The history of administrators Options

[Peter Damian]

In 2001 Larry wrote here http://lists.wikimedia.org/pipermail/wikip...May/000123.html in a reply to Crocker, who had written:

There is no such thing as an "administrator" in Wikipedia in the sense of someone responsible for its content, nor should there be. Nupedia has those (and should); Wikipedia just has us, and we are just as responsible for its content as anyone else. It does have a few folks to set policy, but even they have been very respectful of the community process of content creation and not tried to subvert it by establishing "control" or "ownership". Further, it is obviously impractical to have an infintely scalable content-creation method with non-scalable editing and expect to keep up. Wikipedia CANNOT work unless EVERYONE is an editor and administrator as well as an author. "

Larry replied:

Notice, there is no one claiming to be editor-in-chief or even editor of Wikipedia. A wiki, by its design, doesn't need one. Wikipedia needs people to act as "gardeners" (in Jimbo's metaphor). The reason Wikipedia is so successful at creating content is that there aren't any editors standing in the way of content creation. This means there's a lot of garbage that needs cleaning up, and the whole thing is a work-in-progress, but a lot work *is* done, and we *do* have a lot of very good articles and many that are improving.

It was a little like that when I started contributing in 2003. Quite a friendly place, even.

(1) When was the concept of 'administrator' devised? It seems to be built into the concept of the MediaWiki. Was it part of the original Ward-Cunningham design? Is it related to the administrator concept in computer systems design?

(2) When did the role become important, and when did the divide between editors and admins begin to get bitter?

(3) Who was the first editor to be banned? My research suggests an eccentric IP beginning with '24', who they called '24'.

(4) Who were the first rogue admins?

[Web Fred]

Has there ever been a concerted effort to change the 'job fer life' aspect of sysop to a fixed term (2yrs?)?

[Kelly Martin]

Has there ever been a concerted effort to change the 'job fer life' aspect of sysop to a fixed term (2yrs?)?

Multiple times, but it is always resisted vigorously by people who realize that if this happens they'd never get reelected.

[radek]

Has there ever been a concerted effort to change the 'job fer life' aspect of sysop to a fixed term (2yrs?)?

Multiple times, but it is always resisted vigorously by people who realize that if this happens they'd never get reelected.

Actually, I've come to realize that the problem here is largely logistical. If you have fixed terms of 2 years, you'd have something like 2 or 3 reconfirmations per day. Given how inefficiently Wikipedia functions that would simply result in an insane backlog and it just simply wouldn't work.

Even with 3 year terms you'd have something like 10 reconfirmations per week. (I should give credit to NYBrad for pointing this out).

Rather, what should be done is that "Open to Recall" should be streamlined (so that it's not the total total joke it is presently) and made a required part of the admin deal.

[Kelly Martin]

Actually, I've come to realize that the problem here is largely logistical. If you have fixed terms of 2 years, you'd have something like 2 or 3 reconfirmations per day. Given how inefficiently Wikipedia functions that would simply result in an insane backlog and it just simply wouldn't work.

This is really just proof that Wikipedia has too many admins. Not that it has more admins than it needs to do the work (although it does), but that it is more admins than it can effectively supervise. All the more reason to trim the number back.

However, some ideas could still be possible on-wiki. For example, ArbComm could appoint administrators, and if ArbComm truly represented the community (not just a majority faction or the largest factions), this would be saner than direct community election, because deliberative process could be used, and election and suspension or restriction or removal could be handled swiftly and efficienty. Suspension or restriction, in particular, should not require proof of error, and should not require a proof of reprehensibility, merely a fear of harm pending review. Like an injunction from a judge. It does not require deciding the case.

I proposed, at least twice, that the bureaucrats ought to sit as an "adminship committee" and be empowered to resolve all questions relating to adminship status. This was rejected, of course. The reasons for rejection ranged from the comical to the absurd.

[Abd]

I proposed, at least twice, that the bureaucrats ought to sit as an "adminship committee" and be empowered to resolve all questions relating to adminship status. This was rejected, of course. The reasons for rejection ranged from the comical to the absurd.

Thanks, Kelly. The basic idea is to create a representative body of manageable size that can then make decisions with relative efficiency, by majority vote. My criticism would be that the 'crats don't represent the community very well, but probably better than the general community is represented in the ad-hoc, disorganized discussion process that is routine.

I'd suggest the first step would be the formation of an Assembly that does fully represent the community. I'd use delegable proxy to create a bottom-up hierarchical structure, with Asset Voting (perhaps secret ballot) to select direct participants in the Assembly. Delegable proxy is totally flexible, but Asset (which can use delegable proxy data, or delegable proxy allows the vote negotiations used in Asset to proceed efficiently) produces a peer assembly, where every participant exercises, in the deliberations, the same voting power. It's more traditional than pure delegable proxy., i.e., there is lots of process experience.

Once there is a fully representative body, and once it sets up its own process for efficient deliberation, it can then handle *all* these issues, it would become far easier.

My approach, generally, though, has not been to suggest advanced systems for control, but rather to use them to generate advice. It doesn't need to be so bulletproof. I'm one of the primary inventors or theoreticians of delegable proxy, and I'm quite aware that it's largely untested, so I've suggested advisory functions first, because they are relatively fail-safe. And generating intelligent, trustworthy advice is at least half the problem.