Home » Blog » Media » Systemic Biases in Wikipedia?

Systemic Biases in Wikipedia?

Wikipedia is rapidly emerging as the poster child for peer-produced content. While Linux, Apache and other high-quality open source software convinced most in the technical community that “commons based peer production” is one of the most powerful methods to create robust software, it’s been harder for non-geeks to understand the strange and wonderful processes that turn distributed effort into remarkably large and useful works.

Now one million articles strong, Wikipedia is arguably the largest encyclopedia in the world. Its philosophy of radical openness – anyone can add or subtract anything at any time, though changes can be rolled back – hasn’t led to chaos, but a fascinating system that corrects many acts of vandalism within five minutes. It’s a fantastic gateway project for non-technical people interested in contributing to a large, meaningful Open project.

Amazing though it is, Wikipedia is not flawless. It’s got a problem common to almost all peer production projects: people work on what they want to work on. (This “problem” is probably the secret sauce that makes peer production projects work… which is what makes it such a difficult problem to tackle.) Most of the people who work on Wikipedia are white, male technocrats from the US and Europe. They’re especially knowledgeable about certain subjects – technology, science fiction, libertarianism, life in the US/Europe – and tend to write about these subjects. As a result, the resource tends to be extremely deep on technical topics and shallow in other areas. Nigeria’s brilliant author, Chinua Achebe gets a 1582 byte “stub” of an article, while the GSM mobile phone standard gets 16,500 bytes of main entry, with dozens of related articles.

This caught the eye of Wikipedia contributor Xed, who identified this as a systemic, structural bias in the Wikipedia system. He’s launched a project called CROSSBOW – Committee Regarding Overcoming Serious Systemic Bias On Wikipedia – which is looking for ways to address these biases and increase the number of articles on less-covered topics and increase the visibility of the “less travelled” articles that exist.

His comments have led to a lively conversation about what topics are undercovered on Wikipedia and what steps could be taken to get more coverage. Should Wikipedia attempt to provide richer coverage on nations not featured in mainstream media, using traditional printed material to do research, which, in turn, helps correct the Wikipedia bias towards subjects easily researched on the web? (Xed and I have corresponded on this issue, and I’ve suggested my bottom 32 list of Rodney Dangerfield nations, the ones that get no respect…) Should CROSSBOW focus on new articles on African-American authors, undercovered feminists and other topics – in the developed as well as developing world, that have slipped through the cracks, as wikipedia contributor JMabel suggests? Should CROSSBOW focus on translating content from the hundred non-English wikipedias and their 650,000+ articles?

While I think these are all valid directions, I feel the solution to systemic bias in Wikipedia is the same as the solution to systemic bias in open source software development and in the blogosphere: broaden the sphere of producers. Part of what makes Wikipedia great is that people write about subjects they’re knowledgeable and passionate about. I think it’s possible that CROSSBOW participants will become sufficiently knowledgeable about the civil war in Guinea-Bissau that they can write about the conflict… but I bet the article would greatly benefit from the perspective of someone who survived the conflict. I think CROSSBOW is mostly likely to succeed if it can recruit people around the world to participate in the Wikipedia project – in their own languages and in English – to help start filling the blank spots in Wikipedia and helping it reach its full potential.

I’d love to do some work trying to help determine the “holes” in Wikipedia – I’m very interested in thoughts people have on methodology. Does it make sense to hold Wikipedia up against a traditional encyclopedia like Brittanica, and look for areas Wikipedia doesn’t cover? Or does that miss the whole point of a peer-created work?

While I figure that out, I’m starting work on my first Wikipedia article, on conflict diamonds… keep your eyes peeled.