The Signpost


Has the wind gone out of the AdminShip's sails?

Contribute  —  
Share this
By Kudpung
Not the Admin Ship

Wikipedia Talk:Requests for adminship, once the most lively forum on the project with the exception of ANI, is becalmed. The babble of noise at peak times akin to the background din of a noisy Manchester pub on a Saturday night has dropped not just to a whisper, but to a stony silence. It's become an empty space. Walk through it and you'll make conspicuous footprints in the dust gathering on the floor. Your footfall echoes in the deserted room, "Is there anyone there?" you call halfheartedly before turning round and going back outside into the sunlight. In the street there is also little activity. A kid sits on a doorstep poking vandalism into K-Pop music bios through his cheap Chinese smartphone. A bunch of teenagers sit round their Samsungs seeing who can tag the most new pages in sixty seconds.

The whole place has the feel of a deserted Wild West film set. Across the almost traffic free road, you enter a half open door with a dilapidated sign hanging on one nail: Café Anna it announces in sun-faded coloured letters. Inside the pretty room with many colourful interactive pictures on the wall, there is a stuffed effigy of 'The Founder' in an armchair, but there is no busy bustle in this place either. There's a lady editing an article about a golf complex in Hainan. The article is almost as complete as the complex. She's looks startled when you enter – she obviously wasn't expecting anyone. "Yup, it's kinda quiet here," she says. "April's customers are down to a fifth of the usual month's average."

Related articles
Reforming RfA

Jimbo's NFT, new arbs, fixing RfA, and financial statements
28 December 2021

Editors discuss Wikipedia's vetting process for administrators
26 September 2021

Administrator cadre continues to contract
31 July 2019

The Collective Consciousness of Admin Userpages
31 January 2019

The last leg of the Admin Ship's current cruise
31 July 2018

What do admins actually do?
29 June 2018

Has the wind gone out of the AdminShip's sails?
24 May 2018

Recent retirements typify problem of admin attrition
18 February 2015

Another admin reform attempt flops
15 April 2013

Requests for adminship reform moves forward
21 January 2013

Adminship from the German perspective
22 October 2012

AdminCom: A proposal for changing the way we select admins
15 October 2012

Is the requests for adminship process 'broken'?
18 June 2012

RFAs and active admins—concerns expressed over the continuing drought
14 February 2011

RfA drought worsens in 2010—wikigeneration gulf emerging
9 August 2010

Experimental request for adminship ends in failure
13 October 2008

Efforts to reform Requests for Adminship spark animated discussion
23 April 2007

News and notes: Arbitrators granted CheckUser rights, milestones
6 February 2006

Featured picture process tweaked, changes to adminship debated
27 June 2005

More articles

In this issue
+ Add a comment

Discuss this story

Now, whether those conveniences would be worth running the gauntlet; or, whether RfA voters would even consider a candidate who only wanted to do those things: these are entirely separate questions for which I got nothin. Innisfree987 (talk) 20:55, 24 May 2018 (UTC)[reply]
I think me and Kudpung agreed in principle on most things, although we often differed with respect to method. (My tendency is to take a very bold approach to reform, in contrast to his seemingly more gradual approach. I'm afraid that has led to several unfortunate and regrettable misunderstandings.) I especially supported the idea of separate clerks, but every comprehensive clerking proposal has been rejected. I remember someone, maybe Montanabw, suggesting restrictions on the length of vote rationales. I didn't comment, but I, as well as Kudpung, thought it was a good idea. Perhaps someone could take that up.
Personally, I decided a couple of years ago that I would not spend any more of my time here proposing RfA reforms, and I haven't said much since then. I'm satisfied in the sense that I at least passed something for the first time. I did the very best I could. But again, those reforms were not perfect. If anyone can propose something better and get it through an RfC, that would be great. I would vote for it. But actually passing something (anything, for that matter) is incredibly difficult, to say the least. I applaud anyone who is willing to take up that challenge anew. Good luck. Biblio (talk) 00:14, 25 May 2018 (UTC)[reply]
Thanks for the offer; unless it's likely to pass and will help to prove a point, I'd rather wait until September when I expect to have more time for the project. power~enwiki (π, ν) 16:27, 25 May 2018 (UTC)[reply]
So it doesn't seem to me that there is any particular crisis at this point, even as the total number of admins is still in decline. Dekimasuよ! 05:13, 25 May 2018 (UTC)[reply]
Chart from User:Widefox/editors
I think there's something wrong with Rick Bot's edit summary if you look at the actual chart. Mkdw talk 15:30, 25 May 2018 (UTC)[reply]
The chart (posted above) from User:Widefox/editors tells a fuller story. The green solid line "active admins", is probably the one to be concerned about. I remembered this from the last discussion we had, in the responses to Signpost 2018-02-20 News and notes. ☆ Bri (talk) 15:49, 25 May 2018 (UTC)[reply]
I was looking at Wikipedia:List of administrators/stat table but I realize Rick Bot does a daily count. Mkdw talk 16:25, 25 May 2018 (UTC)[reply]
The graph tells a different story, as a result of its extended timeframe. Things might seem much different if the data was based on the period 2014–present or 2015–present, say. This would also seem to be a reasonable timeframe; we can't expect any year to look like 2007 again. Based upon the monthly averages at the link given by Mkdw, we've been between 523 and 613 every month since April 2014. I have my doubts that we will be below 500 at the end of this year, as would be expected by the graph. Dekimasuよ! 19:20, 25 May 2018 (UTC)[reply]
Probably wouldn't float. Detractors would argue that it would put too much more power in the hands of admins. As one commentator on this page stated: "My default attitude is that an admin is dishonest and/or corrupt. An individual admin has to prove themselves to be otherwise. So maybe their reputation is so slimy that it's hard to recruit any more." So quis custodiet ipsos custodes? That said, there's no harm in trying, nothing ventured, nothing gained. Design your project in every minute detail, launch your RfC.Kudpung กุดผึ้ง (talk) 05:53, 26 May 2018 (UTC)[reply]
I don't think it would float, either. One of the problems that has been alluded to is the poor reputation that some admins have with non-admins. I doubt that having a committee responsible for admin promotion which consisted solely of admins would be acceptable to many non-admins. --RexxS (talk) 00:31, 27 May 2018 (UTC)[reply]
if its an election process like that of ARBCOM then it may not need only admins providing the group was sufficiently large enough and the process clearly defined including those action which with xx period of time & xxxx number of edits immediately exclude someone. The big thing will be clearly defining the characteristics we want from admins, and where the committee has discretion to make judgement calls. It'll be a rocky start but once its running smoothly we could also look at the removal of tools from those not using them, with the bonus of there being somewhere to go should an admin go off the rails. Honestly not everyone is going to agree with and most admins get a "bad reputation" because they doing what is expected from them Gnangarra 23:36, 29 May 2018 (UTC)[reply]
Editors have very high expectations for automated tasks, and thus feel uncomfortable with a seemingly complex task being automated. However, a bot does not have to be perfect. It just has to be equal to or better than manually doing the task. For example, ClueBot reverts about half of all vandalism while only wrongly reverting less than half of one percent of non-vandalistic edits. This is despite it using a relatively simple machine learning algorithm compared to the algorithms that exist today (i.e., Bayesian classifiers and primitive neural networks). A bot that uses the state-of-the-art machine learning methods can achieve great accuracy in tasks more complex than vandalism reversion, like detecting articles that meet CSD criteria. In addition, these algorithms generate a confidence level in their decisions, so the edge cases can just be delegated to admins for manual review. Esquivalience (talk) 01:23, 27 May 2018 (UTC)[reply]
  • No, in order for there to be new admins there has to be editors, and only a small number of those editors would make suitable admins, and a smaller number who wants to be admin. Improving RfA may get a few more to run, but eventually all those who want to run will have run and we are back to where we started. Esquivalience (talk) 01:30, 27 May 2018 (UTC)[reply]
You are putting a lot of faith in our collective ability to pull off several AI projects. So far the track record ain't great. Color me skeptical, not because I think the technology isn't up to it (I do), but the surrounding organization, management, funding, etc. Not to mention another uphill battle against commmunity inertia just to get such a beast authorized to run. ☆ Bri (talk) 01:31, 27 May 2018 (UTC)[reply]
Heya Bri, have you been keeping up with the progress of ORES? Here's a demo from the recent hackathon where machine learning can offer suggestions for something like category suggestions with a "topic" model as an example. Other models can be added and made available to ORES consumers (tools, editors, etc). The code is on-wiki if you want to play with it. CKoerner (WMF) (talk) 16:24, 31 May 2018 (UTC)[reply]
Actually I have only peripherally been involved with ORES -- some folks were looking for tagged training datasets for conflicted/paid editing, which I helped with. Thanks for bringing the phabricator item to my attention. ☆ Bri (talk) 17:01, 31 May 2018 (UTC)[reply]
Nothing is going to be fixed until the crisis comes, and as long as there are a few hundred more or less serious administrators to get the work done, the crisis isn't yet arrived. But it is coming, make no mistake. Then we will see some combination of (a) a serious discussion about loosening standards; (b) WMF taking over more and more administrative duties with paid staff.
The best idea in this thread is that there should be an elected "Administrative Committee" to co-opt qualified candidates, thereby foregoing the wretched, overdramatic, gotcha quiz of overqualified candidates that continues to put up failing grades, year after year. You administrators are yourselves the cause of the lack of administrators. You and your exclusive club... When there's actually a shortage of administrators and you're ready to get serious about actually fixing the problem, let us know. Carrite (talk) 02:29, 29 May 2018 (UTC)[reply]
I once suggested to thank admins for admin actions like protections etc; this was taken by a group of editors against me to prove I am not fit to be admin. (One editor said this is childish, and another one would ping me any time there was a backlog anywhere claiming that I like clearing backlogs).--Ymblanter (talk) 19:01, 29 May 2018 (UTC)[reply]
@Ymblanter:, @Ymblanter:, @Ymblanter:, @Ymblanter:, there is a backlog at WP:RM, WP:AIV, CAT:RFU, CAT:CSD, and CAT:RD1. But anyway, that's silly. Recognizing people for the work they does not mean they are only doing the work in hopes of recognition. Natureium (talk) 19:15, 29 May 2018 (UTC)[reply]
This is of course now possible (for protections and deletions, not blocks). ~ Amory (utc) 19:40, 29 May 2018 (UTC)[reply]
For most of the blocks an admin has to leave the block message, and one can thank the admin for this edit. But I indeed specifically had in mind backlogs such as RFPP and XfD where one can thank an admin directly.--Ymblanter (talk) 19:50, 29 May 2018 (UTC)[reply]
I could add a few more points, but these are the key ones. If admins have less to do in that supra-editorial role, because stuff that's not really dangerous is delegated to a broader range of competent editors, and we no longer expect admins to be godlike in their unimpeachability, yet being impeached for screwing up only has temporary consequences anyway, then it literally will not be possible for there to be a crisis of too few admins.
 — SMcCandlish ¢ 😼  22:59, 31 May 2018 (UTC)[reply]
I can get behind these proposals and would be curious to hear counterarguments czar 00:23, 1 June 2018 (UTC)[reply]
I hate unbundling because all of these people should just be admins in my strange ol' perfect world. With only a couple of exceptions, every single template editor, page mover, etc should just be an admin - and probably would be if they joined Wikipedia a few years earlier. But you're completely correct when you identify "block" as the big stopper here. I'd be fine with unbundling everything but block-related permissions from adminship, and giving them out based on the PERM process. Because you're right - while there is a lot of overlap over using block/protect/delete (especially around counter-vandalism), they do have separate uses. -- Ajraddatz (talk) 00:32, 1 June 2018 (UTC)[reply]
Yep. It's a toolbox. It takes training to use a tablesaw safely. It takes no training, just common sense, to use a hacksaw safely. Treating the tools as exactly equivalent and both requiring the same competence before someone can pick them up is a mistake, and is still a mistake when one says "Well, a professional carpenter often has to choose which tool to use while on the job." The fact that such a choice is made in that particular context is irrelevant to the underlying tool safety question. That logic problem is what led me to post about this again here. Moving on to your actual points instead of recapitulating mine: I'd agree that generally people with advanced bits should just be made admins after some amount of time if no issues have raised about their tool use. Hell, we'd probably have no backlogs if this were the case. But WMF insists that there's a legal reason they have RfA and that it's can't be automatic. They say that WMF's liability would shoot up if the site weren't administered by people who've been through a non-trivial vetting process. IANAL, but I'm not sure I buy this. I think the same legal shield could be in place already, because any administrative decisions can be overruled by ArbCom, and that body is directly elected with even more scrutiny than admins. At any rate, there's likely a middle ground, e.g. a confirmation mini-RfA, that would work. "I hate unbundling because all of these people should just be admins" is kind of spiting one's own face, though. We have to do what we have to do to get the job done. If the old process is damaged, work around it. I hate that I have to take a bus to get to the grocery store, but I won't starve myself in protest, I'll buy the bus ticket. >;-) PS: I agree with your "a couple of exceptions" caveat; there are some (probably under a dozen) people with the page-mover bit who should not have it because of a demonstrable track-record of WP:SYSTEMGAMING at WP:RM, but they got the bit before the requirements and scrutiny level were tightened. That sort of thing is another reason to have some kind of confirmation process rather than a literally automatic sysopping.  — SMcCandlish ¢ 😼  00:55, 1 June 2018 (UTC)[reply]
WMF requires a "community selection process" for access to view deleted revisions. No such requirement exists for the actual delete/protect/block permissions. So we could either have a "deleter" group without the ability to view deleted revisions, or require requests to stay open for a few days of public comment. -- Ajraddatz (talk) 03:36, 1 June 2018 (UTC)[reply]
Ah, yes. Either of those could work, or just create another community selection process for a combined deleter-undeleter bit. WP:PERM processes really already are a community selection process, just less dramatic and awful ones than RfA is.  — SMcCandlish ¢ 😼  04:31, 1 June 2018 (UTC)[reply]
Wouldn't the sanctum sanctorum of unbundled privileges referred to by Ajraddatz – viewing deleted revisions – actually be undeleting, not deleting? ☆ Bri (talk) 05:58, 1 June 2018 (UTC)[reply]
From a legal perspective (or at least the WMF's legal perspective) yes, but from a community perspective I think blocking is the big one. Blocks have the most potential to cause drama out of those three groups of permissions, especially when used against established users rather than vandalism-only accounts and IPs. As an aside, compared to the other two protection looks to be so benign that I'm surprised it hasn't been unbundled already. -- Ajraddatz (talk) 14:51, 1 June 2018 (UTC)[reply]
Yep. With some clear ground rules, and after a few removals of the permission from people who don't follow them, it would do a tremendous amount to reduce stupid, time-sucking drama at this site: any uninvolved, competent, long-term editor could temporarily shut down an obvious editwar, and that would more often than not be sufficient time for the squabblers to come to their senses (if they don't, we have ANI and ANEW). This should at least be enabled for mainspace.  — SMcCandlish ¢ 😼  20:49, 1 June 2018 (UTC)[reply]
Having terms of service would be a hassle because we'd constantly be electing admins. There are a thousand of them for goodness sake. I favor the tactic of lower the requirements of desysoping. If an admin isn't abiding by the community standards, it shouldn't require an act of God ArbCom to remove the permissions. Making desysopping no big deal is the most obvious way I can see to make adminship return to being no big deal. This is not a tenured professorship where we are awarding trusted users freedom to be as crazy as they want once they pass muster. Admins shouldn't be able to go around edit warring and accusing people of being KGB members just because they were seen as competent years ago.
There's a currently running RfA, and some of the oppose votes give the reason that he wants the tools for one project and there's no way to ensure he steps down at the end. While I don't really see the point (if he's not causing disruptions with admin tools, why should he need to give them back because he no longer is working with portals?), I think the ability to remove tools if a new admin proves to be a bad choice would make community members less hesitant to confirm admins. Other perms come with the condition that they'll be revoked if misused, so why should sysop perms be any different? Natureium (talk) 21:20, 1 June 2018 (UTC)[reply]
I have to disagree with the base premise for multiple reasons. The general community impression and complaint is that RfA used to be an always-on process of community engagement and decisionmaking about it's "leadership" to use a not entirely applicable word, while now RfA is charactered as a "ghost town" and on the verge of just being abandoned. Can't have it both ways. Either we want it vibrant again, or we want it tagged {{Historical}} and shut down. The consensus clearly leans strongly toward the former. Second, reconfirmation RfAs would probably be considered to pass at a lower threshold (e.g. >50%). Third, the other hue and cry about admins is that there's an increasing caste-like disconnect between the "admin class" and "everyday editors". The only way to fix that is for more (percentage-wise) of editors to become admins, and admins who do not represent community expectations to get removed. I don't see any other way to get there but non-forever adminship plus a review process, since it would defuse adminship "seriousness" drama and make the reviews actually happen. PS: I haven't seen a single person make the "no way to ensure he steps down at the end" argument you've suggested; rather, the concern is "no way to ensure he does anything useful with the tools after the end" of that cleanup run, plus questions of whether the candidate is actually experienced or interested enough to try.  — SMcCandlish ¢ 😼  22:15, 3 June 2018 (UTC)[reply]
Oppose -- However, I would move to Support if he pledges to resign adminship when he finishes the project.User:DonFB. I think that's the specific phrase I was remembering. Natureium (talk) 14:10, 4 June 2018 (UTC)[reply]


The Signpost · written by many · served by Sinepost V0.9 · 🄯 CC-BY-SA 4.0