Wikipedia talk:Wheel war

From Wikipedia, the free encyclopedia

Shortcut: WT:WHEEL

Archives
/Archive 1 /Archive 2 /Proposal 1 /Archive 3 /Archive 4 - final merger and creation of policy (Oct - Dec 2006)

1 "Wheel" war? Etymology?
2 Once is not a wheel war?
3 Why I removed the picture
4 Biographies of living persons
5 Request for comments: Wheel war
6 Is this wheel warring - question??
7 Flocking wheel wars
8 Wheel war definition
9 Proposed new section
10 Minor links editing re examples
11 Reverting admin actions without discussion is also bad

[edit] "Wheel" war? Etymology?

I'm sitting here trying to figure out why this is called a "wheel" war. The article doesn't appear to give any explanation as to the origin of the phrase. Would somebody care to explain what sort of wheel is being referred to? Matt Gies 18:29, 29 March 2007 (UTC)

See Wheel war. —Centrx→talk • 18:33, 29 March 2007 (UTC)

Alternatively, you should imagine a boat with two people simultaneously trying to turn the steering wheel in the other direction. >Radiant< 08:59, 30 March 2007 (UTC)

[edit] Once is not a wheel war?

In November Radiant inserted this text:

Undoing another admin's action once is not considered a wheel war, although it is still preferable to discuss it with the other admin first. Doing it twice, or repeating your admin action after someone has undone it, is wheel warring, unless a significant period of time has elapsed between the initial administrative action and its repetition.

I find this highly dubious; according to whom is this true? If one reversion of an editor's edit counts as a "revert", the surely one revert of an admin action starts a "wheel war". Was there consensus for this change to the page? Jayjg ^(talk) 16:33, 8 April 2007 (UTC)

Examples of undoing an admin action that are emphatically not wheel warring: unprotecting an old protected page, blocking a user who after being previously unblocked has turned to blatant vandalism, deleting a copyvio despite the page being restored on other grounds... The list goes on and on. Also note, the clear analogy is that wheel war corresponds to edit war not wheel war corresponds to revert. That said, this paragraph does conflict with the policy in defining "wheel war" solely in terms of doing and undoing an action without any reference to consensus or changed circumstances, i.e. any undoing of an admin action can be completely warranted if the circumstances under which the first action was taken change, or if there is otherwise agreement with the specific admin the reverse the action. See [1] & the bolded sentence of the first paragraph currently. —Centrx→talk • 19:54, 8 April 2007 (UTC)

I think the wording could use some tweaking. In general it'd be fair to say that undoing an admin action once is not necessarily a wheel war (and in fact happens reasonably often without much controversy, per Centrx's examples), and undoing it twice (without discussion and so forth) is necessarily a wheel war. Note also that in november this page was subject to heavy scrutiny and debate, which was eventually resolved, so it's fair to say that there were no unnoticed changes. >Radiant< 09:57, 10 April 2007 (UTC)

Several months on and there is still confusion about whether this is the case or not. Considering that there are several ArbCom cases based on wheel wars we really should get this decided. violet/riga (t) 15:33, 22 October 2007 (UTC)

[edit] Why I removed the picture

Wooyi asked me to explain this on the talk page, so I am doing so: I removed the picture he put on the article because, in context, it was stupid. Nandesuka 01:22, 27 May 2007 (UTC)

Thanks for explaining. I will not put it there again unless consensus compels me to do so. Regards. Wooyi^{Talk to me?} 01:26, 27 May 2007 (UTC)

I think it would be nice to have some picture here, just like we have on e.g. the snowball clause or WP:CREEP. >Radiant< 11:42, 29 May 2007 (UTC)

[edit] Biographies of living persons

I've inserted the following, which should be self explanatory:

Beware of restoring any deleted page whose deletion summary indicates that it was deleted under the biographies of living persons policy. The arbitration committee has ruled unanimously that such undeletions must never be performed unless these is an actual consensus.

Basically it's a big sign saying "you need to talk about this and don't do anything unless you've done that and there is consensus." --Tony Sidaway 03:11, 20 June 2007 (UTC)

The problem with this phrasing is that it implies that policy is created by the ArbCom, which to the best of my knowledge is not the case. >Radiant< 11:28, 20 June 2007 (UTC)

Also, this is not directly relevant to wheel warring, and the arbitration case rightfully contains no mention of wheel warring. The restoration of BLP-bad material without some consultation with the deleting admin or other due consideration is unwise and potentially a misuse of admin tools, but the misuse arises from being inconsiderate with sensitive material, not because of wheel warring. —Centrx→talk • 04:58, 25 June 2007 (UTC)

[edit] Request for comments: Wheel war

I propose the following change:

A wheel war starts when a privileged action is reverted without an attempt to form consensus. Admin actions may not be reverted without discussing with the original administrator, or obtaining a clear consensus on one of the administrators' noticeboards, except for emergency situations.

Let's discuss this. - Jehochman ^Talk 14:33, 4 November 2007 (UTC)

It's a bit vague - "emergency situations" is unclear. violet/riga (t) 14:35, 4 November 2007 (UTC)

We can list examples or clarify that. If an administrator goes on a rampage and deletes the mainpage and blocks Jimbo, that's an emergency, per common sense. There's no need to discuss reverting things like that. - Jehochman ^Talk 17:18, 4 November 2007 (UTC)

It may be a bit vague, yes, but I do think we need to move away from the current situation, which seems to be that whoever undoes an administrative action seems to have the upper hand, whether they had consensus for that action or not, and whether they consulted with the original admin or not. That's an unacceptable situation, I'm afraid. Some balance is needed here, we have seen several recent instances of very valid blocks getting overturned without consensus to do so. Rather than have "emergency de-sysoppings" let's get the policy crystal clear here, that one doesn't overturn blocks (or unblocks) just because one does not agree, and especially not in the face of a strong consensus, one seeks consensus first, except in grave emergencies. ++Lar: t/c 18:23, 4 November 2007 (UTC)

I would prefer the placing of the word "appropriate" before priviledged on the first line, so that the upper hand mentioned by Lar does not revert to the originator; poor or bad faith actions should be allowed to be undone without fear of sanction, providing the undoer can give valid reasons upon request, and without reference to the originating admin when it is not an "emergency". I realise that this looks as if it returns the system to as now, but does make both parties to a potential wheel war more or less equal in needing to provide the correct rationale for their actions. LessHeard vanU 19:47, 4 November 2007 (UTC)

Works for me. Except I think I'd spell privileged right :) Or ask for help because I can't keep that one straight. :) ++Lar: t/c 19:54, 4 November 2007 (UTC)

That's fine with me. Version B: - Jehochman ^Talk 20:16, 4 November 2007 (UTC)

A wheel war starts when an appropriate privileged action is reverted without an attempt to form consensus. Consensus can be obtained by discussing with the original administrator, or posting on one of the administrators' noticeboards.

Examples of inappropriate actions include: anything done in error, such as blocking a different user than the one intended; anything done in bad faith, such as deleting the main page; and anything forbidden by policy, such as blocking a user while engaged in a dispute with them or protecting a page while edit warring. Simple disagreements over the length of a block or need for page protection would not be eligible for this exclusion. - Jehochman ^Talk 20:16, 4 November 2007 (UTC)

That actually introduces another subjective term, and I think we may need to include your examples in the policy. violet/riga (t) 20:34, 4 November 2007 (UTC)

Good idea. - Jehochman ^Talk 20:40, 4 November 2007 (UTC)

I was aware that including "appropriate" may leave it open to interpretation/gaming, but considered that debate/drama over the application of that term would be less harmful to the project than the imposition of a bad action which might not be undone until consensus is reached. The inclusion of the three major violations of appropriate action (error/ bad faith/violation of policy) with examples is a good idea. LessHeard vanU 21:47, 4 November 2007 (UTC)

A wheel war starts when an appropriate privileged action is reverted without consensus. Consensus can be obtained by discussing with the original administrator, or posting on one of the administrators' noticeboards.

Inappropriate actions include: anything done in error, such as blocking a different user than the one intended; anything done in bad faith, such as deleting the main page; and anything forbidden by policy, such as blocking a user while engaged in a dispute with them or protecting a page while edit warring. Inappropriate actions do not include the use of sysop tools where reasonable administrators may disagree. Disagreements are to be resolved by discussion.

Revision C, based on the above comments. This may need copy editing and further discussion. - Jehochman ^Talk 22:00, 4 November 2007 (UTC)

Concur with wording (I might not be the best to copy edit - given my mistake previously...) LessHeard vanU 22:09, 4 November 2007 (UTC)

Suggest "Inappropriate actions include (but are not limited to):" as a minor change (and what is with the double indenting, does that give your words more weight? Maybe I should try it! LOL ) Common sense would suggest that "but not limited to" is not needed but you know how it goes. Do we have a consensus forming on the general drift here? ++Lar: t/c 23:15, 4 November 2007 (UTC)

No, this is all wrong. There are numerous situations in which it is perfectly appropriate to reverse a legitimate administrative action. See #Once is not a wheel war? above: "unprotecting an old protected page, blocking a user who after being previously unblocked has turned to blatant vandalism, deleting a copyvio despite the page being restored on other grounds". All of these are reversals of legitimate administrative actions, none of them require discussion, and requiring discussion and some ill-defined consensus would be complete waste of time. The above proposals also have specific internal problems, but regardless they are each wrong in toto. —Centrx→talk • 23:52, 4 November 2007 (UTC)

We're not talking about a new admin action that is triggered by a new event, such as an unblocked user vandalizing. We are talking about an admin who reblocks an unblocked user without any reason whatsoever except for they disagree with the unblock decision. There's a big different between a "revert" and a new decisions that happens to counteract an old decision. . - Jehochman ^Talk 00:26, 5 November 2007 (UTC)

That may not be what you are talking about, but it is prohibited by the wording of the above proposals, and it is not possible to word it to prevent the one while allowing the other. Anything can pass as an intervening "event", and for example in the case of unprotecting a protected page there is no significant intervening event, simply time.

It may seem clear that unprotecting a page that was protected six months ago is not wheel warring, but how about if it was only two days? What if I unprotect a page that was protected two hours ago, because I in my great wisdom consider the original protection to be an "error" or "forbidden by policy"? The loopholes you introduce in order to allow legitimate administrative reversals also admit the same sort of contentious warring the the policy is designed to forbid; each of the contentious parties thinks he is correct or is following policy, and that the other person is in error or is violating policy.

The Three-revert rule is so strictly firm that there be no further reverts unless for blatant vandalism because there is no other way to enforce such a restriction (especially for partial reverts), and because it is so strict it cannot be 0RR, or even 1RR. For the same reason, a wheel war (which, if you read above, is a war, analogous to an edit war, not a wheel revert) cannot be 0RR, even if we ignore the privileged action of editing MediaWiki pages. —Centrx→talk • 00:48, 5 November 2007 (UTC)

Bold-Revert-Discuss is a great technique for articlespace improvements but it works less well for priv actions. We, some of us, are trying to move from where we are now, in which we have seen multiple instances of people undoing extremely recent actions (Zscout's undo of Jimbo comes to mind, among other recent reversions), apparently without any consensus to do so, or even with any attempt to consult with the person doing the privileged action first. We have seen how destructive that can be, haven't we? Right now, the way this policy is worded (and you know people ruleslawyer about the wording, you KNOW they do) whoever undoes the action in quick succession seems to have the upper hand. That seems wrong to me. No one should have the upper hand, but we should be assuming by default that the first admin that acted had good reasons for doing so and that we should by default, discuss first, understand why it was done, and if there is reason to undo, then undo.

Is there general agreement on that point? If there is, then sure, perhaps this wording isn't airtight. Help fix it please! But if there isn't, then discussion of wording is premature because we need to fix the disagreement. Let's address the bigger picture first if we must, or reassure ourselves this is ONLY a wording issue and we all agree with the intent. ++Lar: t/c 01:40, 5 November 2007 (UTC)

The upper hand is the status quo ante, which is perfectly reasonable and is generally the holding pattern with any dispute on Wikipedia. The upper hand belongs not only to the one-reverter individually, but to everyone who decided not to block or unblock the user, who decided not to protect or unprotect the page, who decided not to alter the MediaWiki page, etc.

It may be prudent in many cases to consult with the original acting admin, but that does not mean that failing to do so needs to be defined as a "wheel war". It does not need to be defined as a "wheel war" even in cases where consultation would be extremely prudent, let alone the many cases where consultation is entirely unnecessary. —Centrx→talk • 01:58, 5 November 2007 (UTC)

It's more like a wheel confusion or a Wikipedia:Wheel disagreement. —Centrx→talk • 02:05, 5 November 2007 (UTC)

Sorry, I do not agree. That basically says that no admin can ever act courageously anywhere, that all it takes is one unreasonable admin to undo the action. I might as well take my name off Wikipedia:Admins willing to make difficult blocks. We have had, as I said, far too many really bad reversals of late, and that's not good. It is a recipe for excessive drama and troll enablement. ++Lar: t/c 05:56, 5 November 2007 (UTC)

The same unreasonable admin who is going to undo your courageous action is perfectly capable of initiating unreasonable admin actions that you would not be able to undo under your rule. If you think reversing any admin action should require additional discussion, why shouldn't the initial "courageous" act require discussion in the first place?

All of the bad things would still happen with 0RR, they will simply be shifted from the reverter to the originator. If you think there are too many bad reversals now, changing to 0RR will simply mean that there are too many bad, unreversible admin actions. The same immunity you want for making "difficult blocks" gives immunity to the aforementioned unreasonable admin for making unreasonable blocks. This principle is self-contradictory: By aggrandizing the individual authority of an admin, you enable more unreasonable actions by the very unreasonable admins you sought to prevent. In addition, having blocked users who cannot be unblocked is far more dramatic than whatever present drama you are referring to, and will cause users to leave.

There will be far more blocks, protects, etc. left in bad states by one unreasonable or hasty admin, and collaterally there will be a huge amount of needless discussions--or admins will simply be wary of making certain admin actions at all. You have in your mind the idea that somehow the rule will magically allow reversals of bad admin actions and only allow the good ones to stick; that is not how a revert rule works. As always, the way to deal with unreasonable admins is through dispute resolution or de-sysopping. —Centrx→talk • 06:38, 5 November 2007 (UTC)

I also support some version of 0RR. This page used to advocate 0RR (see here for example, though it didn't use that expression), but at some point it was changed to say that the first reversal isn't wheel warring. This has never made any sense to me. It means that if Admin A blocks and Admin B unblocks without discussion, Admin B's action will always stick, because no one will dare undo it.

The more admins we have, the more likely there will always be someone willing to undo a block without discussion. It therefore makes sense to say to people who want to unblock that they must always either seek the agreement of the blocking admin, or consensus on AN/I. SlimVirgin ^{(talk)(contribs)} 06:11, 5 November 2007 (UTC)

Except, of course, where the blocking admin has made an unambiguous error (e.g. a block for 3RR where there were only two reverts). But it would have to be a very clear error before anyone should unblock without discussion. SlimVirgin ^{(talk)(contribs)} 06:13, 5 November 2007 (UTC)

It was Radiant in November 2006 who added to the lead that reversing an admin's action once isn't wheel warring. [2] SlimVirgin ^{(talk)(contribs)} 06:17, 5 November 2007 (UTC)

No, this page at no time advocated 0RR. The version to which you link specifically states that a wheel war is a struggle that is analogous to an edit war.

I don't see the relevance of the addition of the text in November 2006 and by whom. You were the one who originally took out the verbiage after the policy was initially written.

If your above scenario were correct, then 0RR would mean that if Admin A blocks, Admin A's action will always stick, because no one would dare undo it. Why would that be better? Actually, it would be worse, because under the current policy a third party is capable of undoing an admin's action without needing to dare. —Centrx→talk • 06:59, 5 November 2007 (UTC)

The version I linked to doesn't use the term 0RR, but it implies it. For example, in the lead: "Wheel wars occur when admins get too 'heated up' to discuss something, or when an admin takes it upon him or herself to undo another admin's actions without consultation, or deliberately ignores an existing discussion (usually at WP:ANI or WP:DRV) to implement their preferred version (emphasis added)." [3] SlimVirgin ^{(talk)(contribs)} 07:19, 5 November 2007 (UTC)

If admin A blocks, we presume initially he has good reason. If he cannot be reached in a few hours, and there is consensus to reverse, we reverse. But if we reverse without communicating with the blocking admin and without a clear consensus, we are doing so without full information, and creating disruption. This should result in a desysopping, as it clearly damages the project. Crum375 07:07, 5 November 2007 (UTC)

I think the most critical cases, where the first admin action should not be reversed without either his consent or a very wide and clear consensus, are when the original action is a block, or a deletion due to BLP or copyvio concerns. I can see a block being reversed after a few hours have passed and the blocking admin cannot be reached, and there is consensus to unblock. We need to assume that the blocking or deleting admin may have more information than we do, so we can't just undo his action without all the available data (e.g. in a 2RR case, he may consider it to be edit warring due to past behavior). I fully support 0RR in such cases, where violations would carry an automatic de-sysopping penalty, with possible blocking. Crum375 06:26, 5 November 2007 (UTC)

In practice it can take 30 - 60 minutes to develop a consensus at ANI to lift a block when that needs to happen. Waiting that short time is a small price for maintaining civility and order. If a sysop makes a decision, they deserve a measure of respect. - Jehochman ^Talk 08:00, 5 November 2007 (UTC)

A 'wheel war' is when two or more admins reverse each other's admin actions several times (say four or more). However, I have always maintained that the actual problem that our 'wheel warring' policies seek to solve is (or should be) any single admin action taken where the admin knows, or should know, that it will be widely controversial and is not supported by clear consensus (including clear consensus behind longstanding practice). Such actions are inherently disruptive and lead to the sort of multiple reversals involved in wheel warring. The tricky part is, of course, what constitutes a situation an admin 'should know' is going to be controversial and what constitutes 'clear consensus'? Much of this discussion seems to have been initiated by and/or obliquely refer to my recent reversal of JzG. Personally, I think JzG 'should have known' that his block was going to be controversial and was not supported by longstanding consensus (we do not block users we are in dispute with, we do not indef block users for four reverts over as many days, we do not indef block users without any attempt to first get them to stop the 'disputed' action, et cetera... these are all clear longstanding practice) but maybe he really thought that 'alternate accounts are not allowed to make ANY controversial edits'. I don't think that has ever been practice or that it ever will, but it's a theory. Consensus on the article talk page supported Privatemusings' action, but maybe JzG didn't read it. However, consensus in the community had rejected 'BADSITES' and other such 'all links to anything connected to negative comments about users' theories in multiple extensive discussions over the past year - and JzG was surely aware of that. My action, on the other hand, was taken with the certain knowledge that it would be disputed by several (those who supported JzG's 'side' of the dispute and/or had endorsed the block in the AN/I discussion) but just as certain belief that it was absolutely supported by longstanding consensus... in addition to all the reasons the block went against practice (don't block for disputes you are involved in, first try to resolve disputes then place small blocks - don't go directly to indef with no prior warning or discussion) there was also the simple fact that we do not leave users blocked for edit warring when they have agreed not to edit the page. Blocks are preventative. This is all bedrock undisputed stuff with dozens of previous examples. The possibility that the blocking admin might have additional info is mentioned repeatedly above. That's true and while it would be better to spell out everything in the block log and/or notice to the user sometimes this isn't done and a discussion should be undertaken to get all the info. However, once that discussion has taken place and the admin has responded I do not believe other admins should have to establish a clear consensus in that discussion for reversal - IF there is a clear consensus in standard practice and policy. An admin action should not become the 'de facto' situation by virtue of having been taken first... certainly not when it goes against multiple established consensus practices. --CBD 10:32, 5 November 2007 (UTC)

One issue is spilling the beans. Sometimes a block is based on evidence that we don't want to publicize because it's defamatory, or would give sockpuppeters information on how to better evade detection. In those cases, it's important not to reverse a block without consulting the blocking admin, or at least making an attempt.

A second issue are controversial blocks. A wise administrator will notify the community when making a tough block and invite review. There's no emergency need to unblock somebody while that discussion is ongoing. ANI is a high traffic board, where consensus is quick to form. If the block is truly bad, a consensus to unblock will form in about 30 - 60 minutes, a small price to pay for increasing civility around here. If a consensus does not form, the case can go to Arbcom. They routinely accept cases where one or more administrators have a serious disagreement. Civility is enhanced by having administrators talk to each other rather than dueling with the tools. It's not a question of favoring the first mover or the second mover. We just need to trust each other a little bit. - Jehochman ^Talk 11:06, 5 November 2007 (UTC)

And civility towards the USERS? They can be blocked on a whim, muzzled by protecting their talk pages, subjected to petty insults and unfounded accusations... and they are just supposed to take it calmly? Because we have to show respect for admins? But not users? Yes, treating others with respect and trust is important... but it is (IMO) dead wrong to apply that principle ONLY to admins. Users should not have to remain blocked (or muzzled) if there is a lack of consensus either way. A 'truly good block' won't BE controversial. --CBD 11:31, 5 November 2007 (UTC)

So you're saying that nobody should be blocked unless there's unanimity amongst all 1400 administrators? That can't be right. If an administrator places a block, we assume good faith by discussing it if we disagree, and then we abide the consensus. We don't just revert. Do you think an administrator is blocking on a whim? If you do, take the case to Arbcom, and ask for a remedy.

If there is a lack of consensus to unblock, then maybe there really is a problem, and maybe the user should remain blocked. Again, Arbcom is the solution. The will consider a preliminary motion.

A truly good block may be controversial because sometimes people try to protect their friends. We don't want to create a situation where popular people can do whatever they like because a few friendly admins keep unblocking them without any consensus. That wouldn't be fair. - Jehochman ^Talk 11:40, 5 November 2007 (UTC)

That standard, unanimity amongst all admins, actually IS a common description of the requirement for an indefinite 'community patience' style block, but no... that's not what I'm saying. Above you are essentially suggesting that the 'default' position should be that the user stays blocked unless clear consensus or ArbCom says otherwise. I disagree. You talk about respect and trust and civility towards admins... I apply those same standards to users. If a user has agreed not to do the thing which got them blocked then there is NO reason for them to remain blocked. That is basic respect and trust and civility towards users (not to mention longstanding accepted practice). If they do it again they can always be blocked again.

You say that a truly good block may be disputed by friends of the blockee. I agree, and would respond that, in kind, a truly bad block may be supported by friends of the blocker and/or opponents of the blockee. Is it 'fair' that the user should then remain blocked? Even though they've agreed not to do the disputed thing which they were blocked for, which may or may not have been blockable in the first place? What exactly is the block then accomplishing? The 'punishment' and/or removal of a user that one or more admins dislike? You seek to make this a question of 'fairness'? Which is more 'hurtful'; the dispute of an admin's opinion or the branding of a user as unwanted by and banned from the project? --CBD 12:12, 5 November 2007 (UTC)

Editing Wikipedia is a privilege, not a right. If there is substantial doubt concerning an editor's fitness to continue editing, they should remain blocked. In society, when it comes to criminal charges, we have the opposite situation. We need to be careful not to confuse these two things. - Jehochman ^Talk 12:15, 5 November 2007 (UTC)

Assuming that 'substantial doubt' includes 'no consensus', as would seem the intent from context, then this is the root of the matter and we directly disagree. I think your position is inconsistent with WP:AGF and past practice on Wikipedia. Nor do I think it would be a beneficial change. It suggests that people can be banned, or at least blocked until overturned by ArbCom, merely for holding an opinion that a significant minority strongly oppose. --CBD 13:12, 5 November 2007 (UTC)

I think people are usually blocked for their actions, not for opinions they hold. — Carl (CBM · talk) 13:34, 5 November 2007 (UTC)

Miltopia was blocked for his opinions (regardless of accusations that these opinions were not sincerely held, and debates as to whether or not it's legitimate to block for that; there certainly wasn't any suggestion that there were any _actions_ by him that deserved a block, that he had not already been blocked for). —Random832 14:15, 5 November 2007 (UTC)

[edit] Comment: page protection and time between actions

I often look into {{editprotected}} requests to protected pages. In some cases, it is apparent from the protection log that the article was protected because of some temporary situation, and it has been months since then. In such cases, the routine behavior is just to unprotect the page, and I have never heard anyone call this "wheel warring". If there are changes to the wheel warring guideline, it would be nice for this sort of thing to be explicitly addressed. One way to do this would be to point out that as the time between the action and its reversal increases, the chance that there is a wheel war decreases (without giving explicit timetables). Undoing an actions after 4 hours is different than undoing it after 1 month. — Carl (CBM · talk) 12:54, 5 November 2007 (UTC)

As above, I'd say that the issue should be 'fore-knowledge of controversy'. It should be fine to unprotect even 'after four hours' if the parties have resolved the dispute which led to the protection in the first place... or if it were the featured article of the day temporarily protected due to a huge surge of vandalism. People do these kinds of reversals all the time... because they aren't controversial. Further, if it turns out that there was a secondary dispute being discussed elsewhere (e.g. a centralized discussion page) the unprotector still wouldn't have done anything 'wrong'. --CBD 13:12, 5 November 2007 (UTC)

The problem with that is that I think this entire discussion was started by the Zscout incident, and apparently he did not have a "fore-knowledge" that the unblock would be particularly controversial. Personally, I'm not sure it's possible to come up with anything better than the current guideline to describe our very nuanced practice on undoing admin actions. — Carl (CBM · talk) 13:34, 5 November 2007 (UTC)

If he didn't have foreknowledge that it would be controversial, what did he do that we should be making a rule against? Starting with the a priori goal that we need to—by any means necessary—make a rule against what he did, is putting the cart before the horse. It won't change anything either - the fact that he was desysoped for an action that was absolutely not against either the letter or the spirit of the rules at the time of his action still remains. —Random832 13:48, 5 November 2007 (UTC)

I think the purpose of this discussion is whether to change the written guideline to match practice better. But I don't think it's feasible to switch to 0rr in general, and I don't see any clear way of explaining the difference between blocks that can be undone with no fuss vs. blocks that should not be undone without discussion. — Carl (CBM · talk) 14:09, 5 November 2007 (UTC)

This guideline DOES match practice. Jimbo's actions in the Zscout370 case did not match either. You seem to be confusing this with a disconnect between the guideline and practice (understandable, as Jimbo keeps saying that he did nothing that wasn't common practice for such supposed "wheel wars") [edit conflict on reword 14:21, 5 November 2007 (UTC)] —Random832 14:17, 5 November 2007 (UTC)

I agree that undoing page protection after 4 months, unless the protecting administrator had left a note requesting permanent protection, would not be wheel warring. Let's clarify the wording to make this clear. - Jehochman ^Talk 14:20, 5 November 2007 (UTC)

[edit] "appropriate actions" and upper hands

I carefully weighed the adding of appropriate because admins can and do apply powers incorrectly, and this needs the ability to be remedied without fear of sanction. The upper hand must always be with the sysop who acts legitimately, whether in the initial action or subsequently. LessHeard vanU 14:20, 5 November 2007 (UTC)

It's a very tricky issue. Two admins can look at the same action and call it 'entirely appropriate' or 'completely illegitimate'. That's why I prefer 'consensus' and 'existing practice/policy' to 'appropriate'. I can prove the existence of a very common practice of unblocking people who have agreed not to do the thing they were blocked for. I can't prove it being 'appropriate'... I just believe it is. Others believe it isn't. If a controversial action doesn't have a clear consensus or standard practice behind it then any determination of 'appropriate' comes down to nothing more than the opinion of the person(s) making the decision. If nothing else, I'd say that following consensus and/or common practice should never be INappropriate... even if it leads to the 'wrong decision' in someone else's subjective judgment. --CBD 15:24, 5 November 2007 (UTC)

I've added changed circumstances as an exception. Somebody agreeing not to make the same mistake (assuming they haven't done that before and been proven a liar) is a changed circumstance. This would allow an unblock. See below. - Jehochman ^Talk 16:38, 5 November 2007 (UTC)

Works for me. Would also cover unprotection after time has passed or the parties had resolved the dispute. Situation changes, admin comes along and takes the action which seems appropriate to them under the current circumstances. --CBD 17:11, 5 November 2007 (UTC)

Yes, I think so. - Jehochman ^Talk 19:35, 5 November 2007 (UTC)

[edit] Community Bans

If this rule changes not to allow a reversal without a prior consensus to reverse, then the definition of a "community ban" has to be seriously reworked - The definition as "if no admin will unblock" is fatally flawed in the face if those admins who would like to unblock - even if they are the majority (since a significant minority still defeats consensus-building) - are forbidden to do so. The current definition of a community ban is absolutely predicated on "any admin may unblock". We may have to get rid of the idea of community bans altogether; if there's no consensus to move back towards a CSN-like system. —Random832 13:53, 5 November 2007 (UTC)

And, before anyone even comes in with this objection, yes there is a difference between a banned user and a blocked (even indef) user. [I'd written up an explanation with examples, now removed per WP:WEIGHT - the important part here is that there is one, not who falls into what category]. This distinction has, as far as I can remember, always been made - in discussion if not in written policy - in terms of whether someone would be welcome to come back and make constructive edits —Random832 14:10, 5 November 2007 (UTC)

[edit] Revision D

Proposed, taking CBD's criticisms into account. - Jehochman ^Talk 14:25, 5 November 2007 (UTC)

A privileged action based on new circumstances is not considered a revert, even though it may have the effect of reversing an earlier action. For instance, if a page is protected due to edit warring, and the warring parties come to an agreement on the talk page, the article may be unprotected without need for discussion.

I think tenable [4] is a better word than appropriate, since it has less connotation of correctness. — Carl (CBM · talk) 15:36, 5 November 2007 (UTC)

D2

A wheel war starts when a tenable privileged action is reverted without consensus. Consensus can be obtained by discussing with the original administrator, or posting on one of the administrators' noticeboards.

Untenable actions include: anything done in error, such as blocking a different user than the one intended; anything done in bad faith, such as deleting the main page; and anything forbidden by policy, such as blocking a user while engaged in a dispute with them or protecting a page while edit warring. Tenable actions are those for which a reasonable case can be made; the use of sysop tools in cases where reasonable administrators may disagree is considered tenable. Disagreements are to be resolved by discussion.

I agree, this is even better. - Jehochman ^Talk 15:50, 5 November 2007 (UTC)

Nice work. This better reflects current practice, and allows us to correct mistakes while still respecting each other's actions. Tom Harrison ^Talk 17:01, 5 November 2007 (UTC)

How long would an admin have to wait for consensus (or response from the original admin whose actions are being undone) before acting? Is there a set time, or a set number of concurring !votes on the notice board to justify undoing the privledged action? I ask because I'm sure there will be a borderline case down the line where the question comes up. ZZ ^Claims ~ Evidence 17:32, 5 November 2007 (UTC)

To avoid instruction creep, no time is specified. Based on recent experiences, in obvious cases it seems that consensus at WP:ANI often forms within 30 - 60 minutes. For a contentious decision, it may take longer. Our consensus and "snow" policies provide guidance on this question. - Jehochman ^Talk 17:55, 5 November 2007 (UTC)

That makes good sense. If consensus is slow in coming, then I would think that would be a clear sign that the action should not immediately be overruled. If it's a rapid and unanimous consensus, though, WP:SNOW comes into play. I can't see a situation where there's a time element - someone has to be unblocked NOW - so waiting for consensus shouldn't cause any problems. I like this version. ZZ ^Claims ~ Evidence 18:26, 5 November 2007 (UTC)

I'm concerned that this means that in cases where there is no consensus to either overturn or let stand; the bias is in favor of a user being blocked / a page being protected / etc. Is this really the policy we want? This is like if "no consensus" at AFD meant an article gets deleted. —Random832 19:55, 5 November 2007 (UTC)

There are some cases where this is absolutely what we want... Many of us, including I myself, do strongly feel that any deletion explicitly citing WP:BLP or m:OTRS should be subject to 0RR, though with a clear presumption that the deleting admin must take a full part in any discussion it generates. In these cases we MUST err on the side of caution. ++Lar: t/c 20:06, 5 November 2007 (UTC)

This is where and why admins require the trust of the community - the blocks, protections that should not be overturned without the consensus to do so, even if problematic. It is difficult, but finding a wording that echoes the practice of some areas being more "sensitive" than others is (I suggest) impossible. LessHeard vanU 22:12, 5 November 2007 (UTC)

I do agree that those should not be reverted, but that belongs in those policies, not this one. Such an action would still not be a wheel war, though it would otherwise be a policy violation. —Random832 14:06, 6 November 2007 (UTC)

Unless we have a minimal set of completely logically consistent policies, (that may never happen) some overlap is inevitable and I think mentioning or referring to these situations in this policy is very appropriate. But I mentioned them to show that there are valid reasons for this being absolutely what we want, so I in general don't agree with the concern you raised... the bias IS in favour of trusting the first admin had reasons for doing what they did, and that is appropriate and right. But it is just a bias, consensus can overturn it. ++Lar: t/c 15:56, 6 November 2007 (UTC)

I think that if, in the face of an involved discussion, there is genuinely no consensus in favor of blocking a user, the user should be unblocked by default per WP:AGF, and, in many cases, WP:BITE. Trusting initially that the admin had reasons, sure... but if when those reasons are explained there is not a consensus that it was a good reason to block, then the user should be unblocked. —Random832 16:41, 6 November 2007 (UTC)

Sure. At that point. But prior to a consensus forming, the block should stand. This practice of unblocking because you don't agree, and maybe only then turning up at AN/I is exactly backward. ++Lar: t/c 19:30, 6 November 2007 (UTC)

I agree that tenable is a better term than appropriate, but I wonder if it is understood by a wider readership. A brief review of Rogerts online thesaurus doesn't give anything better, so it is perhaps the default choice. LessHeard vanU 21:02, 5 November 2007 (UTC)

It was the most neutral word I could find. If people don't know it, that might be a good thing, since they will think of it as a term of art here rather than having strong connotations about "correct" or "justified" actions. — Carl (CBM · talk) 21:43, 5 November 2007 (UTC)

We must be careful that the written records of the different policies remain consistent, otherwise there can be much confusion. - Jehochman ^Talk 14:31, 6 November 2007 (UTC)

I disagree with any notion of WP:0RR as the definition of a wheel war, and thus object to all the. That unacceptably privileges the admin who is most hair trigger in acting. It is at least as bad to priviledge the admin that acts first as the admin who reverts. In fact, since the default state for pages is unprotected and undeleted, and the default state for users is unblocked. If there is a disagreement, discussion should occur, but if no consensus is reached the page should be unprotected/undeleted and the user should be unblocked. The policy needs to clearly say that a consensus is required to remain out of the default stage when there is a dispute, which is the exact opposite of the current implication - the current implication is that once there is a dispute the non-default state is privileged (e.g., a user remains blocked until there is a consensus otherwise). The upper hand belongs with every admin who had not taken the action, not with the one admin who acted. If their action was correct, consensus will follow discussion. If such consensus does not follow, their action was wrong and should not stand.

It is acceptable for the policy to discourage reverting prior to discussion, but I think that going to far in this direction is a mistake. WP:BRD works just as well for administrative actions as for any other. A delay in changing the database contents is rarely all that big a deal; a delay in unblocking a poorly blocked user has a real, non-reversible impact on that user's feelings. In the case of blocks, we thus see that it may be better to revert to unblocked and then discuss than to discuss while the user is blocked.

We also know many cases where admins have acted despite knowing that their action is controversial. In no way should we privilege such actions. Above is mentioned JzG's clearly wrong block on PrivateMusings. For BLP, we have determined (though I don't know how well it is documented) that an admin making a controversial BLP deletion should actively seek review at DRV, as I did earlier today with one page.

My basic view is that admins are not privileged users, they are servants of the community who have special tools to carry out their servitude. If an admin is acting, they are saying that they believe the community as a whole is served best by an action, and that the community will support them if discussion occurs. If the community (not just administrators, the entire community that chooses to participate) does not support them, they were wrong and the action should be undone. The proposed drafts all get this wrong, by thinking that an administrative action is a privileged action - no administrative action is privileged, and if it can't be sustained by consensus should be reversed to the default state (unblocked, undeleted, unprotected) as soon as it is clear that no consensus to support it will exist. GRBerry 18:53, 6 November 2007 (UTC)

I am afraid that I don't agree with you in a number of areas. Just dipping at random here: The "clearly wrong" action with respect to PrivateMusings was the unblock, not the block. While it is true that feelings might be hurt from not unblocking, the damage caused by an inappropriate unblock can often far exceed even the complete loss of one good user. No outcome at DRV should be able to overrule a BLP action, as BLP is a foundational policy, not subject to consensus override. Those are just three I spotted at a glance, so I think your view is not reflective of generally accepted thinking. ++Lar: t/c 19:30, 6 November 2007 (UTC)

I'll fork off educating you fully about BLP to another venue, just noting here that in the BDJ ArbComm case the ArbComm explicitly ruled that BLP deletions are subject to DRV review. That they should be is obvious, no admin's claimed basis for any action is validated by the claim. In fact, that seems to be your basic failure to understand how we operate - an administrator's claim is not self validating, no matter whether that claim is a un/deletion reason, a un/block reason, an un/protection reason, or a content item. The logs are open for review precisely because admins can, and regularly do, get things wrong. If a claim was self validating, we could have a 0 revert rule that didn't even allow for consensus overturns, because the self-validating claim would be correct solely because it had been claimed.

I find that your opinion on these subjects is fairly divorced from reality, and it contradicts explicitly documented consensus and policy. I believe that the community has previously rejected multiple times and in multiple venues any attempt to read this policy as privileging the first actor. I've seen it rejected in ArbComm cases, at WP:AN, and at WP:AN/I. Though given the nature of those forums, demonstrating consensus from those forums is essentially impossible. The archives of this page, however, demonstrate both 1) that any bright line test was explicitly rejected when this page became policy and 2) that a zero revert rule was far more strongly rejected than a one revert rule, gathering essentially no support (I found 6-7 between the poll and the discussion versus 30 or more for bright line rules that were more generous). A significant fraction of that support for 0 reverts came from admins that have since been desysopped for cause or otherwise encouraged not to be active. That later desysopped admins were the supporters of a 0 revert policy is a caution, not dispositive, but does need to be mentioned.

There is discussion throughout this archive showing the policy formation period. Neither a 1WW nor a 0WW fits with the spirit of the policy as it was approved, and both should be removed. In fact, that discussion reveals that the only part of the page that was intended to be policy was "No admin shall repeat any admin action in the knowledge that another admin opposes it." As John Reid said in June 2006 "Only this text is intended to have the force of policy; all else is explanation and commentary, with lesser force." Looking at the polling at this archive, 0WW (both of Kelly's suggestions) got a total of 4 supporters (I found a couple more in the talk archive linked above); 1WW or more reverts allowed (Philwelch, Carnildo, John Reid, Locke Cole) got at least 28 supporters (I stopped counting after I couldn't remember which names I'd seen before). When this page became policy, it is very clear that 0WW had been rejected as the policy, that 1WW is closer to what was approved, but that the policy authors intelligently refused to draw a bright line. We should follow their intelligent decision in forming the policy, and remove all bright lines from the page, but if we have to have a bright line, that bright line should be at least as generous as 1WW, as that is the version that had strongest support when this became policy and is the current consensus reflected in actual actions, ArbComm cases, WP:AN discussions and WP:AN/I discussions. GRBerry 21:17, 6 November 2007 (UTC)

I fear that I may not have expressed myself very clearly in what you respond to, which is my fault, but one point you try to make about BLP and DRV really jumps out at me... I really think that consensus can not override our foundational policy. No outcome at DRV along the lines of "this is a clear and present BLP violation but based on who turned up here and commented, we're going to keep the article around anyway and overturn the delete because that's the consensus" can or will be allowed to stand. But that's a side issue. I think your focus on what the authors of this policy felt when it was authored is also a side issue, what matters is that there is a problem today with unfounded overturns of admin actions, apparently on whim, and in contravention of consensus and policy, and that this particular policy needs to be clarified so that it is crystal clear that is not an acceptable behaviour. If you have a suggested wording that will achieve that goal, please suggest it. ++Lar: t/c 22:21, 6 November 2007 (UTC)

I'm not quite following the objections here, but it seems to me that the current version definitely reflects best practise. Admins undoing other admin actions without consultation causes a lot of trouble, and has led to people being desysopped temporarily or permanently, so it seems pretty clear that 0RR is what we should aim for (apart from when an admin makes an umambiguous error or clear policy violation). SlimVirgin ^{(talk)(contribs)} 00:05, 7 November 2007 (UTC)

If that is all we need, zero revert rules and such don't even help. "Any administrative action needs to be supported by policy. If it is unclear whether policy and the facts of a particular case support an action, discuss in an appropriate forum and await consensus before acting. If there is disagreement with an action or proposed action, this policy prohibits the repeated use of administrative tools to revert other administrators prior to the emergence of consensus. If no consensus forms, seek wider input or take the matter to dispute resolution."

There is also a problem, that you seem not to recognize or admit, of unfounded original admin actions taken in contravention of consensus and policy. JzG's block of PrivateMusings is an obvious example of this. JzG was in breach of Wikipedia:Blocking policy#Disputes because he is in a dispute with this user over this topic and even worse was himself attempting to enforce a policy proposal that is very clearly rejected by both the community and the Arbitration Committee, and JzG knew that his position in the dispute is rejected by both groups. (Whether PM should have been blocked by a different admin is a different issue on which relevant evidence is not being offered openly; it was obviously wrong for JzG to block himself.) Even if you continue to claim this particular case wasn't an obvious error, we've all made plenty of mistakes and can recognize that sometimes admins are unwilling to admit their mistakes, since we know that we sometimes have trouble admitting our own mistakes.

The page needs to be aware of two different problems - one is incorrect initial actions, one is incorrect reversions. A bright line zero reversion rule "solves" the risk of incorrect reversions by protecting incorrect initial actions. A bright line one reversion rule "solves" the risk of incorrect initial actions by protecting incorrect reversions. Any bright line rule will "solve" one by protecting the other - and neither should be protected. The solution is to stay far away from bright line rules, as any bright line rule is wrong.

In statistics, the equivalent situation is [[Type I and type II errors], or more broadly false positives and false negatives. In statistics, the solution is to draw the boundaries against type I errors as tight as possible and accept whatever level of type II errors the tester ends up with. That approach isn't appropriate here, as it was the threshold for initially acting far too high. Tightening the standards for reversing an administrative action also requires tightening the standards applied prior to an administrative action. Instead, we should recognize that the willingness of any admin to reverse anothers action is fairly strong evidence that the original action was itself incorrect. In the absense of a clear consensus, it is better to return to the encyclopedia default state (unprotected, unblocked, undeleted), as those states can be assumed to have the highest level of consensus support from non-participants in the discussion. GRBerry 23:29, 6 November 2007 (UTC)

You keep asserting, as part of your argument, that "JzG's block of PrivateMusings is an obvious example" (of a block that sholud have been speedily overturned, presumably). I am afraid I quite disagree. What I find it an obvious example of, rather, is a block that should not have been overturned without a clear consensus to do so. So therefore, if there is disagreement on that, it's a block that should not have been unilaterally overturned at all. There are some admins here that seem to have a history of undoing any action they do not agree with, unilaterally, and even in the face of consensus, which you seem to fail to acknowledge, or fail to be willing to address as an issue. Bringing this around, what specific changes to the wording of this policy do you advocate, if any? ++Lar: t/c 00:14, 7 November 2007 (UTC)

My problem with this discussion - It's not helping me understand whether, under the new policy, the BJADON incident falls within the new policy or not. Georgewilliamherbert 23:07, 6 November 2007 (UTC)

Whether Alkivar's deletion was a violation depends on both the version considered and for the versions in this subsection whether the amount of time constituted a change in circumstances. I'll agree that it isn't clear whether his deletion was a violation. Whether your undeletion was a violation might depend on whether his was, but under the zero revert rule language, I think it is clear that your undeletion was a violation regardless of whether he wheel warred himself. That workshop page is probably the last time the community discussed wheel warring significantly, even if the discussion died out with the dispute successfully being resolved through MFD/DRV. GRBerry 23:29, 6 November 2007 (UTC)

[edit] Consensus

We had a long discussion that resulted in a consensus, and now one editor has reverted the changes.[5] Do we want to restore them, or was all this for naught? - Jehochman ^Talk 13:16, 9 November 2007 (UTC)

WP:BRD? LessHeard vanU 13:39, 9 November 2007 (UTC)

To be honest, I don't see a strong consensus here for the changes (I think this needs more voices and perhaps a longer discussion to make such changes to a policy that can get people de-sysopped), but beyond that the changes to the policy page went well beyond the proposal here. I'd suggest Jehochman self revert and go a little slower. RxS 15:15, 9 November 2007 (UTC)

A more accurate interpretation of the above discussion is: The discussion simply petered out without any conclusion, or it is still ongoing without having yet reached a consensus, or the discussion is finished and there was no consensus for the change. —Centrx→talk • 15:28, 9 November 2007 (UTC)

I think we disagree on the outcome. Perhaps we can find one or more uninvolved administrators to read the discussion and provide further input. If you want to post a neutral message to WP:AN asking for more involvement, that would be great. - Jehochman ^Talk 15:32, 9 November 2007 (UTC)

As long as you mean that more people should come to discuss the issue at hand (rather than to decide by some judicial intervention, which is not consensus), then Wikipedia:Requests for comment is also appropriate (and "uninvolved" administrator does not make sense in this context). —Centrx→talk • 19:36, 10 November 2007 (UTC)

"Previously non-participating sysop"? "Couldn't be bothered until now admin?" I think somebody else need word the recruiting terminology. LessHeard vanU 09:34, 11 November 2007 (UTC)

(Reply to Jehochman) If we had consensus, there wouldn't be disagreement about what the consensus is. That there is disagreement is sufficient evidence to prove that we don't have consensus yet. If more people participate, a consensus may form (and may be different from anything yet discussed), or we may end up with no consensus. GRBerry 19:47, 12 November 2007 (UTC)

I'm participating, not non participating but I thought we, by and large, had consensus for the changes. Something needs to be done so I would like to see what Centrx proposes to address the issues raised that drove the change. ++Lar: t/c 11:30, 12 November 2007 (UTC)

What is wrong with the current policy?

If the problem is that a single sysop was desysoped by a single, singular steward in a single incident, the desysopping was wrong if it was based only on "wheel warring" and the policy ought not be changed for it; the desysopping was right only if it was based on some incidental or other behavior that is not relevant to the wheel warring policy.
If the problem is "drama", any drama prevented by this change would be offset and overridden by the drama caused by temporarily irreversible blocks and other admin actions; this would only increase drama, and collaterally quieting trolls would not be a reason to change a policy, let alone one that can cause desysopping.
If the problem is the reversal of "right" admin actions, any benefit there would be offset and overridden by the fact that "wrong" admin actions would now be temporarily irreversible.

In any case, a single admin action is not a "wheel war", and you would need to rename the page or, because an actual wheel war is more severe than a simple reversal, create a new separate page.

Also, note that the wording proposed actually partly relaxes the wheel warring policy. The reason we have strict 1RR (or 3RR) is because the exceptions necessary to allow any sort of reasonable 0RR open the floodgates because they are so easily interpreted to allow any reverting of "erroneous" or "bad faith" actions, or actions "for which a reasonable case can be made". Any warrior who would be so unreasonable as to reverse an admin action that you think is "right" would be no less unreasonable in interpreting the exceptions to his advantage. Any reasonable 0RR is internally eviscerated. —Centrx→talk • 03:39, 13 November 2007 (UTC)

I think that it's important that if there is a discussion that does not result in a consensus to support either outcome (blocking vs unblocking, deleting vs undeleting) over the other, a no consensus result defaults to the status quo (status quo defined as, in general, whatever was the case _before_ any recent admin actions), whereas the version that was arrived at in the last round (A) did not address this situation explicitly, and (B) implicitly defaults against the status quo. The original version didn't address the situation either (what outcome results from the bright-line rule presented as "0WW" depends solely on how many admins are participating on each "side", with a slight bias towards the status quo, and no recognition of the importance of discussion. The rest of the guideline doesn't provide much guidance at all, even for cases where a consensus is clearly there) —Random832 13:54, 13 November 2007 (UTC)

For the record I support the wording in D2. 1 != 2 00:09, 16 November 2007 (UTC)

[edit] Is this wheel warring - question??

Example (Note: All users here are fictitious!)

Wikipedia admin Joe Public (talk • contribs • blocks • protects • deletions • moves • rights) deletes Joke article No.120404 citing WP:CSD#G1 and WP:CSD#G3.

Then another admin, Joe Bloggs (talk • contribs • blocks • protects • deletions • moves • rights) comes along and undeletes Joke article No.120404 without any discussion, claiming he's exporting it to his wiki (on a free wiki host, e.g. Scribblewiki) under the GFDL, then deleting the article again with the edit summary (Article moved under the GFDL to FooBarWiki)

How should the situation be dealt with?? Taking it to WP:AN or WP:ANI, or WP:RFC??

Thanks, --Solumeiras ^talk 14:01, 7 December 2007 (UTC)

No action whatever. That's absolutely fine. Splash - tk 16:32, 7 December 2007 (UTC)

So the above examples aren't wheel-warring then?? Is that OK to add to Wikipedia:Wheel war?? --Solumeiras ^talk 18:17, 7 December 2007 (UTC)

[edit] Flocking wheel wars

This page has always struggled with a definition, but the bolded version "do not repeat an admin. action when you know another admin opposes it" overlooks some of the most serious cases of wheel warring that have occured, and is not very general. The key thing is that no admin should engage in an action on a disputed point in such a way as to cause another reversal of administrative action. Thus:

Do not continue a chain of administrative reversals on a disputed point.

This allows "delete, undelete" by 2 admins, but not "delete, undelete, delete" by any number of admins during the dispute. I mentally refer to these as 'flocking' wheel wars, where everyone thinks they can be smart by only doing the thing once, even though its plain that admins are fighting over it, one reversal at a time. Splash - tk 17:13, 7 December 2007 (UTC)

Changed to retain the old sentence also, which explicitly forbids a simple action-reversal-reversal by only two admins. A "chain" can imply something longer and more involved. [6] —Centrx→talk • 21:44, 8 December 2007 (UTC)

[edit] Wheel war definition

I came to this page to ask about a definitional statement (possibly to add to tool use in WP:ADMIN), only to find the exact same point being discussed by the community already. Perhaps the two can dovetail. I wonder if these possible approaches to the topic, that approach it slightly differently, might help produce a better wording for the policy:

Administrative tools not to be used antagonistically

General principle —Preceding unsigned comment added by FT2 (talk • contribs) 16:40, 22 January 2008 (UTC)

Administrative tools are to be used co-operatively and within consensus. At the point where it becomes clear that a proposed use of tools would be contentious in a matter, then with very few exceptions tools should not be used by any administrator (even for a desirable reason) and the matter should switch to discussion and consensus-seeking. There are a few, specific, exceptions permitting use of tools:

WP:BLP - material deleted under BLP may be re-deleted if reinstated, if it continues to be non-BLP-compliant.
Privacy - personal information deleted under the Foundations privacy policy may be re-deleted if reinstated.
Emergency - in certain situations there may arise an emergency that cannot be adjourned for discussion. An administrator should not claim emergency unless there is a present emergency (ie, reasonable possibility of actual, imminent and serious harm if not acted upon with admin tools), but in such a case the action should usually not be reverted until appropriate discussion has taken place.
Edit warring - reasonable actions undertaken by uninvolved administrators to quell a visible edit war by protecting a contended page should be respected by all users, and protection should not be undone (or may be reinstated) until it is clear the edit war will not resume or consensus agrees it is appropriate.

Handling of disagreement

Reversion of a clear or obvious error in an administrative action is not usually considered contentious.
Otherwise, disagreements over tool use should be discussed wherever possible with the administrator responsible for the initial action, or with the wider community; unilateral reversion should usually be avoided.
If the action (or a related or similar one) has already been reverted, then it is strictly forbidden for any administrator to re-enact it (or use tools to achieve a similar action) without discussion and consensus, apart from the above exceptions.

Perhaps this definition and approach will cover some of the issues above.

Focuses on desired behavior and actual core problem (admins to use tools co-operatively and not antagonistically to further their "side" in a disagreement).
Allows for clear well-recognized exceptions where even a single revert should not occur without consensus (there may be others, those are likely candidates).
Graduates "Do-Undo" and "Do-Undo-Redo" more in accordance with communal norms, avoiding an "either-or" problem over wording. "Doing+undoing" is to be avoided and discussion used wherever possible but is not cast in stone. "Doing+undoing+redoing" is strictly forbidden.

Also, this possible wording might help:

Administrators are strictly forbidden from fighting over the use of administrative tools by using those tools, whether for desirable reasons or not. With very few exceptions, when an action performed using tools has been rejected to the point that a second administrator has reversed it (or similar related actions were reversed), then there is almost never a valid reason for any administrator to reinstate the same or similar action (or end result) again, without clear discussion leading to a consensus decison, and administrators who do so may risk desysopping for abuse of their access. As a corollary, reversal of an administrative action should also not be undertaken without good cause.

Thoughts? FT2 ^{(Talk | email)} 08:55, 14 January 2008 (UTC)

I like it because this reflects current best practices, whereas the policy as currently written does not. Jehochman ^Talk 09:20, 14 January 2008 (UTC)

I would change the term "second administrator" to "following or subsequent administrator", so that there is no emphasis on the "first" admin action being default wrong and the "second" being default correct - it may be the second (or maybe even third) use of the tools in a matter that initiates the dispute. LessHeard vanU (talk) 20:28, 14 January 2008 (UTC)

With that change, I would agree that this version does well in reflecting current practice. UltraExactZZ ^Claims ~ Evidence 21:37, 14 January 2008 (UTC)

Clarity check - are we discussing both texts (eg as detail and intro), or first only, or second only? I put both forward because each might have uses - the first states the important exceptions and also distinguishes do/undo from do/undo/redo, the second sums up the entire issue in overview and might be good for an introductory paragraph or some such. And agree LHVU's observation/comment. FT2 ^{(Talk | email)} 11:06, 15 January 2008 (UTC)

I was only referring to the second paragraph, the overview. The first para does not determine when an exception might occur, thus does not reference any previous or subsequent action that might be wheelwarring. LessHeard vanU (talk) 13:51, 15 January 2008 (UTC)

The first wording looks pretty good. It eliminates any vagueness so there's no question as to whether one is wheel warring. (I remember during arbcom elections, a lot of questions were asked about what wheel warring meant. Clearly a bit of hardening is needed) The second one is good as well, but the first one spells everything out, which helps. Wizardman 16:39, 22 January 2008 (UTC)

[edit] Proposed new section

I propose to add this section to the page:

It is not wheel warring if an admin undeletes a page deleted by another admin, and then exports it to another site running MediaWiki, per the terms of the GNU Free Documentation License. However, they must use in their edit summary something like (Exporting to FooBarWiki per WP:GFDL, will delete when exported)

Is this a fair suggestion?? --Solumeiras ^talk 18:28, 17 January 2008 (UTC)

(S left me a message requesting a comment) It's true that this is not wheel warring. But it's also not necessary in my view. If someone undeletes to export, they should presumably redelete when they are done, and both edit summaries should show what was going on. We could also put a bunch of other things that are not wheel warring in there but the best essays are short and to the point. ++Lar: t/c 21:02, 17 January 2008 (UTC)

Correct me if I'm wrong, but you don't need to undelete to export. Admins can see the deleted markup from Special:Undelete and copy that ready for pasting, without undelete being needed. FT2 ^{(Talk | email)} 16:08, 22 January 2008 (UTC)

[edit] Minor links editing re examples

I've done some links editing:

All examples exist on the examples page. So I've removed the list of citations from here (no other policy page has a long citations section of arb cases), and just noted that details of cases are at /Example.
Move the examples link to the intro and make clear its examples and cites. Much more normal location and simpler wording.

FT2 ^{(Talk | email)} 21:34, 27 January 2008 (UTC)

[edit] Reverting admin actions without discussion is also bad

From Wikipedia:Requests for arbitration/Pedophilia userbox wheel war:

8.2) Wikipedia:Wheel warring (undoing an administrative action by another administrator) without first attempting to resolve the issue is unacceptable; see Wikipedia:Resolving disputes#Avoidance, "Do not simply revert changes in a dispute."

Could we please record something like this in policy? As currently worded, the policy creates too much second mover advantage, and does not do enough to discourage controversial use of sysop tools. Jehochman ^Talk 22:11, 27 January 2008 (UTC)

Works for me. FT2 ^{(Talk | email)} 09:49, 28 January 2008 (UTC)

Two other quotes jump out from the archives of when this policy was formed (/Archive 4):

Whoever reverses an admin action is responsible for any problems that result.'
Some wheel wars are conducted by admins who each believe they are Wikipedia's sole defender against some threat. An admin needs to remember that he does not stand alone against the forces of chaos; he can enlist the aid of another admin. (which I'd re-edit as "...he can present the matter to other, uninvolved, admins")

FT2 ^{(Talk | email)} 10:19, 28 January 2008 (UTC)

Do note that there are a variety of admin actions, unrelated to disputes between admins, that can and should be reversed without discussion, such as in ordinary maintenance. See above/archives for discussion and examples of this, which is why the policy does not simply say that the reversal of any admin action requires discussion. —Centrx→talk • 06:53, 24 February 2008 (UTC)

Yes, given admins have got tools on trust other admins reverting what we do if we disagree seems to me to be totally ok. For a lot of cases like blocks etc there is an enshrined process for doing this. I would rather not have my mistakes as an admin linger. There are courtesy issue s about telling me etc of course.--BozMo talk 07:05, 24 February 2008 (UTC)

The blocking policy is already clear about seeking further discussion before reversals, I think this should apply to admin acts in general. Reversing each other's action is fine, but not without proper communication. Without discussion it is just edit warring with big tools. _{1 != 2} 14:14, 8 June 2008 (UTC)

No. Read the perfectly legitimate examples of reversing admin actions at section #Once is not a wheel war? above. —Centrx→talk • 18:21, 8 June 2008 (UTC)

I certainly did not mean it should apply to non-controversial reversals like undoing an old page protection. But I do think there is way to much "I think your admin act was wrong so I am just going to reverse it without talking to anyone because I know I am right" going on. Not wheel warring, but not healthy to the Wiki either. _{1 != 2} 18:56, 8 June 2008 (UTC)