Wikipedia talk:RfA reform 2011/Candidates

RfA is a horrible and broken process

(Thread copied from WT:RFA2011. Please continue the discussion here)

When exactly did RfA break? Looking back to 2003, when Jimbo Wales said becoming a sysop is *not a big deal*. the process of becoming an admin was very different to how it is today. It looks to have been an almost entirely stress-free process, with people who would be WP:NOTNOWed today passing with 3 supports and 0 opposes. But then it was a very different encyclopedia back then. Is it realistic to expect RfAs to be stress-free when we also expect a certain level of competence from our admins? Catfish Jim & the soapdish 19:19, 22 May 2011 (UTC)

I suspect the breaking was a gradual process rather than an overnight one. If you look at User:WereSpielChequers/RFA_by_month you might conclude that it broke in the Spring of 2008, but what seems to have happened then is that Rollback was unbundled, and "good vandalfighter" ceased to be a passport to adminhood. I'd agree that RFA should be a different place today than it was in its earliest days, but one of my concerns is that the shift in focus from judging people on their edits to judging them on the Q&A section is less likely to screen out candidates who aren't ready for the mop. Another concern is that the community is seriously divided as to what constitutes the requisite level of competence, and those divisions are played out in RFAs rather than in an abstract discussion as to what the general criteria for admin should be. If I went to a job interview and the interviewing panel spent half the interview arguing amongst themselves as to what the job spec should contain I would regard that organisation as having a dysfunctional recruitment process. So though some of the arbitrary criteria are certainly more exacting than they once were, and it is obviously more stressful than in the early days, I'm not convinced that it is more likely to sift good candidates from bad than RFA did in 2005/6. ϢereSpielChequers 20:28, 22 May 2011 (UTC)

I disagree that the community is seriously divided in the way you suggest. !Voters, on the contrary, seem to unite during RfA, with a few anomalies. There have, I estimate, been more 0-10% and 90-100% supports at closure than there have 40-60%. The latter could be predicted to be equally common (if not more common) if community subdivisions regularly found candidates that certain factions favoured and others did not. Jebus989^✰ 21:25, 22 May 2011 (UTC)

Yes there are lots of candidates that almost everyone can agree are NotNows, and some where support is over 90%. But neither fact is incompatible with my observation "that the community is seriously divided as to what constitutes the requisite level of competence". Division about the requisite level of competence is not the same as having community factions that favoured some candidates and not others. Take for example the issue of tenure, many RFA voters like to see 12 months activity, a few expect 15 months, others might support a candidate with only 6 months activity. But we don't have a rival faction opposing oldtimers for being part of the 2008 wiki-generation. Equally with editcountitis we have RFA !voters looking for minimal edit counts ranging from 3,000 to 6,000; I've even seen editors say they expect at least 8,000 edits, and one RFA this year succeeded despite some opposers citing notNow because of the edit count. But we don't yet have many voters with incompatible RFA criteria, if for example one third of !voters looked for edit count in the 3,000-9,000 range and another third expected over 10,000 edits then it would be impossible for any candidate to get a consensus level of support. RFA is broken but it is still possible for some candidates to get through. ϢereSpielChequers 22:22, 22 May 2011 (UTC)

What we are waiting for, is someone from the task force to volunteer to make a table of criteria compiled from users' essays on the subject, from which we can start to make some recommendations for minima. Currently there are no criteria at all; the only official statements are of the vague 'you should' recommendations such as (contrived example} 'You should have been around long enough to have a basic understanding of policy, and have enough edits to know how to create and delete articles.' RFA is broken but generally those who should pass do get through - at least on their second attempt. The problem is that many potential candidates of the right calibre are refusing to subject themselves to a process that spends more time arguing about itself and with itself, rather than focusing on the candidate and making relevant, objective comments to support their !votes, and that allows candidates to be insulted and attacked with impunity at levels that would attract an indef block elsewhere. Kudpung กุดผึ้ง (talk) 23:18, 22 May 2011 (UTC)

Are you both suggesting a guideline be created which said (for example): ·Candidates should have 5000+ edits ·Candidates should have 12 months of activity etc.? Doesn't that remove any need for an RfA? If we attempt to set some kind of boundary on what opinions are acceptable rationales, they would undoubtedly be used as counterarguments ("What you expect 10,000 non-automated edits? WP:HOWTOVOTE says he only needs 5,000") so then you can then do away with the process and grant administrator rights to those as we grant autoconfirmed status, when a set of conditions are fulfilled. You are essentially suggesting everyone must enter the RfA discussion with the same mindset, else it's a 'seriously divided community' which is constantly 'moving goalposts'. It's just a discussion, and I think you see 100+ editors in agreement over candidates more often than you could get a random sample of 100 people to agree on, well, anything Jebus989^✰ 12:50, 23 May 2011 (UTC)

I'm now looking at getting that very table Kudpung. Might take me a little bit, there's a lot to go through (I'm currently at 106 essays and counting). As for whether RfA will be required, I think this should be used as a guideline to help prevent NOTNOWs... but lets see what the results are before we get too excited ;) Worm^TT · (talk) 13:39, 23 May 2011 (UTC)

Wow, Worm! That's a heck of a task - I had no idea that there really were that number of essays to go through! You have my unbounded admiration for taking this one on - not something that I would have wanted to tackle! Pesky (talk …stalk!) 04:48, 24 May 2011 (UTC)

Well, I've made a start User:Worm That Turned/RFA criteria. The metrics will be fairly easy to judge statistically, but I'm going to go through and have a look at the advice too (when I get time). That'll be the big task! Anyway, hopefully I'll have some stats for us by the end of today. Worm^TT · (talk) 08:12, 24 May 2011 (UTC)

Some of those criteria lists are hopelessly out of date. While those of User:BryanG/RfA criteria (the first one I clicked on at random) may have been indicative of the state of play in 2006, I suspect they're of little use today. Catfish Jim & the soapdish 09:14, 24 May 2011 (UTC)

That's a good point, and I made a note to that effect at the top of the page. I was planning to go through twice, once using all data, and once using only stuff that's been updated since Jan 2009 (from non banned users). I'm just worred that if we use the smaller sample, we won't get quite as representative view. I'm also tempted (when I get time... again) to go through the successful RfAs since 2010 (maybe 2009), and note down the edit count, tenure and mainspace percentage at the point of passing. Worm^TT · (talk) 09:21, 24 May 2011 (UTC)

There are only perhaps 20 or so essays that state clearly defined criteria. There's no need to go trough them all I have, and that why I have left some notes against some of them that are worth reading. There is no suggestion here that "What you expect 10,000 non-automated edits? WP:HOWTOVOTE says he only needs 5,000". If one does what I suggested and makes a sortable table that can caluclate the average criteria, it will most probably come up with some surprisingly acceptable results. Let's get the data before we jump to rash conclusions. Kudpung กุดผึ้ง (talk) 09:37, 24 May 2011 (UTC)

Well, I've been through about 50 so far, and about half have some form of clearly defined criteria. When I get through them all (I've started so I'll finish) I expect we'll have about 60 with numerical data. Take it back down to the ones in the past year or so, I think 20 is probably not far off, I'll get that metric up too. Worm^TT · (talk) 09:41, 24 May 2011 (UTC)

Ok. Early stats are in. General thinking of the essays with metrics (about 50% have metrics) is that you need a minimum of 2500 edits and 6 months experience. NB that's based on opinions, not what actually happens, and includes out of date data. I'll be looking at real data and vetted opinions too. Also, note that about 1 in 6 essays looks for a clean block log and 1/3 requests no blocks in at least the last 6 months. Worm^TT · (talk) 11:01, 24 May 2011 (UTC)

Looking at this year, successful candidates had about 20-30k edits, had been around nearly 4 years and had been actively editing for nearly a year. Only 3 candidates passed with less than 6k edits, and only 5 with less than 10k. Worm^TT · (talk) 12:20, 24 May 2011 (UTC)

Well, 2010 is different. Quite a few less than 10k there, including a couple around the 3000 mark. There is one successful at around 1000 edits, though he had over 1m cross wikis. Worm^TT · (talk) 14:17, 24 May 2011 (UTC)

It might be worth factoring in automated edits in there, as it's something I've noticed brought up in RfA and on various peoples' RfA criteria... For instance, percentages from this year:

• Sadads 58.40
• DeltaQuad 25.02
• Catfish Jim and the soapdish 14.38
• RHM22 0.00
• Bahamut0013 6.38
• Salvio giuliano 49.58
• Feezo 9.23
• Valfontis 7.18
• Fæ 47.51
• JaGa 40.41
• Peridon 0.47
• Gfoley4 41.16
• Neelix Too many edits
• Kudpung 17.28
• Boing! said Zebedee 44.92
• The Bushranger 3.50
• ErrantX 9.18
• Rami R 22.74
• ErikHaugen 0.04
• 5 albert square 71.85
• Acdixon 0.00
• Ponyo 0.01
• Gonzonoir 15.07
• Smartse 12.76
• Gimme danger 0.00
• Lear's Fool 53.30
• Ironholds 30.87

(might be worth checking some of those zero values)Catfish Jim & the soapdish 14:23, 24 May 2011 (UTC)

That's a good idea, unfortunately, it's regularly included on the talk page, meaning that the percentage we get is actually going to be based on their automated edits up to today. Might be signifcantly different from when they gained adminship. (or am I blind and are they there?) Worm^TT · (talk) 14:41, 24 May 2011 (UTC)

It should be possible to get those figures using this tool as you can specify a date range. Makes it a bit more work though. Catfish Jim & the soapdish 15:01, 24 May 2011 (UTC)

Ideal. I'll get on that tomorrow. :) Worm^TT · (talk) 15:04, 24 May 2011 (UTC)

Congratulations - this is excellent work, and the kind of effort that moves this project positively forward. Wales has stated again on his tp yesterday that for proposals with far-reaching board wide implications, data driven arguments are far more important than emotional speculation. We need to get this data into sortable tabular form.

The issue of automatic edits raises interesting points. A candidate who has made 50,000 edits all of which are auto has not demonstrated any knowledge of content or policy. Clicking away at Huggle at 4 pages a minute does not prove anything, while Twinkle also increases the edit count by two for the price of one because it includes not only the tag, but the warning placed on he user's tp at the same time. Manual edits are the criteria to use, and using a percentage may not be the best metric.; 90:10 doesn't look good, but at 90,000:10,000 is still a healthy 10,000 manual edits, whereas 9,000:1,000 would seriously cause me to consider opposing a candidate for lack of manual experience. Kudpung กุดผึ้ง (talk) 04:00, 25 May 2011 (UTC)

"The user should have at least X edits"

Well, the edit count roughly shows the experience in editing. But to have enough experience, you don`t need 10000 edits. Good behaviour, edits of good quality, enough experience(in the article and other namespaces of relevance), Knowledge of the rules, exspecially of Wikipedia:Copyrights and Wikipedia:Neutral point of view, and no indication of possible power abuse; that is what you should look for if someone wants to be an adminstrator.--Müdigkeit (talk) 05:17, 25 May 2011 (UTC)

Nobody is suggesting here that the minimum should be 10,000 edits. The figure is purely hypothetical for the purpose of illustrating the goals of the research that is being done here. Criteria for adminship, if any, will be proposed based on those results. Nevertheless, edit count is still one of the important metrics. I'm an admin and I know virtually nothing about copyright so it wouldn't have been a good criteria to evaluate my work on. A minimum number of quality edits in any fields that represent the main areas that most admins will work on, are important, and of course neutrality and civility. Kudpung กุดผึ้ง (talk) 06:26, 25 May 2011 (UTC)

• I've previously written about this elsewhere, but it seems worth bringing up here too...

  A high edit count is never indicative of how productive an editor someone is. It is incredibly easy to make massive numbers of edits using automated or semi-automated tools. Take for example the Mhiji (contribs) sockpuppet of John254 that racked up 28,077 edits in 4 months (October 16, 2010 - January, 16 2011) using a modified AWB (note) and scripts.

  Similarly, a low edit count should not necessarily count against someone. Instead of making many small incremental changes to an article or page, an editor might instead draft large sections of material offline and add it to Wikipedia in one large edit. I myself tend to do this.

  The best way to sum this up is edit count or quantity of edits should not matter, the only thing which should matter is the quality of the edits. --Tothwolf (talk) 16:06, 27 May 2011 (UTC)

Statistical data from successful RfA and from users' RfA criteria

This page should shortly be populated with information from User:Worm That Turned/RFA criteria (what people say about RFA) and User:Worm That Turned/Successful RFA (stats about successful candidates). Now, the interesting things I've found.

• The edit count of new admins has increased year on year. From a midpoint of ~11k to ~19k.
• Only one person since 2009 has gained adminship with less than 3000 edits (He had over 1 million edits across wikis)
• Editors who passed with less than 10k edits - 2009: 44 (36%), 2010: 20 (26%), 2011: 5 (18% so far)
• The overall user criteria averages bear out as aiming for a minimum of a Journeyman editor - 2000 edits plus 6 months experience
• The more recent (vetted) criteria bears out as aiming for a minimum of 3700 edits plus 9 months experience. (Closer to a Yeoman Editor but not quite)
• Besides edit count and tenure, the next largest concern is the block log - with half of those concerned wanting to see a clean block log or an average of 10 months clear.
• After the blocking issue, edits to specific areas (generally mainspace), civility, policy knowledge and edit summaries were equally important in the criteria stakes. The rest were a bit all over the place.

Now that we have some solid information, perhaps we can come up with a pre-admin checklist, to help prevent NOTNOWS. If we can show that all successful admins met the checklist without issue, I think we can possibly put it forward as a proposal (based on data, not emotion). Worm^TT · (talk) 14:11, 26 May 2011 (UTC)

 Done: Your data has now been posted to This page. Thank you enormously for this detailed and complex research and analysis. --Kudpung กุดผึ้ง (talk) 17:29, 26 May 2011 (UTC)

Data from unsuccessful candidates

I have compiled and posted a sortable table based on unsuccessful candidates from 2011. Kudpung กุดผึ้ง (talk) 14:24, 27 May 2011 (UTC)

TCO opinion

Seems like the actual pass rates are dependent on article-writing and growing more so. Even if there is not a majority that feel this way, there may be a substantial minority that do. And that's what I see in discussions btw. Basically it's different from the essays, which tend to be older and/or tend to have a different point of view on the importance of articles.TCO (talk) 06:28, 18 June 2011 (UTC)

What the other Wikipedias do

French language Wikipedia

There is no consensus on the criteria for a contributor to be eligible. Thus, all registered contributors can access the administrator status.

However, in view of recent results, it seems advisable to have:

• a good knowledge of wiki syntax, rules and functioning of Wikipedia in French
• at least part of the work of Project: Maintenance
• about 3000 contributions to a year of significant activity (these criteria are purely indicative)

The bold text is mine.

German language Wikipedia

• One year and a significant four-figure edit count.

--Kudpung กุดผึ้ง (talk) 15:10, 19 June 2011 (UTC)

More:

Guidelines for Nominations for Admin

• Any user is allowed to run for office, or vote for users. Please bring a proposal for the consent of the candidate. Discussions about candidates can and should involve all users.
• Proposed candidates should have several months should have actively participated in the German Wikipedia and significantly more than the necessary for the entitlement to vote 200 edits in the article namespace can show. Most users expect a clear four-digit editcopunt and activity in different namespaces.
• Please add new proposals on the introduction, as a "template" select one of the preceding sections, possibly from the archive. Please enter the Candidate Involved in the submission.
• Each user can only vote with a pro-or contra-vote or abstain, neutral is not considered in the evaluation. Give your vote in the appropriate box and sign off with - Justifications for longer (more than one paragraph) used the instrument to preserve the clarity, on the talk page.
• Votes should not be commented on by others. This can happen on the discussion page.
• In order for a candidate to be appointed administrator, at least 50 users must vote 'support' within two weeks, with at least two-thirds of the total votes cast will be Support-votes. Neutrals do not count as votes cast.
• The bureaucrats implement the outcome of the vote, however, they have a margin of discretion in determining whether the conditions for the candidates and the voters are satisfied.
• These guidelines are adequate for candidates for special functions (administrator, bureaucrat, oversight, and check user).
  Kudpung กุดผึ้ง (talk) 16:41, 19 June 2011 (UTC)

Comment from Swarm

Note: This thread has been copied from Wikipedia talk:RfA reform 2011/Possible proposals . Please continue the discussion here.

Several good ideas on this page, but the idea I'd most like to see proposed is a set of minimum requirements for RfA candidates. This year, there have been 22 NOTNOW closures, plus multiple SNOW closures due to activity level. In the last year, the successful candidate with the lowest edit count was RHM22, who had about 3,900 edits. This is an exceptionally low edit count, and was the only successful RfA with less than 4,000 edits. While it's good that we can close RfAs early, frankly, it's ridiculous that absolutely anyone can run in the first place. It would save time, effort and stress if there was an edit count level that candidates have to meet before they are allowed to run. It would also filter out trolls. Swarm ^X 02:31, 19 June 2011 (UTC)

We have a whacking great big warning template on the edit notice of the transclusion page but we still get clots who just don't read it . Thre solutions that could be proposed to the commuynity right now. The scripts are easy to instal in the site software and Twinkle. Note that the German Wikipedia requires: One should should demonstrate at least one year of experience as a writer and collaborators in various fields of the project. Candidates with less experience have no chance of being elected to adminship.

Candidates that do not have a significant four-digit number of edits are mostly rejected. The proportions of the edits in different namespaces is often critically examined.:

Suggestions:

1. A checkbox on that transcslusion page that reads something like this:

Before transcluding this page, please be sure to have read all the instructions and advice pages.

[checkbox] I'am aware that my RfA can and will be reverted or closed early if it is obvious that it will not succeed.

If the user does not check the box but tries to save the page, a simple script that will load a notice declining the transclusion: "Sorry, but as you have not read up on all that is required to become an administrator, it will not be possible to process your request this time."

1. 2 Short questionnaire

Enter theses details: 1. Total number of edits to date: [field 1] 2. Total number of recent consecutive months edeiting: [field 2]

if the software detects less than 4,000 for field 1, and/or lest than 6 in field 2, a simple script will load a notice declining the transclusion:

"Please note that candidates with less than 4,000 edits and/or 6 months continuous editing are most unlikely to succeed. If you wish to continue please click here, but be aware that your application is unlikely to succeed." if the software detects less than 2,000 for field 1, and less than 3 in field 2, a simple script will load a notice declining the transclusion:

"Sorry, but you do not appear to have sufficient experience to become an administrator at this time. Please read the pages at xxx and xxx, and xxx, and discuss your request with an administrator before applying again."

1. 3 We simply set the bar as low as possible (e.g. 2,000/3) to ensure that the certain time wasters don't waste our time, install a Twinkle script that deletes the RfA proposal or the RfA if it has already been transcluded, and leaves a message on the applicant's talk page:

"Thank you for applying to be a Wikipedia administrator. You do not appear to have sufficient experience at this time, and your application has been declined. nevertheless, you may wish to read xxxx and xxxx and xxx, and try again when you feel you meet the recommended minimum criteria. In the meantime you may wish to help out at Recent Changes Patrol and New Page Patrol, and taking part in more AfD, RfA, SPI, and AN/I, which will certainly help increase your experience of admin tasks significantly. Good luck, and happy editing!"

Basically, we've discussed setting a bar before, but every time we have, people have assumed us to mean either raising it or loweriing it. This is not the case here. What we are doing here is making both the time wasters not waste their or our time, and encouraging others who may not be time wasters, but have little chance of passing, to take more advice and get more experience - such as those who would pass in another six months or so.
Kudpung กุดผึ้ง (talk) 08:57, 19 June 2011 (UTC)

I'll just add that a lot of people here probably are not aware of the number of RfAs that don't make it to tranclusion, and I don't see a way of tracking those deleted or reverted applications for the stats. Kudpung กุดผึ้ง (talk) 09:06, 19 June 2011 (UTC)

Note: This thread has been copied from Wikipedia talk:RfA reform 2011/Possible proposals . Please continue the discussion here. Kudpung กุดผึ้ง (talk) 03:30, 21 June 2011 (UTC)

Comparison of RfA on other Wikipedias: summary

The table I made was really only to to show what the other Wikis have for minimum qualifications for candidates and voters, but I threw in the rest of the data for good measure. I think our theoretical bar is perfectly alright where it is and generally the right people tend to pass. However, we need to set a low minium to prevent SNOW and NOTNOW. There was another one on 10 July which need not have been if we had one. We seriously need to resolve our civility issues and do something about the electorate. Other Wikis have a similar turn out of voters in spite of their much lower numbers of registered users, and demanding minimum qualifications for voters. This may be due to the fact that they notify their VPs and other venues of current RfAs. Voting is generally a straight vote with very little commenting if any, and no set questions. Neutral votes are not taken into consideration. All in all I was left with a sense of embarrassment at what a mess our RfA often are compared with the orderly fashion the others go about it. --Kudpung กุดผึ้ง (talk) 09:08, 10 July 2011 (UTC)

Unsuccessful candidates in 2011: scores & editcounts

Table has been updated to 10 July 2011 at Unsuccessful candidates. Still shows that in spite of all our advice pages and the blatant edit notice on the transclusion page, there are still some silly attempts demonstrating that a minim number of edits should be set at 1,500 - 3,000 and activity of 3 - 6 months consecutive months. Article space edits are important and should be taken into consideration because at these low levels many edits are to own user space, specially fiddling with own user page. --Kudpung กุดผึ้ง (talk) 10:34, 10 July 2011 (UTC)

Criteria for candidates

A script is available here that could easily be adapted for users and candidate's to check on candidates' eligibility. This script could also be automatically triggered by an attampt to transclude. --Kudpung กุดผึ้ง (talk) 08:54, 4 September 2011 (UTC)