User:WereSpielChequers/RFA reform

This is an essay/shopping list/analysis of wp:RFA from the perspective of one longtime participant and observer of the process.

RFA is broken
RFA has been discussed at interminable length for most of the history of the project. It is widely but not universally considered to be broken for various reasons. I'm one of those that regards it as broken for all of the following reasons:


 * 1) It is unnecessarily harsh in the way it treats candidates, even those who pass don't always come through unscathed. In many cases it drives editors off the project.
 * 2) It appoints insufficient admins to:
 * Maintain the number of active admins
 * Keep the community a healthy self governing one.
 * Keep the site running efficiently (As of June 2012 we have only encountered minor problems due to admin shortages, but as the shortfall grows so the problems will get worse, though it could be years before the problem becomes a crisis, but we don't know how big that safety margin is)
 * 1) It doesn't always exclude the unsuitable candidates, as evidenced by the fact that around a hundred admins have been desysopped or resigned under a cloud. In recent years there has been an increasing emphasis on RFA as an open book exam, and an accompanying decline in scrutiny of the candidate's edits. In my view this makes it less effective at sifting good candidates from bad.
 * 2) It is seen by some as exempt from our normal policies on civility, personal attacks and on some occasions even outing.
 * 3) Many of the best potential admins are unwilling to run at RFA.
 * 4) The Consensus process of weighted majority voting causes unnecessary and arbitrary inflation of standards, because instead of the minority "losing" a vote and thereby being tempted to re-evaluate their position, it is often a majority that "lose". So the psychological pressure is for the arbitrary inflation of standards rather than for convergence to a consensus standard criteria for adminship.
 * 5) It is vulnerable to canvassing.
 * 6) It is capricious and inconsistent. Because the questions and more importantly the scrutiny and standards of the RFA !voters vary greatly from one month to the next, there is a significant element of chance as to which candidates will pass and which will fail RFA.
 * 7) I'm pretty sure that one of the reasons why many longstanding admins regard the system as broken is that they know they wouldn't have passed under current criteria, and in many cases they don't meet the current de facto criteria.
 * 8) It is creating a wikigeneration divide between an active editing community many of whom have been with us for less than two years and a large majority for less than four, and an admin community, none of whom have been here less than a year, less than 1% have been editing for one to two years and the vast majority have been editing for more than five years.

It has also been described as:
 * 1) "Corrupt" (though I'm not aware of evidence to support this accusation).
 * 2) A popularity contest, though considering how many RFA regulars assess all or most RFAs that run the full 7 days, and how many supporters a successful candidate gets who they have never heard of before, this description is at best misleading. More descriptions at User:Jorgenev/"RFA was better before"

And yet, the great majority of those few who do pass do so almost unanimously.

What should RFA be?
RFA is the process by which Wikipedia appoints administrators, it also serves as a de-facto initiation ceremony which sometimes degenerates into a hazing ceremony. Which raises the questions:


 * 1) What are the qualifications needed for administrators?
 * 2) How many admins do we need?
 * 3) Does the process work effectively at sorting good candidates from bad?
 * 4) If we need an initiation ceremony can we decouple it from the adminship process?
 * 5) Can we divert the hazing ceremony elements to Editors for Deletion?

Qualifications
Candidates get opposed for many diverse reasons, some make more sense to me than others. As the pass mark is 65/75% a candidate only needs to show weakness in one area or marginal weakness in multiple areas to fail. RFA is a cross between an online interview and an openbook exam, but unlike most exams or interviews the examiners/interviewers have not agreed in advance the criteria that they are judging the candidates against. These are some of the main criteria and my perception of our divergence in weighting them.


 * 1) Tenure. Candidates with less than 3 months edits are likely to be summarily opposed per wp:Notnow, and many voters will oppose candidates who have been here for less than a year. Personally I'm unlikely to support a candidate who has been here for less than 6 months, and I wouldn't nominate one has been here less than twelve. There are in my experience very few editors who will oppose a candidate for lack of tenure if they have been editing for more than 12 months.
 * 2) Edits. Candidates with fewer than 3,000 edits are likely to be summarily opposed per NotNow. It has been some time since a candidate has passed with less than 3,500 edits and some !voters will oppose below 6,000 edits.
 * 3) High percentage of manual edits. For reasons that mystify me, some RFA !voters will oppose candidates who have a high percentage of automated edits. Personally I think that we should value our vandalfighters and that we need them to be able to block vandals, but I know at least one editor who won't run because he has done too much huggling to get through RFA. (There is a quite separate and much more logical argument that automated edits are not in themselves sufficient and that even if a candidate has 98,000 automated edits they still need 3,000 manual edits to show they can communicate without tools).
 * 4) "Maturity/Legal age" editors who admit to being legally minors, or who are suspected of this, usually attract some opposes either for lack of "maturity" or because some voters consider that all admins should be of legal age. My view and I think that of the consensus is that I will oppose candidates who I consider immature, though my preference is to focus on the immature edits not the age indicated. There is a wider issue here as to which areas of adminship we should steer juvenile admins away from, and a context in which age verification would require that all admins identify themselves to the office.
 * 5) "Consistency of commitment" one of our few sources of recent new admins are the remaining non-admins among the editors who joined us before 2008. However some of these editors have spread their contributions over more months than is fashionable or had recent wikibreaks. Personally I'm quite relaxed about this, though I do expect to see a few hundred recent edits.
 * 6) Article contributions. Opinions at RFA range from admins not needing to have contributed to the building of Wikipedia just to the protection of it, to admins needing to have contributed audited content - in the past there have even been RFAs opposes for candidate's who lack FAs. My view and I think that of the consensus is that that adding reliably sourced content is a basic skill that all admins should have, but GAs and FAs are a bonus not a necessity.
 * 7) Deletion tagging. Possibly the most contentious area in recent RFAs. Admins have the deletion button and many RFA !voters will oppose candidates with a record of overhasty or incorrect tags. Views seem to vary from requiring perfection to "it doesn't matter what the tag was if an admin deleted it". In recent years I've drifted from expecting near perfection to tolerating a small proportion of mistags, especially if the candidate responds positively to feedback.
 * 8) Blocking. One of the most contentious things that an admin can do is to block fellow editors, especially vested contributors. Candidates at RFA can be assessed as to whether their AIV reports indicate that they know which newbies and IPs to block, and occasionally as to what positions they have taken at AN/I re the blocking of vested contributors. I'm hoping that some RFA !voters check candidate's AIV reports and the reason why no-one ever seems to get an oppose for poor AIV tagging is that the people who are active there give good guidance and address errors long before the individuals run at RFA. But I worry that everyone else may be making the same assumption as me. Block duration and to a large extent the blocking of established editors are not areas where RFA can effectively test new candidates.
 * 9) Evaluating consensus. One of the most important roles that an admin has is the judging of consensus, though I'm not seeing much sign of the ability to do this being evaluated at RFA.
 * 10) Transparency and redemption. I'm of the view that as we don't and can't vet who our candidates really are there is little if any point worrying about what a candidate was doing more than 12 months ago. If anything I'd prefer to have a candidate who admits to a problematic past, or at least uses the provision in wp:Clean start to start a new account and inform Arbcom when they run at RFA. But at the same time there are people who I would not want to sneak through RFA... The community is divided on the way it treats candidates who inform Arbcom of prior accounts, and I'm pretty sure that lack of an effective way to screen out past troublemakers is one of the drivers behind the increased expectation for length of tenure.
 * 11) Civility. At one extreme incivility can torpedo an RFA, but there are also some editors who will oppose if they suspect a candidate would become a part of the "civility police".

Need for admins
The community does not agree as to whether we should be a selfgoverning community and that adminship involves some useful editing tools which should be given to all longterm, civil, clueful editors; Or that adminship should be restricted to a small group of very active editors who have "a need for the tools". I'm very much in the former group, but with active admins dwindling by 1% a month and liable to fall this year to late 2005 levels, the pressing issue we face is that at some point in the next few years we will have insufficient admins to keep the vandals in check.

Possible improvements
There are many possible ways to improve RFA, most of the following are not my suggestions but ideas that have emerged in discussions at RFA, or my observations of other wikis. Some are clearly interrelated and some are palliatives that would mildly improve things as opposed to actually fixing it.

Deputise a posse
Unless we reform RFA, then if current trends continue at some point we will hit problems. Looking at other wikis that have hit similar issues, the default solution to a lack of active admins is to appoint a batch of minimally screened candidates. I consider this to be the worst of the workable solutions.

Agree a criteria for adminship
In most real world exams or interviews the examiners or interviewers would try to agree a set of objective criteria for evaluating candidates, and much of the interview or exam would focus on whether or not they meet that standard. A substantial proportion of the discussion in many RFAs is not about whether a candidate meets an agreed standard but about what standard the RFA candidates should be expected to meet. That is discourteous to the candidate, makes us look unprofessional, and is offputting to potential candidates.

Agreeing a set of RFA criteria would reduce the capriciousness of the current system, hopefully persuade many more well qualified candidates to run and possibly even persuade some poorly qualified candidates to postpone. It should also slow the process of standards inflation.

Once we agree to set a criteria we have the equally difficult task of agreeing that criteria. Whilst I don't necessarily agree with all of them, I believe that current expectations include:


 * 1) A clean block log, 3,000 manual edits and at least 12 different months in which they've been active. (Or at least 12 different months in which they've been active and 3,000 manual edits since the latest block).
 * 2) Demonstrated ability to add reliably sourced content to articles.
 * 3) If active in deletion, a record of deletion tagging and AFD votes within policy
 * 4) If active in vandalfighting, a record of correct reporting at AIV
 * 5) If this isn't their first RFA, at least three months activity since their last RFA.

There would of course be some things outside that criteria, and occasions where a nominator was making a case for the community to set aside the criteria. But having a set of criteria for things where we do have consensus should focus the debate onto the candidate's contributions and thereby improve both the tone at RFA and its ability to make the right choices.

It should also slow or hopefully stop the process of standards inflation, as additional criteria would require consensus to add, rather than consensus not to add. It would be much harder to get 70% agreement to add a criteria than historically it has been to get 30% or more to start applying a criteria.

Crat decision
Currently crats have discretion within a 65-75% zone, possible reforms would be:
 * 1) Widen the zone of discretion
 * 2) Encourage crats to discount or strike !Votes that have rationales outside of an agreed criteria for adminship.
 * 3) Have crats appoint admins in the same way that admins appoint Rollbackers or Autopatrollers. So the community would set the criteria for adminship, and crats would then assess candidates against the criteria. This would be one of the simplest ways to fix RFA, but might require a separate user group than the crats, or a crat plus process whereby existing crats would only gain this extra role if they underwent a reconfirmation.

RFA committee
A potential solution would be to make adminship an indirect community decision by electing a committee or individuals to appoint admins, but unless we have a community discussion as to the criteria for adminship this masks the real issue which is that we haven't agreed the criteria that admins should be judged against. We'd have to set a criteria if we elected people to make these choices for us, either explicitly or implicitly by the views of those elected. So why not set a criteria and then judge candidates against it?

Upbundling
Two of the most contentious things an admin can do are to block an established user and to determine consensus on a complex and disputed discussion. Upbundling one or both of these roles to the crats would meet the concerns that some opposers cite at RFA, as all crats are admins this wouldn't increase their remit, though reducing the remit of the other admins would increase the workload of the crats. A rather major hurdle to this is that it would be difficult to get agreement as to the relevant definition of established users and complex disputes.

Unbundling
Unbundling is the process by which less contentious parts of the admin role or tools are made available to a larger number of editors. The largest unbundling exercise so far was probably the creation of wp:Rollbackers. The next one which I think would work and which would enable us to survive many years of a dwindling number of active admins is Vandal fighters (though I hate the name). This would allow hugglers and recent changes patrollers to block IP addresses and editors who have less than 100 edits. There are a number of RFA !voters who are very wary as to who they allow to block established users, so this proposal would allow a larger group of editors to do the >99% of blocks that are uncontentious, but often urgent. At the same time any definition of or special privileges for wp:vested contributors is itself a contentious idea, however setting the threshold very low should resolve this by defining people as Wikipedians before they realise they have become "active".

Recruit more candidates
Good candidates can and do still pass RFA, not enough to maintain active numbers and often we until they are way over-qualified. But there are many people out there who would pass easily if they could but be persuaded to run. I've persuaded a few myself, both directly as a nominator and indirectly through the signpost. More candidates could be persuaded to run if more experienced admins were to look for candidates to nominate. Perhaps what we face is a "crisis of cliqueishness" with even very experienced editors feeling that they are unsure if they yet "belong" sufficiently to run. One of the recommendations from WP:NEWT was that we could reduce the problem at newpage patrol by looking at the editors who actively create articles and appointing more Autopatrollers. From my experience of appointing over a hundred Autopatrollers I have come to realise that this can also serve as a welcome to Wikipedia space with some of those Autopatrollers going on to become admins.

This is far from being a solution to our RFA problems, but it can slowdown the decline.

Training and evaluation
One common reason for opposing is that adminship is not for on the job training, and as the tools are bundled any admin could use any part of them. There is some truth in this, but no good reason why this should remain so. One possible solution would be to create training modules for the various things that admins can do. I would anticipate that once the modules existed RFA candidates would start making declarations that they would "only use the mop in other areas once they'd done the training modules" and it would become easier to nudge admins making aberrant decisions or returning from long wiki breaks to do the relevant training modules. If the training was sufficiently up to date to serve as a briefing of recent changes to returning former admins, then that might reassure those who worry about former admins returning from gaps of several years (up to two years total inactivity, but potentially much longer if one combines a period of a few years doing a handful of edits a year with a complete break of nearly two years).

In the past there have been training systems, either generic as in "the questions" or bespoke as in admin coaching. Admin coaching fell out of fashion during the early years of the drought, possibly because some saw it as training to pass the test rather than training to be a good admin. I'm not sure why User Filll's work fell out of fashion, I think it was useful to me when I first ran.

The broader problem with training and evaluation as an RFA reform is that lack of training in the admin tools is not the issue either in border line candidates or in those admins who make mistakes. The community is very tolerant of anyone who learns from mistakes. Common opposes are more likely to be over judgment, wikiphilosophy, or nonadmin experience.

See This proposal I made in early 2024 which closed "No consensus" with 14 supports and 12 opposes.

"Not unless" candidates
Currently RFA has only two possible outcomes, pass or fail. A third and in my view very positive one would be "not unless". In this scenario the RFA would run much as it does today, but the closing crat would have the option of closing with a statement such as "Due to the mixup of the nominator picking up and responding from the candidate's PC, this RFA can only be closed as a success if a checkuser confirms that the candidate and their nominator are in fact different people." Or "Due to the concerns expressed at the candidate's AIV tagging this RFA can only be closed as a success if the candidate demonstrates an improvement in this area". This would give a crat the opportunity to promote such candidates if they had subsequently met the relevant condition(s), and the discretion not to do so if they had also done something egregious. The candidate would still have the opportunity to submit a completely fresh RFA, and I would hope that in areas where judgement is concerned the crat would consult with the relevant opposers before promoting such a candidate.

One of the advantages of this sort of close being possible is that it would hopefully concentrate attention in the oppose section on the things a candidate would need to do to be suitable for adminship, rather than how little they are trusted or known. It might also make those who oppose for spurious reasons such as a high percentage of automated edits think twice when they worked out the implications of such opposes and saw themselves writing "Oppose 80% automated edits is too high. Candidate needs to give up Huggle and Hotcat and do 60,000 more manual edits in addition to the 20,000 they've already done to bring their Automated edits down to an acceptable 50%".

I'm one of those who failed my first RFA, I remember as the opposes came in, and again when I reread it before my second RFA it was much easier to accept the "not yet" type of opposes than some of the others. I think it could transform the RFA process if we were to focus the opposition section on the things the candidate would need to do to become an admin.

Questions
I'm of the view that an excessive number of poorly targeted questions damages RFA by distracting attention from the candidates contributions and the occasional useful question that is targeted at the candidate. Others have much stronger opinions on this. But I'm not sure what the best way is to keep the question section under control, and I suspect it is merely a side effect of the small number of RFAs running at present.

At the beginning of 2016 questions were limited to two per questioner. This addresses one of the problems, but doesn't stop people asking generic questions that might not be relevant to the candidate.

Clerking
Appoint a non involved clerk for each RfA who would keep the RfA clean and on track, and delete inappropriate !votes and questions. I think this idea has possibilities, but in my view for it to work it would need a much lighter touch than its supporters want. Probably the most effective use of this would be in gently chiding some of the odder contributions. However in order for this to work we would need to get consensus as to what is and is not acceptable at RFA. For example if a candidate has disclosed that they went through Cleanstart and has taken the option of informing Arbcom, then I consider that a deluge of questions which cumulatively or even individually would enable one to identify the former account would be inappropriate. Others consider that a humorous or off the wall question is or should be unacceptable.

Also we need to remove !votes from people who are using this to change policy or have a dig at the nominator;
 * Oppose. Candidate has images on their userpage that are "Not Safe for Work". Is a perfectly reasonable position to take at RFA. But "Oppose candidate uses a commonly used userbox that I take exception to". Should be referred to MFD.
 * Neutral. This nominator has previously nominated candidates whose deleted contributions have turned out to be problematic, so I'm going to wait for a wait for another admin to comment on their deleted contributions." Is helpful and focussed on RFA. "Oppose. The nominator has views I disagree with." Is not really relevant to the RFA.

Clerking was introduced in the reform program in late 2015.

Voting rights
Currently we exclude IPs and there is a vague unwritten rule about "new" accounts. As someone who remembers finding this offputting I would suggest replacing it with a clear and low threshold of say 100 edits in the last 12 months. This would screen out some socks and SPAs, but any genuine newbie who came across that would be more likely to think how close they were to qualifying than think it closed to them. It might also be helpful to exclude RFA edits from the 100, otherwise we could have !voters whose only involvement in the community was to participate at RFA.

This was proposed and rejected in October 2015, with such a clear margin of rejection it probably isn't worth raising again for another year. Though the reason given for the proposal was very different than the reason why I support it.

Optional pre-vetting
One of the most unfortunate aspects of RFA is that often it is the first time that a candidate will be aware of problems in their editing. Expanding WP:editor review or reviving admin coaching would give candidates the opportunity to discover and resolve problems in a less public arena than RFA; and they would be less likely to get opposes for unexpected reasons.

I've persuaded a group of admins who are interested in this to volunteer to review a sample of the tags of Newpage patrollers who request it. This has now been added to the Editor review documentation and could be added to the RFA documentation to ensure it was seen by many patrollers at an earlier stage in their wiki careers.

In early 2016 Requests for adminship/Optional RfA candidate poll started. My suspicion is that some of the reviewers there are overly cautious, and coincidentally the number of RFAs has dipped. But if the end result is that some people postpone but in the end run successfully it may yet work.

Research
Any solution to the RFA problem would benefit from research to define and measure the problem.

My own User:WereSpielChequers/RFA by month has measured various aspects of the problem, including the decline in the number of successful RFAs, the decline in the number of active RFAs and our failure to bring the 2008/9/10 wikigenerations through to adminship.

User:ErrantX/RFA Study is a potential study of the RFA !voting community. Looking at questions such as what evidence these people assess.

User:Alzarian16/RfA participation by year is a study into the declining participation of !voters in RFA.

We understand some aspects of the problem quite well, others badly if at all. More research into the problem would enable better decision making and might resolve some of the policy logjams where part of the community believes one change would be harmful and another part disagrees.

In particular it would help to have research look into:


 * 1) How many non-admin editors out there meet current RFA de facto criteria, and what are their reasons for not yet running at RFA.
 * 2) Is the wiki-generation divide the result of there being very few editors who started in 2008/9/10, or are the active editors who started editing in those years not willing to run at RFA?
 * 3) Is the decline in RFA voting purely a function of there being fewer RFAs, or has the active electorate shrunk, and if so why have some otherwise active editors stopped participating in RFA
 * 4) What do RFA !voters check in RFAs and how much checking do they do?
 * 5) Are there any lessons we can learn from the admins who we've subsequently desysopped or have otherwise resigned under a cloud, were there signs in their RFAs that could have predicted they were not ideal candidates?

Attract more voters with a watchlist notice
A watchlist or site notice, especially one only shown to editors with more than 1,000 edits should attract a broader electorate, it might also tempt more editors out of their particular subject areas and into broader areas of governance.


 * This was implemented at the end of 2015 and has increased turnout. It may take time but hopefully some of the additional voters will start standing as candidates.

Synchronise with EFD
This may seem counterintuitive, but the first trial went well. RFA has unhealthy elements of an initiation rite or hazing ceremony, Editor for deletion was purely a hazing ceremony but in a healthy good humoured way. Running the same editor through both processes in parallel gives those editors who want the opportunity to haze the candidate a venue to do so, and editors who make inappropriate comments in the RFa can have them moved to the EFD or simply be responded to with "this is the serious discussion, the EFD is thataway...


 * Fortunately in late 2013 RFA seems to have got less unpleasant, it still has some people who I consider are too picky, though others probably consider me too picky in my opposition to people who I suspect would be heavy with the deletion button. Editor for deletion has been userfied and appears dormant. I would park this particular proposal at present.  Ϣere  Spiel  Chequers  13:34, 7 December 2013 (UTC)

Lower the pass threshold
Historically bureaucrats had discretion to judge the strength of discussion but very rarely promoted anyone with less that 70% support and almost never rejected a candidate with over 75% support. Lowering the pass threshold should in theory result in the appointment of more candidates. Though not huge numbers as in practice the community is usually strongly for or strongly against a candidacy. At the end of 2015 the crat discretion zone was broadened to 65%-75%. It could take a while for such a subtle change to make much of a difference.

Make the whole process internal
We are used to doing everything in public, but does that add to the stress? Just because the process is on the internet does not mean it has to continue to be on full display. With some software investment, we could render the whole process only visible to accounts that are WP:Extended Confirmed

First proposed in February 2024

Possible ways in which it could deteriorate
Though the number of active admins has dropped by over 300 since it peaked, there are a number of things that could speed up the decline in our number of active admins and bring forward the date when it starts causing substantial identifiable damage to the project.

Extrapolate the trend
Just as one should always be wary about placing false confidence in extrapolated positive trends, it is reckless and unwise to ignore negative trends in the hope that they will go away. We are now eight years into a drought at RFA. That drought we through two phases for five years from 2008 to 2012 it deepened each year with each year appointing little more than half as many admins as the year before. From 2011 to 2015 it has oscillated around a level less than a tenth of what it was in 2007 (2013 saw a little rally, though not even to 2011 levels whilst 2014 dropped to little more than half the 2012 level. 2015 set a new low, at 21 new admins).

Burnout
As the number of active admins falls, we risk burning out our remaining admins or detaching them from the community. If they have only a limited amount of time to spend on Wikipedia and an increasing proportion of that is spent on admin duties then we risk demotivating and losing a volunteer. I suspect we currently have a bit of a safety margin here, but as we are actually dependent on a very small proportion of very active admins for a large proportion of the admin actions, our safety margin may be less than we assume.

Make it an election
Comparisons are sometimes made between the Arbcom elections which many people think work acceptably well and the RFA process which most consider broken. Whilst I share the sentiment I see a fundamental difference between electing a fixed number of people to a committee and deciding whether a large number of individuals met a particular standard. I think that an Arbcom style process might possibly work, if you could agree how many active admins we needed you could hold a monthly election to elect sufficient admins to bring us back to that number. Candidates could of course withdraw if they saw a very well qualified field going for a small number of places, but I still suspect that this system would be even more capricious than the one we currently use.

Agreeing how many admins we should elect each month would require us to decide whether we simply want enough admins to do the various admin chores, or whether we want adminship to be the norm for longterm clueful editors. Then having decided that, put a number to the decision. I honestly can't see that happening by consensus, or even majority.

It would also have the drawback that candidates would not know why they were not supported. By contrast the existing RFA process enables people to rebut misconceptions by !voters, and enables candidates who did not succeed to address !voters concerns before they run again.

Voting as if Admins made rather than enforced policy
There are contentious issues on Wikipedia, but as admins are enforcers of policy not makers of policy in my view it is inappropriate to oppose or support candidates based on the policies that they wish to change. However this does sometimes happen, and if it got more common could bring RFA to a complete impasse. Candidates have been opposed or supported because of their views on the civility policy, admin recall, and IP editing. If one divisive policy issue were to become so contentious that thirty five percent of voters would oppose candidates who took one position and thirty five percent would oppose those who took the other position, then it would cease to be possible to get a candidate through RFA.

Periodic reconfirmation
This is the idea that every admin would have to run a fresh RFA every few years (usually this is proposed to be two or three years). In theory it might not greatly reduce the amount of admin resource available provided all the active admins were willing to run a fresh RFA every few years. However it would lose us a lot of useful admin actions from the large group of semi active admins who use the mop too rarely to get through RFA again. It would almost certainly also lose us a number of good and active admins who aren't prepared to run an RFA again and would rather continue editing but without the mop. The risks of this proposal include the near certainty that a large number of useful and active admins would not be willing to subject themselves to an RFA process that has become far less collegial than when most current admins experienced it.

Obviously supporters of this change don't share my view that the community should be a self governing one with almost all clueful longterm Wikipedians having the mop; But crucially they haven't communicated an alternative vision as to how the community should look like with with admins a scarce elite. Some supporters of this have argued that it would make RFA easier if we were merely appointing an admin to a three year post rather than a lifetime one; This argument rather ignores the obvious point that the fewer admins we have the more it matters who they are - so in reality RFA would get tougher if there were a small number of term limited admins.

Costs: Substantial editor time involved in Reconfirmation RFAs, and loss of a large proportion of the moderately active admin cadre.

Benefits: None

Disadvantages:
 * Fails to treat admins as the scarce unpaid volunteers that they are. Would thereby seriously undermine our efforts to keep spam and vandalism off the pedia.

Term limits
This is a step beyond periodic reconfirmation. Like the US President who is limited to two terms in office, admins would be limited to a maximum amount of time as admins. Term limits are useful in roles such as charity trustees where the number of trustees is necessarily limited, they give no advantage and multiple disadvantages in roles where there need be no maximum number of people in the role.

Term limits mean that admins would be treated as a political group where roles are limited, rather than a wiki style operation where we empower as many editors as possible and aim to share the load. At best it would result in the wiki still functioning, but with a small band of admis who only have the time to be admins and as a result get isolated from the community they were once part of. More likely it would result in: Insufficient admins to keep the site functioning properly. increased status if only through scarcity value of the few admins we have, burnout of the admins and the patrollers who find their vandalfighting work more difficult. Loss of many experienced editors and frustration amongst those that remain.

The proposal is based on the following misconceptions:
 * 1) Mops are expensive and scarce rather than zero cost and unlimited in number.
 * 2) Volunteers can be hired and fired like employees.
 * 3) Volunteers who have some useful tools taken away from them will continue editing anyway, even though they've experienced how the site functions when you have those extra buttons.
 * 4) Blocking vandals and deleting attack pages will somehow happen even if we have far fewer admins to do it.

As most of our active admins were appointed before the RFA drought began in the Spring of 2008, any term limit proposal that would de-admin those appointed in 2003-2007 is pretty much guaranteed to bring the admin shortage to a head. Losing the class of 2007 would be particularly damaging as they are the largest single year group among the active admins.

The best responses to term limit proposals is to suggest to their supporters that they should first get consensus to reduce the number of reasons for which people can be blocked or articles deleted. If we don't need many admins then we can afford to have fewer of them.

Require supporters to give a rationale
Various suggestions over the years to require supporters to spell out their particular reasons for support have failed to get consensus, not least because it would bury RFAs under unnecessary verbiage. There is of course a lack of parity in that an oppose vote does need to explain why they disagree with the nomination, but curiously those who argue that for reasons of parity supporters should precis the nomination in their own words rarely support lowering the threshold to 50% and making a support equal in effect to an oppose.

Another drawback to requiring supporters to spell out their reasons is that !voters who do so are often new to the RFA process, and telling them their !vote is flawed because they omitted the words "per nom" would be as off-putting as it would be pointless. Sometimes an analogy is drawn with FAC where "drive by supports" are positively frowned on, interestingly FAC has a chronic shortage of reviewers whilst one of the few things that works at RFA is that every RFA that runs the 7 days can be guaranteed dozens of participants.

Make it easier to remove the mop
Almost every discussion at WT:RFA about declining numbers of admins will get diverted into a discussion about making it easier to get rid of admins. There have been many such proposals.

Supporters of this proposal seem to be divided between those who think that it is near impossible to get rid of a bad admin, and those who want to purge the admin cadre of scores or even hundreds of admins who they deem to be "bad admins". The former group might well be surprised to discover just how quickly Arbcom can act when needed, and that the list of desysopped or "resigned under a cloud" admins is now about a hundred strong. The latter group have plenty of opportunity to either suggest to Arbcom that they desysop particular admins or to suggest to the community that they elect Arbs who will "purge the admin cadre".

Since the desysopping rate across the life of the project is a rather high 1% of active admins per annum, and one of the biggest difficulties at RFA is persuading good candidates to stand, in my view the endless discussions about the need to get rid of large numbers of unspecified bad admins are a contributor to our admin shortage rather than a potential palliative.

One of the least appealing aspects to the various community de-adminship proposals is the recurring argument that they would be used to get rid of an unspecified number of "bad admins". Most of our currently active admins became admins when RFA was a very different place that judged more on clue and less on editcountitis and answers to trick questions. Most of these admins know that they wouldn't have become admins if RFA's current "criteria" had applied when they ran. Many have gone on to acquire the experience and editcount that today's RFA !voters expect, and personally I doubt that the community has any great desire for a mass cull of existing admins, that may even be why proponents of a cull of "bad admins" are so reticent to define the criteria for being an acceptable admin under the current desysopping regime but a candidate for desysopping under a new regime.

Two proposals to make it easier to remove the mop have passed. The first has desysopped hundreds of inactive admins who haven't edited or logged any action in over a year (though returning admins could get the mop back on request), the second set a 2 year limit on those inactive admins getting the mop back on request. We don't yet have an analysis of the data to show if that made it more or less likely for such admins to return. They clearly haven't succeeded in encouraging more new candidates to come forward. There are still hundreds of admins who edit so rarely that they would snow fail an RFA if they weren't already admins, I'm conscious that attempts to mass desyop semi-active admins have failed to get consensus in the past. Not least because many consider that such proposals do more harm than good, I'm very much in the camp that thinks that we should be encouraging our former volunteers to return, not making them unwelcome when they do.

There is one group of admins who I can see going under a recall or community deadmiship system. I'm one of those RFA !voters who believes that deletions should only be done where there is consensus of the community either for that article or that type of article to be deleted, and many of my opposes have been to candidates who stretch the speedy deletion criteria far beyond the point where it has community consensus. I'm conscious that there are a number of admins whose deletion of articles via speedy deletion would result in their RFA failing if they were an RFA candidate being judged on speedy deletion tags on those same articles. But I would prefer in the main to resolve this with retraining and some sort of policy resolution/reconfirmation as to whether unsourced, "poorly formatted", or "would almost certainly be deleted at AFD" should be formalised as speedy deletion categories. I believe that bringing the community's divide over deletionism and inclusionism to a boil in an attempted mass desyopping of deletionists would be overly damaging to the project regardless of which side won the battle.

Identification to the Office
The WMF already has procedures in place to enable editors to identify themselves to the office, and this is required for Arbcom, checkuser and OTRS. Identification could also be required of all new admins, this would enable the community to set a minimum age for admins. Also if it was done in combination with retaining the data and checking against past records, in the long run it would enable us to screen out returning banned former admins. However there are several drawbacks to this proposal. Firstly we probably don't have consensus to introduce a minimum age for admins. Secondly retaining this data would open our admins up both to potential leaks from the WMF and more seriously to legal writs via the WMF. Thirdly as we probably don't have this data on those former admins who have already been banned, we only benefit in future incidents when new admins get banned and then return. Arguably we have too few new admins for this to be a high risk scenario. Fourthly we would be reducing our protections against misuse of State power. Lastly this would be a disincentive for new admins, and for this reason alone it fails the most basic test of an RFA reform as it would be more likely to exacerbate the shortage than ameliorate it.