Wikipedia talk:Arbitration Committee Elections December 2005/Archive

From Wikipedia, the free encyclopedia

Archive This is an archive of past discussions. Do not edit the contents of this page. If you wish to start a new discussion or revive an old one, please do so on the current talk page.

Contents

Election procedure

  • In the present instance, I agree with Markus. Whereas approval voting may have some advantages of simplicity or where the number of options/candidates under consideration are very small, it is otherwise subject to tactical consensus-blocking and favors the ideological extremes who will prefer "One and only one" candidate. Condorcet methods in general and CSSD in particular are far less subject to strategic (i.e., false) preferences being expressed—no minority group of voters should be able to strengthen their favored result by voting tactically against the majority. One reservation remains, the open nature of Wikipedia votes does open up the possibility of spoilers creating intentional cycles. While CSSD can break these cycles, I believe it would be inappropriate to disregard any cyclical comparison having the support of more than half the ballots cast, as this indicates an extreme lack of consensus and a plurality outcome at best. In these cases, the cyclical winners, along with any tied winners, should be subjected to secondary Approval vote, all other candidates being excluded from this secondary consideration. Whig 06:30, 27 May 2005 (UTC)
  • I suggest we keep it simple and use first past the post.
Find out how many places are on offer, call that X. Then everyone has up to X votes (without being able to vote for the same person more than once). The top X candidates then are elected. Straightforward - we all understand it - fair!
If we complicate things, all we do is end up giving more weight to those who understand the voting system, rather than really electing who we want. Also a condorcet method would make it much easier for trolls to elect trolls, which would very much be to the detriment of WP. Kind regards, jguk 15:38, 28 May 2005 (UTC)
Already during the last elections, it has been proposed by Johnleemk and Nat Krause to change from approval voting to Condorcet voting. But their proposal came to late. Markus Schulze 20 June 2005
  • I see no reason to change from approval voting–it worked wonderfully in the last election. Approval voting is simple and universally understood. Mackensen (talk) 14:12, 25 Jun 2005 (UTC)
Approval voting has several problems. The most severe problem is that approving an additional candidate sometimes hurts, but never helps an already approved candidate. Therefore, the voters will tend to vote only for a very small number of candidates (bullet voting). Just because of this reason, the IEEE (which is the largest organization that was ever using approval voting) abolished approval voting in 2002. Brams writes: "According to the IEEE Executive Director, Daniel J. Senese, AV was abandoned in 2002 because 'few of our members were using it and it was felt that it was no longer needed.'" In other words, only "few members" voted for more than one candidate in IEEE's approval voting elections.
In my opinion, it is important that we change from approval voting to an election method that is less vulnerable to strategies, before the strategic problems of approval voting get blatant in a concrete election. Otherwise, there is the danger that voters get the impression that the election method is changed only to keep a concrete candidate from winning. Markus Schulze 25 Jun 2005
  • I thank you for your reply, but don't think your suggestion is particularly useful. In the forthcoming election we'll be choosing four arbitrators. We aren't voting for any one specific candidate (at least one shouldn't). Therefore whether one candidate is helped or not is really beside the point. Wikipedia works by consensus. Those candidates who receive the most votes are, by definition, those in which the community places the most trust. Given the expanded size of the electorate, which you note, it is important that we use a system people understand and which is fairly tamper-resistant. When Whig attempted to use concordet voting to resolve the "styles" issue, voters were confused as to the nature of the voting system and accusations of strategic voting, vote "corraling" and bad faith were rampant. Many did not view the vote as legitimate. I don't think we would have such a problem if we retained approval voting. Mackensen (talk) 20:31, 25 Jun 2005 (UTC)
  • Dear Mackensen, you are talking about this discussion: 1, 2, 3, 4, 5. I agree with you that that survey was very problematic. But I don't think that this had anything to do with the election method used. I rather have the impression that that survey had the following three problems:
  1. Whig proposed a new election method (here: Cloneproof Schwartz Sequential Dropping, which is a Condorcet method) in connection with a concrete survey. This is always problematic because then voters usually get the impression that the change of the election method is motivated by the fact that the proposer tries to get a concrete result in that survey. Therefore, I recommend that we should discuss the election method for the next elections to the Arbitration Committee now and not directly before or directly after those elections.
  2. It wasn't quite clear how many voters have to prefer the result of that survey to the status quo (i.e. how many voters have to approve the result of that survey) to replace the status quo by the result of that survey. Some voters claimed that Whig proposed Cloneproof Schwartz Sequential Dropping to circumvent this question and to enforce a replacement of the status quo. However, it is important to keep in mind that the Debian project, the largest organization that is using Cloneproof Schwartz Sequential Dropping, explicitly requires a 3:1 supermajority to replace the status quo (That means: To replace the status quo by proposal A, the number of voters who strictly prefer proposal A to the status quo has to be more than 3 times larger than the number of voters who strictly prefer the status quo to proposal A.); I believe that there wouldn't have been such an outcry in this survey if this supermajority rule had been properly applied. It should also be kept in mind that elections, like the next elections to the Arbitration Committee, do not have something like a status quo.
  3. In that survey, the cast ballots were published immediately, so that those who voted later could use this information for voting strategies. However, in an election, always all ballots are published simultaneously and after the deadline so that this problem doesn't occur.
Markus Schulze 28 June 2005
I agree that the survey was problematic, and principally because the votes were necessarily public when cast. If in Wikipedia elections the actual votes are not disclosed until after the deadline, tactical voting becomes less of a concern. I remain concerned that the Debian modification to use 3:1 supermajority creates an exploitable tactic that can be used to spoil the vote by consciously "plonking" for one candidate and voting NOTA ahead of any candidates the voter considers likely to prevail otherwise. This results in a deadlocked outcome, requiring new ballots to resolve. In the case of an Arbitration Committee Election, such a deadlock could be intentionally created to effectively prevent Wikipedia from appointing any arbitrators, harming the project as a result. Whig 29 June 2005 04:25 (UTC)

I would suggest that if Approval voting were universally understood, it would not be used on matters of any importance. In actual practice, the outcomes are very similar to Plurality voting. AV does not differentiate between a barely acceptable candidate and one who is outstanding. I think Schulze's method would be acceptable to many more wikipedians than AV. --Fahrenheit451 29 June 2005 00:55 (UTC)

Again, I don't think my objections are being answered–you haven't explained to me why we shouldn't use approval voting. Surely those candidates who receive the highest approval rating have the confidence of the community? Mackensen (talk) 20:31, 13 July 2005 (UTC)

This page appears to be dominated by those with a financial and academic interest in promoting "condorcet" voting, which is little understood, open to manipulation (to a large extent because of its complexity), experimental, and which has caused vicious arguments wherever anyone has tried using it on Wikipedia. Let's kick this one into touch now - Condorcet voting and Wikipedia DO NOT MIX. Far from endorsing condorcet, we'd be better off developing a policy banning it as a provocative waste of time, jguk 20:41, 13 July 2005 (UTC)
Assume good faith. I have a purely fairness interest in supporting well understood voting methods that promote honesty. Approval voting is far more open to manipulation and polarization, and has little to recommend it over ordinary plurality voting, the latter being simpler and better understood by most people, because in more common use. Whig 21:45, 13 July 2005 (UTC)
I am; otherwise, I'd have mentioned by now that one of the users commenting here (Markus) appears to have devised the vote-counting method in question. But that's neither here nor there. I'd be much obliged if you'd explain how AV is open to manipulation and polarization. Mackensen (talk) 22:23, 13 July 2005 (UTC)
To simplify, let's say there is just a single ArbCom seat, and four candidates: A, B, C, and D. Suppose that A represents a radical extreme faction with 34% of the voters being members of that faction. 66% of the voters would prefer someone other than A, but let's say that of that supermajority, one-half are not so opposed to A as to disapprove him/her. So the A-faction "plonks" by voting for A and no one else (34%) because they are radical and they want to exercise the strategy best calculated to give A the win. A also gets approval by a further 33% (one half of 66%) with a resultant 67% approval. B, C and D have no chance of exceeding this level of approval, as even if EVERYONE not part of A's faction approves of B (and even prefers B to A), the result is 66% for B against 67% for A, and A wins. This is not a consensus outcome, it is the result of strategic voting by the A bloc. Whig 05:39, 14 July 2005 (UTC)

Dear Mackensen, I don't deny that approval voting served well in the past when Wikipedia was an unimportant organization with a small and heterogeneous electorate. But I doubt that approval voting will serve well in the future. This is because approval voting has a very severe problem: Approving an additional candidate sometimes hurts, but never helps an already approved candidate.

There are only very few methods with the property that ranking an additional candidate sometimes hurts, but never helps already ranked candidates. For example, the Bucklin method shares this property. History has shown that, whenever an election method with this property is being used, then, when the organization becomes too large or too heterogeneous, the voters begin to use bullet voting. That means: To guarantee that later preferences don't hurt earlier preferences, each voter will vote for only one candidate, so that the used election method becomes identical to FPP.

The best example for approval voting is the IEEE. With about 360,000 members, the IEEE is the largest organization that was ever using approval voting. In 2002, the IEEE changed from approval voting to FPP because "few of our members were using it and it was felt that it was no longer needed" [1].

The best example for the Bucklin method is its use in some City Council elections in the USA in the 1920s. Hoag and Hallett write that, when the Bucklin method was being used, then usually less than 3% of the voters cast more than a single preference (C.G. Hoag, G.H. Hallett, Proportional Representation, Macmillan, New York, 1926; pages 485--491).

I predict that, when Wikipedia keeps using approval voting, then (as soon as Wikipedia's electorate has reached a certain size or a certain heterogeneity) also here the voters will begin to vote for only one candidate. Therefore, I suggest that we should change from approval voting to an election method that does not share the above mentioned property. Markus Schulze 15 July 2005

Comment

Jimbo, WP would not exist without you, and anyone committed to improving the project, or who reads WP owes you a great debt. However, like any project, this one needs to grow and improve without its founder if it is to succeed in the long-term, and therefore it would be better if you stepped back as far as possible from the detail.

There was a lot of bitterness in last year's election, but that did not result in wholly outrageous appointments to ArbCom - indeed, although there were some disagreements, these were accepted. I'm not sure which particular appointments made last year that you find intolerable, personally I see none as being intolerable.

In the meantime, there has arisen some discontent about some of the members of the ArbCom standing for re-election this time (none of which were elected last year). Now, I have no idea what the community view is about those candidates, but it's perfectly reasonably to allow WPians to have their own head on this, and decide whether such discontent is so great as to render that candidate unelectable.

In short, although last year's process was divisive - it was a growing-up, a process without its founder, which was important to the English Wikipedia. Most WPians who were around from that time wish this year's process to be less bitter, and that in itself will help. Jimbo's proposal is therefore too patrician - let the many WPians who have volunteered hours of their time to make this project a success have a true say. Yes, let's exclude immediately the hopeless candidates, but let the rest run - and I must say that of all the candidates who have nominated themselves so far, none has shown anything other than a genuine commitment to the WP project. I do think that Jimbo's decision is a mistake, and believe the community should override it and proceed as before, jguk 22:50, 21 October 2005 (UTC)

Is this in response to Jimbo's temporary ArbCom appointments? --Merovingian (t) (c) (e) 09:50, 22 October 2005 (UTC)
I believe it is in response to this edit. Talrias (t | e | c) 18:59, 22 October 2005 (UTC)
Well, in that case, I feel that User:Jguk is right. I don't like this move away from election so much. --Merovingian (t) (c) (e) 02:27, 24 October 2005 (UTC)
On the other hand, it could be worse... much worse. --Merovingian (t) (c) (e) 06:03, 24 October 2005 (UTC)

Personally, I'm also disappointed. I thought things were starting off more smoothly this time, and I had hoped that by having settled that there will be no endorsements/disendorsements page, we had already cut off much of the opportunity for rancor. I wish Jimbo had greater trust in the community on this issue; then again, considering the community's overreaction to the Answers.com partnership, I can see why he might have difficulty trusting the community fully about now. --Michael Snow 05:25, 25 October 2005 (UTC)

Let's be fair to the community here - it was due to a confusing press release which gave the strong impression Answers.com had paid for text placement on Wikipedia. Jimbo's comments above are similarly vague. Let's hope he clarifies them too. Talrias (t | e | c) 08:26, 25 October 2005 (UTC)
Lets hope that this happens with less sophistry than the answer to the Answers.com concerns. Personally, if this is not to be a free election, I will withdraw my candidacy. Filiocht | The kettle's on 10:36, 25 October 2005 (UTC)

The current jimbo proposal would be a mistake (because if the comunity does reject candidates the elections could go on all year). A key part of arcom's authoirty is that they are chosen by the community in a fair and open manner. This makes acusations of and "old boy/girl" network less credible. Yes some unsituble candidates will run but they are not going to get elected.Geni 13:05, 25 October 2005 (UTC)

Crossposted from [WikiEN-l]: While I respect Jimbo's opinion I'm not quite sure if this way is the right way to go.

The 'old boys' club' is one of my concerns. The other one (which is slightly related) is that some highly-qualified candidates which could do very well might not be on Jimbo's (and the arbCom's) radar, so how would that work? — Ilγαηερ (Tαlκ) 20:47, 25 October 2005 (UTC)

I believe that Jimbo's decision is sound. Elections and polls here have become highly politicized in a way that, IMO, is at odds with the goals of the project. The AC elections are particularly bad, and we have in the past overlooked well-suited candidates with an eye for fairness in favor of highly active editors who have been able to articulate a campaign message that resonates well. The ongoing problem with resignations is indicative to me that people are seeking election to the AC for prestige rather than service. It is also clear to anyone who follows AC cases that it has always been the case that one or two arbiters have done the vast majority of work. The AC needs more people willing to invest this amount of time, and I believe that an election is a poor way to choose such people.

Those who have joined the project in the last year or two may be unaware of the extent to which Jimbo was involved in choosing and articulating core policies that are the foundation of the project. His leadership has done more for the project than his considerable financial support for it. The fact that he has chosen to be more involved in publicity, fundrasing, and international issues over the last 1-2 years may have eroded the memory of his involvment among editors, but I respect his vision for the project and his involvement in this particular area.

The Uninvited Co., Inc. 22:27, 25 October 2005 (UTC)

This is a wise observation, much of which I agree with. However, there's a need to shift responsibility for selecting arbitrators from Jimbo to the community (for the sake of both, in terms of Jimbo's time constraints and the community developing the skills of self-government). The problem is how best to accomplish this. Elections are a natural method for community selection, and have worked reasonably for the Wikimedia Board of Trustees. Some degree of politicization is inevitable, but I think with an appropriate structure it can be minimized. I have some ideas on this that I will present shortly. --Michael Snow 22:51, 25 October 2005 (UTC)

(I was going to voice some concerns about why this decision was taken, but I hit an edit conflict with UC's post above, which answered most of them to my satisfaction—so never mind.) However, I question the assertion that the ongoing problem with resignations (which is real enough) says anything about open elections as a method for choosing arbitrators. Eight out of the original eleven arbitrators, all of whom were hand-picked, have resigned (and one that I know of turned down the appointment in the first place). One of the five appointed in the interim has yet to particpate in a case. Of the eight who have been elected, three have resigned or not sought re-election. (It is, however, true that the appointed arbitrators do, and have done, most of the work. This is a perfectly good argument against open elections.)

However (and this is the concern that remains unanswered), I'm still put out by the way this decision seems to have been taken in secret and presented as a fait accompli, with no invitation to discussion—not even after the fact. It won't change anything, of course, but I would still like to voice my disapproval of MeatBall:BackRoomDecisions on such important matters. —Charles P. (Mirv) 22:58, 25 October 2005 (UTC)

You are going to get burnout amoung arbcom members whatever. Jimbo's shift away from day to day running of en.wikipedia is a further reason why this is a bad idea. Recent events has shown that the board has difficulty even comunicateing with the community. How does he know who is going to be able to stand the pressure and make fair judgements?Geni 23:11, 25 October 2005 (UTC)
I already know from one of the recent appointments that he doesn't. Sam Spade 23:22, 25 October 2005 (UTC)

I also disagree with Jimbo's decision, whether or not it's the most sensible in theory. This decision has the appearence of excluding the community from the ArbCom selection process, which in the long run could undermine the ArbCom's effectiveness. I support statements above which argue that this round of elections is off to a better start than the last one—even though I probably have a better chance of being appointed than elected. Ingoolemo talk 21:43, 30 October 2005 (UTC)

Please see Wikipedia:Arbitration Committee Elections December 2005/Straw poll. ‣ᓛᖁ♀ᑐ 20:45, 10 November 2005 (UTC)

Proposed overhaul of elections and arbitration process

Several people have indicated that they dislike Jimbo's recent intervention in the Arbitration Committee election process. Although the alternative plan has not fully taken shape, I share some of the concerns. For example, I don't think a situation where Jimbo appoints the Arbitration Committee, and the election is simply to ratify these appointments, is a good solution. However, I understand the concerns about the community's ability to handle this on its own, based in part on the last election. I think we need to find a reasonable compromise position.

Meanwhile, the arbitration system itself continues to struggle. Too many cases are going at once; the arbitrators cannot give them adequate attention; the process moves too slowly. There are a number of other problems, including arbitrator activity and burnout. I have a plan that can address many of these issues. Here's my proposal.

Add a group of users called magistrates as a body below the Arbitration Committee. The disputes that currently go into arbitration get handled by smaller groups of magistrates and/or arbitrators, instead of going before the full body. I would suggest that we keep the four votes to accept a case rule, and simply make those four people the panel for that case. The full Arbitration Committee can review specific cases when appropriate, as a "court of appeal".

Magistrates are appointed directly by Jimbo. Since there is no fixed number, the Arbitration Committee (or anyone else) can suggest names, and people can volunteer directly to him.

While we get the magistrate system up and running, extend the appointments of the arbitrators whose terms expire in December. Hold an election in February (shortly after the next fundraising drive, which should come in January) to fill these seats instead. The newly elected arbitrators begin on March 1, which gives us time to observe the magistrates in action, since I imagine a few of them would run for the Arbitration Committee.

Arbitrators can be elected from the community at large, not just the pool of magistrates. However, many of the best candidates will likely have gained experience and proven their ability as magistrates beforehand.

Benefits of this system:

  • The community gets to decide which of its members are on the main body, which is in most cases the final port of call.
  • New magistrates can be added at any time, without having to wait for an election cycle.
  • Magistrates are not chosen based on unsuitable criteria, such as RfA-style "popularity contests".
  • Elections to the Arbitration Committee should favor candidates who have demonstrated ability (as well as approval from Jimbo).
  • Magistrates who "lose" an election for arbitrator can still continue to serve.
  • Larger pool of people available for any given case.
  • Number of magistrates can be scaled upward as community grows. (I think we could start with around 20.)
  • Decreases the workload for any individual arbitrator/magistrate.
  • Reduce burnout accordingly.
  • A panel can focus more attention on its specific case.
  • Less overworked arbitrators/magistrates may also be able to resolve cases more quickly.
  • Since not everyone participates in a case, magistrates can voluntarily avoid cases in which even a perception might exist that they have a personal interest.
  • In small panels, reaching any decision requires substantial agreement (three out of four).
  • Nevertheless, the Arbitration Committee can modify or reverse problematic decisions from small panels.
  • In cases before the full Arbitration Committee, members of the smaller panel can organize and explain the often unwieldy evidence based on their previous review.
  • For all cases, the pool of magistrates not participating directly is still available to handle housekeeping issues that are often neglected currently.

For all these reasons and more, I hope we can move forward with this proposal. I think this is the best way to balance Jimbo's input with the need to develop responsible self-government by the community. --Michael Snow 04:33, 26 October 2005 (UTC)

Proposals to expand arbcom failed when last proposed as did proposals to set up a two tier comitte. To be honest this looks like an attempt to force a role for jimbo into an area where he isn't really needed. Selecting 20 people is going to be hellishly time consumeing if it is done by wieghing each candidate on their merits rather than grabing the first 20 members of the in-crowd who come to hand. If we look at the workings of you prosed system it fails to deal with the issues that 1. every case will be appealed resulting in even longer time spans 2.every turned down appeal will be protested furriously with the result that in the end either the magistates or arcom will be viewed as a ruber stamping body. Geni 04:50, 26 October 2005 (UTC)
Previous similar proposals haven't been acted on, but a number of people continue to think they would be good ideas. They've never been rejected outright, so I think saying they've failed is seriously overstating the case. We're not talking about "a solution in search of a problem" here; the problems are real and often commented upon. Do you deny they exist, or do you have a better solution? I'm not trying to force a role for Jimbo; I'm trying to salvage a role for the community when the current direction is moving away from that role.
As for the appeals process, yes, the Arbitration Committee would undoubtedly "rubber-stamp" reasonable, well-grounded decisions by the smaller panels. With appeals especially, there is no particular reason to require some specific procedure. Due process, to the extent that it matters here, comes primarily in the original hearing. --Michael Snow 05:05, 26 October 2005 (UTC)
The best solution is to expand arbcom and use the taxi rank system for selecting who handles each case. This has the advantage of not createing more layers of burocracy. There is no need as yet to "salvage a role for the community" defeatism is not helpful. The second paragraph of your statement basicy reduces arbcom to a figure head role which is not acceptable.Geni 14:53, 26 October 2005 (UTC)
Well, what role does the community have right now? Jimbo's making all the appointments. --Michael Snow 15:53, 26 October 2005 (UTC)
That's becuase we don't have the capacity to hold elections every 5 minutes or even every few months. We do have the capacity to hold elections once a year and should not be prevented from doing so.Geni 23:58, 26 October 2005 (UTC)

(Pasted from my own post on the WikiEN-l mailing list.)

What I still don't understand is why you propose a very large pool of magistrates instead of a very large pool of arbitrators.

Expanding the arbitrator pool solves the problems, quickly and simply. From the large pool, the arbitrators self-identify as "available" or "unavailable". Where X is an agreed upon odd number, and Y is an agreed upon even number, appeals go to a random bank of X+Y available arbitrators, from whom, after recusals and a random selection of alternates to monitor the case but sit out the ruling, X arbitrators determine the case. If there's case overload, we add more arbitrators to the pool. If there's burnout, we add more arbitrators to the pool.

Choosing any two trusted Wikipedia members, I have yet to see any trusted pair of Wikipedia members where one is so much more qualified than another at dispute resolution that they should serve in an appeal body over the other. If we trust members enough to issue rulings in arbitration matters, that should be the end of it. We've already found that the current arbitration committee is human, and that they learn from their own mistakes as well. We can already appeal to the abitration committee to review their own rulings.

In your vision of things, what's to stop every single arbitration case from being appealed to the "upper" body? Why even have an appeal to anyone other than those who've already become familiar with the issue? What possible grounds for appeal would there be? Why should any user even be given a right to appeal, especially considering that the mission is not to duplicate a court system, but to create an encyclopedia? Why allow arguments to prolong upward in importance, rather than just getting on with getting them behind us?

(added this bit to my earlier email) I think the simplest solution is the best. More arbitrators. A lot more. Enough that a number of them can take a Wikibreak without breaking the system. Then assign them randomly to case panels as needed to handle however many cases we have at any given time. Unfocused 05:55, 26 October 2005 (UTC)

The reason for the distinction between magistrates and arbitrators is to preserve the election process as an important element. Elections can only happen periodically, they can't be ordered up at will to add more people to the pool. And of the options for the community to select arbitrators, I think elections are the best we have; I don't like the thought of using RfA, or the Mediation Committee's process, to select arbitrators/magistrates.
Oh, and what will stop every single case from being appealed is the same thing that stops every single case from being appealed to Jimbo now - simply that frivolous appeals will be given exactly the attention they deserve. There is no "right" to an appeal; there is an ability to appeal. --Michael Snow 06:11, 26 October 2005 (UTC)
So schedule elections every two or three months instead of annually until the pool has proven to be plenty large enough. I still don't see why there should even be a third party appeal. And even if there were something so important, urgent and unsolvable that we needed an appeal other than Jimbo, couldn't we just select a much larger panel from the much larger arbitrator pool to appeal to? After all, we want judgements to reflect community values rather than decrees from on high, and how better to get community values than to get a wider range of views in the final decision? Why elevate a small group to rule over others when it is contrary to Wiki principle and values? That seems very un-Wiki to me. Unfocused 06:27, 26 October 2005 (UTC)
How do you work an election system without a fixed number of positions? I realize it's conceivable (every candidate above some threshold), but based on the results of previous elections here it will either undersupply the pool or produce arbitrators who don't have substantial community support. Have you thought hard about how to choose an expanded pool in the system you imagine? And I think you attach greater significance to the possibility of appeals than it really warrants. Incidentally, what "Wiki principle and values" are involved in having arbitration in the first place? --Michael Snow 06:44, 26 October 2005 (UTC)
If you have a larger arbcom you are likely to get more people to run.Geni 14:53, 26 October 2005 (UTC)
I never said there wasn't a fixed number of positions, only that it would be a "large" pool. It was my intent to express that it would be grown in increments until it was proven to be large enough, but each scheduled incremental expansion would certainly be of a fixed number of positions. Sorry if I somehow misled you to believe that I expected anything else.
Arbitration systems are far different than court systems. Court systems seek justice. Arbitration systems seek consensus, and where they cannot find consensus, they enforce a solution to solve a problem, not to dispense justice. Consensus is the key principle of the Wiki, and that's why we have an Arbitration Committee instead of a court system.
Finding common grounds through arbitration sometimes means that transgressions will go unpunished. This is especially apt in the Wiki, because the only purpose of the arbitration committee is the good of the Wiki. It happens that what's good for the Wiki is nearly always the consensus of all parties in a dispute. Only where common ground cannot be found and consensus cannot be achieved that the arbitration committe has to issue a non-voluntary ruling, and when that's done, it's done not to dispense justice, but to impose the best possible solution for the good of the Wiki. Now, if the arbitration committee acts in favor of the Wiki, what possible grounds could there ever be for an individual to have a right to appeal? Unfocused 15:05, 26 October 2005 (UTC)
Okay, your view is a little more comprehensible to me now. Incidentally, the answer to your last question is contained in the assumption that precedes it, although again I think it's misplaced to be characterizing appeal as a right. --Michael Snow 15:30, 26 October 2005 (UTC)
No, I'm afraid I don't see the point of creating another level of arbitrators. just have more and create sub-panels of whatever is considered a suitable size for each case as it comes along. Sandpiper 23:53, 27 November 2005 (UTC)

.

Election procedure

I just wanted to express my disappointment for the proposed election procedure. Although I appreciate the contribution Jimbo made to this project, the community has also made a tremendous contribution. I sincerely believe the best way for Wikipedia to promote community spirit and ultimately succeed is through free and democratic elections of arbitrators. Make no mistake, the proposed procedure where Jimbo vets candidates is neither free nor democratic. Cedars 12:14, 30 October 2005 (UTC)

I also find this development rather disappointing, and the idea that the community be allowed only to approve pre-designated candidates gives me a sinking feeling. I strongly second, in addition, Ilyanep's second point above as well as Filiocht's remarks. The Wikipedia instructions for creating a WP in a new language remark that "the free encyclopedia" means free in the sense of freedom, not in the sense of without payment - is this sort of move really in keeping with that sentiment? All that said, I await elaboration as to what process will emerge. Palmiro | Talk 20:38, 5 November 2005 (UTC)
I believe, unequivocally, that elections should be held by Wp members at periodic intervals to install (with set – twice/annually? – terms) and or remove ArbComm members. With authority comes responsibility, and votes dually incite and empower Wikipedians: votes will allow for the installation of members who have perhaps demonstrated meritorious service/judgement on Wp and to ensure true accountability if things somehow go awry. This will also free up Jimbo to participate in areas/issues that legitimately require his input (e.g., creating broad-reaching policy, advocacy, operations, etc.) Let God worry about the big picture, not the details per se. :) E Pluribus Anthony 06:15, 14 November 2005 (UTC)

An alternative viewpoint on member elections

I see the following problems and propose a possible approach that might (with other users collaboration and suggestions) lead towards a solution:

The problem
  • Wikipedia is very large, there are many disputes, ArbCom is going to be busy.
  • As the final recourse for most disputes, ArbCom's panels must be credible, so its members must be credible.
  • The best panels are those with experienced, neutral wikipedians of repute, but this tends to suggest a cabal.
  • Open voting has its own weaknesses, because most users do not know many nominees, and will often vote for friends rather than quality.
  • These should even themselves out but so far it seems highly disputed whether they do.
  • There is an inherent conflict so far, between the desire to get the best wikipedians, but general public consensus, given that this is not like national politics and the most suitable wikipedians are not always well or generally known.
  • A random user tends to know people most, only in articles he has edited or been involved in.
  • People who are most suitable for ArbCom probably have a decent track record in many articles.

I therefore have the following suggestion and rationale to offer. It's not polished, but its a starting point for discussion of pro's and con's.

The suggestion
  • Anyone can apply to be an ArbCom nominee. Its open to all, rather than one person's opinion.
  • Anyone wishing to be an arbcom member, can put to the vote in the articles they have worked on, using standard wording to the effect "I am considering standing for the Arbitration Committee. Guidelines and standards for nominees is here. Please indicate below if you feel my work on this article is neutral anc collaborative enough, my involvement in this article substantial and and my standards high enough to meet ArbCom standards. Thank you." - this could be a template to keep it neutral {{Arbcom-vote|NYNAME}}.
  • Note this is NOT asking if editors on the article would vote for them, but merely do editors on the article feel they are of high enough standing to be a good arbcom member. So any number of editors could ask on an article without conflict.
  • A vote on an article may be counted if >= 7 editors on the article vote (people who have not worked on ther article don't get a vote, usual rules about spam voters apply)
  • Anyone wishing to be an arbcom member can post such a note on any number of articles. (Note that spamming will be useless due to the "...my involvement in this article substantial and and my standards high enough..." criterion)
  • A candidate for arbcom is then someone who meets the following criteria: 6 months active editorship, and approval from votes on at least 7 articles (as many as they like).
  • In their nomination, the candidates can name the articles they received supportive votes on and the voting on those articles, as evidence of the degree of support they have had from other editors on articles they have worked on.
Rationale
  • What we have filtered out here, without much more than a few article-based votes, is, people who have made a contribution of a very specific kind. They have been an active editor, contributing substantial work on at least 7 articles, and whose co-editors feel the quality and size of their work and wiki-policy following was good enough to feel okay about them on arbcom as a result.
  • People who are POV editors as a rule, are POV editors on a very few types of article, as far as Ive seen. People who have really good reputations tend to be wider active and wider known. So this will not be so easily manipulable as a first filter, by POV pushers and vote manipulators, because they may get biased support from a few articles but not from the 7+ (or many) articles that a really good nominee will get support from.
  • People who edit articles which have very few editors (minor additions) will not get "credit" for those, because they lack 7 editors able to vote. this is a good thing because in general you want their contribution to be evaluated by other editors, so articles where there are few other editors may not say much about their quality either way.
  • The support at article level is non-competitive, so the only criteria is "do you think this person is good" not "do you think they are the best editor here"
  • Users voting in the Arbcom elections are not just dependent on "puff". They can look to see how many articles' editors supported the nomination, what the strength of that support was (4-3 or 40-5), what area of articles they have gained support from (more the merrier), and even look back on those articles talk pages to check what other editors said about them.
This achieves the following goals
  1. Arbcom nominees are pre-filtered for their quality as editors
  2. The nomination is open to everyone, but includes a 1st requirement to get support based on actual work done, evaluated by peers on those articles, that will exclude many poor quality candidates and is not overly susceptible to vote pushing.
  3. The process is simple, if I work on Cognition, I post a standard "Arbcom-nominee-vote" template on the talk page asking if people would judge my work on that article of high quality. Easy.
  4. The candidates then presented to the public are not just unknowns with their own speeches of "why i want to be here". They have backers based on factual contribution, on the quality of their help on specific articles, and both the votes and their work can be read, assessed, and checked.

FT2 18:26, 2 November 2005 (UTC)

This proposal overvalues article authors over janitors. Wikipedia needs both. In fact, I think it's a bad idea to pull article authors off authoring articles and put them on ArbCom, because ArbCom is so timeconsuming and because, in terms of sheer numbers, we need article authors more than we need janitors. Finally, being a good article author is not even remotely a good predictor of how good an Arbitrator one would be. Kelly Martin (talk) 22:39, 2 November 2005 (UTC)
What evidence is there of how a person understands wiki values, except those who have seen them exercise those values in practice, in articles? The problem is, I've never (and most wiki editors have never) seen most of any arbcom candidates in action so there is no basis of choosing between them. And ... you know. If wiki editors are trusted by and large to be able to form a consensus able to edit articles, then I'm not sure logically how you can say they aren't the ones who are best placed, again by consensus, to filter out poor choices for arbcom members. If a person can't get support from editors on articles he's worked with, for the quality of his work and standards upheld, he's likely to be a poor arbitrator for others. On the other hand, people who wish to put themselves forward as good arbitrators will likely find that not a hurdle at all. Seems obvious to me. Who else is placed to filter out poor choice candidates? If I was to have a dozen or so people to arbirtrate within the user community I'd want them to also (not only, but also) be capable editors themselves, not just theoreticians about the editing process. Once appointed they may have to pull back from editing, but before them, they're likely to be active editors or at least active mediators. FT2 01:18, 3 November 2005 (UTC)
I'm not a major editor of any significant article. I've written a handful of articles but none of them have ever become the focus of any major editorial activity. I've made thousands of small edits on thousands of articles, mostly vandalism reverts, copyedits, style adjustments. I've also removed non-NPOV language from a couple dozen articles, and mediated a handful of disputes. In many of the mediations I never edited the article in question; so by your standards I wouldn't be able to request support from the editors of that article (despite resolving their dispute) because I am not a "coeditor" of that article. (The ethics by which I conducted mediations when I was a mediator precluded me from editing the article in question, unless the parties explicitly requested it of me.) The only article where I'd have a chance of getting the support level you think I should have is Ward Churchill, which I haven't edited in months. Your nomination criteria would seem to exclude me from consideration from the position of Arbitrator. I'm currently an Arbitrator; was my appointment to that role so gravely inappropriate that the possibility of it needs to be precluded in the future? Kelly Martin (talk) 02:19, 3 November 2005 (UTC)
Which is why I called it a starting point. I think the principle counts - that users who have seen candidates work on wikipedia, should somehow be a filter for poor candidates. Sorry if I left the impression that "editors" meant only those who write articles. I might need to clarify, because I meant "edit" in the broad sense, not just "add content" but exactly as you describe... those who fix, clean, contribute, mediate, policy or community stuff, dispute resolution -- in fact anyone who actively has contributed in a way significant others others have seen and would agree demonstrated a good standard. To me, thats all "editing". I do a lot of that, too.
To take the case in point, I don't know you at all, we have worked on different articles. So your name would be pure random guess for me, one of dozens of unknowns. But if I find that others -- mediators, arbiters, editors who've seen your "thousands of small edits and removal of non-NPOV language" -- recommend you as someone able to do the job, and you point me to articles or pages where most other editors will support that "yep, Kelly Martin helped us pretty well here", or "yep, Kelly Martin has done good work or has a good reputation on this page", then thats very much germane.
FT2 03:15, 3 November 2005 (UTC)

Another proposal to reduce the ArbCom workload

Michael, FT2, I like your thinking, but I have an idea that may simplify the process even further. We already have a body of trusted users: they're called admins! As a first step, why not let our admins vote on proposed arbcom cases? I envision a similar process to AFD, but where only admins can vote, and where there is a defined quorum: perhaps seven or so. Cases which gain a two-thirds supermajority either way will be resolved then and there, so only the really controversial and difficult cases will be passed on to the Arbitration Committee itself. What do you think? GeorgeStepanek\talk 17:05, 4 November 2005 (UTC)

The main problem seems to be perception that this would feel like a cabal to some people, and might cast wikipedia's neutrality and openness in doubt eventually. Even though anyone can be an "admin", it may still carry that feel to some users, and to people who are less aware of WP organization.
That aside, it's a good idea. It could work, provided such a page made clear that "to become a wikipedia admin, see here", to diffuse such issues. I'd have a larger quorum - 9 or 11 - whose votes cover two aspects: whether there appears to be a well defined case to answer or not (or whether it needs better documentation, or is groundless), and whether the case appears to be appropriate for ArbCom, or should be followed up for now in some other manner. That could work well.... it depends whether there are many ArbCom cases which are trivial. But yes, in general, obvious stuff like POV warring shouldn't always need full ArbCom debate, if its obvious, and a way to filter that out, or pre-filter the ArbCom queue, would be no bad thing. Opposing view: There aren't that many ArbCom requests and acceptance/rejection is not a heavy workload. FT2 17:55, 4 November 2005 (UTC) [PS I have edited the title of this section for clarity]
What I *could* see being useful, would be a "fast track" queue (similar to "speedy delete"), where it was not so much "please arbitrate" as "please rule yes/no whether specific policies have been significantly breached". So users could cite a behavior and DIFFs (similar to a case for VfD) and ask simply, "has NPOV been breached" or "is this a personal attack" or "Could I have a ruling that these are original research and user X should not add material without proper citations".
Admins can quickly check the cited references, and rule how they see it, with comments. This can then be used either on the page to bring an editor into line, or as independent prior opinion from experienced users that there is a genuine case for further review by ArbCom if needed.
If you added a rule that a user ruled by this process to be in breach of Wiki policy was then "on notice" that their behavior needed to change, then they could be banned for (say) 72 hrs at a time, if it continued, without accusations of Admin bias, that would probably stop a lot of problems dead in their tracks. FT2 18:37, 4 November 2005 (UTC)
Thank you, FT2, for your kind words, and for your analysis of my ideas. I was actually envisioning more of a lower court / supreme court distinction between this process and the ArbCom. The ArbCom—as the supreme court—would consist of appointed experts that pick and choose difficult and important cases to tackle. The "Arbitration Vote"—as the lower court—would consist of a jury of impartial peers that rule on cases that are obviously valid or obviously spurious. I agree that the quorum needs to be significant to assuage fears that a "cabal" is hijacking the process. It might work something like this:
  1. RandomUser requests arbitration against RandomTroll (as per the current process).
  2. Admins vote on the protected page as to whether RandomTroll is guilty or not guilty of the charges made against him.
  3. The vote is open for a specified period (say two weeks) and until enough votes have been made (12 is the traditional number).
  4. The vote is closed by an admin who has not voted, who simply adds up the votes and determines if RandomTroll is guilty (2/3 for), not guilty (2/3 against) or if there is a hung jury (otherwise).
  5. If there is a hung jury then the case is automatically referred to the ArbCom, who then decide whether they do or do not want to take it.
  6. If RandomTroll is deemed guilty, a new vote for sentencing begins.
  7. RandomUser requests a sentence that she thinks will be supported by the community.
  8. Admins vote on the protected page as to whether the sentence is fair or not fair.
  9. Again the vote is open for a specified period and until enough votes have been made.
  10. The vote is closed by an admin who has not voted, who simply adds up the votes and determines if the sentence is fair (2/3 for) or not fair (otherwise).
  11. If the sentence is deemed fair, then it is applied forthwith.
The advantages of this process are that it will scale well, as both Wikipedia and the number of admins grow, and that RandomUser is motivated to provide a reasonable (and well-reasoned) case—and an appropriate sentence—to make sure that they are both accepted by the community.
Also, it's probably best if those admins who have at any time been involved in the dispute in question be unable to vote on it, to help preserve the impartiality of the verdict. GeorgeStepanek\talk 23:08, 4 November 2005 (UTC)

Don't RfCs pretty much do this already? Flcelloguy | A note? | Desk | WS 23:10, 4 November 2005 (UTC)

The RfC process has no power to impose penalties. The process described above would be used only after an RfC has failed to resolve the issue. GeorgeStepanek\talk 01:29, 5 November 2005 (UTC)

Admins no doubt come in all shades of view. How would you choose the seven required to decide the issue? obviously not the first seven to come along, because one with a particular view might have asked six mates to side with him and form a quorum. It does have the merit of using an existing setof people to do the work. But what is wrong with just having a few more arbitrators and doing it right? Sandpiper 00:06, 28 November 2005 (UTC)

How would you choose the seven required to decide the issue? Maybe randomly? The "elected" Arbcom members could act as judges, yes, but we could have sort of a trial by jury system in which random active admins could be choosen to act as a jury of peers. The only reason that you couldn't extend the jury concept to normal users is the issue of sockpuppets. The Arbcom members would only be their as experts in wikilaw and in determiing when a case should goto wikicourt. This way, people wouldn't think of the arbcom members as an ellite rulling cabal, and the judgement would come from "normal" people. --Rayc 16:33, 28 November 2005 (UTC)

Yes, polls are evil, but,

please see Wikipedia:Arbitration Committee Elections December 2005/Straw poll. Radiant_>|< 19:38, 10 November 2005 (UTC)

In regards to the recent changes

It sounds like we're lacking Consensus here. It would seem as though there needs to be more debate on the Elections procedure before things proceed any further. Voyager640 07:44, 27 November 2005 (UTC)

ArbCom elections are conducted however Jimbo Wales sees fit to conduct them. He is not required to seek or obtain consensus for them. If he waited for consensus to form, there would likely never be an ArbCom election. Kelly Martin (talk) 10:13, 27 November 2005 (UTC)
Close but technicaly no. It is the the wikimedia foundation who have thoeretical ultimate power here. And you know as well as I do that arcom can't operate without a reasonable level of community support.Geni 10:41, 27 November 2005 (UTC)
I'm not talking about who has power. I'm talking about generally accepted guidelines for operation of wikipedia. I don't deny that Jimbo Wales *can* do this. I just don't think it's a reasonable practice given the way things are generally run on wikipedia. Voyager640 16:16, 27 November 2005 (UTC)
I think that asking for consensus is a bit too much since we'd never get consensus as Kelly said, however I do think that he went a litle too far on this and I hope that he'll at least clarify and explain why he's doing things this way. JtkieferT | C | @ ---- 00:26, 28 November 2005 (UTC)
Given the above, it's clear that sufficient discussion or input has not occurred, or consensus reached, on any of this. Patience is a virtue. Even an RfC of limited time (or similar) would have identified overarching support for elections – or other options – from the Wikipedia community: the above indicated poll has been underway for a bit more than two weeks, garnering far more user input than this discussion and a 3:1 margin throughout supporting ArbComm elections. Also, none of this precludes Jimbo from exercising his authority and making a decision based (or not) on community input. It may not be improper that he can do as he sees fit: it's the appearance of impropriety, particularly in light of considerable opposition, that may give Wikipedians pause. E Pluribus Anthony 00:55, 28 November 2005 (UTC)