User talk:Amerique/Community recall

Foundation controls arbcom?

I noticed you mention the arbitration committee in the nutshell. Since they are appointed by the foundation, does the community even have the authority to impose recall? Wouldn't that be a foundation decision? Or perhaps you meant that the foundation could set up a better internal monitoring system? Can you elaborate? _{1 != 2} 21:51, 10 July 2008 (UTC)

I was thinking about that myself. I don't know about that yet. I would think members appointed without popular elections to AC could be exempt from a community recall process, but a rigorous process could be developed for recalling members for whom there is a numerical record of community support. Amerique^dialectics 21:58, 10 July 2008 (UTC)

Or at the very least the result of a recall would be a statement of "no-confidence" on the part of the community in any further "trust activities" on the part of the recalled user, rendering null any further pronouncements on their part. Amerique^dialectics 22:02, 10 July 2008 (UTC)

Perhaps the idea of recalling individuals on the Arbitration Committee should be considered differently from community-granted user-access-level considerations. The idea of involuntary recall, to my knowledge, hasn't been discussed in great depth before. What is your opinion? Amerique^dialectics 22:17, 10 July 2008 (UTC)

I could be wrong, but I didn't think that the Foundation had any part in appointing the ArbCom and that the community has just always had Jimbo Wales appoint them on a de facto basis. Greeves ^{(talk • contribs)} 22:20, 10 July 2008 (UTC)

I think the WMF would be bumping up against Section 230 if they messed around with anything like the Arbcom, possibly. rootology (T) 02:27, 11 July 2008 (UTC)

Do you mean Section 230 of the Communications Decency Act? I don't see the connection. Do you think telephone providers don't make their own rules, but instead let their callers make the decisions? Does that effect their protection from liability? Am I misunderstanding you? _{1 != 2} 14:54, 11 July 2008 (UTC)

Yeah, I agree. I can't see the foundation ever taking any kind of legal action against the committee, rather than simply dissolving it, so i don't see how the point is relevant.

Anyway, for this, I was roughly thinking that a community decision to go to recall could be based on a consensus determined at an RFC, and that the actual recall process could be a popular petition requiring, generally, at least two thirds of however many support votes were cast for the user to receive his or her trust position in the first place, with a simple majority of "no recall" votes overriding the process. For the recall to be effective, proponents would have to win the simple majority with a high proportion of whatever the initial support base was numerically.

Any recall of user access level privileges received through a RFA or RFB, and a "community declaration of no confidence" in ArbCom members, could be worked out along these lines, but I am not sure how to recall people with Checkuser and Oversight who were appointed by the AC without elections. This should definitely be looked into as a part of the current RFC; there seems to be no transparency or way of achieving accountability in the use of those privileges there. Amerique^dialectics 16:56, 11 July 2008 (UTC)

Sorry for any confusion... I meant that the WMF doesn't appoint or sanction or allow the Arbs and has nothing to do with them beyond identify verification, right? If the WMF took any role beyond handing down policy it could risk them losing their carrier/sec 230 protections. I can't see them doing it for that reason. Jimbo signing off on the elections while he's allowed by the local project isn't a WMF thing, it's a local Wikipedia thing, unless I'm way off base. rootology (T) 06:08, 12 July 2008 (UTC)

Rootology is correct. In fact, the English Wikipedia is the only one that has an Arbitration Committee selected in any way but by community decision, and some communities have decided to abolish their Arbcom (Italian Wikipedia apparently). The Foundation's only rule is that all members of Arbitration Committees must satisfactorily identify themselves to the Foundation. Risker (talk) 01:53, 14 July 2008 (UTC)

This would kill the encyclopedia

Think about what it means--to impose the rule of the mob upon those who have been chosen to perform often unpopular but necessary jobs. We must not take this path. --Jenny 22:22, 10 July 2008 (UTC)

I'm considering moving it back to my userspace for the time being. I really didn't know how to work on this by myself, though, so it might not go anywhere over there, but probably it should be left off the table while the RFC is active. Amerique^dialectics 22:25, 10 July 2008 (UTC)

Doing it outright might kill it, sure. It would depend how it happened, but if it happens, its our decision to go down the route of a system like that. We don't know if it would be good or bad until it came to pass (which it probably won't). rootology (T) 02:28, 11 July 2008 (UTC)

I agree the mob can only rule so much, and at some point needs to be told what is what. It is just that no other form of management has ever worked on more than a handful of people throughout the history of mankind, ever. _{1 != 2} 14:51, 11 July 2008 (UTC)

With my system, "the mob" would be more easily able to check a recall than initiate one. Besides, it is better for the community to have a more rigorous system for ensuring accountability on the part of people in trust positions than DR. The ArbCom has not lately inspired confidence in its ability to handle conflicts between high-profile administrators whose problems are well known in the community. Therefore, the community must have a legitimate channel to effect redress, on its own part, on issues that ArbCom can't or won't help out on, indeed in cases like Cla68-SV-FM wherein the individual disputants could have just (theoretically) presented evidence in an RFC for the community to discuss and decide whether anything warranted taking anyone to recall, which if a process had been in place for, all of this would have been long since over and decided by now. There is nothing to fear but fear itself! Amerique^dialectics 17:47, 11 July 2008 (UTC)

We are a community, not a mob, and it is insulting to all of us for anyone to label us as such [1]. A recall process as this proposal attempts to establish is necessary and this proposal appropriately sets a fairly high bar for removal or impeachment of an elected official. The community governs Wikipedia and therefore we as a community should be the ones to issue recalls as necessary or appropriate. Cla68 (talk) 21:47, 12 July 2008 (UTC)

Come now, lets be honest with ourselves, at times we are both. _{1 != 2} 14:53, 13 July 2008 (UTC)

Perhaps. But for any alarmist worries, given how hard it is to get a things to "stick" on even an admin, let alone another higher level, the odds of a recall going through barring anything extraordinary is unlikely. It would be self-policing, as written, I would think. Since everything on Wikipedia except Foundation rules is done by consensus (a Foundation rule itself) including elections and role appointments, this proposal would simply give the community a mechanism to change consensus. Currently the only thing that the community can't change consensus on is appointments of positions. This proposal seems to actually fix the system to be compliant with Foundation principles and rules, especially taken with the fact that I believe all the other "major" projects (Commons, Wikinews, Wikiquote, de.wiki, fr.wiki, it.wiki) have similar mechanisms in place. rootology (T) 16:34, 13 July 2008 (UTC)

Oh, and Tony, the rule of the mob chose them in the first place, and consensus changes. rootology (T) 17:29, 13 July 2008 (UTC)

A group of users can become a mob when some project page or newsletter tells them they must all troop over to some part of Wikipedia and act in unity immediately to prevent some action harmful to the common interest of the editors, or to punish some admin who has acted contrary to the wishes of the person inciting the action. Many parts of Wikipedia are thinly attended, with an influx of 10 or 20 editors having a huge effect on the vote totals. There are several interest groups on Wikipedia which favor some aspect of pop culture, some national, ethnic, religious, sexual or political identity, or some hobby. We should beware of "packing the jury" and allowing a "small mob," say of 10 editors, from punishing an admin for applying Wikipedia policies or guidelines. Edison (talk) 20:32, 23 July 2008 (UTC)

Good idea

It's certainly a good idea; I've thought for years that we need something like this. My proposal is simpler and less demanding: 10 established users (defined as having at least 500-1000 edits and several months on the project) requesting recall within a defined period of time means that the admin in question is subject to a new RfA, even if the admin is not voluntarily open to recall. The admin must then receive a certain percentage of support (I'd favor 55 or 60%) in the RfA to retain adminship. Everyking (talk) 17:07, 13 July 2008 (UTC)

Stuff like that with numbers for who gets in and clauses is just going to lead to fights and swamp this talk page like every other recall proposal ever. The drafted one is neat because it's so painfully simple and to the point. rootology (T) 17:28, 13 July 2008 (UTC)

I actually thought my proposal was simpler. Amerique's proposal says that at least two-thirds of the number of users who supported the candidate in his or her original RfA must support having the recall, while my proposal just requires that 10 established users request it within a defined period of time. I also think the two-thirds number requirement (both for having the recall, and for the recall itself) is so high that it would be almost impossible to ever remove an admin through such a process. Even our most deeply unpopular admins can probably muster one-third support (just think of all those who oppose the very notion of recall, like Tony, who would vote to support the candidate almost unconditionally, as well as those who think admins are supposed to make terrible, painful, unpopular decisions, and will therefore make huge allowances for bad behavior). We need a recall process that will actually function to kick out bad admins; this process, it seems to me, would merely create the possibility on paper, while in reality it would still be a virtual impossibility. Everyking (talk) 17:37, 13 July 2008 (UTC)

I can see tweaking the percentage, but I think the percentage technique is better, because it scales with growth (even though it has the paradoxical effect of making it easier to recall an old admin than a new one). 10 as a number is just too low, and prone to mischief. On any major controversy (homeopathy, television episodes, etc.), either side could easily raise 10 nominators for the slightest of infractions. Maybe a different measure, like 50% of the yes votes at the most recent RFA. That would scale, but keep the same bar for everyone at a given point in time.
Kww (talk) 17:45, 13 July 2008 (UTC)

Exactly. Just like how a good candidate at RFA blazes through and bad candidates get basically pounded in the rear, a good admin getting recalled will get that recall laughed away under the simple percentage system drafted, while a bad admin would get similarly pounded in the rear, as they should be. Nothing is a lifetime entitlement, consensus can change/everything local is decided by consensus (Foundation rules!) and the simpler it is to both curb abuse of the recall process, stop abuse by "bad" folks, and scale is best. Really egregious stuff is for the Arbcom to fire an admin, since they can just go and do it. Constant problems that aren't over the top, though, need something like this. If the community decides they don't trust someone anymore, consensus has changed. rootology (T) 17:47, 13 July 2008 (UTC)

Right, 10 established users is too low a threshold. If 100 users supported someone's RFA at one point, it should take at least 75 users to authorize a recall process. If 76 users don't agree, the process doesn't happen. If only 10 could do it against any opposition, anyone could be recalled over any bad call that gets a lot of publicity at ANI. Amerique^dialectics 17:51, 13 July 2008 (UTC)

(EC) Yeah. Unless someone does something mind-blowingly stupid or evil once off they shouldn't be fired as an admin for it, and the AC can deal with that. This is about demonstrating that the community consensus has changed and that someone shouldn't be an admin/arb/b'cat etc. anymore, due to ongoing actions that individually may not merit a deadmin, is how I read it. If someone was a great guy from 2005-2007 as a non-admin, and flew through RFA in Jan 2007, but from 2008 onward became a terrible admin but not enough for the AC to come in and take the tools, the community should have a way to follow Foundation principles to decide by consensus and change decisions. The Foundation requires this under all decisions being done by consensus, and local rules that no decision is permanent. rootology (T) 17:56, 13 July 2008 (UTC)

You just can't get those kinds of numbers. 75 people so worked up as to demand that a person be stripped of adminship? And you certainly couldn't get two-thirds of voters in an RfA to vote against an incumbent admin. An admin could be downright psychotic and survive under this process. Everyking (talk) 17:54, 13 July 2008 (UTC)

Well, it's never been tried before. If however many people got together to vote for someone at some point, two thirds of that number should be able to "get together" to initiate a recall if the user is egregiously bad. Making it too easy would allow the process to be used against people who may be good admins but only screw up once in awhile. There needs to be a high but not unreasonably high bar for recall to work. Two thirds of however many original supports should be a reasonable and sufficient number. Amerique^dialectics 18:05, 13 July 2008 (UTC)

I'd be willing to scale down the percentage of original supports required to initiate a recall to 50%, but no lower than that. (Striking per below, but 50% seems a good compromise.) Two-thirds support in a recall petition itself is reasonable, and would be more or less like putting someone up for RFA again, except that this time they are running to keep the tools instead of gaining them. Amerique^dialectics 18:22, 13 July 2008 (UTC)

Actually, as a simple majority overrides the motion anyway, the percentage required to initiate can be further scaled down, even to 20-25% of the original supports. Amerique^dialectics 20:14, 13 July 2008 (UTC)

You need 75-80% support to become an admin, wouldn't it take 75-80% to get someone recalled? _{1 != 2} 21:50, 13 July 2008 (UTC)

Yeah, for the terms of the recall petition itself. A motion to have a recall petition could proceed on a smaller consensus, I only put "one to two thirds" of the user's initial supports as a ballpark figure. Even if the number of recall supports exceeds the minimum threshold, the motion still could be defeated if superceeded by a simple majority opposed, so the minimum threshold to have the recall needn't be as high as the minimum threshold for the actual recall itself. Real recalls of state officials in the US are typically initiated on a similar "low threshold" signature requirement of "25% of the votes cast in the last election for the official being recalled." [2] The recall vote itself should only be considered sucessfull if it carries no less than 75% of respondents to the petition, just like how an RFA is run. Amerique^dialectics 00:02, 14 July 2008 (UTC)

Y'all might be interested in reading a similar proposal I wrote up a few months ago. Kurt Weber (Go Colts!) 22:32, 13 July 2008 (UTC)

Thanks for the link! Amerique^dialectics 00:02, 14 July 2008 (UTC)

This proposal stems in inspiration from this idea here: Wikipedia:Requests_for_comment/Arbitration_Committee#IV._Redress and this thread here: Wikipedia_talk:Requests_for_comment/Arbitration_Committee#Nuclear_Option. Amerique^dialectics 01:25, 14 July 2008 (UTC)

I really don't see why we need a whole new procedure, desysopings are uncommon events and arbcom has handled that unpleasant job very well. I really don't see a long list of cases where people who need to be desysoped are not having this done. _{1 != 2} 01:27, 14 July 2008 (UTC)

Why would any consensus decision be immune from change? That would be in violation of Foundation policy and how Wikipedia operates. rootology (T) 01:33, 14 July 2008 (UTC)

Yeah, in any case, RFCs against admins are pretty rare, but the option exists, and even if a motion to recall fails, defeating one would allow the community to reaffirm its trust in the user. And if a process like this was ever sucessfully used, it would save ArbCom the trouble of a huge, drawn-out, bitter drama-fest to accomplish the same thing. If the community gives the tools, the community should be able to take them away. Amerique^dialectics 01:57, 14 July 2008 (UTC)

Precisely. The community giveth, and the community taketh away. Frankly, it's my contention that the Arbitrary Committee has no legitimate authority anyway, so it has no business decreeing that individuals be desysopped. Kurt Weber (Go Colts!) 02:55, 14 July 2008 (UTC)

I honestly don't think 10 established users is such a low bar. You have to make serious, long-term users pretty damn angry for them to demand that you be stripped of adminship. I find it hard to believe that anyone who could accumulate such a number would really be performing his or her duties as an admin properly. Not to say they wouldn't deserve to remain an admin on balance, but there'd very likely be serious problems which would warrant community reevaluation. Everyking (talk) 07:40, 14 July 2008 (UTC)

The "trouble with ten" is that it doesn't scale. It would make it too easy for any so called "cliques and cabals" to initiate a recall against any admin who happened to make a call that pissed them off for some reason. To protect admins who are doing their jobs fairly, the initial threshold should be proportionate to the # of initial supports, not only because that number provides a non-arbitrary way of determining consensus for retaining the tools, but because it allows admins (and others) a necessary measure of protection from a recall based on however much initial support they had. Allowing any ten "established" users to initiate recall would essentially allow only ten to undermine any prior RFA consensus, leading to chaos and the kind of "mob rule" Tony was afraid of. Adopting a scalable process prevents that. Amerique^dialectics 14:46, 14 July 2008 (UTC)

I don't like it being a percentage of supports at the admin's election. I would prefer the percentage to be something like percentage of supports at the last successful RFA. The number of supports for an admin approved in 2004 is going to be much lower than the number of supports in 2009, and it doesn't make sense for it to be easier to recall an older admin than a newer one. I don't care much exactly what percentage of what is chosen, so long as it is the same bar for everyone.
Kww (talk) 15:42, 14 July 2008 (UTC)

It is written in the way it is currently, "the decision to initiate should be considered authorized by the support of at least 55 percent of the original number of supporters for the user at his or her last successful RFA, RFB or other public election," to allow the process to be used against sitting arbitrators. In cases where an admin has a much lower threshold of support, a simple majority of opposes is still all that would be needed to to override a recall, even if by only one oppose. The minimum threshold required would be scalable for everyone, but having a simple majority able to stop a recall at any scale, whether high (>100) or low (<10), means that no one would get railroaded through this process. Amerique^dialectics 16:09, 14 July 2008 (UTC)

Back when I became admin in 2003, there were about 10 people who voted in favor of letting me be an admin. Wikipedia has dramatically changed since then, so how would this work for long-time admins? WhisperToMe (talk) 23:40, 16 July 2008 (UTC)

We are still debating this, but my preference would be to use a high proportion of whatever the initial numeric consensus was to promote as a minimum number to qualify a recall, and allow a simple majority to overrule and reconfirm. Keep in mind that the only way that doing this would realisticly be considered is if someone had reason for launching an RFC. Frivilous motions to recall on no evidence would easily be defeated, using the same metric it would take to qualify. (In your case, if a frivilous RFC was launched, and 6 people supported a recall motion, a concensus of 7 would defeat it.)

Rootology's way, on the other hand, would use the systemwide average pass rate over the past 3 months to set an absolute minimum threshold, which at current levels would trend pretty high, for all users.

So far, this is all just theoretical, but I am considering asking some admins and bureaucrats to volunteer for simulated RFCs directed at testing these processes out. Until some benchmarks are run, we won't know if prior levels of consensus for individual users will hold or can be easily overturned today, or if high numbers can be achieved to "vote out" the most controversial users. But the overall framework needs more development before that happens. Gotta go, Amerique^dialectics 03:03, 17 July 2008 (UTC)

Hi Whisper, under the short method I wrote all admins are held to the "current" RFA standard, so old timers are protected against gaming or frivolous action. Read this section. :) rootology (T) 06:12, 17 July 2008 (UTC)

Clarification

"If a motion to recall is presented during the course of an RFC, the decision to initiate should be considered authorized by the support of at least one to two thirds of the original number of supporters for the user at his or her last successful RFA, RFB or other public election, within a two-week time frame from the moment the motion is time-stamped." Which is it? One third, or two? Or am I completely misreading this? In any event, I think this proposal deserves to be implemented, given its high threshold for withdrawal of the tools (I'm personally recallable if an RFC shows anything but a clear consensus for remaining an admin). And actually, if there came to be less stigma towards the de-adminned as a result of this, that would be a good thing - I think they should be treated more like politicians who have lost elections than like pariahs. Sarcasticidealist (talk) 05:22, 14 July 2008 (UTC)

It's a ballpark figure right now, for the minimum threshold of supports, which in any case can be superseded by a simple majority opposed. I agree, implementing this process would go a long way towards confirming that "adminship is no big deal." (I was think of settling on 55% just not to leave it so ambiguous.) Amerique^dialectics 05:30, 14 July 2008 (UTC)

More or less any figure's fine with me. My preference would actually be to leave it independent of the level of support in the most recent RFA, since that probably unreasonably disadvantages admins who got the bit in 2004 or so (though I guess the counter-argument is that their current community support is likely to be more open to question that that of an admin who broke WP:100 last week). Anyway, I commend you for trying to put this together, though, since I suspect it's doomed to failure, I'm not going to put much effort into defending it. !vote counters should count me as a strong support all the same. Sarcasticidealist (talk) 05:33, 14 July 2008 (UTC)

Thanks. I think Wikipedia is ready for it though. My assumption is that if there is no reason to take an admin to RFC there would be no reason for a recall, and if a guy got the bit in 2004 with, say 15 votes, there is no reason why a consensus of 20-100 shouldn't be enough to recall it if he's consistently messing up with it now. I'm waiting for the ArbCom RFC to be over before I start pushing it, but I think it provides the community with a new avenue of dispute resolution that should significantly lessen the drama that otherwise holds up ArbCom. Amerique^dialectics 05:57, 14 July 2008 (UTC)

55% of the supporters at RfA etc.

I think it worth considering what this means practically given the increased participation figures over time:

• A motion to recall me as a bureaucrat would require 94 people to support it.
• A motion to recall User:Kingturtle as a bureaucrat would require 8 people to support it.
• A motion to recall User:Daniel as an admin would require 128 people to support it.
• A motion to recall User:Ryan Postlethwaite as an admin would require 44 people to support it.
• A motion to recall User:Michael Snow as an admin would require 16 people to support it.
• A motion to recall User:Secretlondon as an admin would require 4 people to support it.

That all seems rather arbitrary to me. It leaves some of the longest serving administrators most at risk of frivolous recall petitions and in the case of those with very high support levels, makes it almost impossible to accumulate the number required for recall. Were this proposal to go forwards, I suggest this element be rethought. WjBscribe 17:10, 14 July 2008 (UTC)

Maybe A + [(55% of the RfX) x Bt], where A and B are arbitrary numbers to be selected (with B most likely to be a decimal) and t is time served since the RfX (to account for inflation)? Daniel (talk) 17:13, 14 July 2008 (UTC)

No need for a more complicated algorithm. With this system, the simple majority of individuals required to oppose would also be scalable, so that if any motion to recall achieves the minimum threshold but has a simple majority in opposition, the user is in effect reconfirmed by that majority and the further recall process does not happen. Amerique^dialectics 17:19, 14 July 2008 (UTC)

Again, why not just set it at a percentage of the support votes at the most recent RFA? That way, it's the same boundary for everybody, and scales with the growth of Wikipedia.
Kww (talk) 17:29, 14 July 2008 (UTC)

As I read it, the proposal is already written to that effect. Amerique^dialectics 18:40, 14 July 2008 (UTC)

Oh scratch that, you mean the most recent RFA someone passed in realtime. I would say no, that the support percentages in these RFAs are pretty arbitrary, based on the popularity of the individual user. It wouldn't be fair to set a threshold for recalling, say Secretlondon, on the percentage of support someone like User:Karanacs recieved on her last RFA. I would say as these individual users are evaluated differently on each RFA, it would only be fair to use some percentage of support at the individual user's last RFA as a threshold for recalling them personally. I like Rootology's suggestion below, though, of using an average of supports for RFAs over the last three months for setting an absolute minimum threshold. Amerique^dialectics 19:29, 14 July 2008 (UTC)

It should be easier to initiate a recall of the longest-tenured servants...it was so long ago, and the community has changed, that they should have to demonstrate that they can still get the support of the community in today's environment. Kurt Weber (Go Colts!) 22:16, 14 July 2008 (UTC)

Good point. If I had been around in 2002 or 2003 or whenever (was it 2004 that RfA started?), and I had been through RfA with 8 people supporting unopposed, then I would, 4 years later, look around and think: "Hang on. Different place now. I should go through another RfA to re-evaluate community support". Indeed, maybe someone should get Category:Admins who will ask for reconfirmation after 3 years as an admin started? Or something similar. Or allow private RfAs in userspace for reconfirmation. Carcharoth (talk) 22:04, 15 July 2008 (UTC)

root's 55% analysis

Based on this: http://en.wikipedia.org/wiki/Wikipedia:Suc...andidacies#2008

• June: 554 supports on 8 successful, 69.25 avg
• May: 1451 supports on 18 successful, 80.61 avg
• April: 1608 supports on 16 successful, 100.5 avg
• Total: 3613 supports on 42 successful, 86.02 avg
• 55% of 86.02 = 47.31 for April-June 08.

Why not base it on a rolling average of the preceding average supports for the past 3? That way everyone is on the same exact standards, fairly. Took less than 3 minutes to pull these with a calculator. Do the same for arbs, cats and everyone else.

So, you would need a dead minimum of 47 to qualify the recall and then it would need to finish passing above that over 55%. You would need a substantial number of people calling for the recall. It would be very easy to "sink" a recall, and harder to get one one done successfully, but it would be possible. Limit the recall to users with x edits/tenure like RFAR voting (heck, just use the same exact metrics) and that's pretty fair and scalable then. Frighteningly so. The ratio needed would update each time and if you're feeling frisky you can adjust the 55% along the axis of the support percentages that are average out over the given months. It would be kids stuff for a bot writer to draft this and the super simple numbers to community recall/deadmin would constantly be current with what the community standards are to sysop in the first place. The numbers would be completely trivial to pull and generate. This literally took me about 3.5 minutes at best to write entirely. rootology (T) 18:22, 14 July 2008 (UTC)

I would support that. Allows an extra layer of insurance to prevent accusations of railroading. Amerique^dialectics 18:40, 14 July 2008 (UTC)

Scratch my old 47 value. That was taking into account "this" month, July 08, which is not completed. The real numbers are:

I can't see how this can be abused... rootology (T) 21:17, 14 July 2008 (UTC)

Sounds good to me. Perhaps we could keep an automated, running meter of the absolute minimum threshold required for authorizing a recall prominently displayed on the process page itself, sort of like a stock exchange ticker. That should relieve everybody concerned with having a minimum bar set too low. Amerique^dialectics 21:33, 14 July 2008 (UTC)

Not even a running ticker. It takes a couple minutes at best to pull these numbers, and its not like these requests for recall are going to be constant. It can be just whipped up "on the fly" as needed, anyone can do it. But as far as the threshold goes, yeah, I think this should assuage some people. For people that passed RFA like 30-40 supports it would require 80+ to get desysopped this month! rootology (T) 21:35, 14 July 2008 (UTC)

Ok, that seems reasonable to me. It could concievably set a bar higher for the motion to recall to pass than for the recall petition itself to succeed, if fewer users actually show up for that than show up to support in the RFC, but I suppose that is ok. Amerique^dialectics 21:48, 14 July 2008 (UTC)

Would you want this rolling average to replace the "one to two thirds" or 55% of support scale based on the individual's previous RFA? I did want the process scaled to prior community support for the individual in question so it could be applied to sitting Arbitrators. Using their individual vote percentages would seem the best way to do that. Amerique^dialectics 22:00, 14 July 2008 (UTC)

Yeah, a rolling average for the last three bearocrat elections that pass (they're very rare, so thats fine), and for arbs just use the last year's average score for the "winners". It'll be hard enough to recall an admin unless they're total non-stop trouble, but it should be similarly hard to recall arbs and all the rest. I can't imagine a b'cat ever being recalled, though. How often do they even screw up? Once every couple years collectively? If that? rootology (T) 22:08, 14 July 2008 (UTC)

Using the rationale for your original proposal regarding 55% of the users RfA, why not make it the number generated above, or 55% of the RfX for the user in question, whichever is higher? That way you preserve the original rationale while putting in the safeguards. I just typed that, and then read "The absolute minimum threshold required to authorize a community recall should not be less than the rolling average support rate for RFAs over the past three months". Do you mean the same thing as I meant? Daniel (talk) 02:03, 15 July 2008 (UTC)

Yes, but I'm thinking setting the bar that high may preclude the necessity for splitting the process into two parts. Achieving that initial threshold of support along with 75% of all votes in total should qualify as the recall itself. So we would need to rethink the process I worked out if we are to use the (fluctuating, but ever growing) systemwide averages as a basis for qualifying a recall. Amerique^dialectics 03:02, 15 July 2008 (UTC)

Until's objections to voting

I think that we should base recall off of evidence applied to policy, not voting. Voting is not always evil, but voting in this case sure seems close. It is like voting someone off the island, this is an encyclopedic project not a reality show. _{1 != 2} 17:33, 14 July 2008 (UTC)

Comparing the idea of being able to change consensus to survivor is a straw man at best and anti-recall propoganda at worst, sorry. Especially calling it "evil". If an admin doesn't constantly do things that are against what the community wants the odds of him getting removed through this are about nonexistant. Is the fact that a community recall may exist that you object to, the fact that the community may be able to undo their RFA decision, or the simple fact that adminship would not be a guaranteed lifetime gig? rootology (T) 18:22, 14 July 2008 (UTC)

Perhaps it is because my comment was moved and thus taken out of context, but where it was originally they were talking about raw numbers and percentages. That sounds a lot like running a democracy which we are most definitely not. My suggestion was not that we not have recall, but that it be based on evidence applied to policy and not a head count of opinions.

Nobody was talking about a lifetime gig, nobody was talking about not being able to undo an RfA, so please read what I say carefully before rebuking me by calling my opinion propaganda. I will take care not to mischaracterize your position, I ask that you do that same. _{1 != 2} 18:42, 14 July 2008 (UTC)

Yes, let's assume good faith of everyone's opinion on what is bound to be a controversial matter. I thank everyone for commenting. Presumably, the only reason anyone would be taken to RFC in the first place is for policy violations. Any move to recall would be based on the discussion of whatever evidence lead to the RFC being initiated in the first place. If it's a frivilous RFC, the community will step in and prevent any recall motion. If the community is behind it, however, and supports a recall, there should be a numeric basis for assuring a dignified and fair process for the user involved. Amerique^dialectics 18:54, 14 July 2008 (UTC)

Otherwise we have what happens at ArbCom sessions, which I would say can be anything but fair or dignified. Amerique^dialectics 19:03, 14 July 2008 (UTC)

Sorry, Until. I totally misinterpreted how you are/were saying it. Sorry, man. Look at my 55% sub section above? I think that will clear it up. rootology (T) 21:26, 14 July 2008 (UTC)

Okay, no problem. I still think voting is a terrible way to deal with recall. We discourage voting in near every other area. I can't support a formula that gives a special number of votes to make such a decision. Consensus based decision making has not failed us so far and I don't think that we should be replacing it with statistical analysis of opinion. _{1 != 2} 03:35, 15 July 2008 (UTC)

RFA is statistcs/voting driven. Beurocrats cannot simpy "promote" anyone unless they meet a ballpark, fluctuating threshold. What is it? 75%? 80%? This isn't voting, it's still consensus. Its just that unless two given numbers that depend on what is expected to "currently" pass an RFA are met based on all RFAs in the past three months, the recall is tossed out. If a recall has 80%, then it passes, the same as a beurocrat can't simply shutdown an RFA over 80% without having hell to pay. Recall obviously should need to somewhat numbers based, since consensus is gameable. Math isn't, and game-proof solutions protect admins or anyone else from bullshit recalls. If 80% of the speaking community decides someone isn't cut out to be an admin anymore, that's a mandate for them to not be. Sorry. rootology (T) 05:11, 15 July 2008 (UTC)

Concur with Rootology. As I see it, Rootology further refined the process I came up with. My question is, if we base the threshold to qualify a motion to recall on systemwide averages, and further attach a 75% of all votes requirement on that motion, do we need a second part to this process at all? My idea was using a simple majority to defeat any motion based on a threshold of initial support records, and then requiring 75% of all votes in a later petition in a 2-part process to effect a recall. We could still make the second stage petition meaningful by keeping the 75% total vote requirement attached to it, but it would seem to me that this would only extend the drama, and that if support for any motion to recall could meet such a high initial threshold and win 75% of all respondents, that is a pretty definite consensus right there and there would be no need for a further process. Amerique^dialectics 17:08, 15 July 2008 (UTC)

Thats why I don't get the second part bit. A recall gets posted. If you need 80 supports that month to qualify it after 7 days, but fail, it gets shut down after 7 days as unsuccessful. If you reach 80 supports within 7 days, but after 7 days it falls shy of the current pass "rate" at RFA, then it's unsuccessful. However, if more people say that an admin has failed his role as an admin is no longer trusted, based on what the current RFA standards are, that's a "successful" recall. One shot, one step. Two tough but feasible thresholds to be met. It's a damning statement of failur on the party of any that fails two such tests of community trust within 7 days. rootology (T) 17:14, 15 July 2008 (UTC)

I'll rewrite it after hearing more comments. (by next week?) I appreciate your input here. Amerique^dialectics 17:28, 15 July 2008 (UTC)

Summary of proposal

Is this a good summary? Posting this here for you to integrate if you want. Change the wording, this is a bit casual, but this seems to be the basic synopsis.

1. Such and such admin has been consistently bad or problematic.
2. Start a recall. If begun today (July 08), you need at least 85 people to support the recall for it to succeed, based on the current RFA standards over the past quarter (three months). See here for an example of how the "numbers" are figured.
3. That rolling average protects "old timers" who may have passed RFA with like 10 support votes. Everyone has the same fair standard--accounts for inflation and the RFA "market economy".
4. Once you have the needed number of "supports", the recall only passes with 75%+ success. Otherwise, it's a fail.

Is that it? rootology (T) 21:39, 14 July 2008 (UTC)

Precisely! Amerique^dialectics 21:42, 14 July 2008 (UTC)

I'd still want to base the minimum support percentage for authorizing a recall against an Arbitrator on his or her personal record of support at their last election, though. Amerique^dialectics 22:05, 14 July 2008 (UTC)

Why not just use the average "winning" arb score from the preceding year's election? That way the arbs are held to the same basic uniform standard and system. On my excel I just used this formula: "=($B6+$B7+$B8)/($C6+$C7+$C8)" But anyone could do the same thing in a minute with a calculator. For arbs the average from any given year is going to be in the low to mid 100s anyway, and it should be much harder to 'recall' them, if it ever even happens. rootology (T) 22:09, 14 July 2008 (UTC)

Well, if there's concensus for that, I'll support it. Using current systemwide averages though does provide a lot of insulation for people who may not have earned that kind of cushion themselves. Tying the minimum threshold to their own support records would seem the most direct way of establishing "how much support" did this person actually enjoy when he or she got the tools, or became an Arb, and how much of a consensus would be reasonable to overturn that on an individual basis today, especially given that only a small majority could override any recall motion to begin with. Amerique^dialectics 22:28, 14 July 2008 (UTC)

Maybe it would be unfair insulation, maybe not. But for something like this to have a chance of becoming active it does have to be uniformly fair and simple, for everyone involved. Some "old timers" may get insulated, but something like this is a long-term thing also. if someone from 2003 gets a little protection, no big deal, since everyone gets that same even protection, meaning that recall ultimately would be based on a user's own actions and the current state of RFA/RFB/Arbcom elections, and nothing else. Recall shouldn't be to "get" anyone, it would be to improve Wikipedia by allowing the enforcement of changed consensus on RFAs, RFBs, and Arbcom elections to be enacted if everyone decides it needs doing. Theres virtually no new process here either, which is another neat thing on your proposal. rootology (T) 22:46, 14 July 2008 (UTC)

Ok, well, I suppose that also means that to override a motion to recall, the simple majority against would also have to be astronomical. In all honesty, I would hope that anyone who was put through this would have the good sense to resign before 84 people agreed just to initiate recall proceedings. I can't see anyone wanting to hold on to the tools after that. Amerique^dialectics 22:59, 14 July 2008 (UTC)

Hold on a minute, are you thinking this could all be done in a one part process? If so, that is different from what I am proposing, which has two distinct parts. Amerique^dialectics 23:23, 14 July 2008 (UTC)

I'll go with whatever version has community support, though. Neither version appears to me to be gameable. It seems to me if we base the minimum threshold to qualify a recall on current systemwide averages, and further require 75% support for it to pass, then having a second recall petition as a consequence of that becomes redundant. Everything could be taken care of within the scope and context of a single "motion to recall." Amerique^dialectics 23:57, 14 July 2008 (UTC)

Unrealistically high bar

I said this earlier, but it bears repeating: no one would ever be desysopped through this procedure. No one! No admin on Wikipedia is so hated that they could be the target of 85 requests for recall, and then manage to attain less than 25% of the vote in the actual recall RfA. I applaud the thought that's going into this, but we must devise a system that could actually succeed in getting bad admins desysopped, otherwise there's just no point in it. Everyking (talk) 06:14, 15 July 2008 (UTC)

If no admin ever gets desysoped after going through this process, that would be fine with me. All I want is that a fair and effective process be available. As I said before, if the community came together in such numbers to support anyone using any of the metrics we are considering, it should take an equitable number to override whatever the prior consensus was. There is no other way to do this that would be fair to everyone. As in the real world, recalls of elected officials are rare, so recalling an admin or arbitrator on Wikipedia shouldn't be an easy thing to do. Amerique^dialectics 17:23, 15 July 2008 (UTC)

What is fair to the community? I don't think having such an extraordinarily high bar for recall is fair to the community, and it certainly wouldn't be effective, because it wouldn't accomplish anything. I'm not saying recall should be easy, but it should at least be realistically possible. A fair system would require that an admin have majority support in a recall vote. That's not particularly demanding; the vast majority of admins would easily survive such a vote, and even most of the controversial ones would probably make it. The number who would be in real danger is probably small enough that you could count them on one hand. A system like that would get rid of the very worst admins, pressure the generally bad ones to shape up, while leaving the overwhelming majority of responsible and civil admins unaffected. Everyking (talk) 18:08, 15 July 2008 (UTC)

As I said below, allowing the community a legitimate means of overturning any recall attempt would only be fair to the community and to any admins or officers in question. Amerique^dialectics 18:39, 15 July 2008 (UTC)

Revised proposal, simpler, fairer, more secure, no 55%

Note--this really IS simpler. The two numbers I have here were stupidly easy to pull.

55% won't work, no arbitrary numbers

I don't think the 55% is going to be totally viable. There will be too much opposition to it as an arbitrary low number, which I can see as a very, very valid concern. It's utterly hard to get to be an admin as a legit user who doesn't get farmed up to get it--and given the crap many of them have to deal with, it would be too easy to lose it. What do you think of really simplifying the wording on the proposal to have the qualifying threshold simply be the average RFA score of the past three months? So, it would be to drop the 55% thing completely, and just have it be the average number of supports for successful RFAs on the preceding 3 months?

Hit that number in 7 days, it's a valid recall (hard standard, but possible), and if it finishes above 75%-80% it's subject to b'cat discretion like a normal RFA, but if it goes over the average percentage of the past 3 months for RFA's passing, the recall is automatically successful. I think thats about as absolutely ungameable as possible, will discourage frivolous recalls (they'll never pass without fantastic evidence) and will truly "bad" users that may not have done anything shockingly bad enough for Arbcom to remove them will be gone. Look at this:

Based on that, if you tried to recall me today, as of July's numbers (again, this was trivial to pull and figure out) the recall will need at least 92 supports (this month) to certify. Don't get that within 7 days, it fails. Average supports are rising all the time--given the extra heat and noise a recall will have, getting this IS feasible for a truly problematic admin. It just is. But, there needs to be a threshold past that. Lots of people scream for heads, but we don't rule by mob. It's consensus.

How to gauge consensus? Same as at RFA but even more fairly. 92 plus people support recalling me. All RFAs in the past 3 months averaged 89.7% for passing. If more than 92 people and more than 89.7% of the responders support recalling me, I'm gone. Alternate: use the same standards as RFA outright. No b'cat in his right sane mind would "fail" any RFA that was at 89% success. Similarly, any recall that is in certified (92 this month) but at 75% (the traditional limit for their discretion) to 89.7% this month is up to the beurocrats to decide. If the recall is at or higher than the live standard for RFA, 89.7%, it automatically passes.

Example Process under this

1. User:WJBscribe wants to recall admin User:Rootology. WJBscribe posts an RFA template there, with evidence and so on. Based on the averages for RFA in the past 3 months, a July 08 recall will require 92 endorsments total and needs 89.7% success to automatically pass.

2. Only "vested" contributors can start or comment on a recall. Use Arbcom election standard or maybe a more rigorous one. To be determined.

3: How it would work. If b'cats are unwilling to take on the task of doing this, then it would need to be the given pass rate based on RFA standards.

3a. Within 7 days, WJBscribe's recall of Rootology is 172/80/0 for 68.3%. It fails.

3b. Within 7 days, WJBscribe's recall of Rootology is 172/40/0 for 81.1%. B'cats decide.

3c. Within 7 days, WJBscribe's recall of Rootology is 172/15/0 for 92.0% Wikipedia has overwhelmingly stated that Rootology doesn't meet the current administration standards that are expected, and is desysopped. He can reapply at RFA at any time.

Thats it.

Summary

Fair, tight, secure, NO cabal has any kind of power to really meatpuppet something that big. Limit recall !voters the same as Arbcom !voters are limited, and you have a precedent based (based on realtime RFA standards) system to do recalls that has virtually zero process. It can even use the exact same RFA templates and page. We'd need an extra third level header I think on WP:RFA. rootology (T) 18:08, 15 July 2008 (UTC)

Thoughts?

• I think this is the only viable way that something like enforced community recall can work on English Wikipedia given politics and size. It's based on easy to get standards that are fully quantifiable and fair based on what the community actually expects of admins. This also makes the RFA process fully compliant with important principles like Consensus Can Change. rootology (T) 18:08, 15 July 2008 (UTC)
• I would support this. I would support any process that had the consensus of the community. If the community adopts this process, and then decides it doesn't have the collective will to ever recall anyone, it shouldn't be looked at as a failure of process, but as a definite decision on the part of the community not to recall that person. As I said before, I would hope someone would have the good sense to resign before such a high threshold was ever reached, but I suppose the reason this is needed is because some people don't, and there's currently no other means short of ArbCom to do anything about it. Amerique^dialectics 18:33, 15 July 2008 (UTC)

• The chance of the ArbCom desysopping anyone is much higher than the chance of that person being defeated in a recall vote according to these rules. If 30% of the community supported an admin in such a recall vote, they'd survive—in no sense does 30% represent the community's will, and it could not possibly be considered "a definite decision on the part of the community not to recall that person". Everyking (talk) 18:41, 15 July 2008 (UTC)

• Arbcom only desysops for stuff that is either what they see as abuse of the tools or implied threat of the abuse of the tools as they see it (what happened to you). They don't and probably can't do it because people may not in general trust the admin anymore. This as Amerique drafted it is really a test of the community's trust in the admin. If an admin isn't trusted by the "current" standards of RFA when challenged, they shouldn't be admins. Thats how I read this. And honestly, if something like this, which gives admins a big benefit of the doubt, is never accepted for a forced recall method, NOTHING ever will be unless someone from on high hands one down. rootology (T) 18:47, 15 July 2008 (UTC)
• (edit conflict) Concur with Root, that is the intention with which I wrote this. This process is based on the same standard it takes for someone to pass an RFA. Recalling someone should not be easier to do than that. If the community really wants to to remove the tools from someone, these standards are not insurmountable if the widespread consensus is there. Amerique^dialectics 18:55, 15 July 2008 (UTC)

But it's not the same standard, is it? The standard is reversed; it requires consensus to desysop, not consensus to make the person a sysop, as is the case in a normal RfA. You are correct that the ArbCom would not desysop in a case of "general distrust" alone, but if some kind of serious abuse or threat was not the basis for a recall vote, how on earth could the kinds of numbers you're discussing be achieved? Either way, I'd say the ArbCom would be far more likely to desysop than this process would. I'd sooner expect the ArbCom to break tradition and desysop people based on lack of community confidence than I would expect 75+% of voters to endorse desysopping. Everyking (talk) 19:22, 15 July 2008 (UTC)

Its a higher standar on purpose. It should never be easier to lose the bit than to gain the bit, but the community needs a safety valve to undo any decision it makes without having to lay out a Power Point presentation for Arbcom listing the Top 10 greatest abuses of the tools by someone to do so. rootology (T) 19:25, 15 July 2008 (UTC)

A few points: first, I appreciate the efforts of everybody here, and I certainly believe a recall process is in order. I just wanted to clarify (as there seems to me to be some ambiguity; maybe not) that bureaucrats can give sysop, but they can't take it away. Stewards would have to be working at these areas in case a recall were to pass. Second, I'm unclear as to how things will go after a successful desysopping. Does the former administrator/bureaucrat/checkuser/oversighter/arbitrator then face community sanction to determine a possible community ban? Assuming the user has multiple rights, do they lose part or all of those rights, or is that determined at the recall? Does the user have the option, assuming loss of administrator or bureaucrat rights, to run again in time should consensus change? —  scetoaux (T|C) 19:36, 15 July 2008 (UTC)

Well, the beurocrats would just decide consensus, like RFA, in my version of the draft. And from what I wrote above: "He can reapply at RFA at any time." This is strictly for "user rights", nothing else. Should they still be an admin, for example. Nothing else. If they were to run for RFA again 3-6 months later and pass, good on them, because it means they regained the trust of everyone. :) rootology (T) 19:43, 15 July 2008 (UTC)

Yeah, in any case, we won't know how this process really works until we have any actual test cases to try it out on. In any case, if I were an admin, I would not want to remain one if there was even half as much "distrust" directed at me as required by this proposal, but no admin or other officer should be able to ignore that level of a community consensus if all these thresholds were met. The aim of this process is to provide a way of gauging whether someone still has community consensus to retain the tools or their elected position. The ArbCom could still desysop someone who survived a recall, in any case. Amerique^dialectics 19:50, 15 July 2008 (UTC)

I don't like the idea of ArbCom desysopping someone that had survived a recall. Depending upon how the recall went, doing so might be in defiance of community consensus, and ArbCom should not have that right. —  scetoaux (T|C) 19:52, 15 July 2008 (UTC)

Well, unless we can somehow take away that responsibility from them, they would still have that power. Community sanctions would otherwise be handled as they always have. The process only has bearing on bits and other elected positions, including individual ArbCom positions. Amerique^dialectics 19:57, 15 July 2008 (UTC)

I think I have to agree with Everyking, this recall standard seems very hard to achieve. The number of people required to support the recall petition is already very high and to then require a positive consensus of participants in the recall makes me doubt that this procedure could recall any but the most flagrantly abusive admins - who would be desysopped by ArbCom in any event. I sympathise with the need to find something that is likely to be acceptable to everyone, but I suspect this is setting an unreachable threshold. If the initial number of people needed to start the process is high, I think one could consider lowering what is needed in the eventual proceedings. 3a above troubles me as a fully worked example. Are we really saying that with 172 people seeking recall and the admin failing to gain the support of even a simple majority of the community, they should continue in the job? WJBscribe (talk) 21:58, 15 July 2008 (UTC)

The flipside is that even if 170-180 people support an RFA it may still fail. Didn't Clown fail an RFA with 200+ supports? rootology (T) 04:35, 16 July 2008 (UTC)

According to User:NoSeptember/RfA voting records (though I note this hasn't been updated much in the last year) the most support someone has had and gone on to be unsuccessful at RfA is 201 (Gracenotes). WJBscribe (talk) 19:37, 17 July 2008 (UTC)

Giggy-Dihydrogen's RFA #3 ended with candidate withdrawal at 299/85. Townlake (talk) 05:43, 21 July 2008 (UTC)

No thank you

As noted by several above, this proposal (while well-meant) has several problems. The key one being numeric (something that's also repeatedly discussed/argued in relation to RfA/B). No need to reiterate the points elucidated above.

One of the reasons that this won't pass (I have no doubts), is also one of the perceived "problems" with the RfA/B process. "Power" is in the eye of the beholder. The perception of what an admin is varies from editor to editor. As such, finding criteria for what an admin "is", is going to be subjective at best. And subjective "voting" isn't a strong way to determine "consensus", merely just a vote tally.

One thing that "might" help at least bring "me" around is if there was a proposal for a majority of admins to be able to discuss desysop of a fellow admin (same with bureaucrats). Though it would have to be a "majority" (RfA consensus determination would apply) of all active admins (987 last I looked, I think). I think that if (roughly) 670 admins out of 1000 decided that perhaps an admin should be de-sysopped, then that defininitely might at least be something worth discussing.

I understand that this proposal is well-meant, but I think you (plural) might be better served in attempting to positively join in the discussion concerning Arbcomm, than this. - jc37 21:12, 15 July 2008 (UTC)

If admins have zero additional authority than non-admins to decide who gets to be an admin, why would they have additional authority to decide who should lose adminship? It's not an ivory tower. rootology (T) 21:17, 15 July 2008 (UTC)

To talk about getting 85 people to endorse a recall petition strains the furthest limits of credibility for me. To talk about 670 admins supporting a desysopping launches into the realm of sheer fantasy—honestly, we might as well consider building exile communities for our desysopped admins on Mars. Everyking (talk) 21:19, 15 July 2008 (UTC)

Mars? There are enough Wiki-clones out there suitable for tortured exile (grin). - jc37 21:22, 15 July 2008 (UTC)

Joking aside, while you may consider 670 to be "sheer fantasy", I wonder if it's worth suggesting in case it could actually happen : ) - jc37 21:24, 15 July 2008 (UTC)

Thanks for your input, Jc37. We are not discussing what an admin is, we are discussing what level of consensus would be required to overturn or reconfirm a prior consensus for any elected position. "Voting" is how consensus on these matters is effectively determined in the first place, and there is not another way to manage the project at this or any scale.

I would not support any proposal that further privileged the administrative classes at the expense of the community. You don't have to be an admin to elect an admin or arbitrator, therefore any process developed should be open to all participants in good standing. I intend to rejoin the discussion at the ArbCom RFC as I find the time. Amerique^dialectics 21:36, 15 July 2008 (UTC)

Fair enough.

Incidentally, I think you may have misunderstood/mis-interpreted what I was attempting to say about subjectivity. (The fault is likely in my attempt than in your reading.)

I was attempting to convey the dichotomy between the "no big deal" camp, and the "But being blocked is a big deal. And being able to delete a page is a big deal. And and and..."

This (among others) is a subjective question based on (honestly) unknown data. This is part of why these have been opposed so consistantly. (Including the threat of "revenge/reprisal".) And since how much of a "big deal" it is to become an admin is "subjective", I think you'll be rather hard pressed to get any sort of consensus to determine the criteria for recall.

And that's just one reason. There are quite a few more.

As for my idle thought/suggestion, it was merely because admins know what it's like to be an admin as well as being an editor. Both sides. And you might be surprised at how aware most truly are. Right now the community pretty much doesn't have any way to desysop except through a lengthy WP:DR process. I offered an alternative that I thought might perhaps at least have a possibility of going forward. Or in other words: "it's a start". - jc37 22:18, 15 July 2008 (UTC)

I appreciate that. I know that RFAs are considered on subjective merits, but that being the case, the only thing concrete to take away from them is the ratio of the consensus used to promote. I know I've said this before, and I know the idea may never gain popularity, but any community recall procedure should take into account the prior consensus used to promote a user for establishing a minimum threshold to recall the same user. "Seniority" in terms of time on the site shouldn't affect the numbers used to determine the levels of consensus individuals have for using the tools. I realize this provides more recently promoted users with "more insulation" than others from a recall, but if we are going to be fair to whatever the prior consensus was, considering each user in terms of their personal context, rather than at a mass statistical level, is the way to do it, if the community actually wants a way to take responsibility for the conduct of people it promotes into leadership positions. Amerique^dialectics 15:16, 17 July 2008 (UTC)

I understand, I merely disagree, and while I may be surprised, I'm also rather doubting that this has any chance of passing.

In addition, thank you for including my thoughts above on the proposal page.

However, it's different enough than the other proposals (at least at the moment), that it's probably better to not propose it (to minimise confusion in discussion, for one thing).

With that in mind, I've removed it.

Thanks again : ) - jc37 21:10, 17 July 2008 (UTC)

Yes please

This is a much needed proposal, is difficult to game, but will produce results. I endorse any of Rootology's suggestions above. —Giggy 04:20, 16 July 2008 (UTC)

I support the concept, but I think it would be easier if the default admin recall process was made compulsory, instead of inventing a new process. The number of editors petitioning could be raised from 6 to perhaps 10 or 12. PhilKnight (talk) 18:43, 18 July 2008 (UTC)

I took a look at it. That process, like Everyking's, provides no means for the wider community to check a small consensus for initiating a recall. The way I've designed this, frivolous recall motions would be discouraged by the consequences of not gaining a majority in support, which in the case of low thresholds could raise the bar for further recall attempts. Amerique^dialectics 21:38, 23 July 2008 (UTC)

Sure, maybe 12 is still too low. Perhaps 20 would be better. PhilKnight (talk) 22:44, 26 July 2008 (UTC)

Just to clarify, there are two ways of running a recall under discussion

• Rootology's way of doing it, as outlined above, would use the systemwide averages over the last three months for successful RFAs or other elections to establish an absolute minimum number to qualify consensus for a recall, and further attach a requirement of 75% of all respondents in support for the recall motion to pass. As he says, "If you need 80 supports that month to qualify it after 7 days, but fail, it gets shut down after 7 days as unsuccessful. If you reach 80 supports within 7 days, but after 7 days it falls shy of the current pass "rate" at RFA, then it's unsuccessful. However, if more people say that an admin has failed his role as an admin is no longer trusted, based on what the current RFA standards are, that's a "successful" recall. One shot, one step. Two tough but feasible thresholds to be met. It's a damning statement of failure on the party of any that fails two such tests of community trust within 7 days."

• My way of doing it, as outlined in the sandbox page, would use the prior support records of the individual user in question to establish a minimum number necessary to qualify a current consensus for a "motion to recall." If that number was reached within 2 weeks, but exceeded by a simple majority opposed, the user in question would be in effect reconfirmed by that majority and no further action with respect to recall would take place. If those support thresholds were reached, a "recall petition" would take place no earlier than a week later, following the format of an RFA or whatever "election" process installed the nominee in the first place, requiring a minimum of 75% support of all respondents to qualify consensus for recalling the user.

There are advantages and disadvantages to both processes. As I see it, Rootology's process would be simpler and faster but also harder to do. My process, I think, would make recall more "achievable," but would take longer to run and there were concerns that using individual support records would make it easier to recall "older" users than newer ones that are currently passing with +100 support records. If some version of my process were adopted, however, I don't see a way out of this last consideration. To me, using a high proportion of whatever the initial support basis was to qualify a current consensus for recall seems eminently reasonable, as well as responsible as it takes into account the "weight" of a prior consensus for "promoting" individuals in considering whether to "demote" them, each taken on an individual basis. RFAs and other elections are considered on subjective merits, and even today, not every RFA passes with +100 support. Rootology's plan, which bases the threshold for recall on systemwide averages, sets a threshold for recall that is probably far higher than the ratios of support of many current administrators received, and basically obliterates the need to consider context in terms of "how much support did this user actually have when they got the tools or the Arb position in the first place."

So, because I think consensus to recall should be scaled to the prior consensus to promote, rather than shifting system wide averages, I prefer the basic outline of my plan to Rootology's. My plan allows everything to be considered on an individual basis, and still sets a high bar for a recall to work. Amerique^dialectics 16:40, 16 July 2008 (UTC)

Oppose

I oppose this on the grounds that it would enable too many frivolous or vexatious complaints to cause admins to spend time defending a recall for making unpopular edits or blocks rather than working on the encyclopedia. Stifle (talk) 13:39, 18 July 2008 (UTC)

I wouldn't say it would "enable," but it may "provide incentive for" filing complaints. The beauty of either Rootology's plan or my plan is that frivolous motions without cause would easily be quashed, effecting a "reconfirmation" of community support for the individual in question. The requirements for successfully initiating either of these processes are high enough that no frivolous RFCs would be brought on by anyone sane. Non-frivolous RFCs are another matter, of course. Amerique^dialectics 15:55, 18 July 2008 (UTC)

Assuming any frivolous motions to recall are inspired by the appearance of a low threshold to qualify, under my plan, the numbers of the consensus in opposition could establish a new minimum threshold for further recall attempts. This should provide a further "dis-incentive" for casual recall motions. Amerique^dialectics 16:06, 18 July 2008 (UTC)

The issue Stifle brings up is why I prefer an evidence applied to policy based system over a vote. This whole idea of having people vote on if a recall should be performed/considered is all very un-Wikipedia. Sanctions against a person should be based on policy and evidence usually in the form of diffs. We don't hold a vote to block people, we expect there to be a damn good reason, discussions are not judged as head counts but by the content of their arguments. I think that should this idea ever be put before a larger audience it will be rejected just like every other similar proposal has in the past. Chillum 16:18, 18 July 2008 (UTC)

A short quote from one of our polices Wikipedia:What Wikipedia is not should put my objection into context: "Wikipedia is not an experiment in democracy or any other political system. Its primary method of determining consensus is through editing and discussion, not voting.". The community already has at its disposal the ability to block or ban someone who is abusing Wikipedia through consensus(not voting), and they also have arbcom to report problems that the community is unable to solve. Any admin who reversed their own community applied block will certainly lose their bit immediately. There are existing forums to deal with this in a manner that is consistent with our method of decision making without resorting to voting which we have established as policy as a way we do not do things. Chillum 16:26, 18 July 2008 (UTC)

The thing is, we already do things this way, whether WP:NOT admits to it or not. Even at a scale of only two users doing something over the objections of one, that's still democratic practice. Consensus says "Consensus among a limited group of editors, at one place and time, should never over-ride community consensus on a wider scale, unless convincing arguments cause the new process to become widely accepted." (I am not sure what that last phrase refers to, though.) Only silence in terms of non-opposition or an obvious preponderance of support over all objections can establish consensus. All my proposal does in essence is apply the same standards of consensus that are regularly applied in RFAs towards recalling the same user that was promoted by them. My proposal is deliberately worded to make what already happens at RFAs more evident, and also more consequential, not only for the purpose of promotion, but for the purpose of establishing a minimum threshold for a future consensus for recall.

In the way I've work it out, the discussion of any evidence of policy violations against a user would take place in an RFC. A motion to recall, if considered appropriate, would follow that. If a recall motion were presented without evidence or cause, it would completely fail, and probably strengthen the position of the user it was initiated against, so there is really nothing to worry about. Amerique^dialectics 20:20, 18 July 2008 (UTC)

I am sorry but that is not what consensus is. It is not a democratic process, if the majority is failing to understand the policy or acting on irrelevant motives then that should be ignored when the consensus is judged. It is not a question of 2 vs 1, and anyone who is going to be judging consensus needs to know that first and foremost. Your proposal turns it into a pure numbers game. This opens the door to abuse, mob justice, and uses a vote to deal with behavioral issues. The worst part is it really does not solve any problem. I ask again, where are all these admins that the current system is failing to deal with? Chillum 20:24, 18 July 2008 (UTC)

Well, I'm not going to name names... but let's just say a process like this should be in place for any problematic users ArbCom itself can't deal with, theoretically. (And problematic users on ArbCom!) But the process uses numbers to provide as much protection for them as would seem reasonable to me, and make it impossible to run against people for whom the support of the community is there. It's not pure numbers, though. I'd imagine lots of rhetorical arguments would occur, pro and con, pursuant to a recall motion in an RFC. What side people decide to support, in an RFC that has any evidential basis, would in large numbers be swayed by strong policy-based arguments, I would think. It's not "the rule of the mob" any more than an RFA or any other formal process for establishing consensus on any matter is. Amerique^dialectics 20:49, 18 July 2008 (UTC)

It may not be more ruled by the mob than RfA, but that is about the only thing on Wikipedia that is mob rule, it is practically a pure vote. But any, and I mean any other formal process is most certainly not based on a count of votes. XfD's for example are based off of the quality and quantity of the arguments put forward and the relevant policies, if 20 people come by that don't get policy and vote to keep that will not have much weight. When we community ban somebody we need a damn sight more than just greater than X percent people wanting it, we need policy based reasoning.

No, as long as this is a vote, I cannot support it. This does not reflect the way we do other things. I want to know why the community does not just put community sanctions against troublesome admins, they have the power. Chillum 20:58, 18 July 2008 (UTC)

This proposal only affects "trust positions" that are already elected in some way, otherwise a non-binding "motion of no-confidence," rather than a recall, is passed. Without numbers though, I would say there is no way to determine a consensus to even issue "community sanctions." One person can make what seems to him or her an irrefutable policy-based argument to do so, (especially if they've written policy to that effect) but if other people don't agree or think they have better arguments to the contrary (or better interpertations of policy or better policies) and it all looks the same to a non-invested outside observer, then without numbers, pro and con, there is no other means of determining things except by fiat, even to concievably pass this as policy. You bring up WP:NOT, I bring up WP:CON, which is more important to uphold? Amerique^dialectics 21:35, 18 July 2008 (UTC)

Community sanctions on vested, long-time users are impossible in practice in any organization, let alone one like this--all it takes is a very loud sub-group of partisans, hangers on, followers, or fans to quash anything and say "there was no consensus". The point of my version of the proposal is specifically to make it a very hard but theoretically attainable threshold that makes it so any one admin, clique, subgroup, or cabal has no ability or power to override consensus. Which is how it should be for any consensus decision. My version is game-proof because of this. If an admin doesn't deserve desysopping, he's not going to get 90+ people saying "fire him" AND get 80%+ support on that. Its simply flat out not going to happen. I'd be curious for your answer to this exact question: If a big of a chunk of the population (80-90+ people, and more than 80% of the total of respondents) says you're not fit or trusted to be an admin anymore, by what conceivable or imagineable right should you still be an admin? rootology (T) 21:48, 18 July 2008 (UTC)

My question is can you envision a situation where there would be such a support for action but existing methods(dispute resolution, community sanctions, blocks, arbcom) would not be able to handle it? Chillum 22:00, 18 July 2008 (UTC)

This would fit in to the existing DR system and basically allow the community another option in the toolkit in addition to the rest. Keep in mind that the DR system leads to the ArbCom and the current lack of institutional procedures for recalling a Arb or user access level privleges is the main reason a universal proposal is needed. Amerique^dialectics 18:46, 21 July 2008 (UTC)

But arbcom is given their authority by the foundation, the same people who have given the community authority. I don't think the community can recall arbcom, all they can do it appeal to the foundation to do so. I also don't think we can gain that authority by coming to a consensus that we want it. Chillum 18:56, 21 July 2008 (UTC)

I know, I thought about that. I'm not exactly sure how the Foundation affects ArbCom, but my process does not get in the way of that and would anyway only produce a "motion of no confidence" on the part of the community against a sitting Arb. It is not a process for "recalling" the committee itself, and the standards of consensus required for recalling a single Arb are high enough that any sucessfull motion would have to be taken seriously, but it would remain in the Foundation or Jimbo's authority to act on or not. Amerique^dialectics 19:10, 21 July 2008 (UTC)

Well, there is this page set up for informing people who want to change the arbitration policy: Wikipedia:Arbitration policy/Procedure for changing this policy, take what you will from that. If all you want to do it show that the community does not like something without being actionable, then I don't think we need a special procedure, we can just say so. Chillum 19:14, 21 July 2008 (UTC)

All for it

Everyking's process is exactly what we need on Wikipedia. Keep it up, Amerique. I'll be supporting this wherever it goes. Mr. IP (talk) 01:04, 19 July 2008 (UTC)

Yes. I agree with Everyking's thoughts that the process must continue to be well thought out to prevent abuse of such a potentially powerful process. I think if an effective and well designed system can be designed, it can be very beneficial to the Wikipedia. I'll be watching this one. --.:Alex:. 17:04, 21 July 2008 (UTC)

I won't have much time to edit WP much over the next few weeks. The problem with Everyking's process is that he puts no check on the "10 established users" who can initiate a recall. However, his threshold to initiate is so small that any potential check, say by a "simple majority," would render his process unworkable, as "10" would likely not be enough to initiate recall against even the least popular admin or an ArbCom member. And if Everyking's process were "voted in" as is it would allow a consensus of only 10 to be sufficient for initiating a recall against any sitting Arb, any unpopular admin, or any good admin who makes a correct but controversial call on some matter. It would produce chaos.

So I can't support Everyking's proposal. I only included it on the project page to differentuate the merits of the various proposals that are being debated here. Both my and Rootology's process provide rigorous but fair and achievable thresholds for recalling every class of trust position, in a manner proportionate to differing scales of consensus. Amerique^dialectics 18:35, 21 July 2008 (UTC)

I just wish to mention that I don't necessarily support Everyking's proposal, just his thoughts on how this proposal must continue to be developed and that I agree with the actual recall idea itself. I think the above comment was mainly directed at Mr. IP but I just feel the need to mention this as text is hard to interpret and discussions like these get confusing. Not to mention that I'm not particularly good at getting my exact intentions across at all. Keep up the good work though everyone! --.:Alex:. 18:41, 21 July 2008 (UTC)

To be clear, I don't favor any kind of "check" in which a recall petition could be thwarted by an equal or greater number of people opposed to recall. If 10 established users think you're so bad that you need to be desysopped, that means you're a controversial admin and your continued adminship needs to be validated by the community. What would actually happen is that if admins felt they were in serious danger, they would clean up their act or at least make promises, in which case few people would support their recall, and many would withdraw their signatures if they had already signed. If you're intransigent and won't listen to ten people's complaints, then you should be willing to face an RfA. Worst case scenario, you lose and get to be a plain old editor again—but if this system was in place, RfA itself would become less demanding because a successful RfA would no longer mean lifetime adminship, and the defeated former admins could prove their worth as an ordinary editor for a while and then get voted back in. It would be a somewhat more flexible system—much healthier, in my opinion, and not at all chaotic. Everyking (talk) 07:13, 12 August 2008 (UTC)

Correlation with RFA

I think that one of our main objectives with regards to adminship was not only providing a process by which an administrator can be formally removed by community consensus as opposed to Arbitration Committee action, but also reforming the Requests for Adminship process so that it is less grueling and unfriendly as it currently is. The correlation between these two processes is this: any process that makes it easier to remove administrator rights consequently makes it easier to gain administrator rights, and as a consequence of that, RFA becomes less like the grueling process we know it as, and more like the friendly, consensus-building process we'd like it to be. This is because the addition of a recall process will put into people's minds that a candidate's fitness for adminship no longer has to be judged solely at RFA; they are now given the ability to make a decision later on. It allows them to give a candidate a chance, whereas before they were completely unwilling to risk the possibility of a misuse. In short, increased accountability leads to a more lenient RFA process. —  scetoaux (T|C) 03:00, 22 July 2008 (UTC)

I never found it grueling. I failed my first one because I was not ready and I received all sorts of useful advice. The second time I passed because I took that advice to heart. I really don't see a lot of people who should pass RfA having a tough time. Chillum 03:25, 22 July 2008 (UTC)

Some more proposals

I decided to be bold and and add two additional proposals. One is a variation of what we use in California. Here if someone can get 5% of the total votes cast at the last gubernatorial election, there will be a recall election. I adjusted that to 10% to get a higher bar and weed out the crap petitions. I am willing to adjust it to an average or the total number of voters at the last Arbcom election also.

The other proposal would allow any steward to summarily remove privileges if an RfC establishes consensus to do so. Thoughts? Geoff Plourde (talk) 06:23, 24 July 2008 (UTC)

The first proposal is a concrete statement of how I would think the outcome of a recall process would be "judged" or decided, by a steward, according to the will of the community. For the second proposal, I honestly wouldn't mind, but I think "10% of prior votes cast is going to be too low a threshold for initiating a recall on WP. (In the real world, thresholds are low because getting people to vote is a difficult thing to do. Also, we have much less respect for public officers than we do for our diligent and hardworking admins and arbs on Wikipedia!) Amerique^dialectics 22:34, 24 July 2008 (UTC)

Jimbo's desysop of Bedford

Interesting case with ramifications for this proposal which I am following but not directly participating in.

• ANI thread here: Wikipedia:Administrators' noticeboard/IncidentArchive454#Bedford_and_misogyny.
• Arbitration request declined here: [3]
• Bedford's RFA here Wikipedia:Requests_for_adminship/Bedford.

It would seem to me, based on this, that numeric thresholds for qualifying a motion to recall should be lower than in my proposal, to be comparable with real world thresholds of 15-40% of prior consensus, but that such motions should still be checked by a simple majority. Although I did not intend community recall as a replacement for Jimbo's right to take action on any perceived "trust issues," or for arbitration process, I still think the community needs an effective way to decide such issues on its own, without having to wait for or expect action from Jimbo or to have to go through a lengthy ArbCom on some issue. Amerique^dialectics 18:45, 26 July 2008 (UTC)

Yes, I agree. The community should have at least as much as (if not more than) power as Jimbo. Kelly ^hi! 18:57, 26 July 2008 (UTC)

Added a link to the arbitration request. Looks like it will be denied. I am interested in the numbers of the consensus in support of desysop. According to Seraphim♥Whipp's calculation on the ANI thread, 7 people requested Bedford be desysoped before Jimbo acted on his own initiative. 8 by my count supported the desysop at ANI after Jimbo already announced it was done. These figures would not have come close to qualifying a recall procedure under my proposal... and I don't think numbers that low should be enough. I'm fine with Jimbo acting on his own initiative, but he definitely acted in the heat of the moment. It can be argued that this curtailed further drama, but still, in absence of action from Jimbo there still should be a community process for archiving the same outcome, but using sufficient numbers so that a small, angry mob wouldn't be enough to initiate, and that is drawn out enough so that rash actions on the part of the community can be prevented. Amerique^dialectics 02:37, 31 July 2008 (UTC)

Geez, if this action shows anything it is that we don't need further methodology for desysoping people. This desysoping was not based on numbers of people objecting or agreeing. It was based on the actions of the person, and what the community expects, not some vote. The fact is that nobody is providing a list of admin that need desysoping but the present system is unable to do so. We are handling desysoping fine already and don't need further methods on top of our already adequate methods. Chillum 04:59, 31 July 2008 (UTC)

Not to get into it, but as far as I know, this is the first time there was ever a "rushed desysop" like this on Jimbo's part, which as far as I can tell, occured for only 2 reverts on the DYK template and some on and off-site NPA violations. I think the NPA violations were heinous, but I've never heard of an admin getting desysoped on that basis before, short of a lengthy arbitration. But the fact that Jimbo can intervene in the cases he wants to doesn't mean that the community shouldn't be able to take action on its own part. Amerique^dialectics 05:49, 31 July 2008 (UTC)

That fact that we handle all our current desysoping shows we don't need this. The idea that the community cannot take action on its own part is just wrong. There is dispute resolution, community sanctions and blocks, there is arbcom, and there is even common sense when it just needs to be done. I have asked over and over who these people are who need desysoping but the current system can't handle it, and nobody provides me with this simple demonstration of need. I don't think this idea will gain the community acceptance it needs until such a time as the community actually needs such a thing. Chillum 14:37, 31 July 2008 (UTC)

Of those, only ArbCom and Jimbo traditionally have had any jurisdiction over removing community granted user access-levels, and there is currently no standardized way of recalling a Checkuser, Oversight, or ArbCom member. This process isn't being developed with any specific person or group of people in mind, but as a means of covering every eventuality. I had never heard of Bedford before Jimbo desysoped him, nor had I thought a couple reverts and NPA violations (however heinous) would be considered grounds for desysoping, before now. So, unless the community prefers having only 2 options, either a spur of the moment action by Jimbo or a drawn out DR process leading to arbitration, community recall presents a viable third way, that would be attached to RFC, that would concievably preclude the need for further DR to accomplish the limited goal of removing the access level privileges of someone who has clearly been misusing their tools. Amerique^dialectics 15:15, 31 July 2008 (UTC)

It sounds to me that the problem this is meant to address has not manifested yet. Perhaps one day our existing systems will not be able to handle desysopings, but today it is doing just fine. Our dispute resolution process is a great thing, it helps diffuse situations. I would hate to see people bypass an attempt to get along in favor of having a vote. These things should be decided based on discussion and evidence, not popular opinion.

As for Checkuser, Oversight, and ArbCom, these things have always been in the remit of the foundation. While they do consider community input they really just pick who they want to. I don't think the community has the authority to do anything in this area. These groups got their authority from the same place the community got its authority, the foundation. We have the same boss. If all you want is to make a process where it can be made clear to the foundation the communities wishes then an RFC will do just fine. Chillum 15:26, 2 August 2008 (UTC)

Jimbo's comments, posted on his talk page

I think any such process should be voluntary, and that there are some good reasons to be deeply concerned about such processes in general. The problem is that we actually want people in positions of trust (the ArbCom for example) to be significantly independent of day-to-day wiki politics. And I am unaware of any cases in which this sort of process has been needed. Perhaps the pointed question has to be asked of the advocates of making this sort of thing mandatory: who do you want removed, and why? If the answer is "well, no one, this is 'just in case'" then I would suggest that this could be an instance of process-creep. And if the answer is "well, this particular person is a good example" then we might want to look at that case a bit harder and see what went wrong... if anything did go wrong.--Jimbo Wales (talk) 02:25, 4 August 2008 (UTC) [4] User_talk:Jimbo_Wales/Archive_38#Community_recall

Active RFCs

• Wikipedia:Requests for comment/Kirill Lokshin

• Wikipedia:Requests for comment/Elonka (Currently deleted, but the talk page is active. Undergoing deletion review here)

• User_talk:Elonka#Recall_Proposal Related to the above. Voluntary recall proposal currently underway at Elonka's talk.

Currently observing these two for implications for this proposal. Based on initial observations of Kirill and Elonka's RFCs, the basic RFC format would not seem to lend itself to a community recall motion, as "oppose" comments to propositions are apparently not allowed. Also, RFCs in themselves are considered "non-binding." The notion of community recall should thus probably be considered a seperate venue from RFC. Amerique^dialectics 18:37, 4 August 2008 (UTC)

Community arbitration?

I had initially thought this could be done within an RFC, but RFC itself is non-binding and due to its structure, it does not seem conducive to "thoughtfull considerations of evidence" per se... It seems that if the initial poster to an RFC does not make a comprehensive case against whoever the subject is, additional voices adding other grievances can create a mishmash of "sound and fury."

Since RFC may be an inappropriate venue, a separate project space should be created if a mandatory recall process were to be implemented. I was looking at the structural outline of arbitration process, and I think, as far as structuring the process, the idea of seperate evidence, workshop, and decision pages may be the best way to go about processing a recall motion. If we were to take arbitration process as a model, the idea of limiting this to trust positions could even be jettisoned and we could call it "community arbitration," with 2/3rds majority deciding the outcome of all proposed decisions. If such a system were implemented, the original ArbCom system could be retained as a "higher court" or a court of appeals of sorts, for decisions made in community arbitration.

Food for thought? Amerique^dialectics 23:56, 10 August 2008 (UTC)