Wikipedia:Requests for comment/Checkuser and oversighter selection

On May 31 the Arbitration Committee announced that the May 2010 CUOS elections had led to an "unsatisfactory result", in that of 4 candidates for Checkuser and 7 for Oversight, only one passed the 70% threshold set for the process.

These roles are important ones for the community and the purpose of the election is to ensure high quality appointees. The purpose of this RFC is to consider the Arbitration Committee's request for community input.

Background

Further information: WP:CUOS § Appointment methods to date

Checkuser and oversight cover two major areas where trust and privacy are required. It is the direct responsibility of the Arbitration Committee to choose and manage these roles on this project. In the past appointments were based upon internal discussion, and then internal discussion informed by communal feedback. Since 2009 appointments arose from internal vetting discussion (Arbcom affirms its trust in each resulting candidate), followed by a community election with a 70% hurdle. 2009 involved public voting and 2010 non-public voting.

Without undue analysis of what went on, the questions at this RFC are:

1. How should the Committee proceed now (the roles need filling), and
2. Is there some change that might help ensure a satisfactory result in future.

Four concrete possibilities to kick-start initial discussion (there may be others):

• Keep as is - but mandate the appointment of candidates who have the highest results (although under 70%).

• Keep as is - but reduce hurdle in future. (The Arbcom election itself only requires 50%)

• Change in future - revert appointment system to being based on communal input rather than communal voting. (Example how this would work).

Advantages - concerns are heard (not just vote count), takes less community time, guarantee of successful results, previous use produced very high quality input ("a very high signal to noise ratio").
Disadvantages - community voting doesn't take place, checking of issues and final decision left to Arbcom.
Note that in both election and input versions, the reasons for a final selection are nonpublic - SecurePoll does not record comments under voting, nor did arbitrators provide detailed reasons.

• Change in future - make appointment system completely community based.

Advantages - point of principle for some that the community should decide all matters directly, gives community full control.
Disadvantages - May lose Arbcom checking of competence and suitability (by users who use that tool daily and know it well), where an Arbcom exists it is always responsible for supervising/removing CUOS, no better result (a 70-80% hurdle is mandatory for community-only elections).

• Other...?

FT2 ^{(Talk | email)} 05:43, 31 May 2010 (UTC)

Discussion

View by Shirik

The majority of the Wikipedia community avoids oversight-level issues and WP:SPI. This is not a fault of the community; this is a good thing, as these are areas which we would rather not have, but as a necessary evil must have to control problematic users. The overwhelming majority of users on Wikipedia do not follow the day-to-day processes in these areas, and thus may not be aware of their daily needs and issues.

At present, the sockpuppet investigations area has been struggling to keep up with its checkuser backlog. There is a desperate need for checkusers, especially in the "quick cases" area, where urgent cases have unacceptably remained open for days without attention. However, the majority of Wikipedians do not realize this dire issue because it is in an area they do not venture into frequently. This is not a fault of the community; the community benefits from leaving only a few, dedicated people to deal with this problem. However, the community must ensure that those dedicated people have the tools necessary to do that job.

For this reason, a fully-open, unchecked election cannot satisfy the needs of the community. Election by users that are not intimately familiar with the current needs of these areas will typically result in indecision, as the roles of checkuser and oversight are understandably met with fierce scrutiny by electors. While such scrutiny is necessary, it is better dealt with by users that are familiar with the area in which these candidates will be working, and by users familiar with the current need for those positions to be filled. The Arbitration Committee is the best group to fill this role. They are familiar with the needs of both the oversight and checkuser areas. While their decisions may periodically be met with opposition, it is perfectly understandable that this decision will not reach 100% support and the primary purpose of the Arbitration Committee is to make those decisions which the Wikipedia community at large is too indecisive to make on their own.

Users endorsing this view

1. As author --Shirik (Questions or Comments?) 06:33, 31 May 2010 (UTC)
2. Support and comment on the Oversight election.

   While on the surface it seems disappointing to the candidates and those that did not pass the vetting process, the need for Oversighters in a couple of different timezones is needed. However, in my opinion as one of the most active non-arbitrator oversighters, we do handle requests in a timely manner. There are slips in the cracks and gaps to be filled, but overall the system is running well. I don't follow checkuser so I can't comment there. As mentioned, though, it does on the surface end up in hurt feelings by all the nominees and editors and supported them. This does not mean a failure of the system, but does imply a bit of malaise. Keegan (talk) 07:00, 31 May 2010 (UTC)

3. I couldn't agree more. In cases where the community is incapable of showing a unified voice such as in this election, (and probably all CU/OS elections in future) I honestly feel that Arbcom needs to play an active roll. Arbcom should handle CU and OC appointments on the proviso they do take note of the community input. No matter what way you look at it, we need more checkusers to deal with the ever growing backlog. As it stands Arbitrators get CU/OS by virtue of being Arbitrators and they can hand that permission out to whom they wish, they may as well handle the elections too.   «l| Promethean ™|l»  (talk) 08:19, 31 May 2010 (UTC)
4. Agree. Clearly, SecurePoll-voting hasn't worked out for us very well. However, it is the most fair system we have out there, and replacing it with public voting a la Wikipedia:Arbitration Committee/CheckUser and Oversight/August 2009 election would not be acceptable to me, as it only encourages herd-voting. Unless we are willing to move entirely to a system like RfA (public voting with comments being judged), this seems like the only way to go. NW (Talk) 13:29, 31 May 2010 (UTC)
5. This is exactly what I said when the results were announced. (X! · talk)  · @671  ·  15:06, 31 May 2010 (UTC)

View by zzuuzz

The high levels of neutral votes, particularly seen for the oversighters, indicate a strong indifference about the candidates among the community. This indifference, neither an acceptance or rejection of the candidates, is mainly due to a lack of information about them. It is often implicit support pending any substantial objections. Several of the candidates are either relatively inactive, or relatively inexperienced, and not much is generally known about them. Despite a section for it, there was only one comment made about any of the candidates in this entire election. More discussion about the candidates should be encouraged and voters should be able to see what concerns are being raised so they can either accept them or reject them. Failing that, an open voting system where concerns and support are expressed openly should be re-adopted. Voters generally react to what has gone before, adding support or opposition as it is required to achieve the right outcome. Arbcom should also consider going by the opposition alone as for these types of roles they should be weighing community opposition more strongly than community popularity. But more importantly they, and the community, should know what these objections are.

Users endorsing this view

1. -- zzuuzz ^(talk) 09:59, 31 May 2010 (UTC)
2. Aiken ♫ 14:58, 31 May 2010 (UTC)
3. --The Evil IP address (talk) 17:42, 31 May 2010 (UTC)

View by BirgitteSB

I believe the largest factor in the non-result of this election was the use of SecurePoll. While public consensus gathering has its own downsides, it is reliable in gathering a greater and more conclusive input. SecurePoll may continue to be effective for Arbcom elections so long as they held so infrequently, but for most other decision people are too ambivalent to skip the consensus gathering part of the process and precede directly to voting. Open voting brings more people into the process. They may come by the page just out of curiosity to see who's voting for who, but once they are on the voting page there is a greater chance they will cast a vote. And those who are nuetral can often come to a conclusion by reading the comments left during open voting. So what if people end up voting by aping those whose judgment they trust, or other less preferred means than a careful review of information leading to a balanced and personal decision. These less preferred means of reaching decisions should still be preferred over the absence of a decision.

For my own part I never voted in the election. When it first started I looked over the names and didn't see anyone that I felt I knew well enough to decide off-hand. There was no discussion about the candidates to glean information from. I thought to come back later and read all the statements and see if there was any useful discussion once the election had some time under its belt. One thing led to another and I never paid attention to when the election was closing so I forgot to come back. I doubt my experience is unusual.

It is common failing of people to design systems that require humans to be more perfect than they are or else systems that try and push them to be more perfect than they are. SecurePoll relies on both these failings. It requires people to devote time energy to informing themselves and coming to a conclusion for not even the filmiest of incentives. As this election has proven the majority of people simply will not devote themselves to this task without such incentives. It requires this because people have decided to try to push the community away from what some may see as "ugly motivations" for making such a decision and instead try to force the decision to rely on "pure motivations". I would call "ugly motivations" instead "human motivation" and "pure motivations" as "angelic motivations". The world is made of humans not angels. If you try to cut out those motivations that make humans tick, you won't end up with a community full of angels rather you will have an empty community. Humans are driven by social interactions. If you restore the social interactions to the decision making process you will gain candidates who attain 70% approval.

1. As author--BirgitteSB 14:47, 31 May 2010 (UTC)
2. Aiken ♫ 14:52, 31 May 2010 (UTC)
3. (X! · talk)  · @671  ·  15:06, 31 May 2010 (UTC)
4. -- zzuuzz ^(talk) 16:07, 31 May 2010 (UTC)
5. LessHeard vanU (talk) 17:33, 31 May 2010 (UTC)
6. Salvio ( ^{Let's talk 'bout it!}) 17:46, 31 May 2010 (UTC)
7. There is a possibility that the secret vote impacted on the voting, and that might be worth considering, though looking at the numbers involved in the 2009 secret ArbCom election compared to the 2008 open ArbCom election, this may not be the full reason. I do, however, support the notion that as many of the candidates were not well known it would have been helpful to have had more open discussion. SilkTork *^YES! 18:57, 31 May 2010 (UTC)

View by Davewild

The idea that open voting leads to more people voting, as opposed to an election held via securepoll, is not substantiated by the evidence of the checkuser/oversighter elections. This election conducted by securepoll saw the highest turnout (support+oppose votes) for one candidate of 325 and a lowest turnout of 172. Compared to either the February 2009 election (highest 123, lowest 57) or the August 2009 election (highest 262, lowest 76), both conducted by open voting, turnout was increased this time. While there is a problem with this situation here due to a lack of successful candidates, reduced numbers of people involved in the election is not a problem.

Users endorsing this view

1. Davewild (talk) 15:21, 31 May 2010 (UTC)
2. Very factual and non-POV view. Suomi Finland 2009 (talk) 17:08, 31 May 2010 (UTC)
3. Statistics. LessHeard vanU (talk) 17:35, 31 May 2010 (UTC)
4. Accurate and well though out. Alzarian16 (talk) 18:29, 31 May 2010 (UTC)
5. Agree. I feel that lack of information and lack of discussion was an issue. It appears to a lack of deciding factors (either positive or negative) rather than lack of numbers that returned an unsatisfactory result. SilkTork *^YES! 19:00, 31 May 2010 (UTC)
6. Purely based in fact. The turnout was fine. The indecisiveness was not. --Shirik (Questions or Comments?) 19:54, 31 May 2010 (UTC)

View by Aiken drum

Clearly, arbcom are out of touch with what the community wants in a candidate. Looking at the results, there are three candidates with more opposition than support, and all except one oversight candidate have less than 60% support - something considered a fail if it was on RFA. This demonstrates, to me, that the choice of candidates was not a particularly great one, as most failed to gain a strong majority, let alone 70%. In future, anyone should be permitted to run (assuming they met basic requirements), and they should not have to be vetted by arbcom. Arbcom can and does get things wrong, as can be seen here. Barring good candidates (as did happen on this occasion) is unacceptable, especially when there were accepted candidates who clearly have a poor endorsement. Also, take note of Xeno's current RFB. He was a potential candidate, who arbcom decided was not worthy of becoming one. And yet, the community is giving him a clear endorsement of trust.

In short, allow anyone to run, and don't bar good candidates for the role, otherwise we'll end up with nobody getting it. Aiken ♫ 15:08, 31 May 2010 (UTC)

Is there evidence other trusted seasoned candidates would have been likely to get 70%+, had applied to stand, got rejected by ArbCom, and the rejection reason was flawed? It seems possible the comment might be over-quick to assume a specific interpretation or "fault". Can you clarify on the talk page if the clearly.. assertion has evidence? FT2 ^{(Talk | email)} 18:00, 31 May 2010 (UTC)

No, because they were not given a chance. However, I gave the example of Xeno's RFB as an example of demonstrated trust. I know they're different roles entirely, but it's an indicator all the same. I've stricken clearly. Aiken ♫ 18:34, 31 May 2010 (UTC)

Users endorsing this view

1. Aiken ♫ 18:30, 31 May 2010 (UTC)
2. Support mostly (do not support statement that ArbCom is "out of touch") Suomi Finland 2009 (talk) 17:26, 31 May 2010 (UTC)
3. Every admin who's been here for some time should be given the option to be a candidate. Bad ones will fail anyway. --The Evil IP address (talk) 17:42, 31 May 2010 (UTC)
4. I wouldn't support the notion that ArbCom are out of touch, though - given the example of Zeno - the vetting process needs looking at. Let us be more open in our processes - allow any admin in good standing to become a candidate, and if ArbCom members have concerns, they should be raised during an open discussion on the candidates. SilkTork *^YES! 19:08, 31 May 2010 (UTC)

View by Suomi Finland 2009

Support for the Other proposal listed by ArbCom. Take the fuel conservation approach to the oil shortage as a partial solution to CUs.

Details

• Perhaps the duties of CU could be reduced so there is no need for so many? Note that I do not imply that the positions don't need to be filled, just that we might do as well if we take a fuel conservation approach to the oil shortage and CU shortage. (When there is a fuel shortage, fuel conservation by having smaller cars and more people taking the bus is a partial solution; drilling for more oil <appointing more checkusers> is not the only solution). Many CU requests do not need to be done. Perhaps we should act like adults and real editors and think whether an edit would be suited for the Globe and Mail/Wall Street Journal/Sydney Morning Herald or Encyclopedia Britannica? This stance would solve many problems that are brought to the CU. For example, DavidYork71 is accused on 29 May 2010 (see WP:SSI) of being the sockpuppetmaster of Noconcept. The evidence is that Noconcept changes "controversies" (section heading) to appraisals. CU was done but was a waste of CU manpower. Nocontact's edit would fail the Globe and Mail/Wall Street Journal/Sydney Morning Herald test without need for CU.

When I have a shortage of milk, I use it only when necessary. Milk with cereal but not guzzling down cups of milk. The same is with money. If I have a shortage, I buy food, not fancy shoes.

Another possibility is to have one paid CU position. There are many computer science graduates in the San Francisco area that might be interested in the job. One paid, full time CU could do a lot of work.

Users endorsing this view

1. Suomi Finland 2009 (talk) 17:12, 31 May 2010 (UTC)

View by Tony1

Summary: (i) SecurePoll community elections are highly desirable and can work; (ii) SecurePoll suppresses the % support results, and a 60% threshold now looks more appropriate; this doesn't equate to 40% oppose; (iii) ArbCom should be free to appoint further down the ranking list than 60% if it sees fit; (iv) ArbCom should continue to short-list the candidates for such critical, skilled, sensitive roles; (v) re-run the election in a couple of months.

Details

• SecurePoll is a major improvement. I strongly support the use of SP, which minimises the potential of the old system for influence and pay-back, and removes the inconvenience factor—a significant contaminant—that plays on how much time voters are willing to devote to scrolling through the morass. Going back to a completely ArbCom-selected wouldn't be the end of the earth, but community voting ensures wide support for an essential group of functionaries. We should allow SP to work.
• Let's make the new system work. Introducing SP to any vote requires a careful reconsideration of the minimum % support vote required for election. As I pointed out during the WP:ACE2009 election, SecurePoll provides new anonymity and convenience for oppose voting. This was born out in the dramatic rise in oppose votes (vertical axis) and the significant suppression of the ranking % (horizontal axis) for each candidate in 2009 compared with the old public, manual voting system used in 2008 (Figure 2). Fortunately, the arbitrary 50% floor in that election was reached by comfortably more candidates than necessary to fill the eight seats vacant. Introducing SP into any election risks overestimating the minimum required vote once voters are given private and convenient access to the oppose button for any candidate for whom they are not hitting the support button. Unfortunately, it's difficult to predict the extent to which the ranking % (i.e., the % support) will change, although the direction is highly predictable.
• Here's how. May I suggest that 60% would be a more appropriate threshold for CU and OS elections in future, with latitude for ArbCom to lower this on the spot, according to the results? This would be in recognition of changed voting circumstances rather than, necessarily, a drop in standards of community support. Why not hold another vote in a couple of months with 60% threshold, now we know where the figures stand. A 60% ranking does not mean that 40% of voters oppose a candidate; please see Figure 2 again on that issue.

Users endorsing this view

1. As author Tony (talk) 15:31, 31 May 2010 (UTC)
2. LessHeard vanU (talk) 17:38, 31 May 2010 (UTC)
3. --Tryptofish (talk) 19:28, 31 May 2010 (UTC)

View by involved user Beeblebrox

When this process began I was in favor of using the secure poll method as it seemed the low-drama route and all the candidates had already been vetted. Now that we have arrived here I see the downside of it. Not because I "lost" but because I don't know why. I was only asked the same general questions as the other candidates, and received no feedback of any kind on my answers to those questions. Among the oversight candidates, my numbers were the third highest in terms of both support and opposition, yet I have no idea why that is. Did I do a poor job answering the questions? Was there some other problem? Is there simply a large group that participated in this process that just does not like me for reasons not related to Oversight at all? I really have no idea, meaning I also have no idea how to improve things for next time. It's very frustrating. Beeblebrox (talk) 15:54, 31 May 2010 (UTC)

Not sure if this is really something one endorses, but this sounds like an argument for RfA-style "voting". NW (Talk) 16:43, 31 May 2010 (UTC)

Yeah, that's is more or less the gist of it. If we don't know why a candidate is being opposed, how do we find a better one? This process is all sizzle and no steak, there's a pile of numbers, but no reasons behind them. Beeblebrox (talk) 16:47, 31 May 2010 (UTC)

Beeblebrox, I appreciate your frustration, but are there not advantages in the larger scheme in a veil of privacy, a distance between voters and candidates in the actual vote? Perhaps we need to encourage more dialogue between the parties before the vote. How can this best be done? I definitely oppose an RfA-style system; too many people look to see how their friends vote and do the same. Tony (talk) 16:51, 31 May 2010 (UTC)

I think it is normal in an RFC to post views on the main page and have back-and-forth discussion on the talk page, I suggest we move this over there. Beeblebrox (talk) 16:58, 31 May 2010 (UTC)

• Users who endorse this view

1. Beeblebrox (talk) 16:47, 31 May 2010 (UTC)
2. Should not be a straight vote, but a consensus based discussion - ya know, like all other processes on Wikipedia. WP:NOTVOTE, WP:CONSENSUS etc. Aiken ♫ 17:13, 31 May 2010 (UTC)
3. --The Evil IP address (talk) 17:42, 31 May 2010 (UTC)
4. Salvio ( ^{Let's talk 'bout it!}) 17:49, 31 May 2010 (UTC)
5. Absolutely, I have the same issues. I have suggested in the ArbCom page discussion that may be the provision of a anonymous comments box, for recording rationales where desired, in some future poll. LessHeard vanU (talk) 18:41, 31 May 2010 (UTC)
6. Yes. Open discussion is the Wiki way. SilkTork *^YES! 19:11, 31 May 2010 (UTC)

View by The Evil IP address

SecurePoll clearly played a role in the relatively low voting people, but there are other things as well. This election, as well as actually the whole Arbcom, is extremely difficult to find. By this, I mean how one person gets to the information pages about it. It took me months until I first found a page where the word "Arbcom" was mentioned, long after I had read the help pages several times already. It's obvious that this organization is in only too few help pages, hardly linked (for example not within all those navigation boxes) and pretty difficult to find. It was also just randomly that I noticed that there was a Checkuser and Oversight election ongoing. So, this is something that really needs some fixing. But also the usage of SecurePoll played a clear role: Things like RfA, RfB have the advantage that they're transparent and you see what other people think. The candidate is able to see what they could make better and potentially improve this behavior in the future. This is not given by using SecurePoll. And I expect from every Checkuser/Oversight candidate that they're able to handle critic on their own person, otherwise they're clearly unable for their job. Seeing what other people oppose for (for example if they give links) makes it also easier to make oneself a judgement. One argument for SecurePoll seems to be that more people might oppose, but honestly, if you don't have the courage to express your opinion publically, then you should absolutely reconsider why it's like that and if your opinion then might not be wrong. People shouldn't be given the possibility to oppose someone if they're in doubt of their own opinion. Concerning privacy, I consider privacy issues to be unlikely, and if there are really some, we have oversighters to deal with them. The last reason for SecurePoll seems to be the influence by so-called friend's votes, but I must say that this is solely the voter's problem. If they're willing to throw their own opinion and vote away for someone else's opinion, so let'em, it's their problem. Not stuff that we have to worry about. --The Evil IP address (talk) 17:37, 31 May 2010 (UTC)

Users who endorse this view:

1. Obviously. --The Evil IP address (talk) 17:37, 31 May 2010 (UTC)

View by Alzarian16

The use of SecurePoll does not appear to have reduced the number of users participating in the elections, so its use should not be discouraged. However, the problem of low support percentages is one which clearly needs to be overcome if the system is to be satisfactory. I suggest that in future elections voting Neutral should not be an option. The incredibly large number of Neutral votes has led to a number of users' valid contributions proving unhelpful in determining a consensus. If Neutral is not an option then users who intend to support or oppose one or two candidates would be encouraged to give a firm opinion on the other candidates, leading to a firmer consensus either way as opposed to the 50-70% support ratios seen in this election. If this is adopted it could also be useful to rerun this set of elections using the new system , although this would have to be at ArbCom's discretion. Alzarian16 (talk) 18:27, 31 May 2010 (UTC)

Users who endorse this view:

1. As author. Alzarian16 (talk) 18:27, 31 May 2010 (UTC)
2. (except "no opinion" on the last sentence.) LessHeard vanU (talk) 18:43, 31 May 2010 (UTC)

View by Collect

It is unclear that "the roles need filling" is perceived as a truism by the majority of voters. If anything, the large amount of negative and neutral votes indicates that a large number of editors do not agree with that claim.

The only area where there might appear any urgency at all is in the area of sock investigations. To that end, ArbCom could institute a new position of "SPI checkuser" as a limited subset of what "Checkuser" entails, and restricted solely to that area. There appears no appreciable need for more Oversighters as far as the general population is concerned.

Changing the normal election percentages or the like is probably unwise, as would be re-running elections until one can convince 70% of the editors that a person must be elected for a position. Collect (talk) 19:02, 31 May 2010 (UTC)

Other Users who endorse this view:

View by LessHeard vanU

A suggestion for "other"; As the process is for a position of trust, then only oppose votes are permitted. Any candidate not receiving an oppose in the SecurePoll is default supported, per WP:AGF by the registered voter. On this basis, nearly all candidates in this most recent vote would have achieved the statutory percentage.

Users who endorse this view:

1. LessHeard vanU (talk) 20:53, 31 May 2010 (UTC)