The expertise problem.

[reddragdiva]

English Wikipedia is allegedly anti-expert. This fails to explain why you can hardly move on the wiki without bumping into someone with multiple degrees, or how it got tagged "unemployed Ph. D. deathmatch."

I submit that English Wikipedia does not have a bias against experts (although there are editors who clearly do), but that massive collaboration is hard. The main problem is how to work with idiots you can’t get rid of, who consider you an idiot they can’t get rid of. "Assume good faith" is not a platitude, it's a warning that someone really can be that clueless and that sincere idiocy is ten times as hard to deal with as knowing trolling; it's a nicer way of phrasing "don't assume malice where stupidity will suffice." Summary of the summary: people remain the problem.

Academia has evolved mechanisms to deal with antisocial idiots (throw them out) and antisocial experts (put them to work in a locked room and keep them away from humans); wikis are still working on the problem. Antisocial experts on a wiki — unquestionably expert, unquestionably unable to collaborate on a wiki — are really special. Thankfully they're usually too weird to then go blogging about it ...

How do other wikis cope with this? Other Wikipedias? Citizendium doesn't seem to have had this yet that I know of, but that could just be early days. Ideas?

Edit: See also this post on my Wikimedia blog, with some comments from people who've been sucked deeper into the thing.

Comments

[vatine]

It's not so much "anti-expert" as "not whoringly pro-expert", from what I've seen (if you're an expert in the subject area, you can certainly edit to your heart's content, but you should consider not creating new article material, since that may fall fould of "no original research", one if not the only reason I've hesitated Wikipedifying an article about Caps, a drinking game quite common at Swedish universities, especially of the engineering sub-breed).

[vatine]

As an aside, have you seen Lauren Weinstein's recent Wikipedia critique</>? If taken seriously enough, it would mean the end of anonymous Wikipedits.

[reddragdiva]

Web 2.0 will destroy civilisation because people can put things into edit boxes unsupervised.

[ciphergoth.livejournal.com]

The article on Trivium (cipher) is currently very misleading, because it's been edited by a rabid lying nutjob with an axe to grind about why his stupid cipher is the best in the world. I doubt there's a Wikipedia editor better qualified than me to comment and correct, but I can't be bothered to fight with this nutjob any more, so I've left it. That's not an anti-expert bias, but if there were a pro-expert bias then it would be harder for him to screw up my changes.

Not that I have any idea how such a thing could be sensibly achieved.

[reddragdiva]

Yuh. The problem is that limiting the bad stuff limits the good stuff, c.f. Nupedia vs Wikipedia.

[m0rbid-princess.livejournal.com]

There's serious business on them internets...

[reddragdiva]

Worse yet, other people prefer that to properly supervised product. We must educate them against this.

[mirrorshard.livejournal.com]

[Y]ou can hardly move on the wiki without bumping into someone with multiple degrees - and tenured professors of theology.

Had to get a quick snark in, but now to the actual substance of comment.

I did write a longish one about user-generated content in virtual worlds, but snipped it because none of the approaches taken there would work for Wikipedia - they're all aimed at reinforcing the designer's paradigm, or promoting individual ownership of content.

So I'd argue that the problem is a combination of anonymity, the very flat official hierarchy, and the very steep, complex, multidimensional unofficial one. The opt-in paradigm doesn't help either, of course, but that one's never going to go away.

I hate saying this, because I'm a very strong non-hierarchialist, but it needs to be managed more explicitly and using more sensible metrics, derived from networked trust systems probably. But that gets us into googlemath.

[hirez.livejournal.com]

Idea? Boot the nonce involved with the Saberhagen bio.

[m0rbid-princess.livejournal.com]

and the very steep, complex, multidimensional unofficial one.

I remember when I first started to read discussion pages on controversial topics, I felt like I'd wandered into an eleborate Mornington Crescent style game with people quoting rules they'd just made up to counter rules someone else had just made up :)

[mirrorshard.livejournal.com]

Wikipedia Crescent is completely different. The rules are all documented, and everything :)

[vatine]

NBot so much "unsupervised" as "with not much in the way of traceability" (admittedly, a self-chosen username without prior validation isn't vastly better). Though I do find the question of editorial responsibility interesting.

[reddragdiva]

Trust metrics are an attempt to solve the grey area problem of social interaction by promising to take away the hard bit, i.e. actually interacting with people. I've yet to see a site where this works on more than a superficial level and doesn't just become a set of playground equipment for trolls.

And since interacting with people is the actual point, I am sceptical of the possibility of an algorithm that would answer what one is interested in better than actually interacting and collaborate with them would: what are they like to interact and collaborate with?

Of course, I could be wrong. Please do elaborate, I can feel a next post coming on ;-)

[vatine]

It's Friday, therefore the Non-fungibility of non-coinage rule is in effect (WP:nfonc) and you are not allowed to discuss Wikipedia Crescent!

[reddragdiva]

Attempting to play Nomic with the rules is one of the stages of development of a Wikipedia editor. It usually kicks in around three to six months if it's going to, as far as I can tell.

[reddragdiva]

No. He was entirely correct. And Scalzi and PNH were being complete cocks.

[vatine]

Most places where I've seen trust metrcis used, they tend towards "using a technical means to stop a social problem" (well, OK, I've only seen it in action on Advogato). There's certain domains where I woudl consider automating that sort of tracking, though.

[reddragdiva]

Although he could have phrased it better, but that doesn't make him wrong. PNH calling him a "psychopath" because PNH couldn't swing his dick around was completely out of order, however.

[reddragdiva]

"the very flat official hierarchy, and the very steep, complex, multidimensional unofficial one."

That's also a problem with society in general. Solving one would probably come close to solving the other. Work on one would probably be applicable to the other.

[reddragdiva]

Individual ownership of content is interesting. On Uncyclopedia, the main author basically owns the article, even though it isn't signed and others can still collaborate. The articles with massive collaboration are almost always a dog's breakfast. Humour is very individual creative writing.

[reddragdiva]

I think a useful approach to the problems with anonymity would be more rewards for consistent identities.

[mirrorshard.livejournal.com]

Points taken, and I don't want to be seen promoting technological quick fixes to social problems - I've seen enough of those attempted in the past. But what I'd like to see implemented is a more visible way to see who someone's connected to and what others have thought of them in the past, without having to delve into deep-wiki to find out.

Being able to game the system may in fact be the entire point of it - after all, as you pointed out, we have seen the trolls and they is us.

[mirrorshard.livejournal.com]

My feeling is that whilst it's definitely a problem with society in general, it gets less bad the more information we have about the other people. If we see all those cues and understand each other better, we can establish positions more easily and don't have to shove and jostle and pose.

[reddragdiva]

The social game (in general) is not optional and it's hard work and some people are just better at it than others. I dislike the idea of trust metrics because they seem to me to be snake oil promising to take away the work and some people being better at it, and those bits are just not optional. But then I'm good at it for my purposes, so I would say that, wouldn't I.

On Wikipedia, looking through someone's contribution list can be useful. You'll see that I, for instance, hate fair use images and spoiler tags ... hmm, not much else in the last 20,000 edits ...

[reddragdiva]

The cues are a language in all given situations and the shoving, jostling and posing just take different forms.