User talk:Roux/RFA-reform

Archive: /v1

Version 2 of XFA process (20:21, 12 July 2009)

This is for discussion of the version dated 20:21, 12 July 2009. Its structure follows that of the previous discussion, which Roux created, as that seemed to work fairly well. I've included an "In general section", as I hope there will be more comments on the principles from a wider range of editors. I've also added specific sections for proposing additions to / removals from the list of grounds for an XFA. --Philcha (talk) 08:26, 14 July 2009 (UTC)

In general (v 2)

Support general idea (v 2)

1. Totally --Philcha (talk) 08:47, 14 July 2009 (UTC)
2. Okay - with addition that I will "drop in" later, when I find the appropriate place. LessHeard vanU (talk) 19:55, 18 July 2009 (UTC)
3. Support - I opposed the previous version, but this is workable. I would recommend adding a comment that the certification requirement doesn't require the administrator or other certifying user to actually support the desysop, just to certify that it could be reasonable grounds for desysop. I would be willing to certify an RFDA, but then vote to keep them as an admin.--Aervanath (talk) 06:27, 20 July 2009 (UTC)
4. Looks to be on the right track, and I do support the general idea being put forward here. Shereth 14:06, 20 July 2009 (UTC)
5. Support. Looks like the two admin or arbcom certification will derail most frivolous filings.--Fabrictramp | talk to me 15:06, 20 July 2009 (UTC)
6. Support general idea. Septentrionalis PMAnderson 20:05, 20 July 2009 (UTC)
7. As before. Jafeluv (talk) 08:55, 21 July 2009 (UTC)
8. Support I agree with the proposal. ~~Awsome EBE123~~(talk | Contribs) 23:24, 22 February 2011 (UTC)

Oppose general idea (v 2)

1. Per Stifle above, and everything else I can possibly oppose. Pzrmd 05:52, 21 July 2009 (UTC)

   • I see no post by Stifle above; can you be specific about why you're opposing? Exploding Boy (talk) 16:27, 21 July 2009 (UTC)

   • I think Pzmrd refers to this. You can see it if you expand the "Discussion on previous version" box at the top. Jafeluv (talk) 07:04, 22 July 2009 (UTC)

2. I'm opposing because I'm not convinced by the current project page that there's a problem and that this new process will solve it; indeed, it seems to me that most of Roux's comments seem to support the current system of dispute resolution ending with ArbCom. Fixing the RFA system--which I agree has become unwieldy and just silly--by creating a parallel system seems like a very roundabout and unnecessarily complicated way of doing things: the opposite of streamlining and simplifying. There's often talk of "rogue admins," yet I never see any real examples of admins gone wild who cannot be brought under control and, if the problem is severe enough, desysopped. Roux himself says "in order for admin abuse to get to the stage where this process can even be implemented" the abuse must be "significant," and that the idea that admins conspire to support each other "just ain't so." What, then, is the reason for creating a new system to deal with the alleged problem instead of using the existing systems? If the real beef is that policies are sometimes inconsistently applied, particularly with regard to blocking, then the solution surely is to tighten up and clarify the policies and guidelines and/or to improve the current dispute resolution process. [ETA:] Additionally, surely simplifying the RFA process, which has been turned into precisely what it's not meant to be, ie: a Great Big Deal, would be a better way to fix problems with it than supporting a flawed process by creating an entirely new process. Exploding Boy (talk) 16:27, 21 July 2009 (UTC)

   It is my belief that creating this process will solve a lot of the problems at RFA, most notably that of it being a trial by fire. If voters know that an admin can be removed by the community should they abuse the tools--without the incredibly lengthy process required by ArbCom--I think they are less likely to be quite as nitpicky as they currently are. In addition, people are less likely to become emotionally invested. More to the point, the existence of a process like this helps ensure that admins cannot get away with abuse; people are leery of starting an ArbCom case (due to time required, chain of prior DR required, etc) over abuse of admin tools. With a relatively simple (but not easy to abuse) community-based option for desysopping, admins will need to be more careful about abusive behaviour. Note that doesn't mean avoiding controversial behaviour, as most controversial behaviour is well within admin remit and not an abuse of the tools. → ROUX ₪ 20:44, 21 July 2009 (UTC)

   Yet it still seems to me that the problem you're proposing to solve isn't the one you're naming. The RFA process can be solved by eliminating the current ridiculous campaign-style Q&As in favour of a better, simpler nomination process whereby users who have a good history are nominated by other users who think they'd be good at dealing with admin-type stuff and everyone gets a chance to comment--the old system, in fact. Anyway, I have yet to see any real evidence that admins are going wild with their tools and escaping the consequences. And if admins are being "abusive," as you say, and getting away with it, then the problem is one of policy and the policies and guidelines need to be streamlined and clarified. We already have Wikiquette alerts; Requests for comment; both informal and formal mediation; the Admin noticeboard(s), and the Unblock request template, as well as Arbitration. I can't see how your proposal does anything beside another element to the DR process. Exploding Boy (talk) 23:07, 21 July 2009 (UTC)

3. Oppose—the idea is well thought-out, and it pains me to oppose a serious proposal made by a serious user. However, I am one of those (maybe few) who believe that RfA is actually one of those processes on Wikipedia that work as they should. There are already many ways to desysop someone if there are clear cases of abuse of tools, blatant violation of important policies, etc. In short, RfA is not broken, and neither is the desysopping process. —Ynhockey ^(Talk) 22:24, 17 August 2009 (UTC)

   Just wondering if you could clarify for me. Do you think this proposal would be harmful, or just unneeded? Thanks!--Fabrictramp | talk to me 23:25, 17 August 2009 (UTC)

At this point, I believe it is unneeded. Whether it will be harmful would be determined in practice, and I see ways in which it can be (e.g. trolls constantly requesting de-adminship of admins in good standing). In all, I am of the very strong opinion that RfA is not broken. If there is a need for more administrators, but not enough people can pass RfA, an alternative could be to create other user groups (like we have rollbackers), who can do specific actions. There is no problem with RfA itself and the current process of de-adminship (RfC/ANI > ArbCom) has worked fine so far, as far as I am aware. —Ynhockey ^(Talk) 16:38, 19 August 2009 (UTC)

Discuss general idea (v 2)

1. This proposal is partly based on the observation "admins are currently essentially promoted for life, and so a great deal of care must be taken--perhaps too much--when promoting them." So why not just change that? Elect admins for a fixed term, say perhaps three years, with an option for re-election. That provides a fairly painless way of eventually getting rid of "rogue" admins, and also prunes away those who have lost interest. SamuelTheGhost (talk) 22:38, 24 July 2009 (UTC)

   I proposed that on this page, and it was shot down, has been removed so we can concentrate on the one proposal that has garnered some support. → ROUX ₪ 13:29, 25 July 2009 (UTC)

   I've reformulated this suggestion at User:SamuelTheGhost/Re-electing admins. Comments there welcome. SamuelTheGhost (talk) 10:53, 26 September 2009 (UTC)

Who may initiate an XFA (v 2)

Support "Who may initiate an XFA" (v 2)

1. I think Roux has considered the right parameters, but I think their content needs tweaking - see Discuss "Who may initiate an XFA" (v 2) --Philcha (talk) 08:47, 14 July 2009 (UTC)

Support: I think these are fine. If anything, the requirements should be less restrictive, not more.--Aervanath (talk) 06:31, 20 July 2009 (UTC) On second thought, per S Marshall's thoughts below, I think there shouldn't be any requirements. I'm not going to oppose on those grounds, though. I will still support this proposal as is.--Aervanath (talk) 06:34, 20 July 2009 (UTC)

Oppose "Who may initiate an XFA" (v 2)

1. Oppose any limitations. Legitimate concerns of countering abuse of recall initiative should be handled otherwise. Anticipated abuse of process is not an excuse for profiling potential claimants. Right now, it's like a real-life courthouse that accepts claims against police only from model citizens over 50 (once a year, or perhaps once in a lifetime). But guess what, model citizens over 50 (like yours truly) confront police very rarely.

   In particular, "no editing restrictions" clause just tempts to be gamed. Right now sysops are free to topic-ban anyone at no cost - quite an effective way of culling the opposing voices. Shut up or be damned ye peasants! The proposal also denies recognition to cased of sysop collusion (or call it consensus) leading to abuse of their privilege ("one per editor per admin"). Etc. Protect everyone, not the model citizens. NVO (talk) 08:26, 20 July 2009 (UTC)

   You're letting perfect be the enemy of good here. There is no way on God's green earth that this can pass community approval without limitations on who can initiate desysop proceedings. → ROUX ₪ 08:46, 20 July 2009 (UTC)

   Perhaps there should be a sort of public ombudsman to speak out as proxy for the unwashed ineligible masses. Someone immune to topic ban. P.S. Appeasing the public results in lame laws. NVO (talk) 09:18, 20 July 2009 (UTC)

   Which then just devolves into a sticky mess about who that person should be. Turtles all the way down, and so on. But more to the point, in order for admin abuse to get to the stage where this process can even be implemented there must already be a community consensus that there is a significant problem that needs to be addressed. And that ensures there will be plenty of people eligible to start this process. → ROUX ₪ 09:49, 20 July 2009 (UTC)

2. Oppose topic ban limitation, as creating second-class citizens. If someone under a topic ban for topic A conflicts with an admin over topic Z, he's still at a disadvantage, but should be able to file. If he files over topic A, he's in violation of the ban, and the complaint, and he, will go away. Septentrionalis PMAnderson 20:08, 20 July 2009 (UTC)

   That's an interesting point. Hm. → ROUX ₪ 01:08, 21 July 2009 (UTC)

   How would you feel with the addition that users not meeting the conditions may ask someone uninvolved to file for them? (This also provides another reality check, the chance for someone else to say "This is probably a bad idea") → ROUX ₪ 01:14, 21 July 2009 (UTC)

   Not really adequate; that is asking someone else to phrase one's arguments. Since editing restrictions will (relevant or not) be brought up during the certification process, there is already a reality check to cover this. Septentrionalis PMAnderson 14:24, 21 July 2009 (UTC)

   My major concern is avoiding "waaaah EvilAdmin topicbanned me waaaaah" and subsequent wikilawyering/gaming. I think it is not unreasonable to say that if this process can be easily gamed, it will be shot down in flames. How would you propose avoiding that sort of gaming? → ROUX ₪ 20:50, 21 July 2009 (UTC)

   What about changing "Never blocked by the admin who is the subject of the RFDA, except where that block was found to be improper or excessive." to "Never blocked or subjected to editing restrictions by the admin who is the subject of the RFDA, except where that block or edit restriction was found to be improper or excessive." This avoids grudges and axegrinding, allows for reasonable participation by those who were wronged, and ensures that someone who is currently topicbanned may both initiate and comment, just not if they are personally involved with the admin in question, and encourages that potentially-abusive current editing restrictions must be dealt with via other DR processes first. → ROUX ₪ 20:54, 21 July 2009 (UTC)

3. Comment. I boldly added the clarification "registered" user, as if you are going to be counting edits, or saying anything about a user, it is really impossible to reliably include IP users, as most of them are using school accounts, work accounts, or dynamically assigned IP addresses, and I believe that the intention in writing "user" was to actually mean "registered user". 199.125.109.58 (talk) 16:12, 22 July 2009 (UTC)

Discuss "Who may initiate an XFA" (v 2)"

• I think 200 edits is too low a threshold, and suggest 1,000. I know I was still flying by the seat of my pants after 200 edits and hardly knew what an admin was. Not all edits are equal - mainspace vs user space vs policy pages etc., major vs minor, etc.; reverted edits; etc. etc. etc. ... OTOH someone who's made 1,000 edits is likely to stick around for a while. --Philcha (talk) 08:47, 14 July 2009 (UTC)

People thought 1500 was too high for commenting threshold, and it makes absolute sense for the two to be the same, so I deliberately went lower than I thought people would want. Hopefully we will be able to move back to a happy medium. → ROUX ₪ 09:02, 14 July 2009 (UTC)

• I think present for 2 months is too low, and prefer the original 6 - for much the same reasons as I'd prefer 1,000 edits rather than 200. --Philcha (talk) 08:47, 14 July 2009 (UTC)

See above; went deliberately low in order to make finding a happy medium easier. → ROUX ₪ 09:02, 14 July 2009 (UTC)

• Roux has a "Never blocked by the admin who is the subject of the XFA, ..." clause in Who may comment on an XFA (v 2) but not in Who may initiate an XFA (v 2). Roux, can you please explain your thoughts on this. --Philcha (talk) 08:47, 14 July 2009 (UTC)

Someone who was once blocked may have an axe to grind. However, a user who has been unfairly or improperly blocked (as defined by the community; note requirement to pursue other DR first) may well have a non-axegrindy case for desysopping. That is the sort of situation where the certification process comes in handy. → ROUX ₪ 09:02, 14 July 2009 (UTC)

• I don't think we need a number. These discussions will be closed by bureaucrats, and mirror the RFA procedure; there's no minimum threshold to !vote at, or act as nominator for, an RFA, but I think bureaucrats can be trusted to know whose !votes to disregard.—S Marshall ^Talk/_Cont 11:21, 14 July 2009 (UTC)
• I would like to see whether the nascent AdminReview process may also be an agreed avenue for bringing forward a request. In this instance some of the criteria for assessing whether there is a case may be bypassed, in that a request forwarded from that page will have already been determined as having good grounds - the processes here will then need to look at whether XfA is appropriate. I will also note this proposal to the AdminReview talkpage to see if they are interested in using this as an option in their process. LessHeard vanU (talk) 20:07, 18 July 2009 (UTC)
  • Hmm, I think there is a really, really logical fit there. Perhaps something like... if an AdminReview case (correct term?) has found abuse, then no need for certification. That would also bypass Philcha's objection about rug-sweeping. → ROUX ₪ 20:43, 18 July 2009 (UTC)
    • It should be noted that AdminReview is a voluntary process, so the likelihood of there being a great deal of traffic from it is negligable (Sysops with a predilection for abusive practices are not going to use it - especially if the end result is a suggestion for RfC, RfAR or this), but an adaption of some of the investigation criteria and methods may be stolen...er, used here. LessHeard vanU (talk) 21:17, 18 July 2009 (UTC)
      • I'm concerned about this becoming overly bureaucratic. As proposed it's fairly loose, and will allow the community to develop norms and guidelines around it as it is used. → ROUX ₪ 21:24, 18 July 2009 (UTC)
• I prefer 2000 edits and 6 months of editing to avoid sleepers and meatpuppets.--Lenticel ^(talk) 04:59, 20 July 2009 (UTC)
• Can we clarify whether an uncertified RFDA counts towards the limits of once per incident, once per admin per year, and once per user per admin? Stifle (talk) 08:15, 20 July 2009 (UTC)
  • I would think yes, yes, no, in that order. We want to prevent frivolous cases wherever possible (alleviating your concern about hounding). → ROUX ₪ 08:44, 20 July 2009 (UTC)

I believe that if a user with less than the required number of edits believes a user should be put up for de-adminship, could they not ask a more senior editor too look into it for them and file it on the less experienced editor's behalf. Is this correct? Malinaccier (talk) 16:38, 20 July 2009 (UTC)

Yes.--Aervanath (talk) 17:57, 20 July 2009 (UTC)

I have added notes to that effect on the proposal page, saying proxying is okay but be careful. I'm going to expand on it slightly to indicate that the request needs to be made onwiki, and that anytime proxying is done it needs to be stated umabiguously ("I am filing this on behalf of.." or some such). → ROUX ₪ 01:00, 21 July 2009 (UTC)

Who may comment on an XFA (v 2)

Support "Who may comment on an XFA" (v 2)

1. Here too I think Roux has considered the right parameters, but I think their content needs tweaking. --Philcha (talk) 08:47, 14 July 2009 (UTC)

Oppose "Who may comment on an XFA" (v 2)

1. Strong oppose to "no editing restrictions" and "never blocked by the suspect" clauses. Hey, if a real court of law discarded statements by involved witnesses then there would be no witnesses at all... Judges and jurors must be uninvolved. A witness does not need to be (he/she signs a "no false statement" paper and is aware of penalties for perjury). Witnesses don't decide the case! As for the 500/3 months clauses ... I'm undecided. They are a good barrier against useless rants, but they also exclude true witnesses that may produce decisive evidence. Under proposed rules, they must find a "profiled" proxy to do so. NVO (talk) 08:36, 20 July 2009 (UTC)

   You're viewing this in far too legalistic terms. Those restrictions are proposed to prevent meat- and sock-puppetry, and to prevent e.g. someone bitching about a perfectly valid block and instituting a request for an admin to be desysopped over it. How would you prevent that from happening? → ROUX ₪ 08:43, 20 July 2009 (UTC)

   I know, I'm the worst subspecies of a lawyer - an old accountant training for law degree :(( I answered it below and reiterate it here: instead of an RFA-style popular vote masked by "comments" with a single crat making statements, desysop should be a panel decision, not necessarily as regulated by Arbcom. It should not be bound by any popular vote (or "consensus" or pileup). Desysop calls are, by definition, "Citizen against Member of Corporation". A popular vote tempts with its simplicity, but with "corporation" on one side, the plaintiff is disadvantaged. Unless the plaintiff is a local celebrity and has his gang ready for action. Seeing the process turning into a Wild West saloon you introduce entry barriers, %500 cover charge (corporate cardholders get free pass) etc. NVO (talk) 09:07, 20 July 2009 (UTC)

   And if it were a panel decision then we'd just be talking about creating another ArbCom, which would be silly, no? Crats are elected precisely to judge a difficult consensus (RFA); this would fall well within their current remit. And if the history of RFA has taught us nothing else, it is that attempting to make changes -- especially barriers to entry -- to entrenched processes is an exercise in sheer futility. → ROUX ₪ 09:46, 20 July 2009 (UTC)

   If Arbcom has other priorities and is already loaded with their own cases, then a second panel is not silly. Not at all. NVO (talk) 15:51, 20 July 2009 (UTC)

   Large sections of the community are already discontented with the amount of power that ArbCom is perceived to have. I doubt that setting up a rival group would do anything other than encourage the already-dangerous level of factionalism that exists within the project. ArbCom's priorities are resolving behavioural disputes where nothing else has worked. It is the final port of call. This proposal is intended to put power more directly back in the hands of the community while avoiding as much as possible those who will inevitably attempt to game the system and keep their pitchforks razor sharp. For some reason you think that admins will conspire to support each other; see my earlier comment regarding factionalism for why this just ain't so. I can think of very few prominent admins (the ones most likely to have someone call for their desysopping, naturally) who don't have enough other admins generally disapproving of them that a certification wouldn't happen. And again--I feel like I am repeating this over and over and over and over because people just aren't getting it--the certification idea is a rubber stamp, ensuring the person initiating the desysop has followed the requirements. It's no different than clerking at SPI or ArbCom or making sure someone has done their homework before taking an article through FAC. But given the potentially disastrous consequences if the system were to be gamed by e.g. sockpuppets certifying each others' desysop requests, we have to limit somehow who can make the final check and say "Yup, you have tried another form of DR, you have shown there may be a good case here." And that is all they are saying; Aervanath has pointed out that s/he could easily see situations in which s/he would both certify a request as valid but vote against desysopping. Please reread the proposal carefully; I have proposed as few limits as possible (and believe me, I had started with significantly stricter ones) while keeping in mind that the only way to get this necessary process approved by the community is to show that it cannot be trivially gamed. Please don't let perfect be the enemy of good, here; politics (which, sadly, is what this will all boil down to) is the art of the possible, and the sort of sweeping changes to the entire project structure that you seem to be wanting cannot be encompassed within the extremely narrow scope of this proposal. → ROUX ₪ 00:26, 21 July 2009 (UTC)

Discuss "Who may comment on an XFA" (v 2)

• I think 200 edits is too low a threshold, and suggest 1,000 - see reasons at Discuss "Who may initiate an XFA" (v 2). --Philcha (talk) 08:47, 14 July 2009 (UTC)
• I think present for 2 months is too low, and prefer the original 6 - for much the same reasons as I'd prefer 1,000 edits rather than 200. --Philcha (talk) 08:47, 14 July 2009 (UTC)
• "Never blocked by the admin who is the subject of the XFA, except where that block was later overturned by the community or by another admin (shortening a block, or granting unblock from an indefinite block, is not 'overturning') found to be improper or excessive" is an improvement, but still fails to provide for blocks too short to be worth contesting and blocks that are ancient history. --Philcha (talk) 08:47, 14 July 2009 (UTC)

See above for my reasoning on editcount and tenure. Perhaps amend this wording to 'Not' for 'never', and '...subject of the XFA in the past 12 months...' → ROUX ₪ 09:04, 14 July 2009 (UTC)

That would deal very well with blocks that are ancient history. What about blocks that that were unjustified but too short to be worth contesting? --Philcha (talk) 10:00, 14 July 2009 (UTC)

In my experience, any time a user feels a block is unjustified they contest them. Well, if they notice them.. I have a terribly punitive block in my log that I didn't bother contesting, as I wasn't even near a computer the entire time it was in effect. I think you are getting a bit too bogged down in what-ifs and maybes; the saying 'edge cases make poor law' is a sound one for a reason. Let IAR and other forms of DR handle the few that might crop up. If it becomes a persistent problem, the process can be changed. → ROUX ₪ 10:03, 14 July 2009 (UTC)

(ec) :::::Your "wasn't even near a computer the entire time it was in effect" was one of the types of case I was thinking of - not such an "edge case" after all. If I were blocked and were off WP for other reasons at the time, but thought the block was unjustified, , I'd probably contest it anyway in order to: clear my record, in case something serious came up later; make the blocking admin think more carefully, as much for others' sake as for my own. However I know people who would not contest such blocks. Quite apart from the unfairness of excluding such people from commenting: they can get their licks in by canvassing, which I'd hate to encourage; a heated XFA might cause other dispute mechanisms such as ANI to be clogged by requests for retrospective unblocks in order to enable people to comment. I think it's better that the rules on eligibility to comment at XFA should ignore blocks not exceeding 2 days. --Philcha (talk) 11:30, 14 July 2009 (UTC)

• I think specific criteria are unnecessary, see my previous reply.—S Marshall ^Talk/_Cont 11:22, 14 July 2009 (UTC)
• I prefer 2000 edits and 6 months of editing to avoid sleepers and meatpuppets.--Lenticel ^(talk) 05:00, 20 July 2009 (UTC)

Grounds for initiating an XFA (v 2)

Support "Grounds for initiating an XFA" (v 2)

1. Goes a long way to eliminating malicious and frivolous "I don't like X" grounds. --Philcha (talk) 08:47, 14 July 2009 (UTC)
2. Supported on the very clear understanding that WP:IAR applies to these grounds — in other words, that where there's a clear consensus to suspend the exact wording of the grounds for one particular case and open a Demand For Desysopping regardless of what they say, then that can be done. Such a consensus would likely come from WP:AN or a WP:RFC/U, but could conceivably come from other project areas such as WP:DRV.—S Marshall ^Talk/_Cont 11:32, 14 July 2009 (UTC)

   I think that (the application of IAR) would come under the remit of whoever is certifying. → ROUX ₪ 07:04, 20 July 2009 (UTC)

3. Per S Marshall.--Aervanath (talk) 06:36, 20 July 2009 (UTC)

Oppose "Grounds for initiating an XFA" (v 2)

1. I agree that desysopping is currently too hard. However, the current wording is untenable, because it puts the bar far too low: "[...] must have abused one or more of the following admin abilities in order for an RFDA to be filed" sounds like a "one strike you're out" policy, which is far too strict and bound to create more drama. I agree that currently the pendulum is too far on one side, but let's not go from one extreme to the other. Instead, we should clarify that there has to be a pattern of abuse. — Sebastian 23:56, 14 August 2009 (UTC) - off topic text removed — Sebastian 02:05, 15 August 2009 (UTC)

   I understand where you're coming from, but defining what a pattern is, well, it's an exercise in impossibility. And I'm relatively sure that a single isolated incident will not garner support for a desysopping; if one occurs as the result of a single incident, my feeling (given the other safeguards in place) is that it will indicate the community's unwillingness to tolerate any admin abuse. And of course the (former) admin in question may resubmit an RfA at any time. → ROUX ₪ 00:06, 15 August 2009 (UTC)

   Hmm, what gives you this certainty? There sometimes is support for the weirdest claims here. But we're not talking about support; this is about the "Grounds for initiating an XFA". Even if a frivolous XFA does not gain support, it is still a waste of time for everyone involved. (And the renewed RfA would about double that!) I've seen enough frivolous calls for desysopping even without an official procedure, and once we open that door, that number is likely to increase. I agree that the term "pattern" is currently not defined, and it may be not be easy to define. But is what we currently have any better? We have a long list of all possible offenses, of which few, if any, are defined. For instance: When is "deleting" an offense? As it stands currently, anyone whose article got deleted has an official ground to initiate an XFA. Compared to that long (and still possibly incomplete) list of undefined offenses, it may actually be easier to define "pattern". And maybe we don't even need to define it exactly. I think we can agree that a pattern has to be at least a small number n of occurrences of the same offense; maybe n = 2 or 3. Filing an XFA based on a pattern requires more research by the claimant C: C needs to find n similar offenses and make the case that they are indeed similar. That would discourage frivolous single-offense complaints. There may be better ideas. — Sebastian 01:53, 15 August 2009 (UTC)    (I stopped watching this page. If you would like to continue the talk, please do so here and ping me.)

Remove from "Grounds for initiating an XFA" (v 2)

• "RFDA may only be used after other avenues of dispute resolution have been pursued" may not be feasible if the grounds is a block. The complaint cannot use other avenues of dispute resolution while blocked, i.e. until after the damage is done. --Philcha (talk) 18:22, 10 February 2010 (UTC)
  • Nor indeed would they be able to use this avenue while blocked, for the same reasons. Turtles all the way down. If necessary. other editors can and do proxy for blocked editors in DR venues. → ROUX ₪ 18:26, 10 February 2010 (UTC)

Discuss "Grounds for initiating an XFA" (v 2)

• Why is "rollback" one of the grounds to initiate and RFDA? Above it says that "An RFDA may be initiated [o]nly for actions that involve use of an administrative function (something an auto-confirmed user or rollbacker couldn't do)". Also, I don't really think that we need a checklist for which admin actions the accused has abused. This would have to be explained by the initiator and backed up by diffs anyway. Jafeluv (talk) 06:18, 20 July 2009 (UTC)
  • Probably because I'm an idiot. Removed. I think the checklist is key because in order for this to get approval from the community, it needs to have a strictly-defined scope. → ROUX ₪ 06:21, 20 July 2009 (UTC)
    • Actually, this is something which should be included. Rollback is still part of the admin package. I would actually modify the sentence to exclude the "and rollbacker" phrasing: any admin can take away the rollback right from a user, but you can't take it away from an admin except by desysop.--Aervanath (talk) 06:38, 20 July 2009 (UTC)
      • Hmm, I hadn't considered that. Probably makes more sense. → ROUX ₪ 06:43, 20 July 2009 (UTC)
      • Agreed, serious rollback abuse should be a reason to consider desysopping. Jafeluv (talk) 08:58, 21 July 2009 (UTC)
• How about replacing the checkboxes with a simple list? Something like: "An admin must have abused one or more of the following admin abilities in order for an RFDA to be filed, including..." (followed by the list of admin powers as before.) Checkboxes should be used for yes or no -questions, not questions that need to be backed up by explanations and diffs. Also, the list of admin powers could be numbered so that people can easily refer to them (like we currently have for speedy deletion criteria). For example, "I think User:BadAdmin should be desysopped for violating #3 (diff), #6 (diff) and three times #12 (diff, diff, diff)." Or something like that. Jafeluv (talk) 11:58, 21 July 2009 (UTC)
  • I think having a checkboxy sort of thing helps ensure that anyone filing one of these must show specifics. I want to avoid even the possibility of "Waaaaah Evil Admin blocked me waaaaah," as the possibility of that will torpedo this. → ROUX ₪ 20:48, 21 July 2009 (UTC)
• There needs to be some kind of statute of limitations. In other words, if a wrongful block was made a year ago and nobody cared about impeachment then, it should not be allowed to be the sole justification for removal today. It could certainly be evidence of a pattern of conduct or a reason that an individual editor would use to determine their !vote, but unless there is RECENT bad conduct, the XFA should be considered out of order. --B (talk) 18:06, 21 July 2009 (UTC)
  • That makes a lot of sense. Three months? Six at the outside, I think. → ROUX ₪ 20:48, 21 July 2009 (UTC)

Procedure (v 2)

Support Procedure (v 2)

1. Supported on the understanding that where there's a clear consensus to proceed with the desysopping, or indeed a clear consensus that it was an abusive case, WP:IAR can overcome the theoretical stages laid down here. We don't need to get bogged down in this level of detail. What we need to do is trust the crats to judge right from wrong.—S Marshall ^Talk/_Cont 11:36, 14 July 2009 (UTC)

   I submit that if there's that much of an egregious/emergency case, ArbCom and its emergency desysop magic wand are a better route to take. → ROUX ₪ 07:03, 20 July 2009 (UTC)

Oppose Procedure (v 2)

• Needs significant amandment before I can support. At present it gives those in power the right to sweep issues under the carpet. There must be a means of preventing this (a "citizen's initiative" process), although I accept the need for moderately demanding conditions as this should not happen often. --Philcha (talk) 08:47, 14 July 2009 (UTC)
  • I think that is extremely unlikely to work. Having no sort of certification process invites widespread abuse, having a certification process open to the community at large invites canvassing and/or socking and/or votestacking. As we have no community standards in place that can effectively deal with socking except as a purely barn-door-after-the-horse-is-gone procedure, there needs to be an element of appealing to those users in whom the community has already vested their trust. I especially think an ArbCom member or clerk is unlikely to sweep cases under the rug, because if they do, ArbCom is where it ends up next. My sense from ArbCom members (posts onwiki, not private conversation) is that they would prefer to have to handle as few cases as possible in order to properly devote time and attention to the really important cases with sitewide implications. The only way to get people on board with this is to ensure as much as possible that it can be neither gamed nor used frivolously or maliciously. Requiring certification from ArbCom/Crat/Admin/MedCom members is a small leap of faith to take in order to prevent the abuse that could otherwise occur. Think of how many postings there are on AN/I every week with someone screaming admin abuse and calling for desysopping. Now think what would happen if those people could start a process for desysopping every time. No, a filter is necessary, and for all the reasons above it cannot be wide open to the community. And we can't be defining 'trusted' groups within the community that have not already been defined as such via community approval processes. Note that I didn't include CU or OS people, as they are not currently selected by the community. I did include ArbCom clerks, as they are chosen by ArbCom and famously display their neutrality on a daily basis. → ROUX ₪ 09:13, 14 July 2009 (UTC)
    • I certainly would not want a "citizen's initiative" process to be come a tool for gaming or for frivolous or malicious de-sysop proposals, and would not it to be too easy to initiate. However allowing officials to have complete control of the agenda is always a mistake. Even if they use this power impartially, it is likely that they will always find higher priorities than dealing with unjust decisions by admins. Given ArbCom's workload and slow response times, giving ArbCom the keys to the door wil cause only frustration and anger. OTOH if admins realise they can be de-sysopped fairly quickly they will think twice - and in some cases they ought to have thought much more carefully.
    • The recent controversy over the proposed Advisory Council has cast doubt on ArbCom's integrity and / or judgement. I don't think giving ArbCom a veto on XFAs will be accepted. --Philcha (talk) 10:24, 14 July 2009 (UTC)
      • The recent controversy is almost entirely manufactured, is based on people not actually paying attention to what is being said, and really has nothing to do with this. Please note that the process states that any two people from those four or five groups can certify. You are being unnecessarily paranoid, to be blunt. Unless you have a suggestion for a non-abusable way for this to be handled entirely by non-admins? → ROUX ₪ 10:29, 14 July 2009 (UTC)
        • I don't think the recent controversy is manufactured. I think ArbCom's motives were honorable, but the lack of public discussion of the Advisory Council's functions and composition before trying to set it up was terrible public relations, giving the impression of rule by fiat and of creating an officially-supported cabal. Like it or not, those in power have to be seen to be pure, and ArbCom collectively forgot this.
        • I accept that you and I have different attitudes to officials of any kind. Mine is based on Lord Acton's famous dictum, "Power tends to corrupt, and absolute power corrupts absolutely. ... There is no worse heresy than the fact that the office sanctifies the holder of it." Means of opposing all official decisions must exist. We can tweak the difficulty of the means of opposing according to whether the community thinks frivolous/ malicious de-sysop proposals are a greater problem than denials / delays in endorsements. We can even go with your endorsement procedure alone for a trial period and then review whether a means of by-passing the endorsement process is required. But failure to mention even this modest possibility for the community to exercise ultimate control will lead to trouble - without any help from me. --Philcha (talk) 11:09, 14 July 2009 (UTC)
          • There are roughly 1700 users who can certify an XFA. Only two of them need to agree with it, and nobody has a veto. Are you saying that all 1700 are so corrupt that they'll always reject XFAs? --Carnildo (talk) 21:52, 14 July 2009 (UTC)
            • What % will watch the "requests to certify an XFA" page? --Philcha (talk) 23:22, 14 July 2009 (UTC)
              • Given the general moths-to-a-flame propensity of Wikipedians, I would wager a lot. Notifications could also be posted at any one of AN, ANI, BN, RFA, Arbcom noticeboard... Also I note that you didn't answer Carnildo's question, which is drectly relevant here. I understand your concern about power and corruption, but you have not actually addressed anything about why certification of already-trusted editors is a good thing, you merely keep stating that power corrupts. Please respond to the points put to you, it would make this all much easier and prevent us from spiraling into the usual bickering over comma placement that torpedoes so many proposals on Wikipedia. Officials do not have 'complete control' of the agenda here. The community (including officials) has control; it just takes two officials to say 'yep, there's a case here.' they may disagree about the likelihood of success, but the only thing they are required to certify is that the person filing the complaint has ticked off the required boxes. That's it. Nothing more. Limiting it to already-trusted users ensures minimal abuse of the system, prevents canvassing, and ensures there's one final check on a user who might possibly be on a bit of a warpath with regards to a specific admin. Further, news of this sort of thing does get around, and it doesn't really matter if it takes two hours or two weeks to get something like this certified; in truly emergency situations ArbCom would step in anyway. → ROUX ₪ 23:40, 14 July 2009 (UTC)
              • Oh, and: if admins start sweeping under the carpet, the issue would go to ArbCom anyway, which means a lot of eyes and a lot of scrutiny. That behaviour wouldn't last very long. → ROUX ₪ 23:43, 14 July 2009 (UTC)
                • Given the response-time of ArbCom, that's not as reassuring as it should be. Note my post of 11:09, 14 July 2009, which implies that it may be enough if the process reserves the right for the community to impose a bypass procedure if endorsements are refused unjustifiably or too slow. There's an old chess maxim "The threat is stronger than the execution", and I'd be happy to rely on that initially - provided there is a real "threat". --Philcha (talk) 05:01, 15 July 2009 (UTC)
                  • I think, again, edge cases make bad law. Hypothetically, if there is a reasonable cause for desysopping and it is declined, not only would the case then move to Arbcom, but ArbCom's gaze would also fall squarely on those who declined to certify a good-faith case. With what results, no clue of course, but abusive patterns would be noticed. Arbcom is always last resort, and I think that we need to keep bureaucracy in this to a minimum. Nutshell: the community bypass procedure is to go to ArbCom. → ROUX ₪ 05:10, 15 July 2009 (UTC)
                    • I have 2 problems w your last post:
                      • Re "ArbCom's gaze would also fall squarely on those who declined to certify a good-faith case", as talk pointed out, "There are roughly 1700 users who can certify an XFA" (21:52, 14 July 2009), so ArbCom can't hold them all accountable.
                      • There was so much discontent with the performance of the previous ArbCom that the recent election was almost a clear-out. An inadequate performance by ArbCom is not an "edge case". To re-quote Acton, "There is no worse heresy than the fact that the office sanctifies the holder of it". --Philcha (talk) 05:50, 15 July 2009 (UTC)

(out) So you're proposing the following hypothetical scenario: JaneTheAdmin does something wrong. Some form(s) of dispute resolution is/are tried. Community discussion at AN or AN/I occurs. JoeTheEditor then files an XFA regarding JaneTheAdmin. 1700 admins, crats, arbs, medcom members, and arbcom clerks refuse to certify the XFA. JoeTheEditor starts an ArbCom case, which ArbCom refuses to hear. After all of that, you think the community somehow should be able to override all of these decisions? That would require a policy change of such sweeping scope that it would daunt even me to consider it. I struggle to believe that is what you are suggesting, but try as I might I can find no other interpretation for it. → ROUX ₪ 06:33, 15 July 2009 (UTC)

I'm beginning to think we need to outline the whole of the procedure for handling incorrect actions by admins, as we obviously seem to have different views of where the problems should be controlled. You say "Some form(s) of dispute resolution is/are tried" without suggestng which - it might help if you could be as specific as in your XFA proposal(s). In the meantime, to get the ball rolling, ...

Let's suppose JaneTheAdmin has something wrong, and it's not a normal judgement call but seriously out of line. In that situation justice must be public, fair and quick, otherwise wounds fester. If JaneTheAdmin is not in the habit of making such errors, an apology and undo will clear up the situation, if done promptly; if not prompt, it will feed the common feeling that admins get more lenient treatment than non-admins - so the first response must be quick, i.e. within hours.

OTOH if JaneTheAdmin has mades several such errors, she should be temporarily de-sysopped as soon as it's clear that there's a case to answer, like a cop being suspended if involved in disciplinary case. The question is who decides whether to "suspend" JaneTheAdmin, how quickly and how is it announced.

NewYordBrad says ArbCom's workload is heavy. That implies that many cases will wait some time before getting any attention; and ArbCom's processes are not quick. --Philcha (talk) 07:22, 15 July 2009 (UTC)

The policy as proposed includes a requirement to pursue other forms of DR first. Things that are seriously out of line are currently handled by ArbCom and/or fast contacts with stewards (I can think of one recent case, last few months, but cannot remember who it was. Point being, stewards will undertake emergency desysoppings where there is a clear and present problem in the absence of ArbCom being immediately available. I suspect crats or similar need to be involved to make it happen). Further point being, emergencies can already be handled. And the idea of making this sort of process immediate/quick is a really bad one. Actions such as removing someone's userrights need to be undertaken soberly after the initial emotional reaction has calmed down somewhat, otherwise the inevitable Wikipedian tendency to pile on with the torches and pitchforks will take over and derail the process. Temporary desysoppings--again, unless there is a clear and pressing security issue e.g. compromised account--are a bad idea. And this sort of process needs to not be that fast, because whether or not it was 100% justified, someone who has been desysopped has a pretty significant albatross around their neck should they attempt to regain the privileges. Now. Can you please directly address the point I made above? It seems that you want the complainant to somehow be able to override DR, community discussions, and an ArbCom case if they are declined certification to initiate an XFA and then move on to ArbCom. Is this or is this not what you are saying? → ROUX ₪ 07:36, 15 July 2009 (UTC)

I've seen some pretty bad blocks in the last couple of months. One was corrected quickly and the blocking admin apologised to the blocked user - as in the first "JaneTheAdmin" case above. In the other the blocking admin made a comment about blocking while not in a fit condition to exercise good judgement - whether that was true or a flippant comment, IMO that deserved instant de-sysop. Really bad cases need to be handled really quickly, both in terms of putting the situation right and in terms of preventing the guilty admin from doing further harm. --Philcha (talk) 08:03, 15 July 2009 (UTC)

That is fundamentally a bad idea, and I guarantee to you that this proposal will fail if you insist that it be in. Just as granting sysop is not instantaneous, neither can removal be, except in clear cases of gross misconduct (e.g. batch deleting Category:Articles) or security issues. Absolutely no good purpose would have been served by allowing the torches and pitchforks to be handed out and a desysop rammed through. These things--both granting and removal--of userrights require calm and sober reflection. You may feel he should be desysopped, that's fine; please take it to ArbCom as this process hasn't been/may well not be implemented. But this will fail completely if you insist on having that sort of thing in here. And you still haven't answered my question. → ROUX ₪ 08:27, 15 July 2009 (UTC)

You're right, I still haven't answered your question, which I take to be "It seems that you want the complainant to somehow be able to override DR, community discussions, and an ArbCom case if they are declined certification to initiate an XFA and then move on to ArbCom. Is this or is this not what you are saying?"

Let's take the case of a block, which in this hypothetical case is unjustified. Where does Dispute Resolution take place and how does the victim initiate it while blocked? How would the victim take part in community discussions while blocked? Waiting until the end of the block is not an option if the block is unjustified - the target duration of an unjustified block is as close to zero as physically possible. The DR-discussion-ArbCom route is too slow to handle bad blocks. There needs to be a court of appeal that is not affected by blocks - and that court of appeal should not be dominated by admins. Are you proposing also to to set up such a court of appeal? If you are, then in cases where a block is found to be unjustified no further certification is needed. If you are not, then XFA / RFDA / whatever is the only court of appeal discussed so far, and its doors should not be controlled by admins. --Philcha (talk) 15:06, 20 July 2009 (UTC)

Bad blocks are generally handled extremely quickly (and with much drama) at AN/I. And again, engaging in a request for desysop while still pissed off is a really bad idea. In general we aim to have a sort of placid feeling when making decisions on Wikipedia, and there is a community norm that making changes while in high dudgeon is inadvisable at best. You want something that will strike while the iron is hot, the pitchforks are sharpened, and the torches are lit. That is never a good time to do anything--for example, that is one reason why RfCs run for 30 days. It is always a better idea to wait for the next morning, see the evidence of last night's mob outside, and think "There but for the grace of God..." There is a court of appeal that is not affected by blocks, two in fact: unblock-l and arbcom-l email lists. And yes, of course those are run by admins--the people empowered by the community to both enact and remove blocks. What you appear to be wanting is an enormous sea-change in how Wikipedia functions on a day to day basis. That is far outside the remit of this proposal. → ROUX ₪ 00:14, 21 July 2009 (UTC)

• Proposed procedure is, basically, a popular vote (love it or hate it, if the crats count the numbers it's a vote). Trial by popular vote is also called lynching. Instead of a single crat I'd rather see a qualified panel making the final pronouncement along the Arbcom pattern (it does not need to be Arbcom itself, or to be regulated like Arbcom; the point is that the panel will have private communications channel for discussion in case of no-clear-consensus). NVO (talk) 08:50, 20 July 2009 (UTC)
  • I fail to see how non-transparency could be a good thing in this sort of process. → ROUX ₪ 09:00, 20 July 2009 (UTC)
    • Do you see it in Arbcom proceedings? In real-life courtroom? Unless the case in black-and-white simple (and needs no panel at all), non-tranparency ensures independence of decision-making process. NVO (talk) 09:28, 20 July 2009 (UTC)
      • Again, this is not a trial. ArbCom is much more explicitly a quasi-judicial body and thus adheres closer to Western notions of jurisprudence. This is very much intended to not be like that. → ROUX ₪ 09:51, 20 July 2009 (UTC)

Discuss Procedure (v 2)

• "The RFDA must be certified by any two members of any of the following groups: Arbcom, Arbcom clerks, MedCom, Bureaucrats, or any two administrators". Redundant. Jafeluv (talk) 06:26, 20 July 2009 (UTC)
  • Effectively true, as there's only one member of MedCom who isn't an admin; all the current ArbCom members, ArbCom clerks, and Bureaucrats are admins. While it's not a technical requirement, anybody who can get enough support to be on ArbCom probably has already gotten enough support to pass RFA. ArbCom clerks generally have to be admins in order to perform any blocks that are necessary as a result of an ArbCom case, and I'm not aware of any Bureaucrat that has even been nominated without being a sysop first. However, there are no written requirements that any of those groups be admins, so I support keeping them on the list, just in case. How about this: "The RFDA must be certified by any two administrators. Non-admin members of any of the following groups can also certify: Arbitration committe members, Arbitration clerks, Mediation Committee members, and Bureaucrats."--Aervanath (talk) 06:59, 20 July 2009 (UTC)
    • That makes sense. Could you add it? → ROUX ₪ 07:01, 20 July 2009 (UTC)
    • Maybe make the part about non-admin members a footnote? It's a pretty special case anyway. Jafeluv (talk) 11:00, 20 July 2009 (UTC)
      •  Done → ROUX ₪ 11:05, 20 July 2009 (UTC)
• I have problems with the requirement that the request be certified by two members of the sysop/b'crat/etc group. I understand that the purpose is to prevent frivolous requests from being taken seriously, but this unecessarily excludes editors in good standing. There has been a (perhaps ill-informed) notion that the administrators tend to stick up for each other, and this requirement could be seen as a weakness that allows sysops as a whole to "protect" their own at the expense of the will of the community. Administrators should not be gatekeepers of the very process that can remove their bit. Shereth 14:16, 20 July 2009 (UTC)
  • If I have learned anything about Wikipedia, it is that users (including admins) will jump on their "wiki-friends'" throats to uphold policy. What I mean is, most of the admins are zealous enough to axe one of their own to remain within policy. Also, since this will sort of be an administrative "power", if this proposal passes, I should think that sysops abusing their "certification power" would be up for de-adminship also (if the trend continued), so nobody would want to jeopardize their position. Malinaccier (talk) 16:47, 20 July 2009 (UTC)
    • It is kind of a hard power to "abuse" since one has to prove that an administrator is willfully refusing to certify a complaint against one of their cohorts. While I agree with you that it's not likely to be an actual problem, it is more a matter of perception that such a requirement gives the admin group a sort of pre-emptive veto against filings. I do not see why other established editors cannot be included among those allowed to certify a complaint. Shereth 16:55, 20 July 2009 (UTC)
      • As I said somewhere above, then it becomes a 'turtles all the way down' problem. Who is an established editor? How do we define that? Suddenly we are faced with a whole new layer of argument which could torpedo this proposal. And the only way in which admins are gatekeepers is in ensuring that the user has dotted the i's, crossed the t's, and attempted in good faith some form of dispute resolution (or there has been community consensus that a specific look at desysopping is warranted). → ROUX ₪ 00:05, 21 July 2009 (UTC)

Outcome (v 2)

Support Outcome (v 2)

1. Yes, especially "An XFA that closes with a desysopping confers no prejudice to the subject standing for a full RFA at any time, at their own judgement and discretion" --Philcha (talk) 08:47, 14 July 2009 (UTC)
2. Support.--Aervanath (talk) 07:00, 20 July 2009 (UTC)
3. Looks good. Jafeluv (talk) 10:53, 20 July 2009 (UTC)
4. Malinaccier (talk) 16:34, 20 July 2009 (UTC)

Discuss Outcome (v 2)

• As before, I think it would be preferable to get desysop rights to the bureaucrats, but if you think that's beyond the scope of this proposal, I'm okay with that. Jafeluv (talk) 06:30, 20 July 2009 (UTC)
  • I think that would be a great next step, but at this time including that would be a poison pill (unfortunately). → ROUX ₪ 06:44, 20 July 2009 (UTC)
    • Roux is correct. There have been proposals for this in the past, but all have failed with large opposition.--Aervanath (talk) 07:01, 20 July 2009 (UTC)
• It needs to be made clearer what the effect of the discussion would be. Does a user have to gain a consensus to remain a sysop, or is a consensus required to remove the flag? (I would strongly oppose the former.) Stifle (talk) 08:13, 20 July 2009 (UTC)
  • I meant for it to be consensus required to remove (as consensus is required to promote in RFA). → ROUX ₪ 08:37, 20 July 2009 (UTC)
    • That is my understanding, as well.--Aervanath (talk) 18:03, 20 July 2009 (UTC)
      • And is now very much clarified here. → ROUX ₪ 00:55, 21 July 2009 (UTC)

Moving to projectspace

Sometime in the next few days I'd like to move this out into projectspace for wider scrutiny and so on. But... WP:RFDA is taken as a shortcut (redirs to Wikipedia:Requests for de-adminship). Anyone have ideas of where to move this to? → ROUX ₪ 09:56, 20 July 2009 (UTC)

How about Wikipedia:Requests for desysopping? Whichever name is decided, WP:ROUX must definitely be a shortcut. Jafeluv (talk) 11:03, 20 July 2009 (UTC)

HAHAHAHAAHAHAAHHAHAA. Oh man, that actually made me giggle. Perhaps Requests Of Unusual eX-adminship? → ROUX ₪ 11:07, 20 July 2009 (UTC)

I think you could reasonably move the current WP:RFDA somewhere else, because it hasn't much to do with the title. Stifle (talk) 14:09, 20 July 2009 (UTC)

I'd put it in a subpage of WP:ADMIN. Malinaccier (talk) 16:29, 20 July 2009 (UTC)

That's a workable idea. My idea was: don't move the actual page. Archive this page and talk page, and copy-paste it to the top of the RFDA page, leaving the current content at the bottom of that page (maybe put a hatnote at the top leading to the bottom section). If this proposal is rejected, we just revert to the prior state, then add a link to this at the bottom with all the others. If this proposal is accepted, then we would probably want to move the original content to a subpage with a see also link, e.g. WP:RFDA/Prior RFDA proposals.--Aervanath (talk) 18:10, 20 July 2009 (UTC)

Or maybe transclude it instead of copy-pasting, so that changes don't have to be made in two places while the proposal is up. Jafeluv (talk) 11:09, 21 July 2009 (UTC)