Architecture is politics: community building and the success of Wikipedia

I find it surprising, almost shocking, that a system as apparently bottom up as Wikipedia works as well as it does. Anybody really can contribute, even anonymously, and without fear of damaging their reputation; while people can be banned, only a tiny fraction of users ever are; it’s easy and very common for people to vandalize pages. A priori it seems like a recipe for disaster, yet for the most part works quite well in practice.

I used to think the explanation for Wikipedia’s success was excellent leadership focused strongly on building a healthy and vibrant community. This certainly does play a significant role – check out this inspiring podcast from Wikipedia founder Jimmy Wales, who has thought long and hard about building great communities.

But great leadership and community building is only a partial explanation for Wikipedia’s success. It is clearly also the case that the design of the MediaWiki software underlying Wikipedia contributes in a major way to the success of the project.

This thought crystallized out for me in a particularly nice way this morning in the form of two fantastic slogans, Mitch Kapor’s “architecture is politics” and Lawrence Lessig’s “code is law” (book, free online).

Kapor’s observation, in particular, explains and neatly summarizes a great deal about community-building projects like Wikipedia. Tweak the software even a little, and one can cause enormous changes in how community interaction is mediated, and thus in how the community functions. Wikipedia is less bottom up than it appears, for it is a relatively small group of people who are effectively responsible for the design and development of the underlying software. Provided those people are competent and committed (I have no reason to doubt this), they can exert an enormous positive influence over the project as a whole, dwarfing the impact of random vandals.

An interesting consequence of this is that, in my opinion, other large projects like Wikipedia are going to need their own independent forks of the underlying software. Different communities have different needs, even during their own evolution, and the ability to change the underlying architecture of the system, and thus affect the politics, is an enormously powerful lever to have in forming a healthy community. In practice, I think this means that for ambitious projects the code used will need to be forked from an existing codebase (like MediaWiki), and developed alongside the community. Successful projects will require not just a healthy community, but a healthy co-evolution of community and codebase. To date this seems to have happened relatively rarely, and I wonder if this isn’t part of the reason why more Wiki projects haven’t succeeded on a large scale – maybe the architecture is mismatched to the needs of the community?


Related question: Mel Conway has observed that creative artifacts produced by organizations inevitably reflect the channels of communication and control in those organizations. What does this principle imply about Wikipedia?

Related posts: Kevin Kelly has recently written some great posts (here and here) about Wikipedia. Theresa Nielsen Hayden has written a nice post about building online communities. Clay Shirky also has some very interesting thoughts about the dynamics of online groups.