Governance, Scaling and Anonymity in Wikipedia

    August 13, 2005

I’m sitting in Jimmy Wales’ talk at OSAF, as though I am his roadie these days, and reminded about anonymity in Wikipedia.

Anonymity is not something commonly valued in the blog world, where it is largely a strong expression of identity, but seems to be an essential attribute within the Wikipedia community. Maybe it’s just the difference of people working together vs. having conversations. Perhaps it’s the initial user experience of being able to edit without logging in, or strong enough social bonds and extreme cases for widespread support for maintaining anonymity.

Link: Open Source Applications Foundation

Jimmy describes the basics of Wikipedia, and then gets on his self-acknowledged soap box. Most social software is designed in a way that makes no sense. If you think about it resurant, serving steak, you need knives, because the customers might stab each other, so, no knives. This creates a culture without trust, with comunity. Most software is too complex from trying to keep people from being bad. Leave things open when you know people can do bad things. Instead of locking pages, leave a note asking them not to damage it — an opportunity to build trust. When they haven’t done any damage in a while, I know Stewart, for example, has not vandalized this page, so I trust him more.

Governance in Wikipedia is a confusing but workable mix of consensus (0.5% of users generate 50% of the edits, a little over 500 people — IMHO, the big scaling and usability issue to be tested), democracy (Vote for Deletion), aristocracy (closer to the statement, “leadership in Wikipedia is knowing where information is) and monarchy (Jimbo as benevolent dictator of all human knowledge, for now). They are “flexible over methodology and value results over process.”

On Wikimania: 50 countries, fascinating experience where he met a lot of them before from traveling and online, the social introductions The goal was increasing cooperation and coordination (it seemed to work from an outsider’s point of view, as F2F often does). Italians found a public domain source of every small town in Italy, created thousands of articles, but the work hasn’t been done in other languages, a big topic of the event.

Stuart Brand asks what the scaling and rule set issues are. The Burning Man organization realized they had a scaling issue when they had to institute a rule of no dogs. Is there a pattern over time? Jimmy responds that communities are inherently scalable from village to town to city to metropolis — but new problems arise. The Soviet top-down method of scaling communities does not scale, but market interaction does. People watch each other. More rules are necessary over time, but they encourage not making rules until they are absolutely necessary. Small languages haven’t made the same rules as large ones. The 3 revert law (in one day) in English version was created to prevent revert wars. But in small languages it’s okay to do 10. What I do varies in many languages, I can’t work on the Chinese Wikipedia in the same way as English, so they vote a lot. Dutch Wikipedia had a civil war, 500 users quit. Problems people think we have are not problems, every one agrees on how to deal with bad users. Dealing with good and bad users is tough. i talk to people to figure out the people and talk to them, id the dutch offender and banned him for 72 hours which shocked some, then the community adapted. Dutch people fight a lot, they become tolerant by arguing.

All we can do is communicate the values that have worked. People are surprisingly good.

Mitch Kapor points out how interesting that there is a back and forth between Jimmy and Stuart, who founded the Well, which had the village model, even with 5-8k people. Wikipedia is a megaopolis. Jimmy notes that conversations still take place around the viliage pump (what a quaint, but civil unlike a hole, notion of a water cooler). Mitch recalls that at one point there were more articles on Tolkien’s Middle Earth than Africa. Jimmy recalls that there were more words about the war on Deep Space 9 than the Congo war and how old time Wikipedians saw that africa was a red link, then someone wrote that africa is a continent.

The Wikipedia community is deeply interested in the systemic bias, that may come with the typical Wikipedia definition. The definers are similar to this room, male dominated (not as bad as free software), tech and broadband access enabled. This used to be worse as it came from the free software world (the first traffic boost was from Slashdot). There are things we need to do to make it more accessible, notably wysiwyg, outeach and keeping the community friendly — something we are good at. Some times the programmer want to automate everything, say give a welcome message after the 10th edit, but what you want is people saying hey, welcome! Something personal.

I asked about anonymity and the great debate on making reputation explicit to aid scaling. Jimmy clearly values keeping reputation implicit and he noted that that works in more market-oriented systems like eBay (huge nod, social capital isn’t fungible and explicit reputation is a barrier to community formation). Notes subcultures on topical areas arise which help scale.

When asked about the early days, a question that alludes positioning for scale, Neupedia had the start of the right community. He personally published paper in option pricing theory, so he tried to write a biography on Robert Myrton, but found the process frustrating, which opened him to the idea. Neupedia had 5k people on the mailing lists, fairly academic, to tap. But software and social model (7 step workflow for publishing), was staying in the way. In the first two weeks 2 weeks they did more work than in 2 years

Don’t have enough academic studies about it, but anecdotally, articles across languages are similar. But there are exceptions. The English Wikipedia said the Wright Brothers invented the airplane, French said otherwise. Now there is a wonderful and detailed article on definitions and a discussion of the issue. Korean and Japanese Wikipedias differ on disputed islands. In the Japanese Wikipedia, incidentally, they use the discussion page for a long time before they make the article.

One thing we haven’t tried is Wikitorials. Skeptical it could work. With the LA Times they couldn’t stand to see a wiki down, so they tried to step in, although they were questioning why they were helping what might be a competitor. they turned off recent changes, a tool for a community to monitor itself, they forked the edit conflict, but a fork is not enough, there are many views.

Someone asks about if SEO gained them a position of being one of the top 50 sites on the web (no) and if a mainstream competitor could create something better, Encarta could open, but it is hardly likely people would contribute, people wouldn’t trust them and a community would be hard to form. Couldn’t say if traditional encyclopedias could come up with a better approach and go all the way. But in the end, there is no such thing as competition for Wikipedia.

Reader Comments

Ross Mayfield is CEO and co-founder of Socialtext, an emerging provider of Enterprise Social Software that dramatically increases group productivity and develops a group memory.

He also writes Ross Mayfield’s Weblog which focuses on markets, technology and musings.