Wikipedia talk:Requests for adminship/Archive 85

From Wikipedia, the free encyclopedia

Contents

[edit] Changing the promotion threshold

I would like to observe that the 75%-80% threshold was never arrived at with any particular degree of analysis or discussion. It dates to a time when it was rare for there to be more than ten total votes cast in an RFA. User:Angela, for example, was made an admin on the strength of three support votes and no opposes. With so few voters and an environment where all the voters knew each other, there was far more pressure to work towards consensus and withdraw objections. Also, the early RFA environment focused mainly on readiness: candidates were not so much rejected as encouraged to wait another month or two. Some early RFA pages may be insightful in understanding the evolution of the process.

In general, at the time the promotion threshold was adopted, it was rare for RFAs to end up being close calls, and so the figure was chosen somewhat arbitrarily. Now the RFA process is essentially a voting system, where no real consensus building takes place (in the sense of people changing their opinions in response to what others have shared). So, the promotion threshold is important. Yet there is no evidence to suggest that users whose RFA pass with 75% are any more likely to pose a problem for the project than users whose RFAs pass with over 90% support. A review of the rare instances where the Arbcom has sanctioned administrators does not show any pattern in this area.

Accordingly, I would like to suggest that, as a community, we work towards a consensus to change the promotion threshold to 65%-70%.

The Uninvited Co., Inc. 19:02, 9 March 2007 (UTC)

The thing that's making me not be overly fond of this idea is that it is possible for over a third of the people commenting on the RfA to oppose it, but it still pass. I realise that's just a statistic, but it's a bit worrying, I think. However, your proposal is well thought out and I will give it some further thought before passing judgement on it myself. --Deskana (talk) (review me please) 19:23, 9 March 2007 (UTC)
I would like also to hear some rational for why it should be 65-70 specifically. You say that the "75%-80% threshold was never arrived at with any particular degree of analysis " but I don't see any analysis backing up your number. JoshuaZ 19:28, 9 March 2007 (UTC)
It works on the German Wikipedia, which requires a 2/3 majority. Kusma (t) 19:31, 9 March 2007 (UTC)
Does the German Wikipedia require people to renew their RFA? If we had something like that, then I'd support dropping the threshold below 75%, but not otherwise. Regards, Ben Aveling 21:46, 9 March 2007 (UTC)
They have a formal recall and temporary desysopping process, which can require people to go through a new RFA. They do not renew the RFAs of admins that nobody complains about. Kusma (t) 07:17, 10 March 2007 (UTC)
Does anyone have stats on how many RfAs had 65-75% support? It think it would be worth examining these candidates as examples of who would have passed with lower standards, but failed under the current ones. ChazBeckett 19:55, 9 March 2007 (UTC)
Assumming no one beats me to it, I could easily work out the change in the number of passes for any arbitrary threshold when I get home this evening. (Of course, one factor that one can't really account for is how people might change there voting patterns in response to a change in the threshold.) Dragons flight 21:25, 9 March 2007 (UTC)
I strongly disagree with this proposal, which drops support ratios needed to pass from approximately 3-4:1 to below 2:1. -- Renesis (talk) 20:00, 9 March 2007 (UTC)
I strongly agree with this proposal, which drops support ratios needed to pass from approximately 3-4:1 to below 2:1. We do not have enough administrators, and this is a quick way to allow more people to pass this process. Do you have a better suggestion how we can increase the promotion rates? I doubt that any easier solution than changing the threshold exists. Kusma (t) 20:34, 9 March 2007 (UTC)
Are you sure we don't have enough? While it has been demonstrated that workload has increased, I'm not sure anyone has demonstrated that we don't have "enough", and I don't believe this to be the case. I think the current growth rate of administrators is satisfactory-to-good. -- Renesis (talk)
We promote as many per month as we did three years ago. The number of admins grows linearly, the workload grows exponentially. See Malthusian catastrophe. Kusma (t) 07:17, 10 March 2007 (UTC)
I'd also like to add that administrators "sanctioned by the Arbcom" are not the only ones causing problems. -- Renesis (talk) 20:02, 9 March 2007 (UTC)

Why wouldn't 2/3 be appropriate? Even a 2/3 consensus would hardly pose a problem for the project in a general sense. It would however, give less weight to those who oppose as a result of personal bias' which sometimes can be arbitrary to the majority consensus. Why keep the few who think 5, 6 or 7 thousand edits aren't enough, stop an otherwise good editor from being able to help the project in a higher capacity. Serious issues with a canididates qualification would come out in the RfA any ways, and like many RfA's, consenus can change due to these issues. IMHO--Hu12 20:47, 9 March 2007 (UTC)

Obviously that would be a problem, but it's not the case. How often do "a few who think 5, 6, or 7 thousand edits aren't enough" bring an RFA down to 75% but above 65%? If there are any RFAs that end up in that range, it is usually because of more serious problems, like civility. And that, to me, is one of the most important qualities of an admin. -- Renesis (talk) 20:59, 9 March 2007 (UTC)
Support of 66% might be appropriate, but isn't just as arbitrary as 75%? Before we make any modifications to a generally accepted range, we should at least have some confidence that the new level is better. Trading one arbritrary standard for another doesn't seem like a wise decision. I'd like to see some analysis before supporting or opposing this proposal. ChazBeckett 21:00, 9 March 2007 (UTC)
Yes, it's arbitrary. Right now we have a dynamic where oppose votes are being given way too much weight. I have for some time sensed a power dynamic in failing RFA nominations. People come towards the end to a borderline nomination and realize that an "oppose" vote may well be a deciding vote but a "support" vote will not be. It's tempting to be the one who tips the scales. The oppose voters aren't trying to convince the support voters of anything because they don't have to. Lowering the promotion threshold would change this and increase the importance of convincing others. The Uninvited Co., Inc. 21:46, 9 March 2007 (UTC)
  • Whatever arbitrary % levels that we wish to assign, the promotion rates will be largely unaffected in the long run. The problem isn't in the thresholds. The problem is in the perception that admins are held up to be god-like figures, and they are a truly separate class from other users. Further, that since there's not an *immediate* way available *right now* to stop someone from being an admin (other than steward) we just can't possibly reduce our standards here because then really bad admins would get through. You know, people like Karmafist (passed first RfA 53-2), NSLE (passed first RfA 71-1, and Yanksox (passed first RfA 104-4). Good thing we had the RfA system in place that we do, else they would have been made admins and we would have been in a real pickle! (apologies to those three people...I do not mean to say they are bad, but rather that RfA is singularly failing to filter out those that ArbCom forcibly de-adminned) --Durin 21:02, 9 March 2007 (UTC)
Your points are well taken, irony and all. Any sort of proposal that doesn't allow the broad community to vote for/against admins is rejected, and absent that, there isn't much we can do other than tinker with what we have. The Uninvited Co., Inc. 21:46, 9 March 2007 (UTC)
  • Yeah. Unfortunately, I don't think tinkering with what we have is going to lead us out of this mess. The pot keeps getting hotter, and far too many people essentially say "What problem?". --Durin 21:55, 9 March 2007 (UTC)
(edit conflict) There's a logical fallacy here. It doesn't follow that because three admins were promoted with strong support and later deadmined, that there is any conclusion that can be drawn regarding whether there would have been many "bad" admins that barely weren't promoted. The observation does not logically lead to the conclusion. —Doug Bell talk 21:59, 9 March 2007 (UTC)
Perhaps. The point I was trying to make is that there is no apparent correlation between support percentage and ultimate performance once adminship is granted. —The preceding unsigned comment was added by UninvitedCompany (talk • contribs) 22:18, 9 March 2007 (UTC).
This is entirely unfair. We couldn't possibly have desysopped people who were not admins. That there are desysoppings is an inherent nature of the fact that we have sysops. Removal of that access right would continue completely regardless of what 'fixed' system we institued in place of the RfA that is currently producing so very many bad apples at such an alarmingly high rate as 12/1100/3 years... Splash - tk 23:20, 9 March 2007 (UTC)
I think that something different needs to be done in the accounting of the "Neutral" votes. A large number of people have a tendency to vote neutral, when in fact they mean "I oppose, but I don't want to hurt your feelings." Right now, neutrals are factored out entirely as if the person made no vote at all. --After Midnight 0001 22:03, 9 March 2007 (UTC)
What difference do you (Uninvited) perceive this causing though? How did you arrive at those particular numbers? Do they solve something, or are we just trying to lessen the weight of those hateful opposers? Splash - tk 23:20, 9 March 2007 (UTC)

The point is the current system is incapable of weeding out "bad" admins (assume for the moment that "bad"=forcible de-adminned). Let's look at the cases of "bad" admins, and their RfAs (all such RfAs where there's a record of their adminship application at WP:RFA; some predate RfA and are not listed here):

That's 12 admins forcibly deadminned for which we have RfA records. Of these 12, only 2 of these fell into the normal bureaucrat discretion range of 75-80% (all admins above 80% on their RfAs have been promoted). 2. That's it. 8 of the 12 got >90%. If anyone still holds the notion that RfA is doing a good job of filtering out potentially bad candidates, and thus we need to keep the standards high, it's terribly flawed reason. RfA does a very poor job of filtering out bad candidates, at least by these above measures. --Durin 22:19, 9 March 2007 (UTC)

Again, a logical fallacy. How about you consider 1,141 admins, 12 desysopped. Seems like it's doing an exceptional job if only 1% of the admins that get through are "bad". Again, since we don't have any way to know how those with say greater than 50% support that were not successful would have fared, the conclusions are not supported by the observations. —Doug Bell talk 22:25, 9 March 2007 (UTC)
Not a logical fallacy: a logical fallacy implies a flaw in the structure of the argument which renders the argument invalid. You may be arguing that Durin's argument itself is invalid, but it is not because of the structure, but rather because of Durin's particular interpretation of the statistics. It is the difference between opinion and logic: you are of the opinion that Durin's argument is invalid; if Durin's argument were fallacious, it would be invalid, no questions asked. --Iamunknown 22:40, 9 March 2007 (UTC)

I think it's a good idea to increase the bureaucrat discretion range, but I think it should be expanded in both ways. What do you think that nominations ending on 60-90% is in the scrutiny of the bureaucrats? So a bcrat could fail a nom with 85%, and pass a nom with 65% etc... →AzaToth 22:42, 9 March 2007 (UTC)

While I think that some additional discretion would be good, too much discretion will mean that we need a dispute resolution process for RfA (much as DRV is for XfD). —Doug Bell talk 22:50, 9 March 2007 (UTC)

[edit] Removing discretion and controversy

  • At the German Wikipedia (the second largest by edit count), there is no bureaucrat discretion. It's 66% up/down vote with suffrage requirements. I had extensive discussions with two people familiar with their RfA and the adminship pool in general there yesterday. In general, the adminship pool is regarded pretty well. They do have de-adminships, but they do so without any functioning ArbCom, and no Jimbo stepping in either, and it is rare.
  • Interestingly, if you look at the history of this talk page, we get about 31 edits to this page per day (judging by the last 5000 edits). There, with 1/5th the number of admins, and 1/3rd the number of articles, they have 1/10th of the traffic (under 3 edits per day) at their RfA. There seems to be considerably less controversy there. --Durin 23:10, 9 March 2007 (UTC)
  • An interesting little paradox that I see at this version of RfA is that we have a very, very limited pool of bureaucrats. We expect ultra-high standards from them to pass RfB, and expect all power of determining consensus for single RfAs to be vested in single bureaucrats to act on their own, independent of each other. Then, we go apeshit crazy when a bureaucrat makes a decision that some think is against consensus. And we're surprised about this?
  • The bureaucrats aren't any better at determining who will make a great admin or not better than anyone else here. The arguments presented with respect to a candidate at RfA are never enough to predict the future. Further, support votes in the vast majority express no particular evidence or opinion other than "support". So, what the bureaucrats are given to work with anyways is very, very limited. And again, the paradox; we complain when bureaucrats do something we don't like.
  • People here are very bound up with the idea that RfA in its current form is something sacred, to be revered, and nothing...not any other system...could possibly do that which it needs to do. It's shocking that the other language wikipedias have miraculously managed to live on despite the obvious rampant problems their RfA system have which are direct, ongoing threats to their very existence since bureaucrats aren't entrusted to exercise discretion. --Durin 23:10, 9 March 2007 (UTC)
The "bureaucrat discretion," to the extent it exists at all, is another artifact. The way it used to work was that there was no concept of "closing" a nomination. Developers would occasionally look through the page and make promotions, noting this fact on the page. Any decision not to promote was tacit. Others would then archive the page eventually. At first, the bureaucrats were like that too, and there was no expectation that an RFA would close at a particular time, just that they would be left open for at least 7 days. Bureaucrats would promote whoever they felt met the threshold, and didn't remove "failed" nominations. That changed around the point when Cecropia became highly involved, at which time the concept that bureaucrats were supposed to "close" failing nominations gained currency. Before long, it became tradition that only bureaucrats could "close" failing nominations, and that's where we are now. When the percentages started to govern, the "discretion" range was a tip of the hat towards m:Voting is evil. We should get rid of it. The Uninvited Co., Inc. 23:45, 9 March 2007 (UTC)
I do agree with many of the points Durin and UninvitedCompany are making here. The exceptional pace at which we are filling up the WT:RFA archives isn't really bearing useful fruit for the Wikipedia project. So at that point why not just eliminate all the controversy and go with a straight vote, perhaps except for discounting disruption such as sockpuppets. The current system does seem to be working as a %1 "failure" rate is about as good as you can get in a human process considering it is impossible to fully predict the future. And while I do currently put a large amount of effort into the few RfA's that are close to the discretionary range to weight the arguments, look for disruption, etc, I can't point to any hard evidence that the end result is substantially better than a straight up down vote, nor that anyone could make it so. But with the evidence that we are significantly wasting edits and time on controversy that we don't need, and no evidence that the result is substantially better, perhaps that means we should eliminate it. But since any threshold is essentially arbitrary, we could just stick with the 75% we have now or move to 66%. I don't see it making much difference which number is chosen. Then the few bad apples that do make it through (and they will in any system) can be dealt with as needed. - Taxman Talk 00:12, 10 March 2007 (UTC)
If it is made into a straight vote, I believe that also would address a large problem with RFB. Namely, that people are scared of making more bureaucrats because of the potential impact a bureaucrat might have via his "discretion". If the expectation is fixed and rigid then there is a lot less ambiguity about what is appropriate. Dragons flight 00:55, 10 March 2007 (UTC)
Oh I think the failure rate of this process is much higher than 1%, unless you feel that promoting qualified candidates is not one of the goals here. In fact, on nominations in the 70-75% range the failure rate is probably closer to 100% than 0%. Christopher Parham (talk) 01:20, 10 March 2007 (UTC)

The following table is based on the closing support percentage for 650 RFAs occuring during the last year. Dragons flight 03:23, 10 March 2007 (UTC)

Support Threshold (%) 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100
RFAs At or Above (%) 85 81 77 74 70 67 65 63 61 58 54 53 49 45 41 37 34 31 26 9

So if everything stayed the same, moving from a ~75% threshold to a 66% threshold would increase the number of successful RFAs by about 20% (ie. (49-41)/41). Dragons flight 03:23, 10 March 2007 (UTC)

Good analysis. I think changing the threshold would change voting patterns somewhat, both by discouraging "power votes" (i.e. oppose votes towards the end that sink a nomination) and by encouraging the "mild oppose" voters to actually oppose rather than declare themselves neutral. I would like to believe that this would not change the overall figures that much but would improve the quality of the decisionmaking. (I also realize that this is probably wishful thinking). The Uninvited Co., Inc. 23:20, 10 March 2007 (UTC)

I had two proposals very recently, one to decrease the threshold to 70% (so very similar to this one), and second one to use voting on a subpage to decide if to adopt any given RfA proposal (since plain discussions by themselves never lead anywhere). So, I propose that we have a formal vote on a subpage as to whether to lower the threshold, and if that fails, let's not resurrect this threshold lowering proposal again. :) Oleg Alexandrov (talk) 04:01, 10 March 2007 (UTC)

One potential issue I see with the straight vote is the "Esperanza effect". You get a block of people that vote for their friends and those votes can overwhelm any reasonable opposition. Maybe this is more of a perceived problem than a real one, but I definitely had the sense for a while there that the Esperanza effect was not only real, but was getting to the point where it might start promoting poor candidates against community consensus. I like the safety valve of the bureaucratic discretion if for no other reason than to dissuade people of the idea that voting blocks would go unchallenged. —Doug Bell talk 05:43, 10 March 2007 (UTC)
I know that's the idea, but we can't really use it now as it is. There is endless controversy any time anything is done outside the strict numerical guidelines. And the discretionary range is so narrow that we can't really fix many problems now anyway. If the type of thing you are talking about goes on, the best we can do is have people point out that such is happening and if there is general agreement there is a problem with an RfA then we can call for special treatment. That would essentially be the same with or without a discretionary range. The good thing about the way RfA works is people can take into account the reasoning of others and formulate or change their position accordingly. That will still be able to happen. - Taxman Talk 15:09, 10 March 2007 (UTC)
Well, if Taxman, Durin, and I are all behind this maybe it's worth a vote. I would suggest that we shoot for a 66% threshold, since it has worked well at de: and there seems to be some support for it here. The Uninvited Co., Inc. 23:20, 10 March 2007 (UTC)

[edit] Deadminship idea from the U.S. government?

Considering that many users are reluctant to grant admin rights because there is no way to easily desysop bad admins, and many admins are reluctant to have a standard process for deadminship due to potential abuse from the enemies made in the course of standard admin activites, I thought of an idea. Admins are elected by the community. (Let's not kid ourselves, it is both an election and a vote). However, the community at large does not think that a deadminship proposal would be free from substantial, prohibitive abuse. I think that for once, the United States Government works in a way Wikipedia can learn from.

U.S. Representatives and Senators are elected by the people, but can only be forcibly removed from either chamber by their own. I think that deadminship should work like this:

Scenario 1:

  • Admin X accuses admin Y of serious misconduct.
  • A deadminship vote is opened on admin Y.
  • A notice of this case is posted on WP:AN. "Example: Admins are invited to review evidence at vote at Wikipedia:De-Adminship Proceedings/RyanGerbil10..."
  • In the deadminship vote (yes, vote, although both both parties, as well as outside observers, are allowed to provide evidence to influence votes) only admins may vote, although other users may comment/give evidence.
  • The vote lasts seven days.
  • Two-thirds of voters must vote to de-admin for deadminship to take place.

Scenario 2:

  • Non-admin user X accuses admin Y of serious misconduct
  • Three non-involved admins endorse that they find the actions questionable enough to open a deadminship vote.
  • Follows same as above.

Scenario 3:

  • Admin X does something terrible.
  • A deadminship vote opens.
  • Arbcom desysops admin X because it thinks admin X's actions were so bad that a deadminship vote is inadequate/too slow.
  • Wheel war erupts over what to do with deadminship vote page (hopefully not) :)

What does everyone think? I'm open to suggestions.

Also, I am the only one who reads deadminship as "dead-men-ship" and not "dee-ad-men-ship?" RyanGerbil10(Упражнение В!) 04:32, 10 March 2007 (UTC)

I read it sometimes one, sometimes the other. I like scenarior 3 a lot. I have an even better idea. Instead of ArbCom doing it let's have a Steward desysop, and then have in addition to a wheel-war have Stewards waring (not sure what that would be called? sphere-warring maybe? ). Seriously, what problem is there that you think this policy will help solve? JoshuaZ 04:50, 10 March 2007 (UTC)
I think one of the reasons the community is disinclined to lower its adminship standards is that there is no simple way to desysop people who do bad things. Theoretically, if we were going to allow many more people to be sysops, we would need a away to desysop many more people as well. For example, if the United States House of Representatives had 43,900 members instead of 439, I think it would be safe to assume that there would be 100 times as many congressman being removed. If we are going to allow a substantial increase in the number of people becoming sysops, we could much more easily desysop them as needed if there were a standard process. Currently, desysoppery usually comes as a post to the AN with meta diffs, showing the latest casualties from the admin corps carried out by stewards whom we have never heard of, oftentimes with little warning. If desysopping were more regular, sysopping would be more regular, which I believe would help the encyclopedia. RyanGerbil10(Упражнение В!) 04:48, 10 March 2007 (UTC)
Just a correction. The United States House of Representatives has 435 members, not 439. Captain panda In vino veritas 04:58, 10 March 2007 (UTC)
Well, 435 voting members... 439 if you count the various non-voting functionaries. RyanGerbil10(Упражнение В!) 05:00, 10 March 2007 (UTC)
I completely disagree with the idea that only admins should get to vote on desysopping. Yes, it works in the government, but for various reasons that don't apply to adminship. For instance, there is no such thing as an inactive Congressman, so they will all be better informed than most other people. Most important, though, is that the government is elected to represent us, not just to clean up bad things. Adminship is not, and should never be, a process where we elect people to represent us in important proceedings. Even if that would work, which it wouldn't, it's completely not the point of adminship. We have one body that we elect to decide things for us, and that is quite enough. -Amarkov moo! 05:07, 10 March 2007 (UTC)
Wow, that's a line of reasoning I never thought of... Still, maybe there's a compromise from this fairly proprietary system and one that's completely open to vandals, trolls, and the regular cadre of enemies an admin gains in the course of their regular duties? RyanGerbil10(Упражнение В!) 05:11, 10 March 2007 (UTC)
Vandals and trolls are easily kept out by a requirement that users who participate must not have been blocked in the past X months (undecided on what X should be), which I think is perfectly fair to expect. As for the regular cadre of enemies, the idea is to determine if the community trusts the admin, and the community includes good faith enemies too. The no consensus result should be to retain adminship, but nothing else is needed to deal with enemies. -Amarkov moo! 05:23, 10 March 2007 (UTC)
I don't think the current system is broken, no need to fix it. HighInBC(Need help? Ask me) 05:38, 10 March 2007 (UTC)

I've got an idea:

Scenario 4:

  • Admin misuses admin tools;
  • User (or admin) attempts to resolve issue with admin;
  • User opens RfC to get community input on admin's behavior;
  • Issue unresolved, user opens ArbCom case;
  • Members of ArbCom, elected by the community, decide if admin should be desysopped.

I don't know, it might work... —Doug Bell talk 11:21, 10 March 2007 (UTC)

We've only had 12 admins de-sysopped, ever (I think the figure is further up the page) and some of the things that went on with the most recent episode (the Brandt wheel war) couldn't have been dealt with quickly enough with any sort of de-adminship committee (or ArbCom), whilst long term misuse of tools is best looked at slowly and methodically through ArbCom, with the submission of evidence by any (and preferably all) interested parties. -- Nick t 14:10, 10 March 2007 (UTC)

What's the problem with having an admin review each year? .V. [Talk|Email] 15:02, 10 March 2007 (UTC)
There are too many. One for bureaucrats would be more feasable. Majorly (o rly?) 15:07, 10 March 2007 (UTC)
I think it would even out once the inactive ones are cleared away. .V. [Talk|Email] 15:08, 10 March 2007 (UTC)
Why should an admin go through a review if they have not done anything wrong? You do realize what will happen if admins start doing what is popular because review time is coming? Admins need to do things that are unpopular sometimes, like choose policy over consensus. Such a review system would clip their wings by forcing them to satisfy the masses every year. I think the current system is just great. HighInBC(Need help? Ask me) 15:37, 10 March 2007 (UTC)
Wikipedia is a consensus reality. If a consensus of Wikipedians agree that an administrator is doing something wrong, they'd boot him out. If they're not, he'll stay. If there's anything that absolutely must be handed down as black letter law, that will come through the proper channels. Having rogue admins is not a good idea. .V. [Talk|Email] 21:23, 10 March 2007 (UTC)
Explain "doing something wrong." Wrong as violating policy? Haven't we all done that? (<cough>WP:CIVIL</cough>) Wrong as not doing some...editors would want us to do? Ludicrous.
Having admins become mere pandering politicians is an even worse idea. Admins are here to keep the encyclopedia running smoothly, with force if necessary, not cater to the trolls that we often find on the boards reporting "admin abuse" where none exist. Yes, some admins do step over the line, but how many? 1%? So the actions of the few poison the entire well? --210physicq (c) 22:10, 10 March 2007 (UTC)
The situation you describe (about some editors not doing what you'd want them to) is very isolated. For example, if maybe five editors felt that way, and the admin was still an all-around good admin, there'd be no problem. If a consensus of Wikipedians are crying out against an admin's actions (keep in mind you'd need a rough consensus to de-admin in this type of review), then I think they should be de-sysopped. .V. [Talk|Email] 00:36, 11 March 2007 (UTC)

This is unnecessary, in my opinion. The ArbCom have proved that they will desysop when necessary. To put things to a "voting is evil" vote is a bad idea. We have ArbCom for a reason, and they will deal with admin abuse. Moreschi Request a recording? 14:17, 11 March 2007 (UTC)

But no sane person likes to bring a case to Arbcom. So unless an admin is terrible, he will not be desysopped, because nobody will be willing to subject themselves to the stressful process. I'm sure there are a few admins who are not trusted by the community, but stay because nobody will bring an Arbcom case. And since Arbcom only really desysops for abuse of admin tools, what happens to the people who aren't abusive, and just don't use them well? -Amarkov moo! 15:40, 11 March 2007 (UTC)
It should be an significant process to de-admin somebody. People should think twice, and they should go through the arbcom process. It is not broken, fixing it will just invent new problems. HighInBC(Need help? Ask me) 18:03, 11 March 2007 (UTC)

[edit] RfB / Stewardship

It occurs to me that perhaps we should just remove RfB from this page. When I look over the requirements for stewardship, and how that system works, I wonder if we shouldn't just do the same thing with Bureaucrats. Granted that there would obviously be a different set of criteria (such as suggesting that the cantidate be multi-lingual to be a steward, for example), but the actual process of gaining stewardship seems rather useful for gaining bureaucratship. I think this would deal with all the concerns and problems listed above. (If wanted, I can go through, point by point : ) - jc37 11:18, 10 March 2007 (UTC)

Well, I don't think your going to need to be multi-lingual on the English Wikipedia :) --TeckWiz ParlateContribs@ 13:26, 10 March 2007 (UTC)
I have a faint grasp on Frenglish and Spanglish - does that help? --Ozgod 16:25, 10 March 2007 (UTC)
Let's hear it for humour and an intentional misread of the post  : ) - See also m:Steward policies#Processes, for more of what I mean. - jc37 22:07, 10 March 2007 (UTC)

[edit] Wikipedia:Requests for adminship/peacefultimes

What's the process for handling fraudulent/inappropriate nomination pages. This user has 3 edits, and obviously didn't intend to fill out the nomination properly. Leebo T/C 20:07, 10 March 2007 (UTC)

I would suggest to close it stating that the user had 3 edits at the time of nomination, and leave a note in the user's talk page. He may be a sockpuppet, I barely knew how to edit by my third edit. -- ReyBrujo 20:24, 10 March 2007 (UTC)
  • I knew how to edit by my third edit. Anyway, I like the whole "bringing justice to vandals" angle, but this is a seriously malformed nom. .V. [Talk|Email] 20:26, 10 March 2007 (UTC)
    Actually, the "bringing justice to vandals" was added by another new account. Leebo T/C 20:27, 10 March 2007 (UTC)

[edit] Predicting RFA, round 2

Probability plot of RFA outcomes
Probability plot of RFA outcomes

Rather than preform necromancy by reviving the long abandoned thread at the top of the page (#Predicting RFA), I've decided to start a new thread to note that I have updated the probability image.

The two main changes are that I have added dashed lines to indicates the points above which all RFAs passed and below which all RFAs failed, and I have extended the time axis to 8 days after discovering that some of the blurring at 7 days was caused by RFAs that shifted a few percent between when 7 days was up and when they were expected to close. This basically confirms Durin's previously remark that no RFA that closed above 77.1% support has failed in the last year, though there are examples of RFAs that lost a few percent support right before closing in a way that likely affected the outcome.

There are 654 RFAs that went into this sample. This is a subsample of the 838 RFAs run in the last year after removing those that were active at times when my RFA summary bot was not working properly or that showed signs of corrupt data from the bot. I also removed the RFAs run for adminbots as unrepresentative of normal RFAs. Dragons flight 22:21, 10 March 2007 (UTC)

Your graph also factors in Carnildo's RfA, which may have somewhat skewed the results. Without his RfA, where would the lower dashed line be? --210physicq (c) 22:51, 10 March 2007 (UTC)
It would move up to basically trace Wikipedia:Requests for adminship/Ryulong 3 promoted at 69.4% support and then Wikipedia:Requests for adminship/Sean Black 2 promoted at 71.6% support, and no one else in the sample was promoted with less than 75% support. Dragons flight 23:25, 10 March 2007 (UTC)
PS. As noted in the discussion below, there were only two RFAs in the sample that failed when closed with greater than 75% support. So in total >99% of RFA outcomes could be predicted solely based on whether the support was greater or less than 75% at close. Dragons flight 19:41, 11 March 2007 (UTC)

[edit] The German RFA Solution

In agreement with the comments posted in #Changing the promotion threshold and #Removing discretion and controversy above, I'd like to seriously discuss adopting a German-like model of RFA. Namely, I propose that:

  • RFA is decided by a straight vote count after 7 days with 2/3 support required to pass
  • Voters may continue to provide the reasons and arguments for their votes, and such arguments can sway others into changing their votes anytime before the 7 days are up.
  • Bureaucrats are election officials whose job is to eliminate sockpuppetry and other disruption, but have no other discretion in deciding what passes or fails.
  • To vote in an RFA, a user must have at least 250 edits starting at least 1 month previously.

[edit] Advantages

  • Straightforward and easy to apply.
  • Increases the number of passing RFAs by ~20% (by moving from 75% to 66% threshold, see discussion above). Hence addressing the percieved need for more admins.
  • Retains the discussion elements, but drops the largely illusory concept of bureaucrat discretion (in practice every close above 77% has passed in the last year and every close below 75% has failed save 3, all of which were controversial).
  • Allows more bureaucrats to be promoted by removing the perception that bureaucrats need to be super-human consensus judging machines.
  • Recognizes that RFA participation is much higher now and hence the need for Bureaucrat discretion is much less than in the early days of RFA.
  • Ensures consistency in how RFAs are handled.

[edit] Disadvantages

  • Promotion is decided based on a vote.
  • Increases the number of passing RFAs by ~20%.
  • Eliminates discretion.

[edit] Discussion

Based on the comments in the sections above, both bureaucrats Taxman and Uninvited Company seem to be open to a change like this, and in my opinion there is no sense in getting worked up over a concept of "discretion" which in practice seems to not even exist any more. I'm sure some people will say that RFA should be less like a vote not more. However, I feel that this proposal is in many ways a reflection of what RFA has already become, but if someone feels that discretion ought to be increased then I would challenge them to make a counter proposal for how to accomplish that.

As a footnote, I would mention that my proposal actually breaks with the practice on the German Wikipedia in that they do not allow discussion to directly accompany votes. Personally, I feel our existing system is superior in that regard. Dragons flight 23:28, 10 March 2007 (UTC)

[edit] Question #1

Would only votes of Support and Oppose be allowed? In other words, do we get rid of all the silly Neutral so I don't hurt your feelings nonsense? --After Midnight 0001 01:17, 11 March 2007 (UTC)

I see no real problem with keeping the neutral section, as the comments appearing there are often helpful to allowing others to make up their mind. Dragons flight 17:03, 11 March 2007 (UTC)
Couldn't this be achieved with a "comments" section? --Ginkgo100talk 03:31, 13 March 2007 (UTC)

[edit] Comments

I really like this idea. For one, it eliminates the hazy requirement of b'crat decision, and makes it so that the decider is ultimately the voters; the swaying lies with the oppose/supports, etc, which I like. You are correct about the percentages in practice, and I think this would be a great step forward. .V. [Talk|Email] 02:09, 11 March 2007 (UTC)

I don't mind this, but may I suggest that maybe we adjust the percentage required to pass to 70% as a middle ground? Even though 2/3 is less than 4% lower than that, it just "feels" wrong (I guess symbolically more than anything concrete) to me that if a third of the people !(?)voting opposed, the candidate could still pass. I think it would be easier to swallow a change from 75% -> 70% than from 75% -> 66.7%. If adjusting the percentage to 70% results in more passed admins with no adverse effects (not that I'm anticipating any), then we could revisit it at a later date. I do like the minimum requirements for voting, although I suspect some might object. Anyway, just my lowly opinion. —bbatsell ¿? ✍ 05:15, 11 March 2007 (UTC)
If this were to be implemented, I would rather keep it at 75% As Dragons flight indicated above, all RfAs with less than 75% support have failed, with the notable exceptions of Carnildo, Ryulong, and Sean Black. Considering such minute chances of attaining adminship with an RfA that a quarter of voters oppose, might as well keep it that way. Remember, this is not the German Wikipedia. We have more people here, and hence have higher standards here. —210physicq (c) 05:28, 11 March 2007 (UTC)
See the discussion at the top of the page. We have a similar number of RFAs as a year ago, but we promote only about half as many and the ones we do promote on average have 50% higher edit counts (10000 edits today vs. 6000 a year ago). The ratio of admins to other active users has been falling steadily for a long time. Many people feel that actual promotion rates need to be higher, and this is one way to address that, just as the threshold moved from ~80% to ~75% a little less than 2 years ago. Though a 20% increase certainly won't address the entirety of the disparity, it would put a dent in it. Dragons flight 17:00, 11 March 2007 (UTC)

I like this idea because I think that the promotion threshold is too high, with the result that we are turning away good candidates. I believe that a 2/3 (or nearly equivalently, 65%) promotion threshold is the right one to use. The elimination of the supposed "discretion range" is really nothing more than a codification of current practice. I support the adoption of specific suffrage requirements whether or not the threshold is changed. The Uninvited Co., Inc. 10:01, 11 March 2007 (UTC)

I don't like it. RFA is not a vote. If the threshold limit has been lowered to compensate for minor opposes, then that makes it far worse. If this was implemented, people may invent all sorts of invalid reasoning to oppose a candidate and get away with it. Remember Boothy? =Nichalp «Talk»= 13:00, 11 March 2007 (UTC)

I agree with Nichalp. RfA is not a vote- nearly but not quite. Crat discretion is in my opinion one of the things that's right about our system. The knowledge that !votes must have some valid rationale to be counted by crats and that a decision will be made by a thinking person in tight cases are what IMO keeps the system sane. If all crat judgment is removed, we may as well have BureaucratBot... WjBscribe 13:41, 11 March 2007 (UTC)
Discretion, i.e. the concept that bureaucrats should apply their superior consensus judging ability to cases in the range of 75-80% support, is a myth. The statistics show and the active bureaucrats admit that it isn't being used, and a 75% throshold is the de facto standard. The core problem is that crats are human, and their judgment isn't really vastly superior to the mob, so once you have several dozen people commenting on an RFA crats generally don't feel comfortable using their power as if it were like ten votes to tip things away from the rest of the community's preferred outcome. Not to mention that many crats seem to agree promotion standards are too high, which gives another force causing things to be compressed against 75% anyway. At the same time, the appearance of dicretion makes people afraid to supprt RFBs because they feel any new bureaucrat must have superhuman discretion judging abilities. It is a concept that is a) not used, b) not particularly helpful (evidence suggests that problem admins rarely have marginal RFAs), and C) creating other problems via the inability to create bureaucrats who would work on other processes. Altogether I believe it is better in the current environment to embrace the de facto situation and eliminate the notion of "discretion". Dragons flight 17:14, 11 March 2007 (UTC)
Am I the only one who sees this as a democracy forming? Vote counting? I think arguments should be weighed on their merits. "Support because I like his userpage", or "Oppose because he sucks" should not have the same weight as well put reasoning. HighInBC(Need help? Ask me) 17:32, 11 March 2007 (UTC)
And what difference would there be if someone voted that way now? However, if people want to have a discussion about valid vs. invalid rationals, I don't mind. I wouldn't mine bureaucrats (as election officials) marking out frivilous votes. But after accounting for sockpuppetry and frivolous votes, I think direct tallying is the most fair, consistent, and realistic way to run a process that has grown to this size, so yes democracy. To say nothing of the fact that it is already being run as a democracy in >99% of cases. Dragons flight 17:43, 11 March 2007 (UTC)
Well, yes, after removing frivolous reasons for opposition, it should probably be a straight democracy. But that's what bureaucrat discretion is; discounting frivolous reasons. I don't think anybody's seriously advocating bureaucrats being able to make the decision, instead of just interpreting consensus. -Amarkov moo! 17:46, 11 March 2007 (UTC)
With respect, no, crat discretion is explicitly supposed to be the ability to make the close calls (where close was defined in practice as 75-80% support). Allowing highly trusted users to make close calls had a lot of logic to it when the average RFA had only 20 opinions, now I feel it is outdated. We talk about them making these calls by interpreting consensus, but there is historically a lot more power there than simply discounting frivolous opinions. Dragons flight 17:51, 11 March 2007 (UTC)
That doesn't mean we allow Boothy type votes to rule the Oppose section. The reason why bureaucratic discretion is hardly seen these days is that very few of the RFA noms fall in the discretionary range near the time of closing. =Nichalp «Talk»= 17:56, 11 March 2007 (UTC)
A little more than 4% of noms end up in the discretionary range, and yet in the last year only 2 noms closing above 75% support were failed (Wikipedia:Requests for adminship/Badlydrawnjeff - 75.7% and Wikipedia:Requests for adminship/Kafziel 2 - 77.1%). There have been dozens of opportunities to use that discretionary range, and yet it isn't used. Dragons flight 18:13, 11 March 2007 (UTC)
Which RFA do you feel that a bureaucrat has not used his discretionary powers to close it? =Nichalp «Talk»= 15:20, 12 March 2007 (UTC)
We can draw the conclusion that because bureaucrats decide to promote users in the discretionary range, they're not using it. However, it could also mean that they decide to ignore frivolous opposition (let's face it, it is much, much easier to demand reasoning from those who oppose than from those who support, due to our AGF culture), or because they agreed with nominators, etc. In short, correlation does not necessarily mean causation. Titoxd(?!? - cool stuff) 04:23, 12 March 2007 (UTC)
Don't ignore the two bureaucrats farther up the page arguing in favor eliminating the discretion as unused and counterproductive. Dragons flight 04:38, 12 March 2007 (UTC)
Well, not completely unused, just enough so that perhaps it's not worth the fuss. So in that sense it's counterproductive (globally for the project), not that it may not be effective for an individual case when used. - Taxman Talk 12:44, 12 March 2007 (UTC)
I like the 80% mark, to lower the bar would cause real problems. If more than 1 in 5 people are opposing you(after discounting frivolous and sock puppet votes), then there is a problem to be addressed. HighInBC(Need help? Ask me) 18:00, 11 March 2007 (UTC)
  • The mark is currently 75%, see directly above. Dragons flight 18:13, 11 March 2007 (UTC)
  • Yes there is a problem to be addressed. However, usually the problem is the terribly poor judgment of the people opposing. Lowering the threshold would have the beneficial effect of giving these individuals less influence in the process. Christopher Parham (talk) 19:23, 11 March 2007 (UTC)

Despite some reservations on whether this is necessary, I do like the suffrage requirements above (these could also then be adopted as minimum adminship requirements to appease that issue somewhat.) I'm not opposed to the rest of the proposal, but I share the concerns regarding the issue that currently people opposing are expected to provide a rationale, and this proposal removes that expectation. I think 75% is working: at 75% I think the problems of promoting marginal candidates and not promoting suitable candidates are fairly well balanced—both occur but neither has proven to be overly detrimental. While this might need to be adjusted if we change the process, I would not change both the process and the threshold at the same time. —Doug Bell talk 21:21, 11 March 2007 (UTC)

  • I think this proposal is very good because it would remove the pointless controversy that sometimes occurs here. >Radiant< 14:55, 12 March 2007 (UTC)
  • This is definitely a step in the right direction. Haukur 12:56, 16 March 2007 (UTC)

[edit] Having discussion before voting

One idea that occured to me was to have a period of discussion before 'voting' opens on an RfA candidate. There's a lot of talk about the standards for RfA being too high, especially regarding people who oppose for trivial reasons (less than x number of edits on [some section of wikipedia]). One other problem i see is that the format of RfA is essentially a vote and not a discussion.

There are defintely regulars who stick around and make votes on many if not all RfAs. But a lot of votes also come from people who vote, and then don't bother watching the RfA they've voted on anymore. This turns RfA into a vote because it makes discussion almost useless. If someone casts a vote on the first day, then doesn't watch the RfA...then no matter what arguments or evidence is presented later on, they won't change their vote because they're not even watcing the proceedings anymore.

If we have a period of discussion - say 3-4 days where people can't cast votes (so opinions aren't counted numerically), but can only discuss, then it should force RfAs to change from being a vote to being a discussion. This is because people would be forced to wait for the period of discussion is over before than can cast a vote that gets counted. Therefore, people with strong opinions regrading the candidate would voice it during the discussion. This leads to a period where editors can cross-examine the candidate, and respond to people who oppose or support. It also means the candidate can defend/explain themselves.

So it's like, the RfA would be listed here for a few days where it's discussion only. And then voting will begin several days later. This means the vote starts after all the information has been presented - unlike now, where you often have people who oppose, then bring out a new (and valid) point about the candidate that no one's picked up before. The discussion period gives a chance for people to try and convince other editors before people at large begin to vote.

At the end, RfA would still be counted as votes (as it is now), but at least this way, we can try and put the 'discussion' back into it. Any thoughts? --`/aksha 05:06, 11 March 2007 (UTC)

See WP:DFA. This was proposed and even implemented for a very short time but then it was (very unfortunately) slammed back down. I loved the idea then and continue to love it today, but it seems that most people are opposed. —bbatsell ¿? ✍ 05:08, 11 March 2007 (UTC)
Huh... I wonder why it was rejected. Seems like a better system to me. What were some of the arguments against it (other than the ones listed on the WP:DFA page)? Leebo T/C 05:26, 11 March 2007 (UTC)
Why was it rejected? Looking at WP:DFA, the two "counter-arguments" i see seem very (to put it bluntly) stupid. "We shouldn't be discouraging people from changing their minds" <-- discussion before hand isn't discouraging people from changing their minds. In fact, it would be the opposite, since it forces people to think and read the arguments BEFORE they vote. And it's pretty clear that in the current system, people rarely change their minds (just see how often in a RfA we have crossed out votes). "Once the discussion period concludes, and a new piece of info is brought up, what then?" <-- well, we just keep discussion. Or people bring up the new piece of information in their votes (which is what happens anyway). This may be a problem, but it can't be more of a problem that our current system. --`/aksha 05:39, 11 March 2007 (UTC)
If I recall correctly (and it's entirely possible if not likely that I don't), WP:RFA was shut down and redirected to WP:DFA rather ham-handedly by a couple of people without much prior notification, and once word got out, everyone and their mother came out of the woodwork to oppose it. It might stand a better chance now if it were to be reformulated and re-proposed (and publicized well ahead of time on AN and VP). It has my support. —bbatsell ¿? ✍ 06:00, 11 March 2007 (UTC)
Well, i'm thinking if people are interested, then we'd need to start off with a few "test-subjects". We'd need a few people wanting to request for adminship in the near future willing to try this out. IMO, trying to do a 180degree turn in putting into a place a new system just isn't going to work. Any changes (and that applies for the other RfA reform ideas too) will have to be phased in slowly, not suddenly. --`/aksha 08:31, 11 March 2007 (UTC)
  • It seems like a good idea and I'm surprised the earlier proposal failed to gain traction. A few questions/concerns:
    1. Who sets the date from when voting opens? Is it automatically transclusion + X days? Who updates the page to say voting is open?
    2. What happens to early votes? Can anyone strike them, or only bureaucrats?
    3. What about durations? WP:DFA would have expanded the total time to two weeks. While nobody should be in a hurry to gain adminship, isn't it a burden to expect a candidate to keep an eye on something for that long?
    4. For any "guinea pig" candidates, the voting period would have to be 7 days as this policy is not yet formalized.
    —dgiestc 09:00, 11 March 2007 (UTC)
  • I'll try and answer those questions with a few possibilities.
    • The voting would begin x number of days after discussion begins. We will need to decide the "x" (i.e. how long the discussion period is). Once x number of days has passed, anyone can update the page to say voting has begun. After voting begins, people can still add to the discussion, or make comments in their votes (like the way we already do). Beauracrates should be allowed to extend the discussion period if they think it's nessasary, in the same way they're currently allowed to extend someone's RfA period.
    • There's no such thing as early votes. During the discussion period, we just have the candidate's statement, followed by questions and answers, followed by a discussion section. When voting begins, three extra sections for "support", "oppose" and "neutral" are added and voting begins. So it's like, when someone wants to count votes, they only count the votes listed under support/oppose/neutral, and ignore the discussion section above it.--`/aksha 09:13, 11 March 2007 (UTC)
      • Someone will create a support/oppose section and vote because they are not familiar with the new system. Their opinion still matters, but do you strike it, merge into the comments section, what? If we don't sort this out ahead of time there will be major drama when it comes up. —dgiestc 09:35, 11 March 2007 (UTC)
        • It would have to be merged into the comments section. They could move it back after the voting period starts, I suppose. The Uninvited Co., Inc. 10:13, 11 March 2007 (UTC)
          • Merge into discussion section. But don't move it back when voting starts - the point of having the discussion period first is so that people can't vote until after they've discussed. We should just move it to the discussion section, and leave the person a talk page note. I think it's safe to assume people voting in RfAs are active enough Wikipedians that they can check their talk page at least once in a few days... --`/aksha 10:41, 11 March 2007 (UTC)
    • as for durations - i don't know. people will need to decide on how long the discussion period is, and how long the voting period will be. If the current RfA is 7 days and it works fine, then i don't see why something like 3 days of discussion followed by 4 days of voting (total 7 days) won't work. But i really don't follow RfAs much, so i don't really know how well our current 7-day length works. --`/aksha 09:13, 11 March 2007 (UTC)
      • From what I've seen of the stats, most RfA are pretty clear by 5 days. Perhaps we can do 3 days of discussion and 5 of votes. —dgiestc 09:35, 11 March 2007 (UTC)
        • I thought you said you wanted to keep it as 7 days? --`/aksha 10:41, 11 March 2007 (UTC)
          • Yeah that was rather unclear. What I meant was: I think a 3+5 period is a good idea. Bureaucrats currently close RfA's according to WP:RFA, which while not explicitly identified as policy, is at least a long-established consensus of 7 days voting. In a trial period, we would need to keep 7-day voting unless there is a consensus to override that which has had broad community input. Otherwise we can have people complaining bureaucrats and policy wonkt have speedy-promoted a few admin candidates. —dgiestc 18:13, 11 March 2007 (UTC)
            • I think 2+5 would work better. In most cases discussion (discussion alone, not voting) would be difficult to sustain over 3 days. IronGargoyle 19:30, 11 March 2007 (UTC)
              • I think 2 days is too short because we need to take into consideration that not everyone checks wikipedia every day. Even for regular editors, not logging in for a 2 day period every once in a while isn't abnormal. So hence i think both the discussion and voting periods need to be at least 3 days to give everyone a fair chance at having their say. --`/aksha 07:44, 12 March 2007 (UTC)

At this point, I support this proposal simply because I believe that any sort of innovation and fresh thinking at RFA is worth strong consideration. I don't believe this proposed change would do any harm. I do see RFAs that suffer from early "drive-by" votes (either support or oppose) with rationale that later proves baseless. The DFA model would help prevent that. The Uninvited Co., Inc. 10:13, 11 March 2007 (UTC)


This seems like a good idea but I'm wondering what it's meant to do. Are people expecting more admins to come out of this process? Or better qualified (whether more or fewer are promoted)? In the first paragraph above it's said that the standards are too high. Now that may or not be the case, but will extended discussion change peoples standards? Will people be willing to discuss their standards on every RFA they !vote on or will they just wait for the !voting period to start? I guess what I'm getting at is that unless there is a factual error (no, he wasn't blocked that time or it wasn't her that made that comment.) I don't know if a period of discussion before !voting will make any difference. What I'm afraid it will do though is to turn up the volume on the mudfights that occur in controversial RFA's. I don't see this helping the culture at RFA and I'm not so sure it'll promote more admins. I'd like to hear what people are expecting out of this. Maybe some examples of RFA's that had new information presented later in the voting that might have changed some opinions if people had come back and reconsidered would help. RxS 17:25, 11 March 2007 (UTC)

What it's supposed to do is reduce the number of opposses and supports for trivial reasons by creating officially a period of time where people can't cast votes, but are able to discussion and bring out evidence and arguments. Hopefully, people will participate or at least look at the discussion before they vote. Also, a lot of people who vote probably never bother to come back and read votes and comments left by voters after them. So if someone digs out some new evidence or brings up a new concern that can change a lot of voter's minds - but it happens late into the RfA, then the people who voted early will probably never see it (because they don't bother to come back and follow what happens after them). And from what i can see, candidates or supports often leave arguments after oppossing votes that generally don't get a response. This could be because the opposer has made up their mind, but it could also mean the person who voted oppose just didn't bother to come back and see the responses to their vote. --`/aksha 07:44, 12 March 2007 (UTC)
I like the idea a lot, a mandatory discussion period for 24-48 hours prior to the polling would really help clear out several issues. People often vote on day 1, then new information comes out on day 2, but they never see it. HighInBC(Need help? Ask me) 17:36, 11 March 2007 (UTC)
Indeed, its a good middle ground between the pure vote and other possible methods. Voice-of-All 18:04, 11 March 2007 (UTC)
I don't much like this (or any 'discussion'-based idea) for purely practical reasons. Even two days' worth of unorganized discussion on a controversial candidate will turn into a hydra of redundant subthreads that's impossible to sort through, probably resulting in everyone just doing whatever they were originally inclined to do in the 'voting' phase due to sheer lack of stamina in plowing through the masses of text. On the other hand, the discussion phase of an uncontroversial candidate will likely be indistinguishable from voting, as a bunch of people post variants of 'yep, he's a swell guy'.
One of the main advantages of the current system is in structuring the subsequent commentary. Within the current system, the single most productive change would be to get rid of this widespread idea that questioning or challenging a rationale is tantamount to 'badgering' and strongly discouraged. 'Consensus formation' can't happen in a meaningful sense if you aren't supposed to initiate extended dialogue with the other participants. This also contributes to the 'drive-by voting' problem because people have formed the expectation that they won't be called on to explain or defend their position. Opabinia regalis 04:47, 12 March 2007 (UTC)
100% agree. Responding to votes (both support and oppose) help show the candidate where they may need to improve, encourage voters to think through their position and help show if a vote is disagreed with by others. Some people seem to take affront when their vote is challenged, in a kind of "I don't need to explain myself" way. That seems completely the wrong attitude. Trebor 07:31, 12 March 2007 (UTC)

(unindent) HOwever, a problem i see is that people don't respond to responses to their votes. Per HighInBC's "People often vote on day 1, then new information comes out on day 2, but they never see it.". I guess the people who regularly vote at RfAs would keep reading RfAs after they voted on them, but i somehow doubt all 100+ people who vote on a RfA are coming back to check what's happening after they vote.

One thing we can clearly see, is that the "discussion" section on the RfA template is (currently) almost never used. It seems to be a good idea to actually have some discussion there in the discusison section... --`/aksha 07:44, 12 March 2007 (UTC)

I've got a feeling that there's usually one or two niggly little things that come out during an RfA and the only thing that stops people from changing from Support to Oppose is the complication of striking out their !vote, rewriting their comments in the other section and changing the tallies. I think if there was discussion to start with, come to the voting stage, there would be a lot more Oppose votes and nobody would be promoted. -- Nick t 12:25, 13 March 2007 (UTC)

We should probably have discussion instead of voting. List the RFA for a few days to gather comments. If no good reason not to promote emerges, then promote. --Tony Sidaway 12:43, 13 March 2007 (UTC)

Then your left with someone (a 'crat) needing to decide what is and isn't a good reason for promotion. -- Nick t 12:46, 13 March 2007 (UTC)
I think we should have discussion or voting, instead of combining the disadvantages of both as we currently do. Both decision-by-bureaucrat-after-community-discussion or bureaucrat-bot-after-community-vote would probably cause less arguing than the current situation where some people vote ('''Support'''. ~~~~) and others discuss, both sides criticize each other ("please provide a reason" vs. "please do not badger the oppose voters"). I don't think we can achieve any change of RfA other than through a cabal decision by the bureaucrats, though, so discussing here instead of convincing individual bureaucrats is probably a waste of time. Kusma (talk) 12:54, 13 March 2007 (UTC)


I think you've gotten it the wrong way round, Nick. We should, rather, look to see if there's a good reason not to promote. Good reasons would be evidence of immaturity, bad behavior, or poor judgement, or simply lack of experience in editing. Bureaucrats, aided by a discussion, are quite capable of spotting this and determining whether there is a reason not to trust the applicant. I don't think these beauty contests serve any useful purpose. They're of little value in determining community consensus and they give a very misleading impression of what administrators are all about. And frankly it's become a bit of a scrum. --Tony Sidaway 13:01, 13 March 2007 (UTC)
Hi Tony, welcome back. While I agree with you that the crats should be capable of deciding what is and isn't a "good reason" not to promote, we'd probably have to determine via community consensus what those reasons are. I say that because people have passed RFA while defending copyright violations, failed for enforcing policy, and been opposed strongly for both spending too much and too little time in project space. Where do we give the crats guidance to make these crucial decisions? -- nae'blis 14:19, 13 March 2007 (UTC)
People should on no account pass RFA whilst advocating destructive activities. If the current RFA system has done that then it's a failure of the current system and an excellent reason to place such decision in the hands of the bureaucrats, who may be persuaded by discussion within the community to alter their criteria. Community consensus is only part of the issue here. There are some opinions, and advocating that Wikipedia should ignore or subvert the copyright law is one, that should automatically disqualify an administrator candidate, and it doesn't matter how many people think he'd still be a valuable admin. And ideally I'd like to see a system that makes bureaucrats more or less immune from coercion, but perhaps that's asking too much. Accountability is important, but I'm not sure we've worked out how to define the responsible community to which bureaucrats should be accountable. Probably not everybody who can edit WP:RFA, but if not, to whom? --Tony Sidaway 18:12, 13 March 2007 (UTC)
There's no good reason to start opening that can of worms. People want more discussion, so go quit yapping on the talk page and go discuss. If someone whines about 'badgering', send them to this talk page. Maybe even put in the 'expressing opinions' section of the front matter that users who comment in an RfA should try to keep abreast of further developments, and should be aware that their reasons for supporting or opposing may be questioned or disputed by other participants. As usual, we don't need a 'government' solution; we just need some ordinary citizens to put their money where their mouths are. The bad rationales will sort themselves out organically. Opabinia regalis 02:51, 14 March 2007 (UTC)
RFA does have a real, though small, problem with badgering. I believe it's correct to say that persistently bad behavior on RFA has sometimes led to blocking. --Tony Sidaway 13:44, 14 March 2007 (UTC)

[edit] Expressing lost of confidence

The reason we should have either of following changes because we cannot descope an admin unless he has done major mistake but we can lose confidence on him even if he has not done some major violation of policies. For example we think he has been pushing POV or being rude few times. It is very painful process to file arbitration. Hence we should have someway to express our lost of confidence on him. Here are two ways to make it easy.

  1. Let say I voted for an admin. However, later he changed and I lost my trust on him. Then I should have right to withdrew my old vote and let Bureaucrats reevaluate his RFA (based on all the withdrew votes since he had been admin).
    Addition:- We should remove an admin only when a large number of people think that he was a bad selection, so that good admins can work without any fear. Let say 50% widthdrew votes make him lose his post.
  2. Otherwise, We should select a admin for a period of one year at most. After that he has to come in front of community and once again go through the process of RFA.

Wikipedia will be a much better place if we have one of the above points implemented. Looking for your reply. --- SAndTLets Talk 19:17, 11 March 2007 (UTC)

  • To the first point; that means a "close call" RfA leaves power of de-adminship potentially in a single user, or two users. This leaves that admin essentially powerless; annoy someone but once, and you're done. Nothing should stand in the way of an admin doing what is right. Yet, what is right frequently angers people. I can't tell you how many people I've seriously annoyed with my work on fair use. This is precisely why I never put myself in the willing-to-be-recalled list. Doing the right thing makes you unpopular around here.
  • To the second point; this has been brought up again and again and again and ... *50. It's never gained traction, and it is unlikely to do so for the simple reason of scale. We have more than 1,000 admins right now. Even if only have of the current admins are active, and we "weeded out" the inactives, we'd still be left with ~12 RfAs per week for re-adminship. Plus, we'd have the additional problems as described in the previous paragraph. --Durin 19:57, 11 March 2007 (UTC)
May be you are right however we have to have something that makes them answerable to large community. May be to remove an admin we should need more than 50% withdrew votes. That will not stop him doing the right things and he will lose his post only when many agree that he was a bad selection. If his selection was really a mistake then we can avoid long and painful arbitration. --- SAndTLets Talk 20:08, 11 March 2007 (UTC)
The last thing we need, as I have always said, are admins who act essentially like politicians, courting people's impressions to pass their next RfA. Admin actions will be controversial from time to time, and most of them have been upheld. This proposal essentially was substantiated due to the abuses of less than 1% of the admins on Wikipedia. To subject admins to a continuous meat-grinder for no apparent benefit is only a waste of everyone's time. —210physicq (c) 20:14, 11 March 2007 (UTC)
That will not work in the case of longer-serving admins: I haven't gone through my RfA, but I am sure that less than half of those who supported me actually contribute with any frequency these days (I was promoted a little over two years ago). Also, those admins who were promoted with huge levels of support would still be immune to this process; there is simply no way that you could get 50+ people worked-up enough to withdraw their votes if the admin's actions weren't bad enough to take them through ArbCom anyway. Rje 20:22, 11 March 2007 (UTC)
I do believe that more community accountability is needed, but the solution to that is to create a de-adminship process that requires a consensus to desysop. Not vote withdrawl, and certainly not yearly reconfirmation. -Amarkov moo! 23:02, 11 March 2007 (UTC)
  • Been there, done that, didn't work... - Mailer Diablo 12:13, 12 March 2007 (UTC)
  • This would only work if you'd also allow people to add support votes later on. Also, it needs a delay factor of some sort, otherwise it's too vulnerable to spur-of-the-moment reactions. All in all I'm not at all sure it's workable. >Radiant< 14:57, 12 March 2007 (UTC)

[edit] Can we agree on a simple thing?

I do not care if we all agree on above two mentioned points. However there is a need to develop a process that makes admins accountable to community. Such a process should not so hard that one person spends many days writing an arbitration case and that case takes months to finish. It should be a process that gives power to each individual and each individual take a tiny step towards removing a bad admin. A good admin should not be worried about annoying other people while taking right decision and only a very large majority should be able to remove a bad admin. If you agree that we need such a process (do not think about what exactly that process should be right now) then please vote below as yes.

NOTE: Please do not vote below. Go to discussion at Wikipedia:Adminship survey/N. --- SAndTLets Talk 16:01, 12 March 2007 (UTC)

  • Yes! we should make them more accountable than they already are (yes)
  1. --- SAndTLets Talk 15:47, 12 March 2007 (UTC)
  • otherwise (no)
  1. Admins are already accountable, the existing arbitration process is fine. I don't see the current system as broken at all. Also, if de-admining is not the desired outcome, then regular community accountability takes effect, just like anyone else. HighInBC(Need help? Ask me) 15:49, 12 March 2007 (UTC)
  • Comments
Don't fork discussions; there is ongoing discussion about this at Wikipedia:Adminship survey/N. --ais523 15:52, 12 March 2007 (UTC)
Thank you for pointing that out. I was NOT at all aware of it. I go there now. --- SAndTLets Talk 16:00, 12 March 2007 (UTC)
And I realised after I wrote that that it might have come off the wrong way (it's hard to read tone in online communications); I meant something like 'please contribute at Wikipedia:Adminship survey/N, to avoid forking discussions'. Sorry for any confusion! --ais523 16:01, 12 March 2007 (UTC)
No problem. --- SAndTLets Talk 16:02, 12 March 2007 (UTC)

[edit] Lots of discussion, and no agreement again.

Is there anyone left who feels this essay has it wrong? http://www.shirky.com/writings/group_enemy.html --Durin 21:14, 12 March 2007 (UTC)

God, that keeps being brought up. Does anyone have a solution, or are we just going to keep hearing "waah waah the group is broken"? -Amarkov moo! 00:16, 13 March 2007 (UTC)

So? Maybe we're wrong about the discussion thing, and it's easy enough to test. Find an intrepid candidate willing to play guinea pig and get him to request that people confine their comments to the discussion section until two days after the RfA was posted. Troutwhack anyone who then opposes over a malformed nomination, and see what happens. Not everything has to change by saltation. Opabinia regalis 00:23, 13 March 2007 (UTC)

Why would it be malformed? --TeckWiz ParlateContribs@ 00:27, 13 March 2007 (UTC)
IIRC, that was done before, and then people started opposing for not allowing the votes. It's not easy to force people to accept a different process. -Amarkov moo! 00:29, 13 March 2007 (UTC)
  • Given the recent flurry for strongly supported passing candidates, I sincerely doubt that "RFA is broken". Rather, the "brokenness" is in the people that don't dare to put themselves forward as a candidate because they believe the process is broken. That is circular reasoning. >Radiant< 07:31, 13 March 2007 (UTC)
    • On the other hand, I wonder if there's been a flurry of new voices in RFA that will strongly support decent candidates to push their percentage out of any danger zone. It seems like everyone in the last month or so has either had >90% and passed or <50% and flamed out. I'm not sure that's a good way to adapt the system, but I'm cynical about our ability to change right now. -- nae'blis 14:16, 13 March 2007 (UTC)

Rabbit season! Duck season! Rabbit season! Duck season! Rabbit season! Duck season! --Durin 00:11, 14 March 2007 (UTC)

[edit] If RfA is revamped...

If the whole RfA process receives a massive makeover, whether now, or in the future, are present administrators willing to stand for reconfirmation of their status under a new RfA process? I am personally open to voluntary dropping of tools, and community re-examination, at any time, but I wonder how other admins (particularly those who have become very comfy with the tools :) ) feel about it all. – riana_dzasta 14:51, 13 March 2007 (UTC)

How do you suggest going about the processing of 1000+ admins in an efficient manner? Alphabetically by username? Is there even a requirement of reconfirmation if no foul play on the part of the admins is detected through an examination of their contribution logs? It would make sense to use the RfA time in an efficient manner and a reconfirmation of established good faith and sound actions wouldn't be the best way of doing this. Better to set up a review panel with criteria for pass/reconfirmation clearly defined so as to not create grounds for equivocation. (aeropagitica) 15:48, 13 March 2007 (UTC)
I'm not suggesting everybody does it. I'm asking if people are willing to resubmit themselves for RFA voluntarily, much like the present recall process. I'm working under the assumption that if the process changes, people's ideas about the process, and hence their ideas about administrators, will change; hence, their opinions about present administrators, and how they attained their status, might change. Obviously, forcibly putting 1000+ admins through RfA would be foolish. – riana_dzasta 16:01, 13 March 2007 (UTC)
I still don't see what you're trying to gain here. A good 30% of those standing for reconfirmation might be declined, which would reduce our active admin corps. A functioning involuntary de-adminning process between CAT:AOR and ArbCom seems like a better use of our time. -- nae'blis 16:26, 13 March 2007 (UTC)
(conflict) :::Most of the ideas seem to be aiming to make RfA easier to pass, so you could probably assume that if they passed the "old" version, they'd pass the "new" one. And if you've been a relatively trouble-free admin, then you should easily pass any test (what better way to judge if someone would be a good admin than by looking at them...being an admin). I'm sure many (most?) admins would be willing to put themselves through the new system, but I don't feel it's that important. (This is all pretty hypothetical anyway, as if present discussion trends continue, nothing will ever get changed.) Trebor 16:32, 13 March 2007 (UTC)
No, I don't see what I'm trying to gain either. I'm thinking aloud (typing aloud?) and wasting people's time :) I know it's not that important, I'm not trying to force people to resubmit themselves or anything like that. – riana_dzasta 17:19, 13 March 2007 (UTC)
While I am for admins going back through the nomination process (maybe not every year, but maybe every three years), current admins should not have to go back through the process again just because the process changed.↔NMajdan•talk 19:08, 13 March 2007 (UTC)
I agree. If you feel the burden, you could possibly do so, but it would really be uncalled for. You earned your mop, why should you be screened to see if you're holding it right? bibliomaniac15 01:13, 14 March 2007 (UTC)
If mass admin recall ever happens it should be 50% the promotion rate as most admins gotten a few enemies. Jaranda wat's sup 03:11, 14 March 2007 (UTC)
On the other hand, admins also tend to make friends and increase their exposure to other long-term contributors. While "there is no cabal", I would be surprised if admins didn't often band together to support each other. Personally I would be concerned about any sitting admin who couldn't get at least 75% support to continue as an admin. Dragons flight 06:09, 14 March 2007 (UTC)
Friendship may fade, but enemies never forget. Raymond Arritt 07:02, 15 March 2007 (UTC)

If RFA is revamped in a way that I think will work, it'd be silly not to trust it. I'd run. Anyone else? --Kim Bruning 03:49, 14 March 2007 (UTC)

Want to be the Guinea pig that Opabinia regalis mentions above? Titoxd(?!? - cool stuff) 04:34, 14 March 2007 (UTC)
To answer Riana's question: I wouldn't mind going for reconfirmation. I would mind 1,100 admins going for reconfirmation, 600 admins going for reconfirmation, or even 150 admins going for reconfirmation. Time is a scarce resource in Wikipedia; let's not waste it that way. Titoxd(?!? - cool stuff) 04:35, 14 March 2007 (UTC)
No, I would not stand for reconfirmation. HighInBC(Need help? Ask me) 04:49, 14 March 2007 (UTC)

I am confused, if the point of revamping RfAs is to make them easier to pass, why would we want admins who went through the harder-to-pass present system to stand for renomination? WjBscribe 05:01, 14 March 2007 (UTC)

That's the point. Why would we? bibliomaniac15 05:25, 14 March 2007 (UTC)
If I felt I didn't meet any criterion set forth in any new system then I would happily seek the communities reconfirmation. Otherwise I wouldn't see the point. Rockpocket 05:41, 14 March 2007 (UTC)
  • This is a WP:PEREN issue. Also, what WJB said. >Radiant< 09:28, 14 March 2007 (UTC)
You are indeed confused. The point is not to make it easier to pass. %-) --Kim Bruning 15:04, 14 March 2007 (UTC)
Ooh, ooh. I know this one! :-) The point is to find those suitable for doing the job, while somehow coping with the problems of scale? Right? Carcharoth 15:55, 14 March 2007 (UTC)
I guess there is an underlying assumption here, that in the encyclopedia that anyone can edit, that there will be enough people trusted to do admin stuff and avoid backlogs, even if everyone in the world really was editing... This is more a problem of scale for the encyclopedia as a whole, rather than purely a case of RfA scaling up. Carcharoth 15:54, 14 March 2007 (UTC)

It occurs to me that the only way a reconfirmation system might ever be implemented is if some of the admins just WP:BOLD it into existence by resubmitting their RfAs and making reconfirmations a reality. If some of them were to prove that admins can indeed be reconfirmed, it might sooth the fears of those who worry it RfDA would lead to mass extermination of admins, and we could then work out the actual details as we went. --tjstrf talk 08:06, 15 March 2007 (UTC)

The process for designating administrators has changed and evolved through many stages. Despite all of these changes, it has never been expected that those chosen in previous iterations must submit themselves again under the new version. --Michael Snow 17:52, 15 March 2007 (UTC)

[edit] Separating out admin powers

I'm sure this is a perennial proposal as well, but would separating out admin powers in a logical manner help resolve some of the issues that are raised here? For certain types of administrative backlogs (those often called 'housekeeping' tasks), there would probably be more chance of getting more people to apply and successfully pass RfA, though really, automation and more bot-work might be a better solution. For the other admin tools, such as blocking and making difficult judgements and decisions, higher thresholds and standards are quite rightly applied. Would there be any chance of achieving some sort of separation of these tools into two roles? We already have bureaucrat, steward, checkuser and oversight flags, so splitting admin into two flags wouldn't really be that big a step. If this has been discussed before, could a link be provided, please. Thanks. Carcharoth 13:58, 14 March 2007 (UTC)

It is indeed a perennial proposal. See Wikipedia:Limited administrators. It's a nice idea in theory, but it's just not practical. It's too confusing, not only for other administrators, but for non-admins who are looking for help and have to ask 5 people before they find someone who can do what they need to get done. If you can trust someone with the tools, you can trust them with the tools. If you can't, you can't. Kafziel Talk 14:05, 14 March 2007 (UTC)
I have yet to see any problem that would require this sort of change. HighInBC(Need help? Ask me) 14:27, 14 March 2007 (UTC)
This could be practical IMO if we kept it very simple. Separate out blocking from the other functions. This is what gets people in the most trouble when they abuse it. The non-blocking admins could be called "janitors" and people with the block button could be called "bouncers". Friday (talk) 14:39, 14 March 2007 (UTC)
Admins should not think of the block button as a bouncer tool, it is also a janitors tool. I don't think it would be a good thing if an admin had the block button, but not the protect button, sometimes it is better to protect than block, but if I only have one button... HighInBC(Need help? Ask me) 14:42, 14 March 2007 (UTC)
Well, I was assuming block included the other functions as well. Since blocking is what gets people in trouble, anyone who's trusted with that is also (presumably) able to be responsible in the use of deletion and protection. A pure wiki deletion system would remove deletion from the admin picture altogether, but I suppose that's a matter for another page. Friday (talk) 14:44, 14 March 2007 (UTC)
I've suggested this before (several times, I think : ) - That the ability to block/unblock be separate from the admin package of userrights, the way checkuser is separate from the admin package of userrights. I think that this would go a long way towards dealing directly with the questions of (dis)trust. - jc37 14:18, 16 March 2007 (UTC)

Could Admins emeritus retain the ability to check deleted revisions? If that were the case, I could get many more people to join the emeritus cabal. (And I've been told that it's technically possible, it only needs community support) --Kim Bruning 15:06, 14 March 2007 (UTC)

Everyone should be able to see deleted revisions (except those that were oversighted, which is a tiny tiny chunk.) This transparent deletion is all advantage with no disadvantage. Friday (talk) 15:15, 14 March 2007 (UTC)
Wow, this is a big day for hauling tired old discussions out of the attic, isn't it? ;) We've got three going at the same time! Kafziel Talk 15:19, 14 March 2007 (UTC)
I like the Wikipedia:Transparent deletion one. It doesn't seem to have got much coverage (and is not really relevant to RfA in any case), but probably deserves more coverage. I remember being shocked when I realised that any edits I made to articles that were later deleted were also deleted from my contributions history. Well, not technically deleted, just no longer visible. This includes prodding articles that get deleted (nominating at AfD leaves more of a trail), and any edits to articles that were later deleted. It is also incredibly frustrating to come across old discussions when following a trail of some sort, and to not be able to see what the discussion was about (when the page being discussed has been deleted). Another problem with deletion is what Brion Vibber said here in January: "Deletion means deletion" (he said more as well). The balance between: (a) oversighting inappropriate stuff; (b) keeping deleted revisions to check on them and restore them if necessary; and (c) getting rid of rubbish - needs to be made clearer. But I should get back to the limited admin idea I resurrected at the beginning of this thread... Carcharoth 15:44, 14 March 2007 (UTC)
Folks, let's just wait for Bitfields for rev_deleted. It is coming Real Soon Now™. And yes, I do mean real soon, I've actually tested the thing. See the gallery at the bottom of that page. Titoxd(?!? - cool stuff) 19:51, 14 March 2007 (UTC)

Re: the janitor/bouncer thing, I'd suggest calling them all admins. That way it is not immediately clear which ones can block you, and people would, well, be more civil... :-) Seriously, I've so many times encountered the attitude among editors that they feel they can go crying to an admin with their problems, when it is actually a content dispute they need to sort out themselves. Re: making it harder to find who you want, the current oversight/checkuser/steward/bureaucrat thing already does that. It took me a while to work out what all that meant. Can anyone bring me up to speed on the history of all this. Originally there was nothing, right? Then there was admin, followed by steward. Then bureaucrat, checkuser and oversight, right? Or was it a different order? Carcharoth 15:51, 14 March 2007 (UTC)