As I write, it's past 00:00 Sept. 24, 2012 UTC, which means that September 2012 will officially be the first month since October 2002 during which there were zero successful Requests for Adminship (RfAs). This is a dismaying milestone for the Wikipedia "community," but it's merely the cherry on top of a dismal year for the RfA process.

As many around here know, the RfA process has producing new Wikipedia administrators in ever-declining numbers since its 2007 peak. In the first three quarters of 2012, only 20 candidates' RfAs have been approved by the necessary supermajority of voters. That's a shockingly small number. There have been 12 additional RfAs where the candidate received a majority but not a supermajority of the votes cast, making the success rate among viable RfAs 62.5% during this period.

Those numbers really aren't surprising for anyone who pays attention to the RfA process, though. What's surprising is the way RfA has been a ghost town during the month of September. It's not just that there were zero successful RfAs -- it's that no RfAs have been initiated at all so far this September. And whereas in the past flurries of concern appeared on WT:RFA during briefer RfA droughts, there's been a dearth of such chatter recently. The so-called "new admin drought" is old hat, and RfA reform efforts are considered hopeless.

Meanwhile, Wikipedia's bureaucrats have flipped the admin switch seven times so far in September, resulting in a net loss of 6 admin toolkits:
*Four inactive admins were desysopped at the beginning of the month because they hadn't edited at all for a full year.
*Acetic Acid (T-C-L), an admin from 2005-6 who was later desysopped for inactivity, was granted re-adminship upon request. But AA, whose last 500 edits go almost all the way back to his October 2005 RfA, is too intimidated by Wikipedia's labyrinthine policies to do anything with his admin tools!
*EncycloPetey (T-C-L) was desysopped by ArbCom motion on September 8.
*MuZemike (T-C-L) decided to go on "administrative hiatus" and asked to have his admin rights removed.

And meanwhile, the number of active Wikipedia admins has dipped below 700.

The graph, Eric. Mustn't forget the graph.

I suspect that of those that are active only a small number are actually using the tools. The rest are just heading for burn-out.

The number of currently 'active' administrators as documented here http://en.wikipedia.org/wiki/Wikipedia: ... nistrators has fallen to 687. I remarked (somewhere on this forum) only recently when it fell below 700. This is quite a significant drop.

There certainly are hints that your hunch might be correct. A number of Wikipedia users are discussing the RfA drought and whether or not it's a problem, here (permalink). According to Hammersoft, 45.2% of deletions and undeletions are performed by a mere nine human administrators (Explicit (T-C-L), Magog the Ogre (T-C-L), Skier Dude (T-C-L), Diannaa (T-C-L), RHaworth (T-C-L), Gogo Dodo (T-H-L), Mark Arsten (T-C-L), Peridon (T-C-L), Plastikspork (T-C-L)) and one admin-bot (Cydebot (T-C-L), which automatically deletes categories and category talk pages after CfDs are closed).

I just scanned that conversation. But my impression is it's among the least self-aware conversations I've ever seen on Wikipedia (and "least self-aware conversation" is one of Wikipedia's hardest fought titles). It never occurs to them that A. They and their collective behavior over many years is a big part of the problem. B. That a solution may lie in a radical redefinition of "Admin" roles (basically, layers of "adminship" with very, very clearly defined roles); and, C. Wikipedia is yesterday's news, and its brief moment of "cool" is past.

The website-wide epistemic closure, represented on one page.

Since getting turned back on as an Administrator by a request on the Bureaucrats Noticeboard (which is the default process for any request), Acetic Acid has contributed a total of two User Talk Page messages. This is not a slag on him, just a fact.

As I have already mentioned on Jimbo's Talk page, where this story surfaced earlier today, while I dislike the RfA process, I am not at all convinced that having a very small and carefully vetted corps of Administrators is worse for the project than a larger and more unfiltered group. The amount of carping about misuse of administrative tools here seems small and lessening in any event... There's a case to be made that 300 or 400 really good, active Sysops might be better for the project than a less focused and larger pool.

RfB

ADDENDA: For example, would this site necessarily be better with 45 Admins than the current dozen or whatever the number is? Might it not be better yet with only the best 5?

I think at least some of them are aware of point "A." Point "B" is a notion that Wikipedia has perennially refused to entertain. Point "C" is something I've never seen mentioned on Wikipedia, probably because they're scared to death of it over there. But it's a proposition worth considering: Perhaps the great majority of adults in the Western world who would be inclined to become active, unpaid Wikipedia editors have already done so.



There is indeed such a case to be made. However, I'm doubtful that Wikipedia has "300 or 400 really good, active Sysops." I'm looking into it as time permits.

Now down to 682.

Still far too many.

I'm with Malleus on this. The sweet spot is probably somewhere in the 300 range. There need to be enough that vandals can be disposed of expeditiously at any hour of the day or night, but the big majority of WP contributors should not have revision deletion and blocking buttons.


RfB

The thing about that chart, is that that puppy lacks curvature, convexity to be precise. It's heading right for zero, rather than looking like it will bottom out at some positive value. The slope seems to be about -75 or so per year so give it four or five years and you'll get your wish.

Yeah, but it's got a "repressed zero" (the Y axis doesn't extend all the way to 0, so the curve looks like its crashing into the sea whereas it is simply declining from massively too many administrators to a super-big bunch).

I wish they'd get rid of the nomenclature and stop spouting words like "promotion" and "Administrator." Call it what it is, a janitorial tool bucket for vandal fighters... Then give it to every single person who actually needs the tools who isn't a douche, while stripping anyone instantly who abuses the gear while involved.

Revision deletion and extreme tools like that should be reserved for bureaucrats, with a very high bar remaining for them.

RfB

Repressed or depressed zeros don't make a difference. What matters is that it looks linear, rather than convex. I don't know, if I had the actual data I could see if I'm just eyeballing it wrong, but a constant slope looks constant regardless of where you put the axis.

You know damn well that will never happen. They will screw it all into the ground first.

Ah give it up, Eric. Randy still thinks it's an encyclopedia, so he'll never see the forest through the trees .

No, I don't agree. Just keep reminding Randy that honest, content-creating editors like himself usually don't go too far at Wikipedia. Hey Randy, no one's offered you a free trip to Gibraltar lately, have they?

The monthly averages of the day-by-day data are given here. If you ignore the first 6 months as clearly not part of the downward trend, let m be the number of months after June 2010, and do a linear regression of Am, the average number of active admins in month m, against the variables m^2 and m, you get Am = 0.033 m^2 — 5.9 m + 830.2 + Rm as the estimated regression. If the assumptions underlying standard statistical analysis of a linear regression model held exactly (which, of course, they don't), then the coefficient of m^2 would be significantly different from zero at the 0.0001 level. Its 95% confidence interval would be [0.018, 0.047].

Thus, if you were silly enough to take this model seriously, the trend curve would be slightly convex. It bottoms out at a minimum of 565 at m = 89 (i.e. November, 2017), and thereafter increases without limit.

If you don't ignore the first six month's data, that's enough to reverse the sign of the coefficient of m^2 in the estimated regression, it then being Am = -0.013 m^2 — 5.5 m + 855.7 + Rm, where now m is the number of months after March 2010.

You think Idaho is part of the UK?!?!

RfB

Try some higher order polynomials (I start getting insignificant coefficients at order 6). That thing is concave, not convex.

Based on a 4th order polynomial, it looks more or less linear up until October 2014, after which there is an acceleration in the decline and "death" (active admins disappear) in September 2015. 5th order it's even worse. Big acceleration in decline summer of next year and "death" by summer 2014.

Of course I wouldn't put too much faith in those predictions but the general conclusion that "it's bad" and that it's "heading for zero" I think is warranted.

(if you want to try some time series process, it's a non stationary process with a unit root. Differencing Am and running the regression in lagged diffs suggest more or less linear decline with 5.64 admins de-activating per month, which gives loosing about 68 per year, which is pretty close to what I said originally)

Edit: with polynomials eventually the coefficient on the highest order significant coefficient will eventually dominate the whole thing. So it's crap for forecasting purposes, especially for a variable which cannot go negative. So actually just a linear trend might be better. Another alternative would be to run it as a inverse polynomial: Am=c0+c1/m+c2/(m^2)+...+e. Haven't tried that

Now fallen to 670 http://en.wikipedia.org/wiki/Wikipedia: ... nistrators .

I wonder if it's going to be 2 months without a successful RfA at this point.

"Gigs" withdrew, with a parting comment that almost makes me think we should have endorsed him:



"QuiteUnusual", who has always struck me as a good lad on WB (he's practically holding the fort down there single-handedly nowadays), apparently has been found to engage in a bit of plagiarism.

"Writ Keeper" apparently accrued most of his edit count using twinkle and participating in something called the "Tea House" (new to me). He seems to have arbcommers at his back, so maybe that one will sail.

If they don't make it, that will make 2 months since the last new admin (August 27).

And now at 668 http://en.wikipedia.org/wiki/Wikipedia: ... nistrators . The Teahouse is some sort of civility/welcoming project, based on the assumption that you can improve Wikipedia by welcoming thousands of new editors, rather than focus on improving the environment for the existing ones.



With all respect, I think this is one of the many areas where mathematical techniques make no sense, at least without an explanatory model. Clearly a continuation of the trend will have an effect on the behaviour being measured, which will in turn feed back into the behaviour. The drop in administrative resources will lead to further pressure being put on the existing resources. This will cause people to discuss a resolution of the problem. This may go two ways. If there is no obvious resolution, the 'shadow of the future' will come into play. This is where foreseeing a resourcing disaster actually causes the disaster to happen earlier than expected due to people jumping ship. If there is a resolution – for example, RfA standards are severely lowered, then the numbers might well rise.

Thus the future behaviour is very hard to predict.

Like having a doorman or greeter at Bergen-Belsen.

Godwin.

Well its a bit more than just that. Teahouse works by welcoming and gently shepherding new editors in 'the ways' of wikipedia. For the most part its pretty good at encouraging and informing new editors on how best to do things. In all manner of article creation. Think of it as more of an after-school workshop for hobbyists.

Bear in mind - one way of dealing with a toxic environment of people that are stubbornly refusing to change is to surround them with nicer ones and hope it rubs off. Done right it can be quite effective in altering a locations 'culture'. Be it in an online forum, or a workplace. You bring in a load of new people, teach them to deal with/ignore the idiots, then let them go. In this the Teahouse might be the best weapon in the long run...

What are these 'ways'?

Yep, quite an unusual editor in many ways Well, the RfA has lasted for seven days now and it's very close so there's a behind the scenes bureaucrat clusterfuck.

Oh, he's a shoe-in.

There's a third candidate but he's a bad joke. There was a WP:SNOW close, but it was re-opened for some strange reason so he's just being trampled into the mud.

Failed.

A) great, a robotic patroller from Essex. He certainly is obsessed with Woodham Mortimer.
B) he looks like yet another sockpuppet to me.

I don't think he's a puppet. Look at his early history on Wikibooks and it's pretty clear that he is who he says he is (or rather doesn't say, but you know what I mean).

You mean this? Ah, now we can see why he was semi-popular at RFA. But Wikibooks is one of the WMF's "ghettoes", and I rarely check users for their Wikibooks activity.
If he's such a good historian and author, why is he a Twinkle patroller on en-WP and on Wikibooks? Something about this is a little "off".

Yeah, actually, looking at the early contribs certainly does point to being experienced elsewhere. I wonder if maybe he was vanished and went off to WB to make a clean start... Wikisource was the more common route for that, but WB works too.

For years RfA has been declining and becoming steadily more dysfunctional. I'm not sure whether that points to a self-destructive failure of Wikipedia process, or if it just means that there isn't much need to reform the process, because there's not much need for new admins. After all, if it were badly needed, wouldn't the community feel the pressure to get its act together and agree on some reforms? It might be better not to promote a lot of new admins, if there isn't much for them to do; it wouldn't be worth the bother.

I hardly use the tools at all anymore, simply because I seldom come across situations where they are needed. It seems like automated tools and a handful of extremely active admins do a pretty good job of handling things, so maybe efficiency has decreased the need for grunt work. Maybe it's even possible that people are learning to behave better on Wikipedia, either out of respect for it--awareness of what exactly it is--or recognition of the futility of misconduct, and so there isn't as much mischief as there was in the early years? Intuitively that seems unlikely, but it's a possibility, I suppose. My perspective may be way off, because I pay almost no attention to admin noticeboards and such these days, but that's something I've wondered about. It may also be that activity levels are declining across the board, and so there's a declining need for admins to manage things--but if activity levels are declining, then that has to be a severe indictment of Wikipedia culture, given the site's tremendous visibility and potential.

But even if that's true, I don't see how it could hurt to promote more admins. It seems like it gives people something to work towards and offers a sense of being recognized and embraced by one's peers. They can certainly find work to do if they want it. Also, if you give the tools out more liberally, you reduce the sense of admins holding some inflated authority position, distinct from other editors.

You're right about that--participation is declining, content generation is unquestionably declining, new articles are
usually worthless stubs created by bots, and even vandalism appears to be declining. People are losing interest in
your magical project, sir, and you can thank your fellow administrators for a major part of it.

After seeing recent candidates exposed for bad behavior, I believe RfA is one of the more functional of the Wikipedia processes.

This, only after eight years of abuse, corruption, vote-stacking, harassment of oppose voters, sockpuppetry, and all kinds of other outrageous hijinks.
They had to clean it up, it was a standing public embarrassment. (List of major RFA atrocities available upon application.)

Welcome to everyking!

Thanks, Greg. I would have showed up sooner, but I didn't know this forum existed.

That sounds interesting. It means you've been avoiding the dramah boards (good) but we've not got the reach into whatever areas of media you follow. I'm genuinely intrigued that you didn't come across us. Perhaps we'll have to discuss why.

I may go again in 6 months or so. My position on sports people microstub notability is sure to draw some opposes even then, and I'm sure posting here isn't helping my odds, but at least I can make the "activity level" people happy.

That was October 22. Now down to 656. Is there a way of retrieving these figures so they can be charted?

[edit]

Here is the chart from 18 Jun 2012, where it was just over 700.



Sep 24: 687
Oct 1: 682
Oct 22: 670
Nov 21: 656

Looks like this month will be a reversal of the trend this year where there were more unsuccessful RFAs in a month than successful ones, though July was just barely an exception. There have been two really bad RFAs (MONGO and Northamerica1000) and so far there has been one successful one (Tommyboy), but there are four more that look to be on the way to passing with flying colors. Nothing earth-shattering about this group. It seems to be the usual crop of patrolling, edit-padding, wikidroids. Tommy seems to have generated a whole lot of stubs on minor politicians, but the only other one with significant content contributions is this Sergecross character, contributions which basically involve creating a few stubs and mehtastic start-up articles about video games and music. With the rest of them it is essentially just a wall of auto-edits.

Who the fuck would nominate Northamerica1000? There are no indicators that that argumentative copyvio fly would amount to anything but a de-sysop that would embroil the entire community, courtesy of Northamerica1000 himself. And yet, many of the opposes seem off target, and such a mofo big! target. How can you not hit it? At least folks didn't let him get away with the major dissembling babble-vomit in response to Q12 (once a plagiarist, always a plagiarist, and this dude is a plagiarist).

At least MONGO was a self-nom.

There is a never-ending "crop of patrolling, edit-padding, wikidroids," so it seems strange they can't crap out a stream of replacements. It's not like there is a qualification minimum what with admins who don't do admin work, admins who don't do article work, admins who don't do dispute resolution work, admins who routinely disrupt editors, admins who live at AN/I, admins who get elected to unblock their own bots and do nothing else.

He's a close pal of the legendary twat Colonel Warden. Together they were nominated for a WR DICK award.
http://wikipediareview.com/index.php?sh ... ntry294080

You can tell the "community" is falling apart: trolls are trying to nominate each other for RFA. The clown who tried to nom NA1000 tried it himself last month.
http://en.wikipedia.org/wiki/Wikipedia: ... cStrikeout

Strikeout appears to be a newbie baseball nerd who registered in April. Every now and then Strikeout has branched into articles related to other sports and in the occasional bit of patrolling makes some minor contributions to more important stuff. It is quite possible that Strikeout was completely clueless about how ridiculously bad that nomination was since the main interactions between those two appear to be outside AfD.

DICK award, yes; en.wiki admin, totally clueless.

I believe clowns' noses should be red:

"I am proposing the initiation of a project that identifies one administrator each day to be the admin of the day. It would be very similar to the Wikipedian of the Day, but more focused and would be a reward given to an admin for a specific admin task he or she performed. The project would list openings ten days in advance, and users could nominate an admin for any of those days, although they would only be allowed to nominate an admin once per admin task being recognized (as in, you would only be allowed to nominate an admin one time in recognition of a specific action, but you could nominate him for all ten days for ten different reasons)."

I'm gonna vomit. Is there a smiley for that?

at least two

Thank you.

New version: