Wikipedia talk:Wheel war policy/Archive

Back when I inforced the 3RR people pulled my blocks all the time. I never saw a problem with this as long as they told me and gave at least some reason (normaly that the person had contacted them by email and agreed to behave from that point on).Geni 03:03, 7 February 2006 (UTC)

Maybe we can define this as falling under "reasonable discussion", although it would be best if they asked permission instead of just informing you. I dunno. — Phil Welch 03:17, 7 February 2006 (UTC)

Nah why should they? If they want to take responcibilty for the behaviour of a user I don't view it as my problem.Geni 03:28, 7 February 2006 (UTC)

Fair enough. You should be able to specify that on your userpage and it would be considered permission to unblock someone you've blocked. (Therefore automatically granting the consensus necessary under this policy.) — Phil Welch 03:35, 7 February 2006 (UTC)

Safeguards

The policy, as it stands, has a pretty obvious flaw: it leaves no room for reversing genuinely bad-faith actions. Thus, User:HypotheticalRogueAdmin can safely block Jimbo and the Arbcom and vandalize the Main Page, since anyone daring to reverse his actions gets desysopped for three months.

On the other hand, any policy that makes exceptions will be open to interpretation, and thus fail to entirely prevent wheel-warring.

Something of a no-win situation, in my opinion. —Kirill Lokshin 03:19, 7 February 2006 (UTC)

Blocking Jimbo and the Arbcom is such an absurd situation that no one would enforce the wheel warring ban. Any genuinely bad-faith action less than that would quickly receive consensus/arbcom/Jimbo clearance to revert almost immediately anyway—not to mention that an obviously rogue admin would be immediately desysopped anyway, if he went on a rampage with the clear intent of locking the DB or somesuch. — Phil Welch 03:28, 7 February 2006 (UTC)

Given that we're already seeing people wheel-warring with Jimbo, I'm not sure if I share your optimism on that ;-) —Kirill Lokshin 03:30, 7 February 2006 (UTC)

This proposal addresses that too: Jimbo automatically wins any wheelwar :) — Phil Welch 03:32, 7 February 2006 (UTC)

Fair enough. Perhaps add a note that Jimbo/the Arbcom/whoever can reinstate anyone desysopped under this policy if they feel their actions were necessary? (There doesn't seem to be any more mention of re-sysopping after the removal of the three-month delay, but I assume it would ordinarily go through RFA?) —Kirill Lokshin 03:46, 7 February 2006 (UTC)

Most of this is addressed in my 1RR edition. Thanks for your feedback. — Phil Welch 04:21, 7 February 2006 (UTC)

It still fails to account for misbehavior. In fact, I would argue that it's even worse. In this revised version, User:RogueAdmin deletes a article clearly out of process (or restores it, or blocks someone, etc.). User:GoodAdmin undoes the action. User:RogueAdmin does it again. Now no one can undo the misdeed until either the Arbitration Committee, Jimbo or the Wikimedia Foundation Board of Directors get around to finding it, investigating it and passing a formal decision. User:RogueAdmin has succeeded in holding the entire project hostage for an indeterminate period of time to his/her particular point of view. Note further that there is no penalty for User:RogueAdmin under this policy, either for the first inappropriate action or for the revert.

People complain that our current processes don't scale. This dependence on such a small group will be impossible to sustain even at current levels of activity, much less as we continue to grow. Rossami (talk) 05:19, 7 February 2006 (UTC)

No, User:GoodAdmin can restore it as soon as he, User:GoodAdmin2, UserGoodAdmin3, and User:FriendlyNonAdmin all agree that User:RogueAdmin has gone rogue. This still allows for consensus decisions (duh). — Phil Welch 05:51, 7 February 2006 (UTC)

If consensus were that easy to achieve, it is unlikely that the wheelwar would have started in the first place. Your existing structure locks in the most recent action during the dispute (rather than the more conservative approach of locking in the status quo from before the dispute started). The rogue admin has little incentive to participate in real discussion. All he/she has to do is be obstructionist enough to polarize the discussion (perhaps by adding snide comments at the end of comments). If I can successfully polarize the discussion, I will be able to credibly argue that consensus has never appeared. (This is because except for a few specific actions like deletion discussions, there are few clear standards for declaring when rough consensus has been achieved.)

In particular, I am concerned that having the explicit appeal to authority available makes it too tempting to polarize the discussion. I don't have to seek consensus because I can drag this out until Jimbo gets involved. I don't see that as a good thing. Yes, the easy cases will be solved quickly but the tough cases will be dragged out forever by whichever side's position is currently "locked in." Rossami (talk) 13:03, 7 February 2006 (UTC)

Given the casualty rate following Jimbo's involvement, though, I suspect waiting for it will not be a particularly popular option. —Kirill Lokshin 17:36, 7 February 2006 (UTC)

Mentioning to a person or two that you are leaving an article that has been abominated can be effective in getting it fixed. It requires accepting that stuff you care about sometimes will stink, and not just for a short time either. I'm not sure this is relevant to your discussion, but I felt like adding it. Metarhyme 04:28, 8 February 2006 (UTC)

Re-sysop clause

I don't think "4. Administrators desysopped under this policy must wait three months before starting the process to regain admin status" is necessary - an ex-admin that has been de-sysoped either is unlikely to get re-promoted through AFD RFA if he tries too soon, or else must have had extroadinary behavior to warrant an exception. —Quarl ^(talk) 2006-02-07 03:24Z

I assume you mean RFA? ;-) —Kirill Lokshin 03:24, 7 February 2006 (UTC)

Yes, of course :) —Quarl ^(talk) 2006-02-07 03:45Z

WP:WHEEL

Phil, do you realize we already have Wikipedia:Wheel war? SlimVirgin ^(talk) 03:27, 7 February 2006 (UTC)

Not only do I realize it, I wrote this proposal specifically as an amendment to it, as it clearly says: "The following is proposed as an addition to Wikipedia:Wheel war, and would be considered policy. Please edit this proposal." — Phil Welch 03:29, 7 February 2006 (UTC)

Okay, sorry. I should read more carefully before commenting. ;-) SlimVirgin ^(talk) 03:41, 7 February 2006 (UTC)

1RR Edition

In response to numerous concerns, I've posted a 1RR edition that's simpler and in my opinion better. — Phil Welch 04:17, 7 February 2006 (UTC)

Both versions contain the clause "1RR is strictly enforced on protected pages." This seems to me to be somewhat pointless because it would be so easy to game. Page protection and unprotection are fairly routine actions which can be taken by any admin. If a page is protected, I can unprotect it, confident that few people will notice. (Who, after all, has time to police the protection logs? And even if they do, it's too easy to argue that it was appropriate for the protection to expire. Remember that page protections are not generally supposed to be permanent but that they do not automatically expire like blocks do.) On the other hand, if the page is now unprotected, I can lock the page, sure that even if reverted, I get the last say and can keep it locked until there is a formal review by those very scarce resources (ArbCom, Jimbo, Directors).

I think you're trying to legislate civility. I can't think of a single example in any context where that's worked. People need to be led to civility. Rossami (talk) 05:41, 7 February 2006 (UTC)

You're severely misreading it: you can revert back as soon as you have a consensus to do so. — Phil Welch 05:53, 7 February 2006 (UTC)

Replacement proposal

Thinking about your underlying goals, I would propose a single, simpler rule.

No one may make a sysop action more than once in any given dispute. That is, all admin actions are subject to the 1RR. If your action is reverted, you may not re-revert it - you must either discuss it or allow some other admin to take the action.

For contentious decisions with lots of participants, this version could extend the dispute a bit but only until one side or the other runs out of people willing to stick their hand into the buzzsaw. I think this has a lot less potential for abuse. I think it would also be more effective at drawing the participants into discussion rather than foisting the dispute onto the ArbCom. Rossami (talk) 05:58, 7 February 2006 (UTC)

How does my version not draw the participants into the discussion? The only reason the Arbcom et. al. clause is even IN there is to allow them or Jimbo to put their foot down if something like the {{user pedophile}} thing happens again. — Phil Welch 07:03, 7 February 2006 (UTC)

See above about how appeal to authority tends to polarize the discussion. Jimbo et al already have the right and ability to insert themselves into the dispute in extreme cases. That executive authority should remain as it is - an unwritten rule rarely exercised. Rossami (talk) 13:06, 7 February 2006 (UTC)

OK, I can see your point. Does my current wording help any? — Phil Welch 16:30, 7 February 2006 (UTC)

Some but the current emphasis on "per-side" counting (as opposed to individual responsibility and accountability) is, in my mind, a fatal flaw. I see it as also more likely to polarize the discussion than to help resolve it. I'm really struggling to understand your insistence on this point. I've not seen any isgnificant examples of wheel-wars that had more than a few active participants. One-revert per admin per dispute is simple to understand and simple to administer. I continue to think it will drive the participants to discussion fastest. Rossami (talk) 21:18, 12 February 2006 (UTC)

In our latest debacle, five admins were deadminned over a single wheel war, and that was only one side. We have to do what we can to stop tag-teaming. — Phil Welch 23:22, 12 February 2006 (UTC)

Not a Board decision

I don't see a need for people apply to the Board to get any sort of decision on their adminship. That should be a completely community based decision and as long as two fifths of the Board are not community members, I'm not at all convinced they should be mentioned in this policy. Angela (disclaimer) 09:18, 7 February 2006 (UTC)

Alright, fixed. — Phil Welch 16:31, 7 February 2006 (UTC)

Too rigid

This proposal is too complicated and rigid. Between the mandatory punishments and the complex rules, the rules-lawyers will have a field day and everyone else will be too scared of being de-adminned to do anything. --Carnildo 00:23, 8 February 2006 (UTC)

How are the rules too complex, and how can they be simplified? — Phil Welch 01:52, 8 February 2006 (UTC)

I've put a simplified proposal up. --Carnildo 04:42, 8 February 2006 (UTC)

I hardly see how being vague will prevent rule-lawyering. We already have wheel-warring admins who say "No, I wasn't actually wheel warring!" Maybe not, but you did break 1RR, and that's enough for me. — Phil Welch 19:50, 8 February 2006 (UTC)

The 3RR allows for tactical editing: making edits that are not reverts, that you don't neccessarily agree with, and that you know the other editor will find absolutely unacceptable, with the goal of forcing the other editor to break the 3RR. Putting any hard limit on the number of times one administrator can undo the actions of another will allow for similar activities. --Carnildo 23:34, 8 February 2006 (UTC)

How exactly do you propose that? Editing a page provides a wide range of possibilities—undoing an admin action is quite binary, making it FAR more suitable to be curbed via an xRR. — Phil Welch 22:28, 11 February 2006 (UTC)

Status Quo should be default position

The appropriate solution, IMO, it to have the default position be the status quo. If AdminTheFirst blocks someone and AdminTheSecond unblocks.. it should stop there. Since the User started out unblocked, they should remain unblocked until consensus is reached. Benefit of doubt should lean towards curtailing sysop powers. --Dante Alighieri | Talk 07:44, 8 February 2006 (UTC)

No, WP:FAITH should apply to admin decisions (it would be nice to have a decree from Jimbo reminding people of this). Hence the decision of the first admin should stand unless there is a reasonable consensus that it was a bad call. Physchim62 (talk) 13:22, 8 February 2006 (UTC)

Physchim62 has it. That is exactly what I've been saying for a while. It's so simple, and would instantly eliminate wheel wars, but people have decided not to listen while wheel wars keep going on all over. I've expanded the idea below too. It's basically like the referee review procedures used now. The call on the field stands unless there is indisputable evidence to overturn it, which in our case would be clear policy and/or a rough consensus. - Taxman ^Talk 16:04, 8 February 2006 (UTC)

I don't believe this would magitastically eliminate wheel wars by any means, though I flattered if you do! The second admin should have the benefit of WP:FAITH as well, i.e. the first admin should consider carefully before reinstating the action (I prefer the 1RR solution to the "never change an admin's decision" Utopia being brewed up at ArbCom). But in the case of further dispute, the actions of the first stand while consensus is reached (consensus would be pretty quick for obvious bad calls). Physchim62 (talk) 16:22, 8 February 2006 (UTC)

FAITH is all good and fine when we're not talking about the blocking of a user. --SPUI (talk - don't use sorted stub templates!) 19:19, 8 February 2006 (UTC)

One certainly should assume that the admin who blocks a user is acting in good faith: not to do so would be a personal attack, and quite a serious one in my opinion. That doesn't mean that all blocks are correct—you can find evidence on my talk page of an incorrect block which I myself made recently—but nor does it mean that blocks are any different from other admin actions such as page deletions. No case for special treatment, nobody has a right to edit Wikipedia (except maybe Jimbo...) Physchim62 (talk) 19:41, 8 February 2006 (UTC)

I very much disagree. If you think undoing a block is a personal attack, you should not be an admin. Period. --SPUI (talk - don't use sorted stub templates!) 19:59, 8 February 2006 (UTC)

SPUI, I never said that! What would be a personal attack would be to say that I only blocked someone because I dislike them: in other words, that I used admin powers to favour a personal position in breach of policy. In the Canadian case, I thought I was within policy, and I unblocked as soon as I realised my mistake. I would not have considered any other admin undoing this block as an attack either–nobody's perfect! Physchim62 (talk) 20:05, 8 February 2006 (UTC)

Oh. Then you're assuming that anyone unblocking someone you blocked without discussion is doubting your good faith. --SPUI (talk - don't use sorted stub templates!) 20:24, 8 February 2006 (UTC)

WP:FAITH applies to the admin who decided to revert the action too, and that they saw an even stronger reason that admin action should not be used. Things should default to the natural state in that case, that is, until consensus says the admin action taking something out of the natural state should be required, it is not required. Blocking a user, deleting a page, or protecting a page, are all things that take the Wiki out of the natural state, which is: all users can edit all pages, and there's a good history of what happened. If admin action was obviously needed, then a consensus should be formed quickly, and the damage due to the possibility of a poor choice of admin intervention is minimized. --Mysidia (talk) 22:20, 11 February 2006 (UTC)

Another simpler proposal

How about instead of the complicated wording in the 1RR that is up now, how about this: No admin action should be undone unless: a) it was very clearly against policy (ie there is a policy explicitly prohibiting the original action) or b) a rough consensus has emerged to undo it.

We're obviously long past where we can assume everyone will use common sense as wheel wars are going on all over the place, even against Jimbo, which common sense would tell you never to revert his ruling. So the easy way is to never revert an admin action without consensus. The reason this should work is there are so few admin actions that are truly urgent. The ones that are, no one would dispute. Then reasonable discussion can come to a conclusion on overturning the action or not. We trust our admins for the most part so we should at least be able to let any of their actions sit for a few hours while it is discussed. This wording is more useful than Carnildo's version because it is specific and less confusing than Phil's. - Taxman ^Talk 16:00, 8 February 2006 (UTC)

No, I don't think that would work. Admin A honestly believes that a template is speedily deletable and so deletes it; Admin B honestly believes that a the template in question doesn't fall under any of the CSDs, and so its deletion is "very clearly against policy" and he restores. That is exactly how Sunday's wheel war started. 0RR is Utopia, and Wikipedia is NOT Utopia: that is why I prefer 1RR. Physchim62 (talk) 16:29, 8 February 2006 (UTC)

No, that's not clearly against policy. Clearly against policy is there is a policy prohibiting the action, such as unblocking yourself. If someone unblocks themselves they should be blocked again. There is no policy that says you can't delete according to the CSD. Insterpreting the rules different where an admin says "I wouldn't have made that block, so I'm undoing it" is exactly what we need to avoid. - Taxman ^Talk 18:30, 8 February 2006 (UTC)

It's not clearly against policy for you because you're a reasonable sort of chap who assumes WP:FAITH. For some admins, an out-of-policy speedy deletion is a heinous offense which would largely justify reversion (usually in the name of process), even if they can accept WP:AGF concerning, say, user blocks. As far as I can see, any version of 0RR simply does not cater for the known variety of (sincerely-held) variations in policy interpretation. Physchim62 (talk) 19:50, 8 February 2006 (UTC)

As noted above, 0RR also prevents routine non-wheel-warring reversions. If User:AdminOne blocks User:MisguidedKid for 3RR, and User:MisguidedKid promises to User:AdminTwo that he'll be a good boy, User:AdminTwo can unblock him. Usually, User:AdminOne doesn't really care, figuring that User:MisguidedKid is now User:AdminTwo's problem. On the other hand, if User:AdminOne has inserted himself into untangling a sticky situation (say an edit war) and User:AdminTwo starts interfering and reverting User:AdminOne's actions, User:AdminOne has every right to say, "I'm handling this, trust me on this", revert back, and continue working without interference.) — Phil Welch 19:58, 8 February 2006 (UTC)

Enforcement

Please, no automatic desysopping. This should be left to the discretion of bureaucrats, although we could have some guidelines here. I would suggest around 24 hours for a first offence and around 7 days for repetition. Anything more serious should go to ArbCom, but bureaucrats could use their discretion not to apply any desysopping at all if they felt it were appropriate to the circumstances (eg, the two parties had since reached agreement). Desysopping should be like user blocking: preventive of disruption rather than punitive of the user. Physchim62 (talk) 16:37, 8 February 2006 (UTC)

OK. I've amended to allow stewards' discretion (bureaucrats can only sysop, they cannot desysop) and to provide for a wheelwarring noticeboard similar to our 3RR noticeboard. — Phil Welch 19:46, 8 February 2006 (UTC)

Bad idea. Discretion only leads to the potential for uneven enforcement. --Dante Alighieri | Talk 08:20, 9 February 2006 (UTC)

Why the exclusion for editing protected pages?

Edit-warring over the interface is very disruptive; the ability to edit interface pages is one of the more dangerous privileges extended to admins. And repeatedly reverting a protected page defeats the purpose of protection in many cases. It seems to me like these privileged actions should be handled like any other. Demi ^T/_C 23:21, 8 February 2006 (UTC)

• Wheel warring over an image copyright tag, for example, could quickly overload the server. Physchim62 (talk) 23:32, 8 February 2006 (UTC)

We already have policies for edit warring. I wanted a simpler policy with a more focused scope. If you want you can float stronger anti-edit-warring policies for protected pages (stronger than 3RR) as another proposed policy, but I don't want to overburden this proposal, which is focused on use of non-editing admin powers. — Phil Welch 01:45, 9 February 2006 (UTC)

Alternative proposal

While i like having a policy that says "Don't do it!" I don't like the policy specifying what will happen if you do. Wheel wars are still rare. Let the AC or Jimbo or in the case of an emergency any passing steward decide. Theresa Knott | Taste the Korn 23:37, 12 February 2006 (UTC)