User talk:EVula/opining/RfA overhaul/archive, round 1

WereSpielChequers comments

• Suggest disenchanted not disenfranchised
• Crats don't need the power to desysop, though it would make this more convenient and could be implemented if this trials OK. But we could trial this with crats making the decision for a steward to implement.
• We need a double jeopardy rule, either no reraising of an offence already debated in this way. Or alternatively only reraise an old offence if new evidence has arisen.

Fully support this idea, once aquitted, always aquitted. fr33kman -s- 03:19, 10 February 2009 (UTC)

• Full desysopping is the penultimate sentence before we start getting into blocks and bans. We also need to give the closing crat the option to temporary desysop or impose some other form of community service or restriction.

Oh and I endorse the section about EVula being smart, awesome and able to leap over small buildings. WereSpielChequers 10:29, 17 January 2009 (UTC)

Aervanath fixed the improper word choice; that's what I get for writing it at 2:30 in the morning while drinking some lemonade. ;)

I think desysopping remaining in the stewards' hands for a trial process isn't a half-bad idea, though not what I'd prefer. But if closing RfDAs is in the hands of bureaucrats, we need to be the ones to ultimately throw that switch, just as we don't go running to the stewards to promote someone. If nothing else, it increases transparency by keeping it on-wiki. Currently, the only place that a de-sysopping shows up (such as an admin retiring) is in Meta's logs, not our own. Lame.

I do not see RfDA as even flirting with the idea of laying down bans or even a trivial block. It is in place strictly to address our need for an easy-to-manage desysopping process. If an administrator is abusing their tools, they should have the tools removed; any other issues need to be handled via another process. EVula // talk // ☯ // 20:00, 17 January 2009 (UTC)

I suspect you see the world in sharper contrasts of black and white than I do. My fear of a system that could only inflict a maximum punishment, is that much like a nineteenth century British jury being reluctant to hang shoplifters so RFDA would work badly if it could only choose between desysop or declare innocent. Yes we should restrict this to incidents where the bit is at stake, but the system could only work if it has grey outcomes for grey situations. WereSpielChequers 17:50, 2 February 2009 (UTC)

The grey outcomes "including the restriction of their use of certain powers or placement on administrative probation" (quoted from here) could either then be referred to ArbCom for consideration, or meted out by the closing bureaucrat in the same way that WP:Articles for Deletion also allows the closing admin to close it as "merge"/"transwiki"/"userfy" etc; not just "delete"/"keep". -kotra (talk) 02:49, 3 February 2009 (UTC)

Davidwr's comments

I'd prefer time-limited adminships, with existing adminships expiring in a staggered manner over an extended period of time.

For example, if we went with 2-year terms, existing admins would stay until 12-36 months after the proposal became effective, based on a lottery. As their terms expire, they would have to run for RFA again if they wanted to keep the bit. You would have 4 categories of expiring admins: Inactive editors who simply lapse, those who choose not to run again who simply lose their bit and regain any previous bits such as rollbacker, those who will be "shoe-ins" who will effectively be running again pro forma, and those who aren't shoe-ins who will face what amounts to a real RFA. You will also have those who previously chose to let their turn expire but later run again. Some of those would be shoe-ins, some would not.

With a known expiration date, people who oppose may be more understanding.

I'm not fixated on 2-year terms, 1- or 3-year terms is fine. The shorter the term, the more paperwork involved, the longer the term, the more like the current situation it will be.

Does this actually address the problem?

At present, arbcom has no problem efficiently desysopping bad admins for cause.

Also, ideally any user who can be trusted not to abuse the tools should in fact be an admin.

I submit that the problem with RFA is not because it's hard to get rid of an admin - rather, the requirements keep ratcheting up because the habitues can ratchet them up - it's a species of instruction creep. I've watched this happen. "No, I don't think three months is enough. No, I don't think five hundred thousand edits is enough."

The problem is that the present RFA gives too much voice to axegrinders - David Gerard (talk) 01:33, 18 January 2009 (UTC)

Although arbitrators are elected (I think), their decisions are not consensus. Furthermore, does it make sense to force every request through ArbCom? ArbCom is supposed to be for dispute resolution. Well, what if you have a very well behaved admin, who is polite etc. but who is no longer trusted by the community. If a dispute never arises, it could be argued that said admin is beyond the scope of ArbCom, much as content disputes are now. --Thinboy00 @981, i.e. 22:32, 29 January 2009 (UTC)

Familiar

This all sounds very familiar. Chillum 02:35, 28 January 2009 (UTC)

That's why I added it to WP:PEREN and WP:RFDA.--Aervanath (talk) 02:42, 28 January 2009 (UTC)

Personal agendas

Whilst I believe that there should be someway for the Wikipedia community as a whole to revoke a sysop bit, my main concern here is that there is massive room for abuse. I could nominate someone for desysoping simply because they didn't delete something I wanted deleting, or because they said something that hurt my feelings, or else simply because I just didn't like them. There's no criteria. Where's the checks and balances? Some editors just don't like sysops and might go around nominating all of them one-by-one; the sheer number of nominations could be stagering and that amount of disruption to Wikipedia would be very harmful to the project as a whole. Even a system whereby an editor nominated a sysop and polled for the support of, say 10 others, would be disruptive and harmful if noms happened ten-a-penny. fr33kman -s- 03:16, 10 February 2009 (UTC)

The reasons you cited are exactly the reason nothing like this has gotten underway. I personally do not feel this fear of abuse; I feel that the community would quite quickly develop methods to prevent abuse of the system. The current RFA system has WP:NOTNOW, which allows early closure of RFAs if there's no chance of it succeeding; the same logic would apply to frivolous RFDA's. However, enough people feel as you do that there is no chance of this sort of thing getting community consensus anytime soon...or ever. See WP:RFDA for a long list of rejected proposals (of which this is simply the latest).--Aervanath (talk) 18:29, 10 February 2009 (UTC)

Consensus

The flaw in this proposal is that you are talking about RfA and consensus in the same sentence. RfA is a vote, it is not a consensus forming exercise (you can tell by the fact that it is very unusual for people to change their opinions after their initial vote - consensus is all about changing your mind or making other people change theirs). Yes, crats discount obviously flawed votes and they sometimes try and account for unusual circumstances, but at the end of the day they make their decisions based on percentages because if they do anything else it causes an outcry. That means any RfDA based on the same system will also be a vote, therefore this proposal is fundamentally incomplete without a proposed percentage required to pass. --Tango (talk) 18:05, 11 February 2009 (UTC)

Actually, I haven't seen many RFAs where there weren't any vote-switchers. Granted, a majority of people don't change their vote, but that's true for most RFCs, as far as I can tell. People swoop in, take a quick look at the subject, leave a comment, and go their merry way, without becoming really invested in the discussion. Also, the percentage required for promotion (or, in this proposal, demotion), that it's as close to consensus as we're likely to get on an issue like this. Consensus is not unanimity.--Aervanath (talk) 18:24, 11 February 2009 (UTC)

Compared to the number of voters, the number of people changing their minds is pretty insignificant. A true consensus is not quite unanimity, but it's close (I like to think like this: Unanimity is everyone supporting, consensus is nobody opposing). It's pretty much impossible to get a true consensus when a large number of people are involved, so RfA works on "rough consensus" which is just a posh way of saying "supermajority". It's a vote, pure and simple. --Tango (talk) 18:30, 11 February 2009 (UTC)

When RfA was created, was there a "proposed percentage required to pass"? If the answer is no (I don't honestly know, that was way before my time), then why would we need such a proposed percentage in this case either? Can't we just let the percentage coalesce through practice, like it has with RfA? -kotra (talk) 18:47, 11 February 2009 (UTC)

When RfA was created the project was very small and it was normal for only a handful of people to comment on each RfA. When you have a small number of people involved, it is possible to get a true consensus. --Tango (talk) 18:48, 11 February 2009 (UTC)

Honestly, that was my hope; I wanted to write as few hard-and-fast rules for this process as possible, so it could evolve on its own (similar to how RfA developed). Sadly, that wasn't well-received, as too many people perceived the lack of hard rules as a lack of insurance against axe-grinders. *sigh* EVula // talk // ☯ // 19:26, 11 February 2009 (UTC)

I appreciate the principle of keeping things simple and letting processes evolve organically, but I just don't think it will work now the community is as large as it is. --Tango (talk) 19:50, 11 February 2009 (UTC)

Ok, well, if we need a percentage up-front, I suggest just copying WP:RFA:

At the end of that period, a bureaucrat will review the discussion to see whether there is a consensus for promotion. This is sometimes difficult to ascertain, and is not a numerical measurement, but as a general descriptive rule of thumb most of those above ~80% approval pass, most of those below ~70% fail, and the area between is subject to bureaucratic discretion.

This general percentage area seems to be widely used to gauge "consensus" throughout Wikipedia (with a few exceptions), so I'd imagine it would be a good starting point here as well. -kotra (talk) 21:24, 11 February 2009 (UTC)

If an admin still has 80% support after a few months with the mop, they haven't been doing their job properly. The very nature of the job means you end up annoying people. When you block someone, you'll annoy the person you blocked. When you delete a page, you'll annoy the person that created it. When you protect an article, you'll annoy the person on whose version you didn't protect it. Etc. Etc. The only way to maintain that level of support is to avoid any admin actions that are even slightly controversial, and we need people to take controversial actions from time to time. --Tango (talk) 21:32, 11 February 2009 (UTC)

So are you proposing a lower/higher percentage, the !voting being restricted to admins or autoconfirmed editors in good standing, closing bureaucrats being given more leeway in their decisions, or what? Your point is valid, and I agree we don't want to allow vengeful editors to gang up on an admin who is just doing their job appropriately. I think, though, there are a few options open to us to counteract that problem. -kotra (talk) 21:53, 11 February 2009 (UTC)

I'm not proposing anything. I think leaving it to ArbCom is our best bet (it's not ideal, but as you say, we don't have many options). --Tango (talk) 22:19, 11 February 2009 (UTC)

If it proves to be worse than leaving it to ArbCom, we could always stop it (and maybe re-sysop any admins who had been killed in the process). I'm willing to give it a try. Though we'd probably need a volunteer-only test first, to prove the concept (see my suggestion below). -kotra (talk) 22:48, 11 February 2009 (UTC)

When I first proposed it on WT:RFA, I stated that I was willing to put myself up a the first "victim;" that offer still stands (and there are enough people that don't care for me that I don't think it'd be a unanimous decision for me to keep my flags). EVula // talk // ☯ // 23:50, 11 February 2009 (UTC)

Vocal minority?

It's easy enough to say "I would support this, and loose some respect for any admin who felt they were above the need to maintain a community mandate", so I'll say it.

“ I would support this, and loose some respect for any admin who felt they were above the need to maintain a community mandate ”

Of course, the enormous mistake that is AOR probably scuttles it, but what can you do? WilyD 22:18, 11 February 2009 (UTC)

That is the very reason that RfA has some of its current issues, in my opinion; administrators should serve the community, rather than being a "ruling caste" the current appointment-for-life system. I've always considered RfA the community's process, but really the entirety of Wikipedia's administration should be the community's process; that includes who administers it and who doesn't. Currently, the community only has a voice in the beginning, with the end being up to ArbCom (thru a long and arduous process that can burn out editors even if they've been cleared of any wrongdoing). I'd prefer a sysop system that is as easy to get as it is to lose; then, the community might loosen up on some of its more ludicrous requirements for its administrators. (right now, people are too scared to take a chance on someone) EVula // talk // ☯ // 23:58, 11 February 2009 (UTC)

Administrators are here not to serve the community, they are chosen by the community to serve the project of building an encyclopedia. Most of the time the two goals overlap, but sometimes you need to do things unpopular with the community such as enforcing fair use criteria or not allowing large groups of people from acting disruptive toward a particularly political article. Chillum 00:22, 12 February 2009 (UTC)

I think even in those two cases, they overlap if you look at the "greater" community. Particularly in those two cases, there are sometimes smaller groups of people who go against the greater community and create misleading consensuses (that word sounds wrong). Consensus doesn't scale evenly, unfortunately. -kotra (talk) 02:25, 12 February 2009 (UTC)

Indeed, I think administrator's serve the community's needs, not their whims. If Editor X wants the article they wrote about their pet dog Mr. Fluffles kept but Admin Z deletes it, tough luck for Editor X. EVula // talk // ☯ // 06:24, 12 February 2009 (UTC)

Well, if I'm far more unpopular/abusive than I'd guess, I would accept it if a process like this was forced upon me by the community, although I am emphatically not open to recall. WilyD 13:22, 12 February 2009 (UTC)

Trial run

I'm not sure what the creator of this proposal had intended for the next step, but it seems to me that there would need to be some live demonstrations of the process before it would be accepted by the community. The best way would be to perform some actual RFDAs on actual admins open to recall. If there are any such martyrs willing to put their flag on the line (preferably including some controversial admins in the mix), and a bureaucrat or two to close the RFDAs, we could do this. That might be a tall order, though.

Is this the logical next step once this RFDA proposal is finalized? -kotra (talk) 22:42, 11 February 2009 (UTC)

In order for it to be a useful trial you need there to be genuine calls for the admin to be desysopped (although, some trials without such calls would be useful for determining where the percentage cutoff should go). Do we want to test this out on real life dramas? I'm concerned it might just make things worse. You would need to agreement of both the admin and the people calling for the admin's mop, which would be hard to get (if they were people that tended to agree with eachother, we wouldn't be there in the first place). If you don't have both parties' agreement then you risk the result being disputed and it all just being a waste of time. --Tango (talk) 22:53, 11 February 2009 (UTC)

I think agreement could be found. For example, Jéské Couriano is an administrator open to recall, and recently there was a bit of a brouhaha surrounding a block summary he made, including a proposal that could have desysopped him for it if it had been policy. To my knowledge, nobody directly called for his recall, but some may have considered doing so. This is just an example, I'm sure better ones could be found. But I'm more concerned with if this would be a good idea at all, disregarding for a moment if the participants could be found. -kotra (talk) 23:37, 11 February 2009 (UTC)

The problem with a live-fire exercise like that is that some people are going to object to the entire process itself and not the actual candidate. The exact role of bureaucrats in the process was also called into question, which is why I never so much as pushed it past the RfA talk page. I'd be more than happy if this gained enough momentum to be pushed thru now, though. EVula // talk // ☯ // 23:53, 11 February 2009 (UTC)

I think the closing bureaucrat will just have to discard any "oppose" !votes that are merely opposing the process itself. That may not be necessary, though. We won't know until we try. The role of the bureaucrat will have to just be a "closer" for now, not enforcer. For that, they'd have to ask a steward. As for momentum, I think seeing it in action is what most of the community will need to support or oppose this proposal. Otherwise, I don't think it'll get any farther than all the other proposals. I may be getting ahead of myself here, but I think we have two options:

1. Find someone who wants you desysopped and is willing to use this process, or

2. Find someone who wants a different admin open to recall desysopped, with both parties willing to try this process.

If #1, I think someone else other than you should compose/organize the RFDA, including finding a closing bureaucrat and canvassing the community for !voters. Otherwise, some paranoid individuals might think you've rigged the system in your favor. I offer to do this, though I'm a humble non-admin and still after three years a bit wet behind the ears. So if someone else volunteers that may be better.

If #2, you could do it no problem (if you're so inclined). -kotra (talk) 01:28, 12 February 2009 (UTC)

Non-binding initial RFDA?

I've been considering transcluding a non-binding reconfirmation RfA for myself once I hit my one-year promotion anniversary (Nov. 15), as sort of an WP:Editor review of my use of the bit. My plan for it would be to make clear that it was non-binding, but ask for a bureaucrat to evaluate the consensus as if it were a real RfDA, and see how the bureaucrat closed it. It would be formatted exactly like a regular RfA. However, this is 9 months off, and people probably want a test run sooner than that. EVula, if you really do want to try it out on yourself (or other willing guinea pig) first, why not just do something along those lines? That way we won't need to start any new process, we can just use the existing one, and the first few will be non-binding anyway, so no one would actually fear losing the bit. You'd only need to find someone willing to tread the hot coals of RFA a second time.--Aervanath (talk) 17:36, 13 February 2009 (UTC)

That's the thing, though; I think for it to be a true test-run, it has to have teeth. :) EVula // talk // ☯ // 17:52, 13 February 2009 (UTC)

Bit of a question, though. How can one volunteer for a binding occurance or such? I will argue that the community has a mandate to force me to undergo a reconfirmation or a RFD if it wants to, but it's useless if one volunteers.

Okay, so let me say this, then, if you're looking for a guinea pig, I would accept it if I was forced by a consensus to undergo such a procedure, and there was actually a bureaucrat willing to close it (and a steward willing to -sysop, if that was what the community wills). But I will not volunteer, nor would I consent to an optional/nonbinding one. (Is this clear enough, or do I need to double the space between my lines?) WilyD 20:13, 13 February 2009 (UTC)

I think basically what this boils down to is the admin voluntarily offering to abide by the wishes of the community, even if they think the community happens to be wrong. Doesn't seem useless to me, in fact, it's how consensus works. -kotra (talk) 22:33, 13 February 2009 (UTC)

No, voluntary/nonbinding makes it the same sham as AOR. What I am saying is that if such a thing were proper, you would need an admin who would non-voluntarily abide by the wishes of the community, regardless of what the community decides. WilyD 22:48, 13 February 2009 (UTC)

Ow, my tiny brain. Are you saying that AOR is a sham because the admin can say they'll abide by the wishes of the community, but when it comes down to it, refuse to give up their adminship? I.e. it's "nonbinding"? If so, I'd have to agree. However, I'm not sure that this trial run would also be nonbinding. My reason: the !voting, if done correctly, would be a reflection of the community, and therefore, the Steward would have to desysop according to the Stewards policy. However, the Steward handbook might conflict with that, since it says to consult "the local wiki's policy on removal of rights". Currently, WP:DESYSOP (which, while not technically a policy, seems to be the closest we have on the subject) doesn't specifically describe a way for the community to directly ask for a desysopping. However, it does say "The community also has, in theory, the power to remove administrator access from any contributor." So, I'd argue that the trial run, if judged to fairly represent the community, would have the power to involuntarily desysop the admin. It would be unprecedented, but valid.

But none of this matters as long as the admin isn't a liar. As long as they stay true to their word and voluntarily steps down if the community asks them to, none of the above bureaucracy needs to be waded through.

Man, I hope I understood you correctly this time, and didn't write all that in vain. -kotra (talk) 00:21, 14 February 2009 (UTC)

Err, I think this is more or less correct. The problem with AOR is that it's voluntary, and you can withdraw from it, or change your terms anytime - or simply refuse to comply with the outcome. Realistically, there's probably little or no need to have a procedure to desysop those who'll honour it when they're told by the community to step down. But we can't trust in that. This is probably part of the high "trust" standards at RFA - you can never rely upon being able to fix mistakes there. A true trial would need to not be a volunteer, but would need to be someone who recognised the legitimacy of the process, if you're trying to prepare for that. Otherwise you're trialling something different. WilyD 00:36, 14 February 2009 (UTC)

That makes sense. Then, you're advocating the trial run be "forced" upon an unwilling admin? The sort of "bad-case scenario" RFDA would be most useful for? I think this is possible. The admin doesn't even need to participate if they don't want to, but they'll need to know that the RFDA may result in their desysopping, even involuntarily (assuming the bureaucrat and steward determine that everything checks out).

However. That situation (admin unwilling to participate or accept the ruling, getting a steward involved, a huge argument about how valid the "trial run" would be) would explode drama everywhere, and I don't want that stuff on me. I propose a volunteer-only run first. If that's successful, a kicking-and-screaming run could come next, since at least by then the community might have softened up to the idea a little. -kotra (talk) 01:16, 14 February 2009 (UTC)

No, this is why I've suggested an admin who is willing to participate and accept the ruling, but is explicitly not a volunteer. Above, I say

“ if you're looking for a guinea pig, I would accept it if I was forced by a consensus to undergo such a procedure, and there was actually a bureaucrat willing to close it (and a steward willing to -sysop, if that was what the community wills). But I will not volunteer, nor would I consent to an optional/nonbinding one. (Is this clear enough, or do I need to double the space between my lines?) ”

.

I'm not sure what else I can say. I don't think an explicitly voluntary one would demonstrate much of anything. It would have to be explicitly non-voluntary. WilyD 01:28, 14 February 2009 (UTC)

The "forced by consensus" bit is the hard part. How do we get consensus to do a "trial run" where everything is the same as if it were the real thing, including consequences, if there's not yet consensus for the real thing? It really wouldn't be a "trial run" anymore, the community would be giving consensus for a single, full-blown RFDA. I think that's why we want to start with a toned-down version. It would demonstrate, at the very least, that the process works or fails in certain situations, and no consensus is needed to do it. -kotra (talk) 01:44, 14 February 2009 (UTC)

A single instance, rather than a "usual process", would strike me as a trial run. WilyD 04:58, 14 February 2009 (UTC)

Maybe so.. but I doubt people would give consensus to force someone to be possibly desysopped with a system they haven't seen in practice yet. Unless they just don't like you. Are you generally hated? -kotra (talk) 19:03, 14 February 2009 (UTC)

No idea. I did having the worst showing among ArbCom candidates who were sysops in the last election. That certainly could be read as people having little faith in my abilities. WilyD 21:58, 15 February 2009 (UTC)

Hmm... I don't think you're hated enough yet. You should abuse your powers more. -kotra (talk) 22:53, 15 February 2009 (UTC)

If I got less than 20% support in a RFDA, I would expect to be desysop'd. WilyD 01:02, 16 February 2009 (UTC)

Process thought

I'm still kicking around the RfDA idea in the back of my head. I still believe that the best way to accomplish a feasible system is for it to be simple, but perhaps a bit of additional bureaucracy is needed to provide an extra level of protection.

So what's my current revised thought? Maintain the parallel between RfA and RfDA.

In an RfA, there are standard questions that, while "optional," will sink an RfA if either unanswered or have poor answers. Similarly, to prevent axe grinders, RfDAs will require three examples of abusive or non-productive behavior, coupled with some evidence of an attempt at dispute resolution for each (ie: Vandal X can't be blocked by Admin Y and immediately call for his resignation as a result). An attempt at dispute resolution in some manner (a talk page discussion, an AN or ANI thread, ArbCom, mediation, etc) will establish that there's a pattern of negative behavior, and that filing the RfDA isn't a knee-jerk reaction.

I can think of a few requirements, such as evidence needing to be as recent as such-and-such a time, or limiting the number of times a piece of evidence can be brought against an admin, but for now, I think leaving it open is sufficient. (and for those that may find the concept familiar, it's because I'm adapting it from my insanely bureaucratic recall criteria).

Thoughts? EVula // talk // ☯ // 21:39, 19 February 2009 (UTC)

I had thought of including desysopping as part of it, but was unsure at the time, maybe some of the stuff at WP:RFBAN might be of interest. MBisanz ^talk 21:41, 19 February 2009 (UTC)

What if the admin does just one blatantly desysop-worthy action, like blocking someone with their social security number and credit card info as the block summary? Or two moderately desysop-worthy actions, like deleting the AN/I page on two separate occasions, each time without supplying a reason? A requirement of three incidents could prevent legitimate RfDAs. Frivolous RfDAs could be quickly closed as per WP:SNOW, as frivolous RfAs are already. -kotra (talk) 22:05, 19 February 2009 (UTC)

Fair points; I think your "1 action" example is a bit extreme, but I do believe you're correct that sometimes someone should be demoted over two bad actions. I think making a two-instance minimum is the best compromise; it will still restrict the viability of axe-grinders, but still allows for bad apples to be justifiably taken to task. EVula // talk // ☯ // 23:00, 19 February 2009 (UTC)

I think that could work. If any one action was blatantly desysop-worthy, then that's probably something that you could post as an emergency motion to the ArbCom. Two is probably a reasonable compromise.--Aervanath (talk) 07:23, 20 February 2009 (UTC)

I suppose that's ok, a minimum of two would weed out the majority of frivolous RfDAs, and I admit that a single action being enough for de-adminship is extremely rare and usually falls under current emergency desysopping measures. I still think this is unnecessary bureaucracy since we already have the snowball clause, but perhaps it would get invoked too often (it's my belief that IAR and its applications like SNOW should be rarely used). -kotra (talk) 22:03, 20 February 2009 (UTC)