Wikipedia talk:Arbitration Committee/CheckUser and Oversight appointments

Is this the right type of election?

The proposed method is approval voting. In practical terms, this means "first n past the post", in this proposal, based on a ratio of three candidates to each appointment. Several questions arise.

1. Is this wasteful of good candidates? Out of a pool of three, only one will be appointed.
2. Should the ratio instead be, say, two candidates to each vacancy?
3. Should we consider an altogether different system? for example, would a "support/oppose" system (similar to that used at RfA) be better? Under this system, any candidate with a very clear consensus for promotion – for instance c. 85% – would be promoted.
4. Is it better to have formal approval voting elections every three or four months or to run a series of "support/oppose" elections for individual candidates as vacancies arise?

Your thoughts? --ROGER DAVIES ^talk 18:18, 8 January 2009 (UTC)

If any candidate receives clear community support, and the committee has already vetted them, why not promote them? With that, this process needs to steer as far away from RfA as possible. Requests for CU and/or OV should not be a popularity contest, nor should it create any amount of drama. Tiptoety ^talk 18:34, 8 January 2009 (UTC)

Yes. It was the desire to minimise drama that led to the voting proposal here. What would you regard as an indication of clear community support? --ROGER DAVIES ^talk 18:37, 8 January 2009 (UTC)

I think we can know that there is community support by including a minimum number in the approval voting process. I think 150 is a good gauge? We discussed this too late to put it in the Committee's final draft FloNight♥♥♥ 18:46, 8 January 2009 (UTC)

Yep, I was just about to say something similar. I think we it comes down to it, the committee has enough common sense to say, "It is clear that consensus favors promoting" and visa versa. Tiptoety ^talk 18:53, 8 January 2009 (UTC)

I think we want to have something clearly spelled out to make it clear who has the Community support so the Committee is not doing the deciding. Also, the users voting might alter their votes if they know the minimum number required to pass the Community part of the process. FloNight♥♥♥ 19:01, 8 January 2009 (UTC)

To be honest, I liked the way the board elections were held. The vote itself was held off-wiki using software that did not reveal peoples votes or the results until the vote closed. To me, that reduces a ton of unnecessary drama. At the same time, there needs to be some room for discussion about the candidate and that is why I think a page (on-wiki) with a nomination statement (either from the candidate or ArbCom) along with space for questions from the community would be a good idea. Thoughts? Tiptoety ^talk 20:12, 8 January 2009 (UTC)

What if someone doesn't like any of the candidates? They just don't vote? The problem with approval voting is that we don't know the proportion of the community who are against a candidate. Imagine in one election the winner gets 240 votes, then in the next, the winner only gets 160. Were fewer people interested in voting, or did 80 people not vote in order to oppose all the candidates? Is a vote for a candidate a vote in support of that person, or a vote against the other 2? We can't really know. Mr.Z-man 21:01, 8 January 2009 (UTC)

There are problems with "support/oppose" voting, too. If multiple candidates are standing in the same election, a person may oppose one of them in order to support another one more strongly (see tactical voting). I confess that one of my votes on the ArbCom election was motivated by this thought process; had I known that 10 candidates would be appointed, I would have supported the candidate in question. I don't feel sorry for doing that, and I'm happy that both the candidate in question and the one I preferred ahead of him made it on to the Committee. My point is that there will be problems with any system. My sense is that FloNight has the right idea. ArbCom can be flexible: they can say "We will take the top two candidates by approval, or the number of candidates who get 100+ supports, whichever is larger." This would answer Tiptoety's concern. Again, this can be gamed, but in any voting method we choose the lesser among evils. Crystal whacker (talk) 22:17, 8 January 2009 (UTC)

I read further down this page that there may be a limited number of candidate positions for any one election. That makes no sense to me - checkuser positions are no more limited in theory than ArbCom positions, and ArbCom just got expanded by three members for no particular reason - but if that's the case, then my "either/or" proposition does not hold water. Crystal whacker (talk) 22:23, 8 January 2009 (UTC)

In this case we know that about "n" positions will be appointed, and we know we have "m" candidates, each and all of whom the committee would be fine with (as they have been vetted). So the question is which candidates have the highest communal endorsement (for trust or otherwise) in the wider community. Approval voting is well designed for that purpose.

As a side view, it also kills off the poisonous problem of people opposing perfectly good candidates (a bit hit percentage-wise) merely to enhance one's own preferences. A user can choose to vote only for "n" candidates, but this way they cannot claim that the other candidates are actively unsuited or communally undesirable (the purpose of an oppose). If there is a genuine concern, or a genuine reason a user is not communally desirable, as opposed to simple tactical voting, then it needs to be put on the talk page and if others agree (in the course of 10 days) they will act accordingly and withdraw support as well. If the oppose is not actually a reflection on the candidate, then simply say nothing. FT2 ^{(Talk | email)} 22:28, 8 January 2009 (UTC)

But when we only officially take into consideration the opinions of the people who support the candidate, we have no idea how many people we're ignoring. It would be entirely possible for someone to "win" despite having the support of only a small fraction of the active community as everyone else can't do anything but comment. If their comments don't sway the support voters, they won't matter. You could have 10 times as many people complaining on the talk page with valid reasons as supporting, but the person could still succeed, and the supports don't have to be for good reasons. The system also doesn't take into consideration the possibility that none of the candidates might be satisfactory. Wouldn't it be better to have another election with different people than force the community to choose between the lesser of 3 evils? Mr.Z-man 00:59, 9 January 2009 (UTC)

One of my concerns is that with each election, the less preferred , but still community liked candidates will gain stigma for having "lost" elections making each subsequent election harder for that individual.--Tznkai (talk) 23:30, 8 January 2009 (UTC)

• Approval voting is a Really Bad Idea for this kind of election. This table should point out why: for the 2007 ArbCom election, out of 14 candidates with sufficient "approval", only 6 had sufficient consensus to be elected. I'm sure there's plenty of similar outcomes on RFA. What is being proposed here is essentially the same to having WP:AFD delete any article for which 20+ people vote "delete", and restricting all "keep" comments to the talk page (and yes, that is the proper metaphor, because when in doubt the default is to NOT delete an article, and to NOT give a user CU rights). >Radiant< 21:30, 9 January 2009 (UTC)
  • Except that the default IS to give checkuser to people whom arbcom think should have it, so the analogy doesn't quite work.--Scott Mac (Doc) 21:42, 9 January 2009 (UTC)
    • Yeah, I'm sure that making it a popularity contest is a great idea. >Radiant< 22:09, 9 January 2009 (UTC)

If the purpose is to discover which candidates, out of a finite set, the community would most like to see elected, why wouldn't we use a preferential voting system like instant-runoff?? It's as simple to explain as saying "please write the candidates in your order of preference, favourite first", and not significantly more difficult to comprehend the counting process (although it is admittedly somewhat time-consuming to actually perform the count, it is a fairly transparent count when published). It allows editors to approve any number of candidates and, while not permitting explicit disapproval, makes not voting for someone at all a more overt and explicit indication of a lack of support than in an approval vote (as the votes for all candidates are presented together). Sounds good to me. Happy‑melon 22:00, 9 January 2009 (UTC)

Let's have a hypothetical

First let's get one thing out of the way. There is no perfect election system. Only what will meet our needs while minimizing the specific problems we face. There may in fact not be a 'best' system among those. Only a least worst. and maybe not even that. :)

That said, I want to imagine something. Say we have editors Alice, Bob and Carol. Each of them has been selected by the AC to run as checkusers, but (crucially) each would face some significant community disapproval in that function. Pick any number of reasons this could be the case. They might have been to cavalier with someone's private information, they might have mucked up a SSP here and there, or they may have just stepped on too many toes. Regardless, our hypothetical editors all stand for election and one (or two) will be approved. How does the community tell the committee that they have failed in each choice when they have only the option of choosing among them? What sort of mechanism would exist for disqualifying a candidate should "talk page opposes" (my phrasing) present significant concerns?

I know that demanding this is a big jump (right now no election system exists). I also know that the AC is not likely to bumble into this hypothetical.

Thoughts? Protonk (talk) 23:53, 8 January 2009 (UTC)

Err..

14 supporting, no objections, and no abstentions? That doesn't add up to 17--Tznkai (talk) 18:42, 8 January 2009 (UTC)

Holiday period absences, I think, but it does read oddly, I agree :) --ROGER DAVIES ^talk 18:44, 8 January 2009 (UTC)

Not everyone voted. FloNight♥♥♥ 18:47, 8 January 2009 (UTC)

Members of a body who have not voted are the ones "abstained" in traditional parlance, thus my confusion.--Tznkai (talk) 22:47, 8 January 2009 (UTC)

In this case the more closely analogous phrase is "not present" - which is not the same as abstained, but reduces the total potential for votes. Avruch ^T 00:08, 9 January 2009 (UTC)

Variation of number of appointees

At this edit, [1], I indicated my broad agreement with the proposals, but also stated my objection to Arbcom varying the number of appointees during or after an election (as in point 8 of the "Election" section of the proposal). Roger Davies here [2], asked me to expand my comment. I object to such variation because an election in which the number of appointments is unknown to the electorate is wide open to allegations of Arbcom gaming (or worse) the results in order to get a favoured candidate in. Even a squeaky clean Arbcom enjoying the confidence of the great majority of users, would always be open to such allegations in any case where they varied the number of appointments. This proposal, along with the proposed Arbcom policy update, represents a very welcome and I believe very real and honest attempt by Arbcom to address recent issues which have undermined confidence. It would be unfortunate if in the process they were to adopt a policy or procedure so open to at best misunderstanding (and at worse to abuse). DuncanHill (talk) 18:54, 8 January 2009 (UTC)

I asked you to comment here because this is bound to crop up again. Thanks for doing so.

First, the idea is to announce the number of vacancies up front. The very real problem arises with a tie. There are a couple of possible scenarios:

• Two vacancies with the first three candidates all scoring the same number of votes. Who gets passed over?
• Three vacancies. The first two candidates score very highly. The next four very poorly, with third and fourth place tying and fifth and sixth only one vote behind? What do you do here? (Probably only appoint the top two?)

However, in either scenario, Arbcom would certainly explain the rationale behind the decision. Does this help? Or can you suggest a better way forward? --ROGER DAVIES ^talk 19:04, 8 January 2009 (UTC)

Should we hold another whole additional election if mid-election we get one or more resignations from highly active people with CU or OS access? We were trying to fore see the problem of delays. I'm not sure we should get to rigid if the people are meeting the minimum vote threshold we set. Do you see my point? FloNight♥♥♥ 19:08, 8 January 2009 (UTC)

I do see your point Flo, and Roger's too. However, I do feel that say a run-off (even with the delay that would inevitably bring) would be preferable to a system open to such misunderstanding. Confidence is the key - an election for, say, 4 positions, which ends up with 6 winners is not one which will inspire community confidence. DuncanHill (talk) 19:12, 8 January 2009 (UTC)

If we immediately hold another election, is that going to be seen in a better way? Plus, we really do need to be sensitive to over taxing the Community with elections and other administrative tasks. We are trying to reach a balance. FloNight♥♥♥ 19:18, 8 January 2009 (UTC)

Yes, I think it would be seen in a much better way. "Over taxing the community with elections" or "having an election and then telling the community it was a different election than they thought it was". I know which I prefer. DuncanHill (talk) 19:30, 8 January 2009 (UTC)

We've gotten feedback from others that we are making Wikipedia too heavy with bureaucracy. That is the objection for some people the new Review Board. Let's see what others say. FloNight♥♥♥ 19:47, 8 January 2009 (UTC)

The issue on this one is, we might feel the need for "around 3-4 checkusers". For example last August we knew we needed at least 4, but we could have chosen as many as 6. Or, if the candidates turned up to have problems via communal feedback, as few as 3. So we enter an election knowing we want 3-4 (or 3 but willing to consider a 4th), and we might make the decision on the 4th place based on the results. If the top results are 150-145-137-133-104 we might opt for the 4th place. If they are 150-145-137-116 we might not. Consider it an election for n places with an option to maybe appoint 1 or 2 more when considering the results, a bit like this last Arbcom election in a way. FT2 ^{(Talk | email)} 22:33, 8 January 2009 (UTC)

Well, why not make clear beforehand that "this is an election for whatever number of posts that we decide after the election that we want", unlike the last arbcom election which was claimed to be for a certain number but turned out to be for a different number. At its simplest, it is a question of honesty - if you claim the election is for n posts, and appoint m (where m != n) it does appear less than honest. DuncanHill (talk) 22:41, 8 January 2009 (UTC)

That's true. I had in mind that we might say "We are considering 3 places, perhaps 4" or "Between 2 - 4 checkusers will be appointed". Something like that, an indication. Not a completely different number that is a complete surprise though. Would that work? FT2 ^{(Talk | email)} 22:57, 8 January 2009 (UTC)

It still wouldn't be ideal, but it would be less bad (assuming the number of appointments did turn out to be within the range given beforehand). DuncanHill (talk) 23:00, 8 January 2009 (UTC)

I would expect not to go much outside any stated numbers anyway. Another example might be, 4 candidates, 1 position anticipated, but the voting comes in at 190-177-120-96, with strong support for the first two. We might think "okay, the 2nd user seems very trusted by the community, we were debating a 2nd position anyway for a while; we could probably do well to appoint them", rather than waiting for a further election some months away. FT2 ^{(Talk | email)} 23:11, 8 January 2009 (UTC)

"Urgently needed"

Most of the proposal looks pretty good. The only part that made me do a double take was this part:

In the event appointments are very urgently needed and there are insufficient immediately available candidates, the community will be consulted as to its wishes.

Personally, I can't envision any situation in which there would be a lack of CheckUsers / Oversighters that would warrant an emergency-type situation. This clause just seems odd and I somewhat worry about possible future abuse of it (getting "consensus" on obscure subpages, for example). --MZMcBride (talk) 19:23, 8 January 2009 (UTC)

Agreed, while this may be an issue for smaller wikis with only one or two CUs or OVs it is not an issue here. I have yet to see and doubt I ever will a situation that requires emergency promotion of CU or OV rights. Tiptoety ^talk 19:25, 8 January 2009 (UTC)

Largely I agree, but in the case every person with Checkuser of Oversight access quit, it would be nice to have a back up plan spelled out. Going to the Community for approval to do something quicker than using 2 week process seemed wise to us. FloNight♥♥♥ 19:53, 8 January 2009 (UTC)

I am not sure that is anywhere near reality though, I do not think all the CUs and OVs are going to quit at one time. Tiptoety ^talk 19:58, 8 January 2009 (UTC)

(ec)It may be useful to reference Wikinews' experience when all its Checkusers were de-righted because under 25 people turned out for its arbcom elections. It had to hold new two week long elections with a turnout of 25+ before the rights were restored. Obviously if our Arbcom itself was incapacitated for the same reason, it couldn't appoint new CUs, but it can occur that all CUs are de-righted. MBisanz ^talk 19:59, 8 January 2009 (UTC)

Technically, I read this as "if we lack pre-vetted candidates, we might ask the community if they got names we could consider". This doesn't seem to be adding any loophole to the community vetting. Am I wrong? -- lucasbfr ^talk 20:00, 8 January 2009 (UTC)

In the event emergency checkuser appointments are deemed necessary, could the ArbCom (or indeed, in the exceptionally unlikely set of circumstances that results in a mass resignation/rights removal of all Arbitrators, The Foundation/Jimbo/Brion) not request assistance from the Stewards and worry about additional checkuser appointments in due course. In an emergency sort of situation, I believe we would be better served by users who are familiar with the tool and identified to the Foundation (I know there's the downside that they might not be fully up to speed with our policies here). An emergency situation could be related to numerous sockpuppet / open proxy attacks on the site in addition to some sort of mass CU resignation, so it seems like the wrong time to appoint new checkusers would be during a situation where experienced checkusers are needed. Nick (talk) 20:10, 8 January 2009 (UTC)

Yes, I think we could get Checkusers from other projects or stewards to fill in. That would be what I think that the Community and the Committee would decide during the discussion that the Committee would start. FloNight♥♥♥ 20:16, 8 January 2009 (UTC)

I get the feeling this language was put in as a CYA "just in case" the unlikely scenario happened. rootology (C)(T) 22:19, 8 January 2009 (UTC)

Partly, but it has a more serious basis. There are not that many users who will be appropriate to appoint to checkusership. If 2 leave at a similar time, then the team might well notice the gap in person-power. Or if we start getting oversightable edits from a time zone with lower coverage then we might quickly want to appoint an extra couple of coversighters, who live in those specific time zones, to increase the responsiveness. So urgency can exist.

Now... for each place we need 3 vetted candidates. Take it that this is not easy. We may only have 3 candidates, and we need 3 more checkusers, 2 at a pinch. What to do? Wait until we find 3 - 6 more viable candidates, for a total of 6 - 9? Or instead, tell the community "We think the community badly needs at least 2, maybe 3 checkusers, and we only have these 3 candidates we're really happy with. What do you want us to do? Here are some options for a straw poll, or maybe there are options others will propose." We don't know what situations might arise, and therefore we can't be sure what would be the best solution in advance; "ask the community" seems the best general-purpose solution. FT2 ^{(Talk | email)} 22:43, 8 January 2009 (UTC)

Candidate to appointee ratio

I'll just swipe Roger's two questions to the community from above:

1. Is this wasteful of good candidates? Out of a pool of three, only one will be appointed.
2. Should the ratio instead be, say, two candidates to each vacancy?

This is an important point where community feedback is needed, so I'm splitting it off. Some points that have come up during discussions amongst committee members:

• There's a limited pool of potential candidates for CheckUser in particular - active administrators with the necessary technical expertise, plus the willingness to expose themselves to some of the abuse that checkusers get.
• If a maximum of one in three candidates that make it through Arbcom vetting are appointed, will that encourage or discourage qualified people to apply?
• If we add a "minimum number of supports" criterion, should we also keep the requirement of three candidates for every open position?

The feedback thus far on this proposal has been very insightful. Risker (talk) 20:23, 8 January 2009 (UTC)

I should add that I am one of those who originally insisted for the more-candidates-than-seats system; presenting 3 candidates to fill 3 seats isn't a real community choice at all. I agree that approval voting alleviates this somewhat. — Coren ^(talk) 20:43, 8 January 2009 (UTC)

I don't agree with one in three. If there are three slots open, then the top three in voting should get it. If 5 are vetted by arbcom, then the top 3 voted upon by the commnunity should get it. — Rlevse • Talk • 20:46, 8 January 2009 (UTC)

It would be wise to always supply more candidates than seats. The exact ratio could vary. The Committee should not appoint a candidate who evidently has less support votes than typical, even if there is an open seat. Those who are not appointed in one round can be recycled as candidates in subsequent rounds until they are either appointed or withdraw from the process. Jehochman ^Talk 20:49, 8 January 2009 (UTC)

Good points, Jehochman; there was pretty strong agreement that there should be more candidates than positions, but we weren't quite sure how the community would feel about set ratios, or whether establishing a minimum approval level might change the ratio. Risker (talk) 21:02, 8 January 2009 (UTC)

Some of the appearance of "wastefulness" seems to arise from an unstated assumption that the "losing" two thirds will never be appointed. But I see no reason why these won't ever stand in an other election. In fact, if the arbcom were to game the system, all this would amount to is that a steadfast community would be able to veto two candidates. Not the greatest of wastes. ;-) — the Sidhekin (talk) 20:55, 8 January 2009 (UTC)

Also a good point, Sidhekin. The challenge is that, with a 3 candidates to 1 position ratio, that would mean 9 candidates for 3 positions, which is a lot of people to vet (for both the committee and the community), and is incidentally approximately 1% of the active administrative corps. Once vetted by the committee, though, unless something significant arises in the interim, a candidate should be able to stand in the next election without an issue. Risker (talk) 21:02, 8 January 2009 (UTC)

However, if elections occur infrequently, a candidate might lapse into inactivity in the meanwhile, and their services might forever be lost. It's best to judge whether a person is needed to help, and able to help, at any particular moment, without searching into the future possibilities. Crystal whacker (talk) 22:25, 8 January 2009 (UTC)

"Vacancy"

My sole substantive issue with this proposal is the wording "vacancy", and I would like to know in more detail how this is to be considered. There is a danger that the resignation or retirement of a member of either the CheckUser or Oversight teams would lead to a vacancy that would automatically require an election. It is, of course, better that as few users have access to private and sensitive data as possible; it is important, therefore, that the proposal does not bind (or appear to bind) the Committee to holding elections where they would be unnecessary.

I should like to know, therefore, the criteria and process that is to be adopted to determine a need for new Checkusers or Oversighters, in particular

1. what factors affect the judgment,
2. who is to be consulted,
3. how often such assessment shall be made, and
4. if 3 is "as and when necessary", what shall prompt the assessment?

[[Sam Korn]] ^(smoddy) 21:06, 8 January 2009 (UTC)

Par 3. of "Candidacy and vetting" goes into more detail – "ArbCom will normally call an election (i) when there is consensus for further CheckUser and/or Oversight appointments ..." – though it doesn't spell out all the answers you seek. There are no particular plans (yet?) to regularly re-evaluate needs. --ROGER DAVIES ^talk 21:19, 8 January 2009 (UTC)

Yes, I saw that. Perhaps there should be explicit (even if non-binding) plans for how the Committee intends to proceed? Furthermore, consensus among whom? Presumably the Committee? [[Sam Korn]] ^(smoddy) 21:25, 8 January 2009 (UTC)

This may be one point where the Review Board/Audit Panel may be helpful; they can keep on top of statistical usage and may be able to recommend to the committee that more appointments should be made. I will work a bit on this tonight to see what I can come up with, and will post a link to this point. Risker (talk) 21:31, 8 January 2009 (UTC)

Good idea. --ROGER DAVIES ^talk 21:48, 8 January 2009 (UTC)

One thing to consider is how the people who are doing the job feel about it. If they say they feel overloaded and burned out, then it's time to get someone new on the job. Crystal whacker (talk) 22:27, 8 January 2009 (UTC)

I drew up the stats on Checkuser last spring (see WT:CHECK#Checkuser usage); our top checkusers were doing immense amounts of work. Broadly I'd trust the checkusers themselves to say if they need more hands. My own view is they should agree an informal guideline on "usual usage" and if they are constantly going over it then be aware and ask for help if it's excessive. (Although if they're enjoying it or "into it", then I wouldn't stop them; the aim is they have a guide what's usual.) I would let the checkusers set that guide themselves, to roughly correspond to a mid-way active relaxed checkuser. As Crystal says, "Trust the checkusers who are doing the work". One thing for sure, we don't want to operate in a way that encourages burn-out over self-pacing. FT2 ^{(Talk | email)} 22:52, 8 January 2009 (UTC)

I appreciate the responses. I am eager to ensure that there is the right number of Checkusers -- neither too few nor too many. As I spell out above, "too many" is problematic in itself (though more so for CheckUser than for Oversight, as urgency is a factor where Oversight is concerned). I recognise that the current situation (no rules or guidelines whatsoever) is just as bad or worse in this respect than the current proposal. Now does, however, seem a good moment to also examine what the procedures are for making these decisions, in order that the right ones be made. (And yes, I very much hope that the answer to #1 above would include the teams currently doing the work.) [[Sam Korn]] ^(smoddy) 23:16, 8 January 2009 (UTC)

Voter eligibility

As currently written, "Any unbanned editor who has made at least 150 mainspace edits by the first day of the calendar month before the election may vote." This would parallel the rules in place for the 2008 ArbCom election, where users needed 150 mainspace edits before 1 November 2008 for an election that took place in December 2008.

I do not think such a stringency is needed for ad hoc elections as opposed to annual elections. I think it would be sufficient to allow an editor with 150 edits on the day that the election is announced to the community, assuming that the election will take place shortly after it is announced. This would prevent what appears to be the motivating concern, namely, that an editor might try to gain 150 edits after an election is announced in order to vote for a candidate, where that voter would otherwise be ineligible. I see the countervailing viewpoint, namely, that maybe a voter should be experienced to a minimum threshold not only at the moment the election is announced, but should have "150 edits plus one month of experience" to combine edit count with months of experience. I recommend a more lenient policy, but since I see it both ways I'm also willing to support the criterion as it stands. Crystal whacker (talk) 22:09, 8 January 2009 (UTC)

• support the criteria as it stands because we want to stop people from 'hearing' about the election (which could occur before its announcement) and responding, while still not impacting too many new users. Protonk (talk) 23:16, 8 January 2009 (UTC)

Vetting procedure

Very quick question, but am I correct in my understanding that the vetting procedure has already begun for new CU's, with some people already selected and contacted to go through the process? Ryan Postlethwaite^{See the mess I've created or let's have banter} 00:24, 9 January 2009 (UTC)

Yes, that is correct. In addition, there should be (or should have been) a public call for people to apply for vetting. Actually, the proposal says it better than I have: "ArbCom will periodically invite applications from the community for CheckUser and Oversight permissions, although any editor may apply for pre-vetting at any time." As FloNight has said elsewhere: "We wanted to get something in place for the next appointment which will be fairly soon". So yes, the process is in effect in motion and any editor wanting to be considered should apply. Roger Davies and Rlevse have been handling most of the admin stuff for this, so contact them. Of course, if this proposal runs into trouble, then this may have to be reconsidered, but that is what I understand is happening. Carcharoth (talk) 00:35, 9 January 2009 (UTC)

Tidying discussion

Some of the straw poll comments are turning into threaded discussions. Would anyone object if those discussions were moved here, with links from the comments to the relevant section of this talk page? Carcharoth (talk) 00:37, 9 January 2009 (UTC)

Please do. Tiptoety ^talk 01:39, 9 January 2009 (UTC)

Funny

Does this really differ that much from this, which I tried to push through about a year ago? I don't think it does much. But I was basically slapped in the face by Arbcom when I proposed it: we were told the whole thing was illegitimate; I had "sour grapes" according to David Gerard; it was "impossible" for the community to vote where there is an arbcom; FloNight did not believe that "CU access consensus discussion scales with a wiki as large as Wikipedia English"; and that the community apparently preferred ArbCom choosing people secretly. Mackensen, an Ombudsmen claims on the archive of the talk page that "The present system works just fine, barring sour grapes", a highly rude comment, and inaccurate also. Sam Korn claims that an RFA system would essentially be a popularity contest, yet he supports this poll.

So, what's changed in a year? No changes policy-wise as far as I know. This should have happened last time it was proposed, almost one year ago. Perhaps this is why 2008 arbcom was so poor (and all the ArbComs before it). Too many arbitrators hating the idea that their special little "gift" (as Mackensen, Wikimedia Ombudsman puts it) would be stolen from them. It's really pathetic - but on the other hand, thank you to the arbcom of 2009 who have finally done the right thing and listened to the community for once. It sure makes a change, and a good one at that. Majorly talk 00:53, 9 January 2009 (UTC)

I think the difference was in perception (i.e. who presented it), though there are subtle differences in what was presented as well. A change presented by the group that has been charged with dealing with something can sometimes have more chance of gaining traction than something presented by a single person. No comment on the rest of what Majorly has said. Carcharoth (talk) 01:08, 9 January 2009 (UTC)

It was not just me, and the proposals are essentially the same, minus minor differences such as number of votes or whatever. They're both involving the community to an extent that they can voice their opinion on a candidate and have it listened to. This should not have taken a year to get going. Majorly talk 01:15, 9 January 2009 (UTC)

There are many things that shouldn't have taken a year to get going. Which one next? Carcharoth (talk) 01:18, 9 January 2009 (UTC)

ArbCom vetoed it though, saying it was a bad idea and would never work - it was ArbCom's gift, according to checkuser, Ombudsman and former arbitrator Mackensen. If ArbCom had bothered to listen to the Community (which I recall, were not completely opposed), this would have happened a lot sooner. Why the sudden change of heart? Majorly talk 01:25, 9 January 2009 (UTC)

Nevermind, it's happened now. Majorly talk 01:37, 9 January 2009 (UTC)

Regime change? ;) - Mailer Diablo 05:46, 9 January 2009 (UTC)

I would say the three substantive differences between the two proposals is the advance vetting by Arbcom (thus meeting the due diligence expectations of the WMF without the drama of rejecting an elected candidate who did not meet requirements), the voting method proposed (which is simpler and does not require assessment of consensus by a group that indicated it didn't feel that responsibility was within their purview), and the acknowledgement of the WMF having the right to refuse proposed candidates. I think Carcharoth may also have a good point. Two, actually, as I don't have any comment on the rest of what Majorly has said. Risker (talk) 01:21, 9 January 2009 (UTC)

You're looking at it too closely. The whole thing was proposed, with ideas for how it could work. The issue is ArbCom shot it down without even looking at the details. It could have been identical to this. The current version is not the original proposal either. The point is, it was vetoed by ArbCom, without even being given a chance - and of course ArbCom called it illegitimate, "sour grapes" and all sorts of incorrect things. I'd still like to know why the change of heart - why is ArbCom suddenly interested in what the community thinks? Majorly talk 01:25, 9 January 2009 (UTC)

Actually, it doesn't matter why - it's what they should have always been doing, so no point in questioning :) Majorly talk 01:36, 9 January 2009 (UTC)

The recent appointment of ten new arbitrators has completely changed the landscape at ArbCom. That might be a factor :) --ROGER DAVIES ^talk 04:54, 9 January 2009 (UTC)

You will find firstly that I "broadly agree" with the proposal. I feel it is a workable procedure, unlike the one previously proposed. The differences, as briefly outlined by Risker above, are substantive and important. Finally, I would suggest that the manner by which this proposal was created is significantly more likely to engender support than the one you adopted. [[Sam Korn]] ^(smoddy) 11:07, 9 January 2009 (UTC)

Discussion from Lar's comment

Copied over from the main page. Discussion should be continued below. Carcharoth (talk) 01:46, 9 January 2009 (UTC)

I am not sure about the "arbcom decides when more are needed" part, I'm thinking that may not be ideal. There should have to be an election at least every X months (duno what X is but at least once a year?) The rest seems sound enough to broadly agree with. ++Lar: t/c 23:33, 8 January 2009 (UTC)

Hmm - is a surplus of users with access to otherwise confidential information really a good thing? Holding regular elections would just keep giving more people access to this information even though there might not be any more people needed. The way I see it is: there should only be as many checkusers/oversights that are needed and ArbCom should have a pretty good handle on when things are falling behind. Your thoughts on that? - Rjd0060 (talk) 00:06, 9 January 2009 (UTC)

I suggested that we go to regular elections if needed, for example on the first Sunday of each quarter, and we are discussing doing it, still. This was one of the few loose ends that we need to decide but not urgent. We wanted to get something in place for the next appointment which will be fairly soon. FloNight♥♥♥ 00:21, 9 January 2009 (UTC)

Copied over from the main page. Discussion should be continued below. Carcharoth (talk) 01:46, 9 January 2009 (UTC)

Well - I still think that we should not have any extra users around with access to this sensitive data. If we have a need for additional personnel then the Committee should plan an election. If things are getting done within a reasonable amount time and the existing CUs/OSs aren't "overworked" then that should be a sign that no more are needed at the time. - Rjd0060 (talk) 03:47, 9 January 2009 (UTC)

Yep, it should be based on supply and demand. Tiptoety ^talk 04:49, 9 January 2009 (UTC)

I strongly and fully agree with this. [[Sam Korn]] ^(smoddy) 11:25, 9 January 2009 (UTC)

If this is the plan (to have a surplus of people) then I may oppose the proposal. I'd appreciate further input from FloNight, or any arb, as she said it was still being discussed. - Rjd0060 (talk) 15:29, 9 January 2009 (UTC)

• I'd prefer we just have an election rotation and the committee ask inactive CUs/OSs to stand down or seek re-election (though if is is three seeking reelection and we use simple approval voting that won't work). No one here has a capacity to judge whether or not supply or demand for checkusers will be high >1 month from now. More to the point, I don't think we have an excess of CUs right now. There is CU work to be done at SSP (even if requests aren't explicitly asking for a CU) and the backlog there has a great deal to do with the administrators expected to work on sockpuppet cases being denied technical tools necessary to make a big decisions (indef many accounts) on anything but WP:DUCK cases. Protonk (talk) 05:22, 9 January 2009 (UTC)
  • This lack of attention might be helped (a) if there were requests for CU attention (I know I could give more time to the task!) and (b) by the soon-to-arrive WP:SSPI, which should make the system more streamlined. I would prefer to make any decisions about an excess or a need in a few weeks' time, when this process and the new CUs on the AC have bedded in. [[Sam Korn]] ^(smoddy) 11:25, 9 January 2009 (UTC)

Removal of rights

As brought up by a few editors at the straw poll, there is a feeling that a similar process should be in place to "de"-CheckUser or "de"-Oversight someone. While I agree there needs to be some very stringent checks and balances, much of what CheckUsers and Oversights do is not viewable by the public. Many of the reasons for their actions are justified through emails, mailing lists, or off-wiki discussions with ArbCom members, other CUs/OVs ect. For this reason, a similar system for removing CU/OV rights (where the community would !vote) is not a good idea. As such, a totally different system needs to be put in place, possibly something like the review board? Tiptoety ^talk 04:57, 9 January 2009 (UTC)

I agree. I think a more effective method for auditing Checkusers and Oversight needs to be implemented. I think that requiring a No Confidence style vote for an CUer or OSer might be used by the new review panel as one way to remedy for a complaint in certain situations. But in other situations, direct permanent removal of access would be the most appropriate way to handle the matter. Because of the varying types of problems that may occur as well as concerns about privacy, no single method will fit the bill. FloNight♥♥♥ 14:15, 9 January 2009 (UTC)

Re:My oppose

Roger Davies asked me to expand on my oppose in talke. OK. The proposal claims to split scrutiny between the "community" and "ArbCom" without specifying how this is to be done [beyond making some candidates go through an election process]. I'm worried that this will be used in order to turn this into the Review Board that appears to be having problems under its own name on the other page. Other problems:

• 1) doesn't specify if ArbCom will have its own rules for vetting beyond whim. It would be nice if such selection principles are stated, even if they are obvious.
• 2) Editors who are arbitrators, or who have been arbitrators in the past 12 months, are disqualified from voting. - what difference does this make, all the candidates will be pre-vetted anyway?
• 3) Simple CU and OSers "elected" pose political problems not necessarily conducive to the smooth operation of the site, and it might prove awkward to remove them given the legitimacy they can claim. Not sure really if this is a good thing or bad thing actually, but it seems strange to create more potentially entrenched problems when we still haven't set something up that keeps them in line. Deacon of Pndapetzim (Talk) 12:00, 9 January 2009 (UTC)
  • The Committee very much want to hear the views of people that see problems with the proposal. Our intention is to have one or more FAQs that covers the specifics in more detail. But before we wrote it and the other detailed on-wiki pages, we wanted to get the general thoughts of the Community. So far the comments are inline with our internal discussions. The Committee largely supported moving to a process that included a Community election. We started out with several good ideas. They needed to be merged in a way that brought the best concepts from each one. Our intention was to present the proposed policy in a form that could be voted on, but with the understanding that we were going to make changes to it to smooth out some of the final issues. I share your concerns about #3. But since most of the people that currently have OS and CU access were elected in a similar process (ArbCom election), I don't think we are really adding an new element. FloNight♥♥♥ 12:42, 9 January 2009 (UTC)

How do the Arbs themselves fit into this?

One issue that the proposal doesn't consider is the fact that a significant number of checkusers and oversighters are sitting or former arbitrators. It notes that it is current practice for CU and OS to be given on request to these individuals, but not explicitly whether this practice is to change. If this process is intended to become the sole channel for acquiring CU/OS rights, how will the vetting process accomodate the fact that a significant number of candidates are nominally on the reviewing panel? Naturally arbitrators asking for permissions would have no direct say in their own 'clearance', but how would the vetting process be conducted in that situation to ensure that is it A) transparent to accusations of cabalism, and yet B) private in the sense that other Arbitrators are free to voice concerns? It's not fair on either the candidate or the other Arbitrators to expect a (potentially controversial) discussion of the candidate to take place on a mailing list or internal wiki that the candidate has access to; yet it would be unacceptably opaque for such a discussion to take place outside the normal ArbCom channels.

There is also the question of how this proposal meshes with Arbitrators' need to have occasional access to the Checkuser and Oversight logs for the purposes of evaluating Arbitration Cases. Is there an implication that ArbCom will ask the stewards to 'slice and dice' the CU/OS permissions to give Arbitrators log access only unless they take this new route to gaining the full permissions? If so, should the fact that these permissions are acquired through a separate process be mentioned? Happy‑melon 13:02, 9 January 2009 (UTC)

Because it has been the past custom for Arbitrators to get access to Oversight and Checkuser by asking for it, during the recent ArbCom election the Community took that into consideration when they made their selections for the 2009 Committee. IMO, this process will continue to be used since I think that Arbitrators will need to do checks for themselves when deciding some cases. Mentioning this custom in the policy would clarify the issue, yes. My hope is that we will move to a different form of auditing Checkuser and Oversight access where ArbCom will not be directly involved. The current proposal for a Review Board does not separate that function from ArbCom, yet. I think we need to move in that direction. FloNight♥♥♥ 13:55, 9 January 2009 (UTC)

Philosophical problems

My first annoyance with this is that there is a "poll" prominently on the project page here, whilst discussion is relegated to the back. I am REALLY concerned about this, because it seems to be indicative of the way things are headed on wikipedia. We are moving away from discussion, where the strength of the case is what carries the day, to "let's poll/elect/vote". I suspect a minority of those voting will bother to read the discussion, and so they vote on the strength of their gut reaction to the proposal, without listening to what (perhaps more experienced) users think before coming to a view.

That procedural concern - that numbers now matter more than clued discussion - is at the heart of my difficulty with the proposal itself. Yet again, arbcom respond to a minority who have been badmouthing it and expressing a lack of confidence in it, by expressing a lack of confidence in its own judgement. (It was the same with the review board - not only are arbcom presenting themselves as unfit to review, they also expressed themselves as unfit to choose the reviewers!). Now, let's think how this works. We need checkusers, say. Arbcom invites people to submit their names. Out 15 arbs (who are supposed to be out most experienced and trusted users) vet the names and removed anyone unsuitable. Then, rather than make a decision, they submit the choice to an RfA-like process, where anyone with 150 edits can jump in and say "oppose - he's not used enough edit summaries/ written enough DYKs," or "he blocked me once" or "he deleted my userbox 4 years ago and I've hated him since". The one thing the community seems to agree on is that RfA sucks, why are we preferring it to a discussion by out 15 chosen best members?

(aside - how on earth can anyone with 150 edits assess what's needed in a checkuser on wikipedia, know the history, and assess a candidate's suitability in a rounded way? No really, that's a serious question?)

I'd also ask: "what's broken about the current system?" It seems to me that the most worrying bit of checkuser oversight recently has NOT been those recently appointed by arbcom, but older hands who are not sensitive to recent concerns. We'd be better addressing that. It seems more urgent to end the notion that once you've got the tools you've got the tools, since that leads to the sense of power. Why not allow arbcom to appoint checkusers in the first instance, for an initial one year term. If they wish to have the tools beyond that, they need the approval of both arbcom AND the community - renewed every 12 months (or maybe 24). Yes, that would still involve an election type process, but people be voting on the concrete question of how they'd used the tools and responded to any concerns raised rather than just whether they liked the person in general. Better that than another popularity contest.

OK, I'll admit my bias here. At times in the past I was doing a LOT of BLP related work, I considered asking for oversight. (I've no interest in it now.) I think I could have probably convinced arbcom that I'd use it reasonably and with enough clue. However, I'm not fool enough to think for a minute I'd come through any mass participation beauty contest. But then, maybe a system that rules out people like me is not a bad system at all.--Scott Mac (Doc) 19:32, 9 January 2009 (UTC)