User talk:Biblioworm/Desysopping proposal

The problem

This proposal would lead to major abuse with a consequent withdrawal of admins from all contentious areas. Have a look at the block log of some active admins and estimate how many haters they generate in a month, then contemplate how anyone can create as many accounts as they like, do ten trivial edits and then start attacking the admin. They would tick whatever boxes are required, then pile-on with a wall of meaningless diffs with claims that would take an hour to properly evaluate, and longer for a response. Consider WP:GS/GG where lots of inactive users have returned to Wikipedia in order to tell the world how the gamers are correct and the female victims are to blame—they would love a system whereby they could tie the admins defending the encyclopedia up in knots, and there would be never-ending disputes where the admin would have to spend three hours every week doing nothing but defending themselves against POV pushers. The number of "good faith" pro-gamergate editors is much larger than the number of good editors who oppose them, and any admin taking action against a pro-gamergate editor would be voted out immediately. The situation at WP:AE is similar—only a handful of admins patrol there and there are dozens of people who hate them. There are no examples of an admin needing a desysop where the matter has not been handled by Arbcom (there is one pending case which will resume when the admin returns—that is a very reasonable outcome). Johnuniq (talk) 00:43, 4 December 2014 (UTC)[reply]

This is the very reason why I have implemented a protection mechanism. For the official voting to take place at all, at least four people must agree (the original person bringing the case, and the three uninvolved certifiers). After this, a majority of users must agree that the admin should be desysopped. Sure, ArbCom could always take care of this, but the process is slow, and the community has little say in the final result. My philosophy is that if you have to ability to give something, you should also be able to take it away. --Biblioworm 01:01, 4 December 2014 (UTC)[reply]

Again, there are no examples of a case which was handled slowly and where there could be any expectation that the community would have handled it faster. Those who have paid attention to the gamergate issue know that there are lots of returned users who could easily declare themselves to be uninvolved. One such returned user had his admin bit removed due to inactivity, then regained it on return. They coordinate off-wiki and could easily get four users who have old accounts at Wikipedia to return and follow the songsheet to harass admins who monitor the topic. The gamergate articles must follow the standard policies, so the fact that there are more SPAs and POV pushers than good editors does not matter (apart from the pain it causes). However, they would have well over 50% of the voters at any forum to desysop an admin who impedes their path, and they coordinate off-wiki. The same applies to several other contentious topics such as WP:ARBPIA where there are numerically more POV pushers on the I side—they would stop any admin who acted against them at WP:AE. Johnuniq (talk) 01:33, 4 December 2014 (UTC)[reply]

(edit conflict) Quite honestly, there's a huge host of "terrible" problems that you could find with anything around here, and I won't be able to give an answer to every single possible exploit that comes up. For example, I could say that RfA could be exploited, as a POV pusher could pass RfA if enough supporters of the POV were canvassed to support the candidate. Or, ArbCom might decide to form a dictorship and impose draconian sanctions for minor violations (they do have the ability to make binding decisions, after all). Or, ...well, I think you get the idea. Sometimes, I think we need to try out something new and stop blocking every proposal because there's a loophole that might lead to abuse of the procedure. Any procedure that exists here can (and will) be abused in one way or another. This system might turn out to work out really well, or it might not. If it works, great! If it doesn't work, we can always remove it. --Biblioworm 02:05, 4 December 2014 (UTC)[reply]

Biblio, if this gets to be official (I hope it does), you should have your own day. Because if you don't have the "certified" users in the proposal, then people will say "a bunch of haters could get together and all vote for a desysopping", if you do have them, then people will say "it's too much unneeded bureaucracy and corruption could easily happen". It's not only impossible to please everybody here, it's impossible to please anybody. --AmaryllisGardener ^talk 01:52, 4 December 2014 (UTC)[reply]

Discretionary range

I strongly believe if such a system is put into place that the system should have both a discretionary range and that other early close types similar to what is allowed at an RfA be allowed. The reason that a discretionary range is important is first because admins will make enemies if properly fulfilling their duty and such things need to be considered when judging consensus. Likewise in highly fractionalized cases it can work in the opposite direction. For example what end of a hard boiled egg should be cracked? An biased admin that abused their powers to support the small end, might have a large group of like minded supporters even if they really crossed the line. By having a discretionary range, such issues can be considered. It also allows the gravity of the issue to be taken into account. The other bit of discretion should be the ability to deal with things like closing improperly certified cases, or snow closing if certain extreme cases. For example, consider an admin that voluntarily returned their bit ( which of course would likely "under a cloud" ) should in result in the discussion being closed immediately and the bit withdrawn. Likewise, consider an arbcom case being open involving the admin and the issue; in some of those cases it would be best to suspend or stop this process as arbitration can deal with more than just the admin bit including blocking or banning involved parties. PaleAqua (talk) 03:10, 4 December 2014 (UTC)[reply]

@PaleAqua: Perhaps the discretionary range could be 50-60%? --Biblioworm 03:42, 4 December 2014 (UTC)[reply]

Reasonable to start with something small like that, which can be adjusted over time in practice. In theory an admin that got even near but below 50% number ( ignoring other biases as mentioned earlier ) might be too controversial to stay an admin. I can see admins that get only like 40-50% or so of the calls for desysoping in some circumstances to still probably need to loose the bit and I can see really good vandal fighters / arbcom enforcers getting 60-70% even though they are a great benefit as an admin. Hard to tell in advance what will happen. Such a system might help future RfAs, but really needs good safeguards to prevent it from being used to drive off good admins. To be honest, I'm not 100% convinced that this well help with the new admin shortage and there is a chance it could backfire. While there are hat / role collectors, for the admins that the project really wants I imagine many view it as a chore albeit a noble one. Passing the RfA is difficult / exposing. Getting challenged when making hard decisions is difficult. Having responsibility is difficult. And now having to worry about random vandals making you go through the motions of a desysop procedure is difficult. The benefit of an easy desysop procedure is making !voters more comfortable electing less qualified admins. The hazard I see is that it might make editors that would make good admins less likely to try. It's really going to be a tricky balance to make a system that is easy/quick enough to make the community more supportive and yet not too onerous to potential admins that they don't even try. PaleAqua (talk) 04:27, 4 December 2014 (UTC)[reply]

Doing your homework

At a minimum, editors who wish to propose a new desysopping scheme should read Wikipedia:Requests for de-adminship, and the preceding proposals linked therefrom. Otherwise, they risk wasting a tremendous amount of time – their own and others' – reinventing a square wheel.

As a freebie, I'll even point you to Wikipedia:Community de-adminship/RfC#Flaws in this process noted by TenOfAllTrades. While the current proposal isn't exactly the same as the previous proposal, it's close enough that most of the same problems apply. TenOfAllTrades(talk) 03:56, 4 December 2014 (UTC)[reply]

Thanks for linking me to those pages, but I don't think we should quit just because we haven't yet found the right way to go about it. Put bluntly, I have no plans to give up. --Biblioworm 04:11, 4 December 2014 (UTC)[reply]

Your chances for success will be greatly increased by doing as TenOfAllTrades suggests; read the other proposals and try to discern why they failed. What may seem like a great idea may have had several opportunities in the limelight and been soundly trounced for excellent reasons. Proposing such ideas, or close variants, will very likely lead to the same trouncing. --Hammersoft (talk) 15:57, 4 December 2014 (UTC)[reply]

I've spent many, many hours reading a lot of historical proposals (but I am still reading). From what I can see, most of the proposals similar to mine were brought up quite a few years ago. Consensus can change, especially with the way RfA is now. I don't think there's any harm in trying again. It's also worthy to note that the concept behind this proposal got a large amount of support when it was on WT:RFA. --Biblioworm 16:58, 4 December 2014 (UTC)[reply]

The basic concept did. But, there was very little in the way of details. If you provided the actual details and ran it again, I would guess it would be a significant failure. As I said there, people in general want some sort of system. But, when you get into the details it gets tricky. Very tricky. --Hammersoft (talk) 19:53, 4 December 2014 (UTC)[reply]

Could admins take on the role of certifiers?

Although this question was noped in the previous discussion, I think it's worth asking again. 1) Although there's a question of whether it's clique-y, if you can't find two or three admins to certify a request as legitimate, there's not a snowball in a gasoline suit's chance in hell of the request being successful. 2) There's an enormous (but I'll suggest wholly specious) concern that the process will be tilted against the admin in question by people they've blocked/whatnot in the past, this is strongly alleviated by requiring admin certifiers. Wily D 11:35, 4 December 2014 (UTC)[reply]

Just a thought, throwing it out there (haven't thought this through myself, just want to put it out there); Why do we need certifiers? Ok, to stop frivolous cases. I get that. But, we don't need certifiers for RfA. Frivolous RfAs are removed per WP:NOTNOW, and sometimes per WP:SNOWBALL. This might actually be preferable. If a user gets angry about an admin's behavior, and doesn't get their case certified, they may think the system is tilted against them. Without certifiers, they 'have their day in court'. If a case is truly frivolous, it'll be run out of town anyway. Thoughts? --Hammersoft (talk) 15:55, 4 December 2014 (UTC)[reply]
- The short answer is that a lot of people are concerned the system will be strongly slanted against the admin in question (for instance, see the concerns linked in the section above). Neutral certifiers are supposed to ensure the system isn't abused. Now, I think these concerns are pretty nuts; my guess is that the system will be overwhelmingly stacked in favour of the admin, like cops getting dragged in front of a grand jury. But both suppositions are guesses, so a bit of caution ain't necessarily bad. Wily D 08:43, 5 December 2014 (UTC)[reply]

Community has authority over admins they do not use

As it stands the community is allowed to come to a consensus to block or ban any user including admins. If someone proposes a block because an admin acted contrary to community expectations and a consensus formed for this block then that would happen.

The community rarely if ever does this. Why create a new process when the community already has access to regular enforcement? We should not be treating admins as a special class of people who need special enforcement. The community already has authority over them.

The fact is that the vast majority of the time when someone goes to a noticeboard to complain about admin abuse the community either finds it is not abuse or they do not find it actionable. Always the option to take action is available. There is no "don't block admins" or "don't ban admins" policy. It is not done because admins generally don't act in a fashion that calls for action.

Where are all these admins that are acting with impunity that this is supposed to solve? Until someone actually points out a specific problem that is not being addressed by the current system then this is a solution in need of a problem. Chillum 16:59, 4 December 2014 (UTC)[reply]

The big driver I've seen for a desysoping procedure is that it might indirectly increase the number of candiates that try for RfA on the theory that !voters would be more accepting of conditional candiates if they felt bad choices could easily be corrected. I.E. increase the perception that making someone an admin is no big deal. In my opinion the best hope would be a low-stress balanced process that is well known for comfort reasons of RfA !voters, but either almost never actually used or has enough safe guards ( including SLOW closes not going through the motions in cases that clearly won't result in a desysop ) to avoid discouraging admins from dealing with difficult problems. PaleAqua (talk) 19:31, 4 December 2014 (UTC)[reply]

Well, you said at the parent discussion, I would support an administrative conduct noticeboard where issues can be reported and discussed, the discussion there could be used to indicate to arbcom if the community wants the matter pursued. Also there is no rule against admins being blocked, topic banned, interaction banned or just plain banned by the consensus of the community. (Chillum 8:01 pm, 2 December 2014 (UTC−8))

Also, Dirtlawyer1 said there, The community needs a vehicle by which the community, not Arbcom, may desysop administrators in whom the community no longer has confidence.

Finally, Mellowed Fillmore said the community elects admins and the community should be allowed to remove admins as well. Even if this proposal fails, I would recommend the creation of a page where users can recommend that an admin be given a non–binding vote of no confidence.

I think these were the highlight of the discussion. Let's say admin BadAdmin makes a serious and basic numerical calculation error and is so stubborn that he refuses multiple request to acknowledge it or explain it away, despite overwhelming, clearly presented and unambiguous evidence. (Serious in the sense that it leads to a temporary block.) That sort of stubbornness, if part of a pattern, would IMO, be solid grounds for de-adminship. But there's no place for such an issue to be logged so that a pattern can emerge. This is "a specific problem that is not being addressed by the current system". I can create Wikipedia:Sockpuppet_investigations/BadAdmin if there's evidence, and it'll get archived. But if I were to create Wikipedia:Admin Conduct Notices/BadAdmin and log such an issue there, will that get deleted? Perhaps we should create the structure, e.g. Wikipedia:Admin Conduct Notices and just be BOLD. --Elvey^(t•c) 02:33, 5 December 2014 (UTC)[reply]

It is true that the community has the authority to block or ban anyone, but just because an admin is displaying poor judgement doesn't necessarily mean that he should blocked. In the same way, a simple block would not always be sufficient. Blocks and bans are not substitutes for deysopping a bad admin that is otherwise a useful contributor. --Biblioworm 03:06, 5 December 2014 (UTC)[reply]

There are also topic bans, interaction bans and general sanction that the community can levy. If Mr. BadAdmin kept poorly interpreting CSD the community is in their rights to ban the admin from using CSD as a reason for deletion. We should use existing measures to deal with admins instead of treating them like some special class that requires some special process to deal with. Treat them like any other problem user. Chillum 07:32, 5 December 2014 (UTC)[reply]

Again, are there any actual current examples of problematic behavior by admins that the current systems cannot deal with? This problem is often claimed but I rarely see real examples and when examples are presented it is normally sour grapes rather than actual abuse. Chillum 07:35, 5 December 2014 (UTC)[reply]
- Even if there was, do you really think I would say it out loud, right here, where the whole wikiWorld can hear it? I'd also probably find myself blocked for supposed personal attacks. --Biblioworm 15:56, 5 December 2014 (UTC)[reply]

Not if you provided evidence. There is no "protect the admin" conspiracy where we block people for bringing up legitimate grievances. I find the suggestion that admins are so corrupt that they will block anyone who critizes them despite merit to be a little out of touch with how things run.
You say "Even if there was" which makes me wonder if you even have a single example. Since you joined in August have you seen any examples of admins acting with impunity while the community had consensus for action?
You are not alone in failing to provide an example, I ask this question every time someone proposes a new de-sysop process(about 5-10 times a year) and I never get an example and it never passes. I never get an example because the really problematic admins have been desysoped by arbcom or they improved their behavior after the community pushed them. I find your answer to be a cop-out, if you want people to take your solution to a problem seriously you need to show the problem is real. Chillum 17:18, 5 December 2014 (UTC)[reply]

There is a distinct difference between personally insulting someone and calling their behavior into question. Further, you do not have to use current examples. A good starting point might be to review the cases brought before ArbCom where administrators were removed from administrator status. Dig into the histories leading up to the cases. I find myself in strong agreement with Chillum. Problem solving requires identification of a problem before you seek to solve it. If that is not done, you have no metric on which to assess success of the solution. --Hammersoft (talk) 16:02, 5 December 2014 (UTC)[reply]

Actually I would like a current example. 4 years ago there were 4 admins I thought should not be admins. Arbcom has de-sysoped 3 of them and the remaining one has improved their behavior to the point where their behavior is appropriate of an admin.
If an admin was a problem and then stopped being a problem then the community successfully exerted pressure and the system works. I want to see example of current failures, past examples do nothing to demonstrate an existing problem.
The problem with a popular vote de-sysoping system is that in my 8 years I have blocked thousands of users and about 70% of them thought it was unfair, in exactly zero of those blocks(or any other admin action I took) was there a community consensus that I had acted contrary to policy. Any popular vote will be helplessly biased towards ax-grinders and doused with sour grapes, what is worse the arguments will not have to rely on actual policy or community expectations.
De-sysoping needs to be done based on interpretation of policy as it relates to the communities expectations of admins, not a vote. Enforcing community consensus results in the cumulative spite of countless people who found themselves on the wrong side of consensus, under the proposed system I would be afraid to do my job. Chillum 17:42, 5 December 2014 (UTC)[reply]

This statement under the proposed system I would be afraid to do my job specifically is one of my big concerns. The only reason I could support a reconfirmation RfA like thing is if it's indirect effect is to increase the number of viable RfA candidates. The more I think about it though I fear that the impact will be exactly the opposite. Of the problematic admins that I've seen is similar to Chillum, most have been dealt with though other channels. To be honest I'm beginning to think ( and mostly in jest ) that the best way to increase the number of good RfA candidates is to have a pool of unlikely to pass RfA reforms that a bot randomly changes a few words and then post somewhere as a suggestion whenever the number of RfAs is too low. Followed by a bunch of people that echo chamber it so that it gets visibly discussed before ultimately failing. Seems like every time an unsuccessful proposal is made the number of good RfAs go up. :) PaleAqua (talk) 18:23, 5 December 2014 (UTC)[reply]
All right, fine. There are no current examples (with the exception of DP). However, there have been in the past, and there is nothing wrong with planning for the future, is there? --Biblioworm 18:26, 5 December 2014 (UTC)[reply]

If there is a demonstrable problem in the future I will support a promising solution. Until then, if it ain't broke, don't fix it. Chillum 18:31, 5 December 2014 (UTC)[reply]

And yet, you said at the parent discussion, I would support an administrative conduct noticeboard where issues can be reported and discussed, the discussion there could be used to indicate to arbcom if the community wants the matter pursued. Also there is no rule against admins being blocked, topic banned, interaction banned or just plain banned by the consensus of the community. (Chillum 8:01 pm, 2 December 2014 (UTC−8))

Have you changed your mind? If so, why?--Elvey^(t•c) 03:04, 6 December 2014 (UTC)[reply]

Let's give this proposal a trial period

Instead of trying to implement this permanently, I think we should ask for a three case trial period when this proposal is brought to the village pump. This way, the community can get an idea of how it would actually work without the danger of irreversibility. If it is working well, we keep it for good. --Biblioworm 15:53, 5 December 2014 (UTC)[reply]

No. Look, please take this in the spirit it is intended. This proposal is NOT ready for primetime, even for a trial. There is considerable ground work that has not been done. If step 100 is the point at which a trial might be a good idea, we're somewhere around step 14. There's a long, long road ahead. If these steps are not taken, I guarantee the trial will fail, and fail badly. Do yourself a favor; do the homework, do the ground work, and make this thing sing. --Hammersoft (talk) 16:05, 5 December 2014 (UTC)[reply]
- I never said that I thought this was ready for primetime. I'm just saying that when it is, we should go for a trial period first. --Biblioworm 16:35, 5 December 2014 (UTC)[reply]

How exactly does that work? Do we use the trial process to de-sysop some people and then see how we like it? If we don't adopt it when the trial is over do we re-sysop those effected by this test? I think the best you can hope for is to run some "mock trials" with no authority and see how much of an ax-griding shit fest it turns into. Chillum 17:23, 5 December 2014 (UTC)[reply]

Although I think this is a bad proposal, that sort of trial run is worse than nothing. More details would be needed as to the nature of the "trial period" to determine whether it would be significantly worse than just implementing the proposal, which may or may not be horrible. — Arthur Rubin (talk) 16:32, 8 December 2014 (UTC)[reply]

Actually, it could be worse than nothing. To date, no trial has been triad of any system similar to this one, though many have been proposed. If one IS tried, it could become a monument forever referenced in future discussions as to why this should not be attempted. If a trial is going to be attempted, at least some effort needs to go into fleshing out the idea in full first. There are considerable obstacles right now to this system, with a great many unanswered questions that need to be resolved. As is, it will fail, and fail hard. --Hammersoft (talk) 16:36, 8 December 2014 (UTC)[reply]

Why a trial won't/can't work

I've deliberately waited until there was a lull in this discussion before chiming in. As anyone can see, very often taking such a discussion away from the generally very noisy page at WT:RfA has some distict advantages and I would like to thank Biblioworm for starting something that is not as daft as it sounds. It will not succeed of course, but it has generated some serious discussion, with particular emphasis on the excellent input from Hammersoft, from a non-admin's point of view, and Chillum's as an admin. Surprisingly for some (but not for me) their views both on desysoping and trials are in concert with much of what I have been loud-mouthedly saying for a long time.

There are two basic tenets of a requirement for desysoping:

One upheld by a small but distinct and partly cohesive group that by default considers all admins to be badmins who seek to disrupt the work of established, prolific content creators by stalking and persecuting them. By no mean coincidence however, these users also happen to be the ones who are frequently the targets of criticism or even sanctions for uncivil behaviour, personal attacks, and/or simply using socially unacceotable words or expressions. Their long-term goal is to force the collapse of the entire Wikipedia system of management by its elected peers.
One larger contingent of the community which contends that while there may well be some admins who in spite of the high bar practiced at RfA, manage to get the mop and then either use their 'powers' less intelligently than expected, or are discovered to be quietly abusing some of the fundamental policies they are expected to promote and uphold such as COPYVIO, Socking, Civility, and paid advocacy. This consensus admits that such occasions are rare, but that a more efficient desysoping mechanism needs to be put in place.

Discounting the first group, data clearly demonstrates that 'drama' desysopings are very rare: desyoped for cause, and then a few who resigned not entirely without a cloud. So rare in fact that either we are not being sufficiently effective in bringing badmins to book, or that group #1 is indeed harming the collaborative spirit of the project by disseminating false claims of the magnitude of badminship. So rare in fact that in order for a trial to be of use, it would need to be based on literally dozens of desysops over at least a year or two. We don't have that many cases. And since the watershed year of 2007 when the community began to significantly tighten its criteria for admins and the turnout for voting increased dramatically, the number of candidates plummeted exponentially to a level where we are no longer even replacing attrition.

I don't have a silver bullet either (I tried hard enough in the 2011 project), but RfA is certainly a lot kinder to its brave contenders than it was then. Most importantly, I am relieved to learn that like here at Biblioworm's proposal, occasionally some intelligent dialogue is possible. We just have to keep trying to find a solution that has at least a 51% chance of being carried before it is launched at RfC. Such RfCs need extremely careful fine tuning. --Kudpung กุดผึ้ง (talk) 05:14, 21 December 2014 (UTC)[reply]

You're almost certainly correct that short of a miracle, this proposal will fail. I'm still planning on coming back to make a few changes after I'm done with what I'm in the middle of now. Once I'm finished with my changes, I suppose I'll throw it out there as a final Hail Mary pass, so to speak. After that, I'm planning on taking a break from the RfA reform madhouse and focusing more on content building, improvement, and review. Once my mind has freshened up a bit, I may come back with a few new proposals. --Biblioworm 05:54, 21 December 2014 (UTC)[reply]