Wikipedia reaches a turning point: it's losing administrators faster than it can appoint them

Useful article in The Telegraph:

http://blogs.telegraph.co.uk/technology ... oint-them/





Some of these points seem to parallel those Eric has made over the past few days.

That's an astonishing graph. They've had a 30 percent decline in active administrators (generously defined as making 15 edits a month) since 2007, as the number of articles has continued to expand. That trend points to a crisis at some point.

But has Wikipedia gotten worse as the ratio of administrators to articles has plummeted in the past five years? My suspicion is, not really. It's still awful where it's always been awful, and incomprehensible on complex topics. But I'm sure the Star Trek articles are just fine. Maybe it doesn't matter, from the perspective of maintaining the status quo, if you have 700 administrators doing vandal fighting or 1,000. Maybe the status quo would be almost as well served by just 100.

Of course, from the perspective of people interested in improving the product, their administrator system is a major obstacle. You would need to blow it up and start over completely (add that to the long list of things "the community" would never stand for).

Just eyeballing that graph, Wikipedia has been, on net, losing about 70 admins a year. At that rate, they'll be out in another ten years (although obviously I would expect the rate of decline to slow as they get closer to running out entirely).

It's fairly obvious that being a Wikipedia editor or Wikipedia admin is no longer "cool", and that the alternacool kids are now doing something else instead of playing the Wikipedia MMPORG.

We are already seeing the average-time-to-revert-vandalism increasing. Before long, Wikipedia will dissolve into a mass of vandalism. Although it is probably also becoming less cool to vandalize Wikipedia, and so the rate of vandalism will also probably drop. So it'll be spam instead of vandalism. Wikipedia's insanely high pageranks and view counts make it a very delicious target for spammers, and without effective administrators it'll fall very quickly to that.

Of course, as Dan points out, 700 admins may be way more than sufficient... but that assumes that their admins are well-managed. They're not: administrators do what they please, and some areas of responsibility are grossly overstaffed and others are left neglected, because of the personal predilections of the administrative corps. Wikipedia could well get by on 100 admins, if those 100 admins were conscientious individuals willing to do what needs to be done. But, quite simply, they aren't. And the odds are that when Wikipedia does get down to 100 admins, that'll be even more true.

Honestly, I have been wondering if they should just pay people to do the routine maintenance in which most admins are currently engaged. In other words, leave all the vandal-fighting and other uncontroversial actions to paid volunteers and have the devoted volunteer admins focus on more difficult work. Of course, that isn't going to make any of the unpaid volunteers better admins, but it would at least prevent Wikipedia from turning into a haven of vandalism.

You're just fantasizing about the possibility of being paid to bully people on Wikipedia. Don't worry; it will never happen.

For me the routine maintenance stuff is an occasional stress-reliever rather than something to which I would actually want to devote my time, even if I could get paid for it.

Thank you. I will now show one of the charts I prepared from my administrator study.


Another thing I discovered: three years ago, the average administrator was either a writer of content or a "gnome", patrollers and others were in the minority. Now that almost 2/3 of those administrators have left (very quietly), we now find that the ones quitting tended to be content writers and gnomes. So now the patrollers and "Facebookers" are dominant, along with the worst kinds of gnomes--the kind who use bots habitually to make worthless changes.

The funny thing about the bots is that they tend to get most of the job done most of the time. They don't necessarily do it well, and are useless for "subtle" vandalism, but they tend to catch the "Joey sux big cox!" crap that actual humans used to bash with rollback buttons. The problem with the bots and borgs is that they make watchlists useless for everyone else, and so everyone else loses interest.

This might have even been an evolutionary thing if the encyclopedia had gotten to the stage where it needed maintenance rather than mostly needing to be re-written, but instead it's become one of the major barriers to building and maintaining the kind of community you need to write articles.

A couple thoughts...

1. I wouldn't equate "administrator" with "vandal fighter." While many or most administrators are vandal fighters, not all vandal fighters are administrators.

2. I don't consider 15 edits a month to be "active." I think the official threshold for so-called "very active Wikipedians" is 100 edits per month. That should be the benchmark for administrative "activity." The number of active administrators could be far fewer than the number cited here.

3. Vandalism fighting is increasingly automated. Better automation means fewer active vandal fighters are necessary.

4. In a pinch, if the vandalistic barbarians start to swarm over the gate, Wikipedia can always raise the gate. Shutting down IP editing, for example, would slash a major fraction of total vandalism. Requiring real name registration and sign-in-to-edit would slash that minor fraction of remaining vandalism by a major fraction.

RfB

I just missed a watershed here - I've been watching this page http://en.wikipedia.org/wiki/Wikipedia: ... nistrators for a while, waiting for the number of active admins to fall below 700. Which it just has.

Obviously there is no statistical significance between the change from 701 to 700, and the change from 700 to 699. But a lot of emotional significance.

Eric, thanks for that chart. But what is the definition of 'inactive'? Presumably, no edits for a defined period of time. What is the defined period of time?

And now, for my next magical trick.....two charts showing how the balance of administrator activity has changed in the past ten months.

This inspired me to check the list of administrators active (> 30 edits in last two months) in October last year. I have released this page http://www.logicmuseum.com/x/index.php? ... nistrators .

What struck me was that those still editing are editing apparently as addictively as ever. Those who are editing less have simply stopped editing. That is, there is no gradual decline in editing, just an 'emergency stop'.

Click on the contributions field, which gives you the last 30 edits. The closer you are to 8 July (i.e. 2 months ago) the closer you are to being a 'not active' editor. There aren't many July ones. It's either still in September, or stopped altogether. A brief sample however, and not meant to be scientific.

It's an interesting ranking criterion you use and equally interesting that a number of hate figures are at one or other end of the list. (David Gerrard with few edits per page, Slim Virgin, Gwen Gale and FT2 with relatively many.

As a matter of interest, how do you count a fuckwit patroller who is also a bloody thief, or a content creator who is a thief? Don't tell me they don't exist.

Additionally, the change from 700 to 699 yesterday was not because of an administrator being reclassified as semi-active or inactive; rather, it was because of EncycloPetey (T-C-L) being desysopped by ArbCom motion. So, this milestone resulted from the patrollers on the ArbCom stripping a content contributor of his admin rights. Some might find that symbolic.



That's very interesting; thanks for posting it. You don't have to name them if you don't want to, but are your three "thieves" from your 2011 sample still at it?



Regarding overlap: Eric, in another thread you indicated that EncycloPetey was both a content creator and an evil patroller. How would you have classified him for this sample? And surely there's an overlap between evil patrollers and "facebookers"?

You betcha. Wknight94 disappeared for several months, then returned. I suspect they threatened to desysop him.


Admins who did a lot of actual adding-of-content work (more than 50%) and who were not obvious robot gnomes were called "content" admins.
EP would go in that category, he's not a typical "evil patroller" as he isn't going around blocking people for kicks. A true "evil patroller" doesn't
add ANY content, or very damn little. Xeno was a perfect example, until they put him on Arbcom in November. Since then, he's been a "Facebooker".

MuZemike is a more typical "evil one". Did you know the idiot blocked himself in 2010? That, sir, is the true "evil patroller".