User:Geogre/RFA-Derby


 * 1) Some people will tell me I'm wrong, but one fable I heard to explain the "Carnildo affair," where user:Carnildo was appointed an administrator despite having polled 60% approval on his fourth RFA and despite very vocal dissent from "respected Wikipedians" as well as disrespected ones, was that it occurred because "RFA is broken."
 * 2) You see, RFA promotes some unworthy people, and there are no reliable procedures for getting rid of them, and it fails to promote some worthy people.  We all agree with that, I think.
 * 3) RFA is distorted by the IRC lensing effect, where someone chummy and non-threatening on IRC all day can get huge landslides of approval without having (believe it or not) more than a few hundred Wikipedia edits and no article edits.  The IRC lensing effect can work the other way, too, when long time and tireless IRC personalities can voice a dark view of a candidate and see an avalanche of "oppose" votes come in.  Whenever you see pro or con votes in the 100's, that person has spent some time on IRC, even if he or she had no desire to influence the vote at all.  (This is not an accusation or a blame laying.  This is an observation.)
 * 4) RFA is influenced by "RFA week conversions," where people who flaunt process regularly suddenly fly right, speak nicely, and advocate policies for a week or two before the RFA opens.
 * 5) RFA is influenced by edit counters and by editors who rack up hundreds of edits by putting a tag on every article or talk page warring or doing RC patrol a few days.
 * 6) RFA seems to require saints, as a single "fuck" in an edit can derail a perfect editor's chances, and grudges last an eternity.
 * 7) "Consensus" is a word with nearly no definition, but it has been held to be 75-80% approval on RFA, and it doesn't take much to drop such a percentage.
 * 8) Because Carnildo kept failing the RFA's (and he kept failing for his own fault in my opinion, but others disagree), the bureaucrats, with the persuasion of two current ArbCom members and two former ArbCom members, decided that they never needed votes in the first place, that who got the buttons was at their discretion anyway.  Therefore, despite never having defied the community before, they were going to "fix" the "broken" RFA by just deciding to give out the powers anyway.
 * 9) This has made a great, great many people angry.  Just ask people how they feel about it, and see if you don't consider the number of angry people a great number or not.  If you don't, might I suggest some accounting texts for you?

Therefore if "RFA is broken" is the problem, why not ask the community to fix RFA instead of appropriating more power in fewer hands and reducing the level of debate? Why not have a derby?

Here is my proposal: I would like for each person who has the time, energy, and wit to come up with a whole or partial solution to the assumed premise that "RFA is broken." Let's let this run for a week or so (an eternity in Wikipedia time, I know). No one bashes any other idea. No one takes credit for any. No debate at this juncture. At the end of the week or so, let's just vote for the ones we like best. Of the various ideas, the top 3-5 will be chosen and presented to the general community for a vote. We ask people to choose which of them they like best: 1, 2, 3, 4, 5, or status quo. If we get consensus on any, that's how we go. If we get majorities, we eliminate the low-finishers and vote again. If we get a majority, even a simple majority, for status quo, then we answer back to the 'crats: "No, RFA is not broken: some people just don't need to be promoted, despite your 'better judgment.'"

I will offer proposals myself, yes, as I agree that RFA is ailing, though I hardly think the Carnildo case was a symptom of that disease. Geogre 02:07, 21 September 2006 (UTC)

How to enter your horse in the derby
We can do it a bunch of different ways, but the best way that I can think of is that each person craft his or her solution and lodge it in his or her own user talk space. (E.g. this is at User:Geogre/RFA-Derby, and yours could be at User:YourNameHere/RFA-Reform. This way, you can keep your candidate no matter what.  If things don't go well here, your own good ideas remain where you can refine and offer them without any more intervention here.

In the space below, just put your link:

The Horses!

 * This is where the link to your talk page space and candidate lie


 * Okay. Since our current RFA system is a hybrid, I'm going to list both extremes as a thought experiment. Comments welcome. Please don't assume that either extreme is my actual opinion; this is just brainstorming. Also, please don't be offended if the language seems harsh, they're called extremes for a reason, and I mean no disrespect to anyone, bureaucrat or otherwise.  &gt; R a d i a n t &lt;  22:22, 21 September 2006 (UTC)
 * Since the current mash-up is not well loved, why not use both of the above methods where appropiate. - brenneman  color="black" title="Admin actions">{L} 07:38, 27 September 2006 (UTC)

RFA must be a vote
The basic definition of an admin is someone who can be trusted to not abuse the tools, and RFA serves to gauge the trust our community places in the candidate. Trust is, by definition, a subjective issue, and very personal. While most people can explain quite reasonably why they do or do not trust someone, other people can understand the reasoning fully and yet disagree equally reasonably. No editor can rightfully claim that another editor is wrong on this or that his trust (or lack thereof) is invalid.

As such, it is impossible for any third party, such as a bureaucrat, to accurately interpret community consensus based on arguments - doing so will inevitably lead to the bureaucrat counting or discounting reasons based on his personal opinion, which, because of human nature, isn't any better or worse than anyone else's opinion. Thus, any RFA that is interpreted by a bureaucrat in any such subjective way, is by definition biased.

The only fair way to judge an RFA is to make it an actual vote, discounting sockpuppets, of course, but nobody else even if their reasoning seems spurious. We should create a strict limit of two-thirds or three-quarters of the participants, and use that as an absolute divider between who passes and who fails. The role of a bureaucrat is a strictly formal one (which gives them more time for renaming users, checking bots, and writing articles). Incidentally, this system is in use in most other-language Wikipedias that have a formal adminship system, and it works fine there. Most dissent on RFA is because of its vagueness; a strict borderline may be somewhat arbitrary, but at least it is open, egalitarian, and fair. Nothing else is.

RFA must not be a vote
The basic definition of an admin is someone who can be trusted not to abuse the tools, and RFA serves to cauge the trust our community places in the candidate. Trust is, by definition, a subjective issue - however, there are certain reasons that are more valid than other reasons. A candidate who was a hostile edit warrior in the past or escalated a dispute by handling it wrongly is likely unsuited for adminship; a candidate who has a limited field of interest perceived as trivial or ludicrous, or who is in disagreement about an issue important in the real world but unrelated to adminship, would likely be a good candidate.

Many Wikipedians are quite capable of judging other people rationally, and estimating their efficacy in adminship based on past actions. However, several editors are less rational, and can be seen to hold a grudge against a candidate, or to oppose for reasons of principle, or to support because of reasons that may be laudable but have no bearing on adminship whatsoever.

The only fair way to judge an RFA is to consider the given arguments (both support and oppose) on their merit, and discount all that have no bearing on adminship. This is a difficult task, but that's what we elect bureaucrats for. RFA should not resemble anything like a vote; instead, a few people should make brief, sourced arguments on why the candidate may or may not be suitable. People who have nothing to say but "me too" or "support per that user" should stay away, they're not contributing to the process. Also, comments should not be signed - it's the argument that count, not the person who makes it.

After some flexible amount of time, several bureaucrats should discuss with one another whether there are, in their best judgment, any compelling reasons not to promote the candidate - and if not, perform the promotion. No single bureaucrat should make this decision, if only for the sake of perceived fairness. To ensure accountability to the community, bureaucrats should serve terms of up to one year. Depending on bureaucrat judgment may be somewhat subjective, but at least it ensures that people are promoted or demoted for a good reason, which is only fair. Nothing else is.

RFA first a vote, then not so much
The discussion for adminship proposal was (at one stage) a quasi-article writing section firt followed by the existing pseudo-vote.

Instead, begin with a straight vote. Any nomination with outside certain ranges of oppose/support stays a straight vote. This means that for the uncontroversial "certain-to-fail" and "certain-to-pass" nominations no further thought is required. However, once a certain number of oppose or support votes are gathered, or when the ratio is within certain bounds, it passes into a non-voting quasi request for comment. This should be a neutral and collaborative account of the nominee's actions.

Then at the end of this, another straight vote, hopefully with clearer consensus.

This is intended to streamline the bulk of the process, while allowing for more careful (and slower?) analysis of more complex cases.