User:WereSpielChequers/RFA changes

To solve or mitigate a problem, you must first define it. This page is an attempt to answer the questions: "What are the RFA problems?" and "How can they be solved"

(Suggested changes on talk page please, but do feel free to copyedit)

RFA is the process by which Wikipedia appoints new administrators, - a group of editors with a few extra powers and arguably some extra status within the community. There seems to be a common consensus that there are problems with RFA, comments such as RFA is broken & RFA is a joke are frequently made and rarely disputed.

Up until the spring of 2008 there were usually 100 or so admins appointed by RFA every quarter, however the last three quarters of 2008 saw only 116 successful RFAs in 9 months - well below half the historic rate; and there were fewer active Admins at the end of 2008 than at the beginning.

All other signs are that Wikipedia is healthy and growing, with the RFA drought starting just when an all time peak of new editors should have been producing a record crop of admins.

Problems
While we don't have agreement as to how many admins we should have, or even how many active ones we currently have, numbers of candidates and successful RFAs are both at a low - September 2008 saw fewer new admins than any month in years, but by Jan 2009 things have deteriorated sharply even from then.

RFA is an abrasive process that loses us some editors. A cursory trawl of Unsuccessful adminship candidacies doesn't take long to find candidates who ceased editing during or soon after their RFAs.

It may be deterring or delaying some good candidates - September 2008 had our lowest haul of new admins for years and looking at the November 2008 new admins I see a consistent pattern of users who should have been adminned months before.

Wikipedia is an adhocracy (and probably couldn't have to where it is today without being one), and as such is subject to the downsides of adhocracies such as: all of which are very obvious in the RFA process.
 * lack of common standards can lead to inconsistent results;
 * differences between written policy and current practice, which can lead to cliqueism and raise barriers against newbies;
 * difficulty in resolving conflicts;

Can't do attitude
While there may be consensus that RFA is broken, there is also a widespread perception that despite recent changes (and the ease with which they were done) RFA is somehow beyond repair and that any attempt to fix it is ultimately futile. Hopefully that can't do attitude can be worn away by demonstrating that change is possible. But if that doesn't work I think we should be prepared to cut the Gordian knot by deleting the archives for Wikipedia talk:Requests for adminship. This may seem a radical step, but after 36,000 edits these archives have grown beyond being usable or useful and instead have become part of the problem, any proposal to reform RFA risks the response "We've discussed this before". This could be useful if it was in terms of "we discussed this x weeks ago and consensus was not to do this, have a read of this dif, if you think that consensus or something else might have changed lets reopen it". But that has not been my experience. I think we probably have consensus that RFA is broken, but in order to fix it we need to change the can't do mindset that has built up re this problem. Deleting the archives for Wikipedia talk:Requests for adminship and starting afresh might seem extreme, but if anyone has suggested this before, I suggest that the growth in the archives since then is grounds for reopening the debate. However I could live with a rolling three months or a policy of only reopening old discussions at three month intervals.

Differing standards
In a job interview, an exam or an approval process the interviewers or examiners would coordinate and benchmark so as to try and achieve a common standard, ideally the job ad and any shortlisting should be influenced by that standard. RFA has two levels of reviewers, the !voters and the crats, and neither seems to have benchmarked their approach, the !voters in particular behave more like voters in a public election than an appointment panel; criteria don't merely diverge with different voters looking for different things from a candidate, expectations for things such as edit count are widely divergent and other criteria can even be contradictory. It is quite possible to see !voters in an RFA disagree whether:
 * 1) The candidate is experienced or inexperienced.
 * 2) The candidate is to be trusted or not.
 * 3) The candidate's contributions are substantial/Valued or not
 * 4) The candidate's answers to the questions are good or bad.
 * 5) The nomination is "brilliant" or "poor".
 * 6) The candidate has or has not waited long enough since their last attempt.

Getting the crats to discuss their criteria and attempt to harmonise their approach to RFA may be possible as they are a relatively small group, I'm not convinced that the same is true for the !voters, and we may need a system for RFA that can work despite standards differing - a monthly election on similar lines to the Arbcom one might achieve this, with rules such as:
 * 1) The monthly RFA process will run for one week from 00.00 on the third Sunday of the month.
 * 2) The number of admins to be appointed will be 1,000 minus the number of Active admins as of the end of the previous month.
 * 3) Unsuccessful candidates may not re-stand for a month for every 10% less support they had than the least successful winning candidate.

Alternatively is there a way to get the regular !voters to discuss and harmonise their criteria?

Ratcheting standards
Revisiting this in January 2009 it strikes me that the problem could be inherent in the system, there is a natural psychological process that we undertake to normalise our behaviour with our peers, and if we are in the "losing" side we reevaluate and learn from the experience. If the benchmark was 50% then this would tend to normalise to the mean, but with consensus at 70 - 75% I believe this process is overvaluing oppose votes and thereby ratcheting up our standards at RFA to the disservice of the project. Shifting to a 50% threshold would readily resolve this and lower the bar, though perhaps not as low as it was in 2006, and partially stabilise the criteria of RFA voters. Hopefully it would also tempt more of our longstanding contributors to run.

Hazing ceremony
Part of RFA's current role is as a hazing ceremony before initiating some editors to admin roles within Wikipedia.

In the last decade or so I've changed my views on initiation ceremonies from strong distaste to awareness that they are a widespread feature of human societies, and providing certain pitfalls are avoided they can be a natural and healthy coming of age ceremony.

Hazing ceremonies at their worst can lapse into cruelty and/or become an escalating cycle with each clade of hazers being just that little rougher than the hazing they went through themselves, which could explain the mysterious process whereby expected standards at RFA have been inflating over time.

Wikipedia being a global organisation we have many different cultures and backgrounds, and while some potential admins are deterred by this others clearly relish it, so I suggest that the hazing ceremony aspects of RFA be hived off to Editors for deletion.

This would have the following advantages over hazing as part of the RFA:


 * 1) It's humourous.
 * 2) It's optional.
 * 3) It's unrelated to anything else on Wikipedia.
 * 4) It shouldn't greatly matter if the standards of behaviour at Editors for deletion tend to oscillate over time.

Civility
There is a fine line between saying that you don't think an editor is suitable for the mop for a particular reason and criticising that editor. If the oppose is because "I don't trust this candidate" it is especially difficult to phrase things in such a way that it isn't a personal attack, and while not now and insufficient experience opposes are easy to phrase it would be wrong to give an unsuitable candidate the false hope that they simply need more experience. However I doubt if I'm unusual in seeing RFA as an area exempt from wp:civil, and I suspect that that is putting off some potential candidates.

I would like wp:civil to be reintroduced to wp:rfa, I think this one could be done quite easily by crats striking !votes that use language such as "The candidate's perspective on the deletion process is best described as "completely out to lunch". The dump I took this morning was, I assure you, a sincere contribution. Unfortunately, it was still crap." as incivil, ideally whilst the RFA is in process.

Candidates failing per snow
Hopefully the October 2008 changes to Guide to requests for adminship will delay some of the less experienced candidates and reduce the number of closures per snow. But we could create and code a rule that no candidate can self nom unless they have done 1,000 edits, which should sharply reduce the snow candidates. This might also reduce the editcountitis effect at RFA as only one RFA in recent months has succeeded with fewer than 3,000 edits.

Handling failed candidates
Rejection sucks, and even a civil RFA can be a bruising experience; hopefully the proposals I've outlined would lead to fewer candidates being rejected, and make the process more civil. But it will still happen, and we need to consider how we treat losing candidates.