Wikipedia:WikiProject Administrator/Five Problems with a Single Solution

"Five Problems with a Single Solution" is an attempt to explain what has prompted attempts at RFA/ADMIN reform and how reform could help Wikipedia function better.

RFA is widely described as being problematic. Large inter-admin disputes are extended and drama-filled. Arbcom is overloaded and overworked. Concerns are sometimes raised about the status of "admin accountability" or "double standards" for admins.

What looks like a diverse set of problems may all share a single cause. And therefore, many different problems might all be helped by one single fix.

The right person for the right job
Some people are best as editors, some people are best as admins. They are not "ranks" as much as they are "roles" or "job titles". Someone who works most efficiently as an editor is no better or worse than someone who works most efficiently as an admin.

An ideal editor is someone who works best in areas that don't require the tools-- a heavy duty content contributor,  or someone who works best being a strong advocate for Wikipedia policies rather than a neutral enforcer of policy.

An ideal admin is someone who works best in areas that do require the tools-- a kind wikignome who cleans things up, a wise adjudicator who closes out deletion debates, a trusted broker who calms things down, or an unbiased enforcer of policy.

(Of course, in reality, the two roles aren't mutually exclusive.  Many admins spend the bulk of the time editing away, only needing the actual tools occasionally.   Many editors spend their time mediating disputes and cleaning things up, thereby acting as tool-less admins.)

The Role Assignment Question
Once upon a time, the administrator population consisted entirely of people hand-picked for the job. Adminship was not a big deal. Most everyone, regardless of their aptitude, worked in the role of editor.

But of course, that meant a huge pool of talent was going untapped. Lots and lots of potential admins didn't have access to the tools. At the same time, you didn't want the tools to go to anyone who would misuse them.

Wikipedia faced the Role Assignment Question. How would it decide which users belonged in which population? 

The Creation of RFA
In order to sort individuals into their ideal roles, Wikipedia created RFA.

At RFA, an editor could be examined by the community. After examining the editor's record, the community would come to a decision. Some candidates would be retained in their current role as editor, while others would be reassigned to the role of admin.

RFA, in theory, would quickly and efficiently put individuals to their best possible use. Users with the aptitude to be admins could be given the tools more quickly than before. Users who could best serve as editors would be identified in order to prevent tool abuse.



The "RFA Problems"
An ideal RFA would neatly and efficiently sort candidates into their appropriate roles.

All and all, RFA does an excellent job-- but it is widely described as having serious problems.



Problem #1: RFA is intense
The first problem people often note is that RFA is a very intense process for the candidates.

The intense scrutiny applied to candidates may discourage people from considering adminship. Editors who are considering adminship may intentionally avoid helping resolve disputes for fear of sabotaging their chances at adminship.

If we could make RFAs less intense while still ensuring that we have high-quality admins, it would make for a better project. 

Problem #2: RFA rejects a lot of good talent
Right now, RFA rejects a lot of promising talent.

In some cases, a RFA is rejected because of explicit concerns about the candidate. But lots of failed RFAs come from not having enough data yet to reliably predict how the person will function as an admin.

Every RFA is a gamble-- nobody can predict the future. But there's wide agreement that it's better for many good candidates to get rejected than for a single bad candidate to get the promotion. After all, a rejected candidate can always come back once we know them better-- but in practice, adminship is essentially for life.

Indeed, it's not unusual for a good candidate to undergo multiple RFAs before finally being given a chance to become an admin. Some of our admins had to endure four or five RFAs before going on to serve admirably.


 * To avoid the possibility of a problematic admin, RFA initially rejects a lot of good talent.



Problem #3: Nobody's Perfect
Occasionally, RFA promotes someone who, in retrospect, probably shouldn't have been promoted.

Nothing will ever work flawlessly, RFA is no exception. There's nothing directly that can be done about this-- it's just a fact of life that no matter what standards are imposed, no matter how difficult RFA is, sometimes promising candidates just don't work out as admins (see also: the Peter Principle).

RFA's imperfection is reflected in a survey where a majority of responders (72%) endorsed the statement:
 * Admins are doing a good job, but there are some bad eggs that should be removed.

Problem #4: Post-RFA Wiki-stress
Over time, wiki-stress builds up. Wikipedians who initially did well as admins may develop frayed nerves. Even users who used to be best as admins may, through the wear and tear of admin life, start to become prone to incivility, newbie-biting, drama-philia, or other problems.

This was also demonstrated by an analysis of 20 recent desysoppings. . Most of those desysopped had gotten more than 85% support at RFA-- 7 out of the 20 were unanimous support. One even had over 150 support votes. So just making RFA even more difficult isn't the answer.


 * Even when RFA is perfect, a few admins will still develop problems later on.

Problem #5: Drama City


Inter-admin disputes are often far more difficult to resolve.

Demoralized admins stir drama and create factions. They demoralize others around them and sometimes start to see Wikipedia as a battleground.

Demoralized editors often do these things too, of course. But inter-editor conflicts are often far easier to resolve since editors have the luxury of having dedicated admins who can help them resolve their conflicts. But when a group of admins have a conflict, they generally have nowhere to go but Arbcom.

And going to Arbcom usually means several weeks or months of entrenched conflict, resulting in hurt feelings and demoralization all round.

Meanwhile, Arbcom is overloaded and overworked. As the only effective form of dispute resolution available to admins, Arbitration currently has to handle all inter-admin conduct disputes-- disputes that the community might be able resolve on its own, if given the chance. 

=*** The Fix ***= Fortunately, there's a growing understanding of how to fix these problems. All the problems can be solved or improved by the creation of a "mirror-image of RFA" that can re-assess adminship at some point after it has been granted.

One potential name for such a process would be "RfDA".

Features of RfDA


 * High barrier to RfDA-- RfDA must be difficult to initiate.
 * Resistant to mere retribution--- RfDA must recognize that even the most level-headed and consistently civil admin will still attract a certain amount of controversy.
 * Community-driven-- The community retains the ability to assign users to the various roles.
 * Not a trial-- While misuse of tools is one potential problem, there is also widespread agreement that a good admin also exhibits other traits like civility, impartiality, and cool-headedness.
 * No Stigma--- Unlike a forced desysopping, RfDA must not imply any wrongdoing, abuse, or any other "guilt".
 * Appeals to Arbcom-- As a final safety net, the Arbitration Committee must be able to over-rule decisions made by RfDA

Why it works

 * RFA won't have to be as intense because the community will be able to revisit earlier decisions.
 * RFA won't have to reject as much good talent.  Editors who show significant promise can be reassigned to the role of admin, since the decision isn't permanent.
 * RFA can afford to slip up every now and then, since mistakes can be corrected.
 * If an admin's behavior completely degrades due to wiki-stress, they can be reassigned to editor.
 * Overall drama can be reduced because admins who tend to inflame disputes can be reassigned back to editor.
 * Ultimately, errors at RFA or RfDA can still be corrected by the wisdom of Arbcom.

Why it is inevitable

 * Arbcom is maxed-out.
 * The Arbitration Committee in its current form can only handle a small, limited number of cases. Relative to changes in the overall admin population, the number of cases arbcom can take is essentially constant.


 * Problems increase with population and time.
 * There is no fixed limit on problems and disputes-- they scale with population and time. As new admins get appointed and the total admin population increases, there will be more and more problems that need to be addressed. Over time, a few formerly exemplary admins will inevitably develop problems.


 * Problems may then increase the rate of new problems.
 * The rate of admin-decay may also increase with the total number of "admins-who-would-be-best-as-editors". Which is it say, problematic admins make it all the more likely that other admins will also develop problems. Incivility tends to be met with incivility, and personal attacks tend to engender more personal attacks.


 * Left unaltered, things will continue to get worse.
 * RFA and Admin problems will probably continue to increase until changes are made to Wikipedia's system for evaluating adminship (RFA & RFDA).

What RFDA might look like
A lot of different structures or processes could accomplish the RFDA function:


 * The number of seats in the Arbcom could increase in magnitude to allow for far more cases.
 * Lower courts, with judges elected by community or appointed by ArbCom, could be created to handle additional arbitrations.
 * A "ethics & standards committee" could be formed to evaluate overall admin performance, issue warnings, and in some cases remove the bit.
 * A "mirror-RFA" process could be created to use discussion or !votes to determine admin status.
 * Reconfirmation requests or requirements could trigger a reconfirmation RFA.
 * Candidates at RFA could be asked to agree to a "reconfirmation contract" that will force them to return to RFA at a certain time or under certain circumstances. In this manner, many "promising maybe" candidates could be supported rather than opposed.