Wikiversity talk:Recusal

Conflicts with proposal
The Privacy policy states that users can remove use of their own personal information. The proposal for a Recusal policy would contradict this, causing a major conflict.

Also, Wikipedia principles of recusal only require an admin involved in the content editing/crafting of a topic to be recused from using admin ops regarding it - therefore, situations such as an admin deleting something then blocking a person for recreating it, etc, would not apply per that standard. What would apply would be actions like Jtneill's where he uses admin ops to make non-controversial actions to help his students put forth information.

The latter example would seem to negate any possible recusal policy, as custodians are, often, teachers who use the ops to help in their classroom activities.

Finally, in mentorship, mentees, by traditional, allow their mentor to reverse all actions, which makes recusal in those situations null and void. Otherwise, there could be no true mentor/mentee relationship. Ottava Rima (talk) 18:33, 22 November 2010 (UTC)
 * Removal of personal information could be considered an emergency. Ottava is not reading the proposed policy correctly. It allows the use of tools in an emergency, and sets out what a custodian should do if faced with an emergency. There is no "major conflict" as Ottava asserts, merely caution. Removing your own personal information by revision deletion obviously creates what the policy calls a "recusal requirement," but because allowing the information to stand could cause immediate and irreparable harm, this would easily fall into "emergency." Therefore the custodian could use the tools, but would be required by the proposed policy to report this at RCA. If the very report could be damaging, then a possible conflict would arise, but that is hard to sustain as a real problem. It would not have applied to any of the cases that have actually arisen, as far as I've seen. The attention called to the edit would be only of import as to custodians who can see deleted edits.
 * This particular issue might be avoided entirely by requiring Wikiversity custodians to use their real names, that's a long term solution that could be very appropriate for Wikiversity. That's not relevant here, it is a different kind of proposal. (It does not negate anonymous editing, but only anonymous assumption of administrative positions, it's already an established principle, and the only issue is how deeply the requirement goes.)


 * The discussion about deletion/blocking is not relevant. Such an action, as part of an original sequence, would not create a recusal requirement in itself. And even with a recusal requirement, a custodian can still act, the policy makes that clear.
 * An example would be this sequence:
 * User creates page with copyvio.
 * Custodian speedy deletes page as copyvio.
 * User gets angry and insults custodian.
 * A block at this point has the appearance of retaliation for protest, and there is therefore a recusal requirement. Generally, a custodian should not block an editor for insulting himself or herself. That's recusal failure, and is usually not an emergency. Another custodian might, but would normally warn first. So suppose this sequence happens:
 * User creates page with copyvio.
 * Custodian speedy deletes page as copyvio, warns user about copyvio.
 * User recreates page.
 * Custodian deletes page and blocks for copyvio.
 * There is no recusal requirement created, except maybe as to future actions with this particular user, it depends on whether or not a reasonable user, reading the history, might suspect bias. Generally, continuing the same action doesn't create a recusal requirement until intervening events have set up a possible appearance of bias.
 * User creates page with copyvio.
 * Custodian speedy deletes page as copyvio, warns user about copyvio.
 * User recreates page and claims that the custodian is biased.
 * Custodian deletes page and blocks for copyvio.
 * This enters the gray area where the policy really kicks in. There is an appearance of bias possible for the user and anyone who supports the user's point of view, perhaps. So the custodian should treat this as possible recusal failure, but there is a clear justification for considering the situation an emergency. The custodian should report the action to RCA. In this case that report would be very simple, nothing more than would already be proper for the block notice on the user talk page. The deletion would already allege copyvio, and the block reason would be simple: repeated creation of copyvio. The custodian should also rigorously refrain from actions likely to inflame the situation, and should apologize for the necessity. It's very quick to do this, unless the custodian gets truly involved and loses his or her own temper, which is what often happens. The custodian, to the extent possible, should encourage the user, and I'd expect a custodian often involved in such actions to create templates that do the job quickly, telling a user how to appeal, offering suggestions for avoiding problems, etc.


 * Generally, Ottava's objections seem to be to absurd interpretations and applications of the policy. I've never seen recusal failure alleged for someone doing something friendly or helpful, only when the action actually offended others and caused harm. If a friend asks a custodian to do something, such as provide a copy of a deleted page, it would only create an appearance of bias if the provision of the copy violated policy, as an example, or was known, in advance, to be offensive to another, in which case the request should probably be presented to RCA. It is difficult to imagine that a marginal case would result in serious consequences to the custodian, unless "marginal cases" became frequent, reversed upon review or resulting in some admonishment, and the custodian did not respond by becoming more careful.


 * I've seen this kind of response from administrators on Wikipedia, and that's the reason why Wikipedia's recusal policy is only primitively developed, admins have resisted it tooth and nail, to the extent of denying it, before ArbComm, so vigorously that I've claimed that ArbComm should suspend the admin privilege for those who deny the requirement. They haven't, and as a result, blatant recusal failure accumulated for years until the situation became so toxic that a massive RfAr resulted in a handful of reprimands, topic bans, and still no peace, continued conflict. We can head that off, right now, by clearly defining recusal failure while we are small. It's already a problem, as the current CR shows.


 * Good recusal policy *protects* administrators who follow it, from becoming needlessly involved in controversy. If an admin is neutral, and acts neutrally, people may protest, and if they do, the admin will, absent emergency, curtail activities in the area. An administrator who blocks may often be attacked as biased, it's routine. By recognizing recusal requirements, the admin will conduct himself or herself in a manner that is designed to avoid escalation of conflict through personal insistence on being right.


 * User is uncivil.
 * Administrator warns.
 * User rails against administrator.
 * Administrator does nothing but respond civilly, but also points out that the warning remains effective.
 * User attacks another user.
 * Administrator blocks, providing evidence, and reports conflict to RCA, a link to the user talk page would probably suffice. Custodian informs user of the report, makes sure that user is instructed how to use an unblock template, and drops the topic. And lets others decide.
 * Accusation of bias or recusal failure is never a reason for unblock, in itself. This, as well, is a common misunderstanding about recusal policy. It's feared that it will be used by trolls to make it impossible to block them. That is actually preposterous, once the policy is understood.


 * Mentorship is a special case, but recusal only applies to reversal of the mentee actions, which is generally understood to apply to other actions. Ottava has in mind a very special and very unusual incident; he'd called a user a "liar," I warned him, he effectively wrote, "who are you to warn me?" and he repeated the incivility, and I blocked him for two hours, to make the point. I was his mentee, so he technically was allowed by that agreement to unblock himself, but only on a finding that my action harmed wikiversity. However, he had other options, clearly, and there was no emergency. He did not report the incident, in spite of the obvious conflict, and he revision-deleted the block log. Recusal policy is for the protection of the community and for the protection of administrators against accusations of bias.


 * A sensible administrator would probably have waited out the two hours, and would have laughed about the event. This sequence, besides the earlier "liar" edits, is what showed me that there were serious problems with Ottava as to his sense of entitlement. But Ottava is not the point here, the policy is. The existence of a possible exception, even if that were acknowledged, does not change the fundamental issues. In any case, what the allowance to Ottava as being my mentor was was not a blanket permission to reverse any action of mine, but a permission to do so without wheel-warring, whereas if I reversed his actions as wheel-warring insistence, I'd have been violating the policy and might have been emergency desysopped just for that. If the action of reversal was itself improper, the permission from me would have no effect, since I have no right to give permission for improper actions. Unblocking himself, absent emergency, was an improper action, and I deliberately made the block very short to avoid a claim of emergency on his part. I also would not have blocked if the cause of block were not clear: repetition of an offense after warning. That's why Jtneil later approved the block as within legitimate custodian discretion.


 * But in no way would recusal violation, itself, have been the sole reason for a complaint about this. Custodians can make mistakes. Far more serious, in fact, was the failure to respond to warnings of incivility, and Ottava continued to repeat that, over the ensuing months. --Abd 20:10, 22 November 2010 (UTC)
 * You know, your accusations of Wikilawyering fall flat when you have been blocked on other projects for it and your responses are mostly long and off-topic. If you don't like it pointed out how your proposal is incompatible with others, find some way to fix the incompatibility instead of attacking others like the above. Ottava Rima (talk) 20:17, 22 November 2010 (UTC)
 * I responded in detail to Ottava's claims of "incompatibility." Is there any other objection? As to what may be seen as an attack above, please point it out, and I assume I will strike it. Even if it is "true." The purpose here is to discuss the policy, and examples are given only to clarify the real situations behind policy proposals and consideration, not to accuse. In particular, the sequence about mentorship was in response to the issue raised by Ottava, when there is only one known example in the entire history of Wikiversity that this could be about. --Abd 20:23, 22 November 2010 (UTC)
 * You made claims about "emergency" when the word is vague and subjective. Your claims to the rest also leave vague and subjective matters. Then, you don't actually remove many of the incompatibilities.
 * Also, it is odd how you accuse me of not discussing a policy proposal before proposing when you, yourself haven't done that by writing this proposal. Ottava Rima (talk) 20:32, 22 November 2010 (UTC)
 * This proposal was first made at the custodian page, with plenty of discussion. It was suggested to put it here as a proposed policy. Which we are now discussing. Ottava, please don't raise non-issues, it tempts response that takes us more off-topic. Had you proposed the inactive custodian policy in this manner, there would have been no problem. Instead, you proposed it with an immediate site notice. Yes, "emergency" is subjective. It's subject to discretion, and the community, if necessary, will judge whether there was abuse of discretion. This is really standard regulatory stuff, and it's only lack of real-world experience that makes it seem unfeasible to you. "Emergency," in practice, isn't actually vague. It means that damage not easily repairable is reasonably likely if there is delay. The only problem would be if a custodian routinely claims emergency when there is none. If the custodian nevertheless notifies RCA, whenever "emergency" is necessary as a claim to avoid censure for recusal failure, making a frivolous claim, there is no "emergency" about dealing with the improper claims, because the custodian has avoided harm from unexamined actions, and has disclosed the potential conflict of interest that is behind recusal failure.
 * I have blocked user SoAndSo, in spite of this user claiming that I'm prejudiced against him or her, because the user was posting copyright violations after warning, and it is my opinion [per precedent] that this is adequate as an emergency to allow disregard of recusal requirements. I do not believe that I'm actually biased, but to avoid the appearance of bias, per recusal policy, I am reporting this here, and suggest neutral review of the block, as well as neutral review of any unblock template the user puts up. I have also advised the user how to appeal the block, and consent to any other custodian unblocking if the custodian believes the block is unnecessary, it is not necessary to consult me. Necessary evidence is already placed on the user Talk page, so that both the user, any custodian reviewing, or any other person, may understand the purpose of the block. Thanks. --(Custodian)
 * A custodian may claim emergency under IAR. I don't advise doing it frequently, but the community, if there is objection, will, in later review, decide if the claim was justified. I would expect that only continued failure to respect recusal policy, perhaps by claiming emergency frivolously, would lead to sanctions against a sysop.
 * It's not difficult to understand. The purpose is to increase community confidence in how administrators function, by avoiding the appearance of bias. An additional purpose is to avoid actual bias, the "appearance of bias" is a hedge around the law, so to speak. Nobody is going to get creamed for wandering too close to an edge, just maybe some reprimand or warning not to keep repeating it. I.e., if a custodian does not anticipate an appearance of bias, nobody has alleged bias, and the custodian acts and in hindsight it's possible for someone to assert an appearance of bias, that's not really a problem unless any reasonable person would have recognized the problem. Like, on Wikipedia, an administrator blocked me for violating a ban he had declared, during an RfAr over whether or not he had the right to ban me. No reasonable person would have failed to see that there was an appearance of bias! And there was no emergency, the comment I made allegedly violating the ban was nothing more than a reference to the archive for the article, where a question was answered that had been asked. The edit wasn't offensive or uncivil, there was no copyright violation, outing, BLP violation, etc.
 * No policy substitutes for IAR. IAR is, in fact, the common-law principle of Public Policy. It is not necessary, then, for a policy to cover all contingencies, and the purpose of this proposed Recusal policy is to make explicit what recusal failure means, how to avoid it, and how to deal with emergencies, i.e., where a custodian believes that an action is important, and urgent, and delay could cause damage, more than a temporary appearance of possible bias. By following the suggestions in the article, which include rigorously avoiding incivility with a user, even if the user is hopping mad, any possible damage from an appearance of bias is minimized.
 * I could also give a really good example from my own block history on Wikipedia, where the blocking admin conducted herself perfectly. She was wrong about the block and the underlying situation, but by carefully stepping aside, she avoided any impression on my part that she was truly biased, and turned the situation into the reality: I had to convince the community that there was no harm in unblocking me, and if I'd started screaming "Unfair! Bias! Stupid Admin!" I'd have simply been shooting myself in the foot, since she'd recused. It depersonalized the issue. And I'd made no claim that she was biased before she blocked me, so she really had no special recusal requirements. It was a highly visible block, though, so there wasn't any need to take it to a noticeboard in any case, there was plenty of attention on my Talk page.
 * None of this should be construed as a criticism of a specific sysop. My own view is that when a policy is implemented or clarified, if a sysop has been violating it, an opportunity should exist for the sysop to understand the policy, and prior violations would not be an issue unless the sysop argues against the clarified recusal policy, coupled with an apparent intention to disregard the policy in the future. That might be a problem. Normally, though, there would be specific warning before anyone would be sanctioned for violating recusal policy, unless the violation was egregious. We do not expect custodians to be error-free. --Abd 00:50, 23 November 2010 (UTC)

Proposal was gutted, with some consequences
With, Darklama, with the edit summary, (simplify/condense this and try to address issues with the proposal), gutted it. This page was started because a detailed coverage of recusal policy, which Wikiversity badly needs, was too much for the page it was originally proposed on. The core of this recusal policy is the concept of emergency action.

As written with Darklama's changes, the policy will not be able to obtain consensus, and if it did, it would be, essentially, an insult to the custodial community, depriving them of the discretion they need to be able to deal with emergent situations, which, by definition, cannot be anticipated, and on-wiki discussion process takes way, way too long, and tends to be disruptive in itself; thus, in reality, custodians, where they seek consensus for an action, do so on IRC or privately, thus creating the impression of a cabal, making the situation worse, rather than better. Discussions to seek consensus may be private, initially, but if users act on this confined consensus, as if it were a community consensus, serious damage can be done, and this has been a relentless problem on Wikipedia. Many of our users have suffered from it.

Custodians will take actions as they see fit, without prior discussion. It's human nature, and it's a necessity. However, this is a wiki. Any action can be undone. A sensible recusal policy insures that any actions that might involve bias will receive a fair review. That is what the language of "emergency" is about. Recusal rules are designed to avoid the appearance of bias. Connected with them is the avoidance of wheel-warring. Wheel-warring is the custodial tool equivalent of revert warring, and the simplest definition means the re-assertion of a prior position by a custodian, after another has reversed it. It is not a simple reversal, and a lot of confusion has been caused by this. Reversing a custodian's action without discussion is possibly a problem, and requires discussion, but it is not necessary that the discussion precede the reversal, if there is fear of immediate harm.

The policy, as I'd written it, requires that the custodian, to bypass recusal requirements, declare an emergency. An emergency doesn't mean that the wiki is on fire. It means that delay will cause harm, in the view of the custodian, harm that might not be easily fixed. Incivility is, in fact, such an emergency, because when incivility is allowed to continue, it drives people way, it entrenches conflicts. It is not enough to prohibit incivility, dispute resolution process must also be provided, but that process cannot proceed in the presence of continued incivility.

We must face the issue of emergency action, not sweep it under the carpet by pretending we can discuss everything in advance. I reverted Darklama, but he has again edited extensively, removing the consideration of emergency, and directly contradicting it. This was not a mere simplification, it was a radical removal of the most important part of what I'd written.

During the recent flap, I referred to this proposed policy many times, I hadn't noticed Darklama's change. No wonder people didn't understand what I was doing! But what I was doing was more respectful of community consensus than routine practice at all the WMF wikis. I was consulting in advance and post-facto. I was explicitly allowing reversal of my action without prior discussion. And I was not supported, but was blamed. And for that reason, if this lack of support had continued without abatement, I'd have resigned, because I'm not willing to support a community as a custodian if the community is not willing to support me. It's mutual.

Just two weeks or so, before my action, SBJ had blocked a user for two days for incivility. That user had previously been blocked for three days. The user had been warned, and had ignored the warnings. And then the user started threatening members of the community with being blocked merely for participating in discussion! (No longer a sysop, the user seems to have an inflated view of his influence, and said, about my comments, that there were two sysops who had promised to block me, and they would do so as soon as they 'came on.') The user was filing massive process, and this user is a skilled wikilawyer, files process on any sysop who warns him, that will come out in a future review. I warned, and immediately a topic ban proposal was filed, from him. Immediate support appeared from users known to be close to him, it's clear that the support was arranged before the proposal was filed. This was a true emergency, and massive disruption was being created, accompanied by more incivility, directed at many users, and here I was, trusted with the tools by our most solid servant, Jtneill.

The three-day block was by an uninvolved custodian. But SBJ was very much involved. I advised him to consult, he might have done so anyway. The block was allowed to stand. If my later block violated the policy, so did his. But when I blocked, SBJ reversed it, rejecting it as excessive. Wikiversity doesn't have an "indef" block option, for some strange reason. I thought "infinite," which it does have, excessive. "Indef" blocks are used on Wikipedia, properly, to indicate a situation without some clear ending point, where discussion may be required. It's not a ban. On Wikipedia, as here, bans require community consensus, and, my opinion, they are a last resort, represent a kind of community failure, as well as individual failure. My one-year block, explicitly stated as indefinite, merely a recognition of the seriousness of the situation, not a ban or a ban proposal, was treated as if I'd bashed his brains out with a banhammer.

Probationary custodianship is designed to test the performance of a candidate. Well, folks, I was tested. How did I do? That's not a question for here, except as it relates to process. To determine if my actions were correct would require careful review. I was involved and therefore under a recusal requirement. I requested review. Was my action reviewed? No. I was not desysopped for that action. It was hardly mentioned. What was the cause of desysopping? It's clear: a certain number of people voted in a way that could be interpreted as lack of confidence, and the comments of people showing something different, made in the original site of the topic ban proposal, were not copied over. The true problem user was allowed to run the process, and then SBJ signed on to it, and why he did that, I truly don't know, except that he'd been opposed to my custodianship from the beginning, for unclear reasons.

Without a clear policy, we get nothing but politics and erratic action, unpredictable. This is causing ongoing damage. I know that my own participation in the business of this community, developing learning resources, has been seriously dented. I know that, without the protections of a stable community, my plans to bring in the whole alternative education community through connections with AERO are on hold. I know that highly competent (and trusted) users from other wikis are staying away from Wikiversity, which has a reputation as a collection of nutbags. If we don't fix this situation, it will not fix itself. --Abd 15:42, 3 February 2011 (UTC)


 * I agree without a clear policy, we get only politics and erratic actions. What purpose a policy servers and why a problem needs to be addressed should also be clearly explained. Although Jtneil may not have had recusal in mind, I believe what he wrote, so far describes what the purpose of Resucal is the most skillfully. Recent discussions have been scattered over many pages, but I think the issue of acting under the guise of an emergency required this proposal to be rewritten to address when people can act in order to avoid the appearance of acting rashly and risking consequences from the community. Also the need to recuse and the problems that arise from not doing so are not unique to Custodians, so I generalized it a bit more to apply to anyone. -- dark lama  18:19, 3 February 2011 (UTC)


 * What I don't think has been understood is that emergency action is sometimes needed. Any "official" with discretion will act, and I've seen you act, in advance of "consensus." It's only a problem if the action isn't reviewed, or is in itself seriously harmful, greater than the potential harm of not acting. If the policy does not recognize and guide what to do, if you, as a custodian, perceive an emergency, defined as a situation that, allowed to continue pending discussion and normal process, is perceived by you as more harmful than waiting, then you will be at risk if you act, and you may refrain from using your personal best judgment. And so a dangerous situtation can be allowed to continue. The possible damage of acting includes the possible harm of being perceived as rash. So review process must
 * Consider the action itself, was it legitimate?
 * Determine whether or not to undo the action, or to redo it if it was undone by another custodian.
 * Determine if the harm from an appearance of bias or rash action was greater than the harm from allowing the situation to continue without action.
 * Determine if the action of the custodian was outside community norms for how to handle perceived emergencies.


 * Some have thought that "emergency" should be defined. I've defined it subjectively, as to the perception of the custodian. There is also the perception of the community, upon review. If, on review, the declaration of emergency allowing the custodian to act, in spite of recusal requirements, is found to be "rash," not based on any actual harm being done, for example, or being unnecessary (as with the situation already being handled by another custodian), then the custodian would be reprimanded and advised. If it was grossly out of line, with serious risk of repetition, desysop might be immediate.


 * "Emergency," however, is probably not possible to subjectively define. Particular emergencies may be defined, and they are then not emergencies, as such, but rather anticipated situations, with specific guidance, but real police have discretion. If they act improperly, they can be sanctioned, even to the point of prosecution for egregious violations. But if they see someone about to harm someone else, they do not have to discuss what to do with the community. They may act. That they may act includes the possibility that they may err. In the end, they will be judged on whether or not their decisions were reasonable in context.


 * The determinations are separate. An action might be, in itself, justified, but if there is no emergency, then recusal rules should be followed in themselves. In the other direction, an action might be unjustified, in the end, but the custodian's action found reasonable, given what the custodian, exercising due caution, knew.


 * Review of what happened with me, what I actually did, the situation I was faced with, and how the community responded -- or did not respond! -- all could provide valuable information about the actual problems with recusal policy or lack of same. So I hope to set up that process, but after some time, so that the heat of the moment isn't affecting it. --Abd 00:31, 4 February 2011 (UTC)


 * I understood you think emergency action is sometimes needed and people need to be able to act on their own discretion. I think people need to be advised to exercise due caution, and need to be provided with practical guidance on how to be discrete. I think people should not be advised to do things which the community will likely reject, such as act first and ask questions later.


 * People can agree that quick action is a good idea even if they may be unable to agree what actions should be taken. People can agree that a situation is harming Wikiversity, even if they may be unable to agree what should be done about it. I try to do my best to look for common ground, even if I may not be under any obligation to recuse, before I act. I think the common ground to look for here before doing anything should be:


 * *People appear to agree that a situation is harming Wikiversity,
 * *People appear to agree that any or some action is better than allowing the current situation to continue unchanged, and
 * *People appear to agree that quick or immediate action should be taken.


 * I think whether "agree" is the same as or similar to "consensus" may be a distraction. I do not necessarily agree or disagree with the conclusion you drew that by agree I mean consensus. One small possible distinction that I guess I need to express is an agreement can happen without it being intended and without it being necessarily what people were focused on when trying to establish a consensus.


 * I hope you can recognize that what I suggest are things to review before acting, while what you have suggested are things to review after acting. I think without reviewing before acting, any review afterwards is more likely to be answered with "No, Undo, Greater harm was caused, and actions were outside community norms." I think by encouraging people to self-review the above things before acting there is no need to mention emergency at all. -- dark lama  15:14, 4 February 2011 (UTC)