secret ballot. Dislike unanticipated delays in announcement of result (schedule says 14 December 2010: close of voting... 15 December 2010: certification and announcement of the results). This wasn't expected, something has gone wrong, and no-one is saying what William M. Connolley (talk) 11:13, 18 December 2009 (UTC)[reply]
    I don't go so easily with a conspiracy theory: it looks intead that for the first time the votes/voters will be checked properly for validity. Tony (talk) 11:46, 18 December 2009 (UTC)[reply]
    The 'overrun' compared to the timetable reflects a hopeless optimism on the part of the timetable, not any problem with the process. Compare the date of announcement in previous elections: 20 December in 2008, 26 December in 2007 (but with voting closing on 16 December), 26 December in 2006 (with voting closing on 17 December). The average time before the nominations are announced is thus around 8 days. Given the limited data that Jimbo will have on which to base his nominations, that stage of the process should be extremely quick, perhaps a day at most to account for timezone dislocations. So compared to other elections, we are still well ahead of schedule. The only difference is that, instead of everyone waiting with baited breath to see how Jimbo interprets the results, everyone waits with baited breath to see what the results actually are. Happymelon 12:24, 18 December 2009 (UTC)[reply]
    What went "wrong" is that our insistence on neutral and distant scrutineers had the unfortunate, but predicable consequence that they were unfamiliar with enwiki jargon, history, community norms, and so on. They were, quite frankly, and with no disrespect intended, flying blind. Checkuser is not magic pixie dust because all it reveals is technical data - technical data does not say. "Hi! I'm a sock of a banned user." Rather, it works the other way around, you can follow hunches and technical data can buttress or weaken them. Delays are the inevitable result of unfamiliarity.--Tznkai (talk) 20:55, 19 December 2009 (UTC)[reply]

Feedback from election volunteers

edit

The volunteer co-ordinators, scrutineers, and election administrators of the 2009 elections, and Tim Starling, who prepared the software, are invited to give feedback and make suggestions in this section.

Things seem to have gone well, overall. Many of the hiccups we had (like the start and end times, period for questions, etc) were artifacts of the Open Voting system used in previous years - which might be my fault, a bit, as I started setting up election pages under last year's open voting system prior to the RFC. Next year will be vastly simpler if we know we're using one system from the beginning. Congrats also to the volunteers who pitched in for ACE2009 - well done. UltraExactZZ Said ~ Did 13:20, 15 December 2009 (UTC)[reply]

Some comments on SecurePoll, may have thoughts on the rest.

  • SecurePoll needs to be documented here
  • SecurePoll needs to log tally generations.
  • Consider obscuring enhanced list data for scrutineers/election admins unless it is deliberately accessed, and again, logged.

--Tznkai (talk) 21:51, 15 December 2009 (UTC)[reply]

Scrutineering needs to be explicitly privileged. SecurePoll doesn't leave voters very much on-wiki privacy; voters should be reassured that, unless electoral fraud is involved, any skeletons that are found in their closets will be left in the closet by the scrutineers. Happymelon 12:29, 18 December 2009 (UTC)[reply]

SecurePoll needs:

  • Proper logging of administrative actions
  • An API interface
  • Better filters and sorting on the list page
  • A CAPTCHA on the vote page
  • Finite lifetime on the checkuser data
  • A configuration interface
  • Optional memory on the vote page
  • Optional vote receipt

Should be enough to be getting on with :D Happymelon 10:46, 19 December 2009 (UTC)[reply]

Let's not leave it until October to ask Tim to do these things. I wonder whether a committee of the coordinators plus administrators plus anyone else from the community should be convened in the new year to liaise with the developers on these matters. :-) Tony (talk) 13:14, 19 December 2009 (UTC)[reply]

Requests from the scrutineers:

  • Checkuser log extract on /details page.
  • A clearer checklist of things to do to complete the scrutineering.

Happymelon 13:22, 24 December 2009 (UTC)[reply]

Nice work, HM; this is extremely valuable feedback straight from the horses' mouths. Tony (talk) 15:28, 24 December 2009 (UTC)[reply]

There needs to be a little more of an explicit process on how to do administrative things, especially once the election starts. For example, if a change to an interface message needs to be made, such as the number of open seats changing, a candidate withdrawing, etc. there was no set process for how to go about requesting and implementing changes, it was just done on an ad-hoc basis on the talk page. Logging of message changes will help, but there really should be an explicit process for requesting, discussing, and improving administrative tasks. Mr.Z-man 22:18, 26 December 2009 (UTC)[reply]

Feedback on the organization of the election

edit

(New section because it doesn't seem to fit in an existing one.) For next year, we should remember to:

Cenarium (talk) 16:26, 17 December 2009 (UTC)[reply]

Just a few odd notes from me:

  • Start/end times should be expressed 00:01 and 23:59 to avoid confusion.
  • Check next time that SecurePoll start/end times are correct.
  • Make it very clear when the periods for inserting questions end.
  • Mention banned users' status under "Voter eligibility". Tony (talk) 12:10, 11 December 2009 (UTC)[reply]
  • Clarify somewhere in the instructions to voters that they should not leave it until the last minute to vote, lest there be software lag. Tony (talk) 07:11, 16 December 2009 (UTC)[reply]

Feedback on the statistical method

edit

I’m guessing this has been thrashed out in years gone by, but I think the ranking done by (S/(S+O)) is flawed. This is especially so when neutral voting is an available alternative on the ballot sheet. For a Reducio ad absurdum example, take a candidate who is so ‘meh’ they get only neutral votes, this should put them right in the middle of the field, but instead they will divide by zero, perhaps creating an eddy in the space-time continuum. Alternatively, someone who gets a few votes of support (from their mates, say), and otherwise neutral votes, will top the field at 100%. Likewise, a largely unknown, average candidate who managed to annoy some special-interest-group will appear less popular than they actually are. Discounting the neutral votes pushes the candidates with more neutral votes outward from the centre of the field. To be sure, under real-world conditions, this effect gets largely smoothed over, but not entirely.

I argue that the method for ranking should be ((S-O)/T), taking account of those who make a decision to come along on ‘polling day’ to vote, and actively tick the neutral box on their ballot for a portion of the candidates. This year, the effect is apparent in a few places, but I will take places 8 and 9 as an example. The candidate ranked 9th has 44 more supports and 31 more opposes that the candidate ranked 8th. The candidate ranked 9th has 0.2% less support under (S/(S+O)) but would have 1.3% more support under ((S-O)/T). ‒ Jaymax✍ 01:16, 19 December 2009 (UTC)[reply]

Just to note I can't even remember how I voted for those two candidates, much less have any connection or bias towards either of them ‒ Jaymax✍
Also to note that, only because candidate votes are balloted together, and there is no difference between a non-vote and a 'neutral' (whereas they should be distinct as they were with non-secret-voting), this equates to the same thing as the net-vote result. ‒ Jaymax✍ 03:55, 19 December 2009 (UTC)[reply]

IMHO, the steward election process with instant run-off balloting is the appropriate way to go. ScienceApologist (talk) 23:34, 19 December 2009 (UTC)[reply]

I assume you're being imprecise when you say "instant runoff". Instant runoff is a single-winner method. It can be generalized into a proportional representation method called the single transferable vote, which would be good in a real political election, but (as I have argued elsewhere) I don't think we want to elect an ArbCom that is a microcosm of Wikipedia and all its disagreements.
The steward election process used the Schulze method, not IRV. The Schulze method with multiple winners, as well as the range voting method we use now, are both reasonable majoritarian methods of electing multiple winners. I don't have a reason to prefer one over the other, but I would not oppose a move to the Schulze method for next year. rspεεr (talk) 18:16, 20 December 2009 (UTC)[reply]
I definitely preferred certain candidates over other candidates in more than a three-tiered system. That's why I think that the Schulze method is better. ScienceApologist (talk) 23:48, 20 December 2009 (UTC)[reply]
Um, I think you both might be talking about the Board election. The steward elections in recent years have used a metric of a support/oppose ratio of at least 80% with at least 30 supporting users. Note that this is a bar to pass and not a ranking, as in this scheme every candidate can be elected, or no candidates at all can be elected (if none can muster the 80% and 30 support)... there is no certain way to predict in advance how many will be. This works well for Stewards which don't have a fixed number, but may not work as well for a committee in which one desires to set the number of seats in advance. (that's not to say that it might not be a good method to use... why have fixed numbers of arbitrators after all???? the number of arbitrators over time is a sawtooth function anyway) ++Lar: t/c 23:23, 21 December 2009 (UTC)[reply]
  • I think a better way to vote is to use the plurality system. People could vote for as many candidates as there we're open seats. This makes more sense to me at least. ~ DC (Talk|Edits) 05:06, 21 December 2009 (UTC)[reply]
    • DC, interesting: whom would that favour and disfavour? Tony (talk) 05:37, 21 December 2009 (UTC)[reply]
      • I'm not sure if it would favor anyone over anyone else. But it would eliminate an anomaly where a candidate who gets less support votes wins because they get less oppose (ie, KnightLago winning this year, despite having less support votes than 5 unseated candidates). ~ DC (Talk|Edits) 05:52, 21 December 2009 (UTC)Note:I'm not contesting his election this year, since he won according to the rules used for the election. He's simply an example.[reply]
        • Given that oppose votes are intended to mean something, why is that an anomaly? rspεεr (talk) 08:04, 21 December 2009 (UTC)[reply]
          • If I had known that it would come to the result it did, I'd have opposed everyone I was neutral on just to make sure that the people I supported were more likely to succeed. Alas, that strategic voting didn't occur to me until after voting had ended, but if we have this system next year you can be sure I'll be opposing far more candidates - even if I don't have a reason. ScienceApologist (talk) 13:21, 21 December 2009 (UTC)[reply]
    • That's actually called "plurality at large". It has few desirable properties. What reason would there be to restrict the votes like that? rspεεr (talk) 08:04, 21 December 2009 (UTC)[reply]
      • II would be against any move to only have support votes. It leaves the strong possibility of candidates who would have more opposition than support getting elected. For instance in this years election one candidate who got 78 more oppose votes than support votes would have been only one place from being elected based on support votes only. The ability to prevent candidates who have strong broad opposition from being elected is a strength of this system. (my preference remains to change to a system where candidates can be ranked but thats another argument) Davewild (talk) 08:12, 21 December 2009 (UTC)[reply]
        Concur - oppose votes should negate support votes 1:1. That is in the spirit of consensus. A divisive candidate who gets twice as many supports as anyone else, but also twice as many opposes, is not going to be as good an arbiter as someone who gets only half as many supports, but no-one opposes. ‒ Jaymax✍ 08:44, 21 December 2009 (UTC)[reply]
        • That seems to ignore the fact that the current system rewards, then, people for opposing just for the sake of opposing in order to make the preferred candidates look more appealing in the end. ScienceApologist (talk) 13:23, 21 December 2009 (UTC)[reply]
          • The only effect that would have, SA, is to change everything to (in effect) a simple support vote tally (which, interestingly enough, would have given nearly the same result this year— as far as I can tell only the ninth place would have switched) and depress the apparent support percentage even more (in proportion, in fact, to the number of candidates which is downright perverse).

            The secondary effect is that people would be left with no way to really oppose someone they feel would be bad for ArbCom. Those are all arguments in favor of switch to a preferential voting system, really. — Coren (talk) 13:52, 21 December 2009 (UTC)[reply]

            • You are assuming, Coren, that everybody votes strategically. However, we know that's not going to happen. More likely that we actually experienced some strategic voting in this cycle and it's almost impossible to say what its effect was. I used "neutral" as a viable option, but I bet there were some who didn't. Their support votes, then, ended up being more powerful than mine. Definitely perverse. ScienceApologist (talk) 14:48, 21 December 2009 (UTC)[reply]
              • At the risk of repeating myself again redundantly another time once more: this just highlights the importance of moving to a sane (preferential) voting system.  :-) — Coren (talk) 22:44, 21 December 2009 (UTC)[reply]
                • Support/neutral/oppose is neither novel for Wikipedia nor is the mathematics of the best "strategic voting" tactics difficult to figure out. Although the Schulze method has been used for some meta elections, it's a very inscrutable system on first blush. Certainly people who understand the system will be able to lodge "more powerful" votes than those who don't. Thus the definite perversion is preserved. --JayHenry (talk) 02:48, 23 December 2009 (UTC)[reply]
                  • There's a bit of FUD there that I'd like to quash. The only sane way to cast a "more powerful" vote than others in a Schulze election is to rank more people, an option that is available to everyone. This is all a voter needs to know in order to cast the most powerful vote they can.
                    To go into some details: yes, it is true that opportunities to strategically flip some of your preferences must exist somewhere in any preferential voting system -- but the only way to do this in Schulze is to exploit a Condorcet cycle. Doing so requires successfully predicting how everyone is going to vote on the whole, being lucky enough for it to be one of the 5% or so of cases where such a cycle is even possible, and figuring out how to make the cycle actually help or hurt the candidate you want, a chaotic system that grows more complex with the number of candidates and which you may not even be able to alter with your single vote anyway.
                    The analogue in a non-Condorcet-based election, which is only a bit less probable, would be to deliberately create a tie, and while they flip a coin to pick the winner you turn on a nearby fan to make the coin land on the side you want. I assure you, nobody is going to do this. The option of just voting what you mean works perfectly fine, and there aren't going to be people with advanced degrees in game theory conspiring to manipulate the election. rspεεr (talk) 09:30, 23 December 2009 (UTC)[reply]
                    • But indeed, that's really the only way of casting a more powerful vote in the current system. Ranking more people, which, in our current system, means casting opposes instead of neutrals. Guessing at the margins. My point is that both Schulze and our current system are open to some trivial level of "strategic voting" and we shouldn't be making our decision off this criteria; Schulze, however, has the added defect of inscrutability. Look, FUD? Imagine if you had studied classics instead of computer science and someone told you the outcome of the vote would be determined by this. It's not about FUD. It's about: why would we use a voting system that most of our voters aren't going to be able to comprehend? If people saw the votes in Schulze they wouldn't even be able to determine the outcome themselves. I don't think you appreciate how this could undermine the legitimacy of the whole process. --JayHenry (talk) 14:16, 23 December 2009 (UTC)[reply]
  • Endorse having neutrals count for something. Support/Neutral/Oppose/Abstain with default=abstain is a good way to do this. A neutral would count as as half a vote for and half a vote against, pulling the total towards 50%. Or to put it another way, support%=(S+N/2)/(S+N+O). davidwr/(talk)/(contribs)/(e-mail) 23:14, 22 December 2009 (UTC)[reply]

Feedback from ordinary voters

edit

My experience is that is was incredibly hard to make a reasonable vote without taking an age. Perhaps this explains why so many people ended up casting "neutral" votes. Wikipedia:Too long; didn't read springs to mind! Although the voting guides that various people produced were very useful, it took a while to find them and they were highly subjective and incomplete. It was very hard to even make up my mind as to what criteria I should be using! Should it be based on experience (number of edits, use of admin tools, other rights etc.), quality of judgement (how could I see?), stance on various issues (where is it shown?), a combination of these? All in all, I left with a feeling it was a very unsatisfactory election. Providing better, more focussed, information would enhance the election considerably. AndrewRT(Talk) 22:56, 20 December 2009 (UTC)[reply]

I disagree. There were a few candidates I unequivocally support because I have seen them in action as admins, a few who I would unequivocally oppose, and the balance took some time to read around and make up my mind (which was neutral in some cases). I don't think that's a bad thing, the arbitration committee has a high profile role so it is not a problem if you have to think about it. There's always the ability to go back and change your vote anyway. Overall this was a much more satisfactory approach than the old method where votes were not secret so there were issues of pile-on and poisoning the well in vote comments. I'm sure it could be improved further, but reducing the amount of information about the candidates does not sound to me like an improvement. I would always recommend a review of contributions before making up your mind anyway. Guy (Help!) 16:46, 21 December 2009 (UTC)[reply]
  • I find this election incredibly problematic, although I appreciate we can't really counter my flaws, since it would mean reducing ArbCom's size. I find it worrying that we have "top 8" candidates out there who wouldn't even pass RfA with those support percentages, with no offence meant to said candidates. While they can be said to have community acceptance, it's hardly a strong one in some cases. I worry that we've potentially exhausted the mine of well-liked and supported candidates, and we risk appointing arbitrators which have nowhere near universal support. Ironholds (talk) 01:50, 22 December 2009 (UTC)[reply]
  • Bear in mind, people's standards for support for ARBCOM are probably much higher than for admins. For RFA, someone's attitude might be "they seem to know what they are doing, they won't break the wiki," for ARBCOM they may want to see that the candidate has a much deeper understanding and level of experience with tasks like mediation, uncovering and understanding facts, balancing all issues for the common good, deliberative thinking, etc. davidwr/(talk)/(contribs)/(e-mail) 23:18, 22 December 2009 (UTC)[reply]
Introduce SecurePoll to RfA and the percentage boundary for passing would need to be adjusted: many more voters will cast an Oppose than otherwise. There would need to be a careful trial period, with crats willing to feel their way towards something lower than the current 70–80% (possibly 60–70%). Look at the graph of the ArbCom oppose levels: it's a dramatic shift. I must say that anything but private voting for RfCs is unacceptable to me—the current system encourages herd voting; it is superficial. Tony (talk) 03:15, 23 December 2009 (UTC)[reply]