Joi Ito's Web

Joi Ito's conversation with the living web.

Ethan explains that although Wikipedia tries to maintain an neutral point of view (NPOV), it is inherently systemically biased by its demographic to pay more attention to articles that the contributors know about and research from sources which are available online. Xed, a Wikipedian has tried to address this systemic bias with a new project called the "Committee Regarding Overcoming Serious Systemic Bias On Wikipedia" or CROSSBOW.

From draft CROSSBOW manifesto
Wikipedia has a number of systemic biases, mostly deriving from the demographics of our participant base, the heavy bias towards online research, and the (generally commendable) tendency to "write what you know". Systemic bias is not to be confused with systematic bias. The latter just means "thoroughgoing bias". Systemic bias means that there are structural reasons why Wikipedia gives certain topics much better coverage than others. As of this writing, Wikipedia is disproportionately white and male; disproportionately American; disproportionately written by people from white collar backgrounds. We do not think this is a result of a conspiracy - it is largely a result of self-selection - but it has effects not all of which are beneficial, and which need to be looked at and (in some cases) countered.

Wikipedia is biased toward over-inclusion of certain material pertaining to (for example) science fiction, contemporary youth culture, contemporary U.S. and UK culture in general, and anything already well covered in the English-langauge portion of the Internet. These excessive inclusions are relatively harmless: at worst, people look at some of these articles and say "this is silly, why is it in an encyclopedia?" Of far greater (and more detrimental) consequence, these same biases lead to minimal or non-existent treatment of topics of great importance. One example is that, as of this writing, the Congo Civil War, possibly the largest war since World War II has claimed over 3 million lives, but one would be hard pressed to learn much about it from Wikipedia. In fact, there is more information on a fictional plant.

They are planning a variety of projects to try to address the bias. If you are interested and can help, you should.

Our good friend Andrew Orlowski points out that as Wikipedia tries to get more distribution on smaller devices such as mobile phones, they need to be wary of the size of the database and the framework in order to be more inclusive than just web oriented techies or in his words, "Californian techno-utopians, wiki-fiddlers."

So the most useful thing the Wikipedia project could do is not write another adoring 20,000 word article on our good friend Joi Ito (the spiritual leader), or "memes", but nail down a simple lightweight framework that librarians, schools, churches and small businesses could use as an annotation and broadcast channel.

This is another way to address the bias. Move to non-web devices too, although in this article Andrew is talking about "Questions like 'What's the kid's soccer schedule?', and 'Is Thursday street cleaning day on Geary?'" I do agree that Wikipedians should be spending their time writing about the Congo Civil War instead of writing a 20,000 word article on me.


Don't worry, nobody knows enough about you to make it to 20,000 words.
That is reserved for subjects that come with schematic diagrams and an array of
customizable options.


I wouldn't worry.
Systemic bias tends to be self-correcting in any truly effective information distribution system.

(Good lord, did I really type such a sentence at 4:00 in the morning?)

Around 1993 or 1994, I was telling a friend about the internet, Usenet in particular. He's a rare book dealer, dealing almost exclusively with the very high end stuff.

I was showing him the newsgroups where books get traded and he truly wasn't impressed. "It's all science fiction and Star Trek stuff."
Well, not really, but it wasn't the class of stuff he was used to dealing with, where the dealers would often not only know the book being discussed, but often could guess what particular *copy* might be coming on the market.

Remember, this was before eBay. Now, of course, the barriers to entry are almost non-existent and anyone who wants to can connect with any other like-minded people. It's hard to imagine the Internet *not* being the place where the most important work in any niche like this happens

Wikipedia, as big as it is, is still in its infancy.
The buzz is growing and people are starting to grok the concept, but it's still a bit foreign to most people who encounter it.

As people figure it out and find it via the search engines, it will grow and change and fill out in areas that it now lacks. Right now, the data is being populated by people not so far removed from the original builders of the thing. Of course it has bias.

It won't really fill out until people forget what it's trying to *be* and only see what it *is* at its heart. Around that point, I think people will understand that this is a two-way medium and start to offer information.

No, on second thought, I think it will happen when people start to *correct* bad information that they find.
Just like Usenet.

"Systemic bias tends to be self-correcting in any truly effective information distribution system."

Yes and no :) the problem is the adjective "effective". I think Wikipedia does a faire job even with its problem but yes there's a bias. The same bias that you will find for example in search engine.

Search Engines are not anymore search engines but source of information. Basically most of the people do:

1. A question
2. type words in a search engine
3. look at the first page of results and take the most relevant information in this first page of 10 answers,
4. most of them don't check if there's a better answer 40 or 100 links below.
5. The karma of this page has increased again
6. The next user will find the same information, etc... and so on.

The good information get lost.
Another bias is English. How many people, Joi included asked me to publish in English... Why should I? :) I understand why they ask, but I don't think it's a good idea. Language is one of the shell of ideas and cultures and many non native publishes in english loosing little by little the nature of some cultures and ideas.
For example I would love to know how many articles of wikipedia have been translated from english to another language, more than the opposite.

Systemic bias in the civil society is usually regulated by making anti-discriminatory regulations. The laws of the community trying to not forget everyone, and to force the dominant group to make effort for the minority.

The good thing with wikipedia is its localization, I see a huge possibility here. In the sense that people will be able to create their own thing in their languages. It's not the case of most of the open source project out there. Look at Movable Type, Technorati, Friendster, Orkut, etc. How many of these tools are really localized, almost none. MT starts a little bit but far from what it should be.

:) yes for the infancy, yes it can be wonderful, but it will not be done without politically pushing it. I have chosen languages because it's easy, but it can be applied to many things, for cultural differences between profession when addressing a topic :)

Example of a problem with MT.

if you type a title in English, the URI can be transformed in english to have the title into it. Google index it. The summer is hot = the_summer_is_hot

if you type a title in French, the accentued characters are removed alltogether. L'été est chaud = l_t_est_chaud . No algorithm to transform it as l_ete_est_chaud

if you type a title in Chinese, everything disappear.... pooofff

Here you would need a volunteer politic from SixApart to develop the internationalization of it and the international standard on IRI (International URI) soon released at IETF and the implementation of it. Many things ;)

But I'm like you, I tend to be positive it will be possible to succeed.

What is a neutral point of view? We can be objective about the birthdate of a politician but what room is their for revisionism in all its forms. How come the wikipedia hasn't forked (perhaps it has?) where particular groups like feminists, ecologists, neocons or anarchists haven't given their own spin on it.

NPOV is a good thing to aim for but we should be clear that
a. NPOV may mean very bland articles
b. Very comprehensive articles that lose their concision and read like literature reviews
c. do not represent anything akin to an 'objective' point of view

4 TrackBacks

Listed below are links to blogs that reference this entry: Addressing Wikipedia's systemic bias.

TrackBack URL for this entry:

Le débat sur l'intérêt du concept de wiki, et surtout sur le niveau qualitatif atteint par Wikipedia, a redémarré, notamment... Read More

(Via Joi Ito) Interesting posting on how the Wikipedia, as it matures both in content and access technology, needs to address issues of systemic bias. I am currently fiddling with using a wiki in a corporate research context, and have quickly found tha... Read More

Joi blogs about an inherent demographic bias at WikiPedia and a plan to address the problem. He also mentions Orlowski's comments, but (as usual) he's more tactful than I was.... Read More

Joi Ito提到Ethan Zckerman憂慮wikipedia偏見的文章。


Read More