Wikipedia Empire Based on Open Source Worst Practices?

The way that Wikipedia is implemented and managed is an example of Worst Practices for an open source project. In the conclusion of this essay I'll comment on that and how it could be fixed but first, whey does it matter?

Why Does it Matter How Wikipedia is Run?

Isn't it interesting, as Nicholas Carr reminds us, that Wikipedia often ranks high in search results? Mr. Carr did his own little experiment by Googling for terms like "World War II", "Genome", and "Magna Carta" finding Wikipedia in the top 10 in every case. I was able to replicate his results and my additional searches ("peak oil", "war on terror", "william s. burroughs", etc.) had similar results.

And, of course, Wikipedia's popularity does not originate simply in its search rankings. One must also look at who promotes it, where, and how. But the search rankings are a nice tangible indication of Wikipedia's prominence.

Sadly, cheerleaders can (and probably will) claim such evidence helps to support the hypothesis that Wikipedia is simply a great resource.

Meanwhile, those of us looking critically have to ask about social impact and quality and wonder how those factors relate to the management structure and technology structure of Wikipedia. In other words, what are the consequences of giving so much search-engine driven attention to information of such broad scope from a single project? How reliable is that information? How complete? How biased? To what extent are those metrics determined by the way the project is run and the way the software is implemented?

Social Impact

One example....

The social impact of Wikipedia now promises to go far beyond the net. The one laptop per child project aims to distribute millions of $100 laptops to children in underdeveloped areas. Because Internet connectivity will often be absent or sporadic for these children, the laptops are designed to be useful in isolation and in small clusters.

Mr. Wales (Wikipedia founder and chairman of the board) has announced that static snapshots of the Wikipedia will be included on those laptops.

In other words, the One Laptop Per Child project is, additionally, a One Encyclopedia Per Child project and that Encyclopedia is the Wikipedia. For those children, search engines are beyond the horizon. Wikipedia will be their primary source of information on many topics.

Management

The Wikipedia community, such as it is, is all a-titter about the degree to which Wikipedia is a centralized effort rather than a distributed effort. I think that concern is slightly off the mark.

It is certainly true that, at the highest levels, Wikipedia is thoroughly centralized. A small group (dare I say clique?) resolve disputes and establish global policy.

At the same time, individual articles are are more like fiefdoms, at least the interesting articles. Wikipedia is less a centralized bureaucracy, and more of a semi-anarchic feudal state. For example, and in my experience, if you want to make an addition to a controversial article you will quickly come up against an "article squatter" -- the de facto local lord. An article squatter is someone who is ingratiated to the central Wikipedia authority and who monitors, in more or less real-time, all changes made to the article. Article squatters are the feudal lords of Wikipedia and the only way to displace someone is to spend a great deal of time fighting with them, possibly escalating through the central authority. No doubt there exist article squatters who make it there mission to work with others to improve content but, in my samples, the trend is more towards censorship.

The key point is this: while the Wikipedia Central Authority lays down some ground rules, areas where controversial content is likely to arise is largely dominated by the intersection of two sets: those who have too much time on their hands (and a vested interest in the article), and those who know how to ingratiate themselves to the central authority (who, by in large, would rather set broad policy than study the issues of any one article).

Returning this to social impact, to name just one example: our One Encyclopedia Per Child user base will, on many controversial articles, be listening to little more than the those writers in the Wikipedia community who have the greatest resources to spend defining a so-called neutral point of view.

Wikpedia Technology v. Open Source Best Practices

To a large degree, for open source projects,

• technology infrastructure determines...
• management structure determines...
• social behavior determines...
• quality and degrees of freedom.

I could write a book on that although it might be partially redundant.

In the case of Wikipedia, we should compare Wikipedia to, say, the way that complete (GNU/)Linux operating systems distributions work.

Unix systems are highly modular. That is to say, they are made up of many separately developed and maintained components that add up to the complete system. For each component, there is a potential of choice.

A very broad and obvious example is desktop software. Some Linux distributions run the KDE desktop software, others use Gnome. The desktop software is a module. There are competing and separately maintained candidates for desktop software. There is no central authority for desktop software.

Why does that de-centralization and competition-for-quality happen? Because unix (including linux) is designed to for it. Unix is designed to break down into modules like that. One has to work hard (and to be sure, some do) to eliminate the naturally competitive, naturally decentralized environment created by the software architecture.

Now, why isn't Wikipedia like that? Encyclopedias are also, quite naturally modular. Isn't each article a separate module, more or less? To be sure, all articles have to be written so that they can be combined with others, but there is no good reason that we shouldn't have multiple competing versions of, say, the article about Jesus, or the article about George Washington.

The Unix architecture separates and democratizes both aggregation from participation. For example, one can participate in the unix desktop competition by working on Gnome or working on KDE or starting some completely new desktop software. There is no Central Unix Authority to decide what participation is legitimate or not. Now, once there are competing versions of each module, the next question is: which do we use? If I buy a Unix system, do I get KDE or Gnome? Once again, the architecture of unix (and open source best practices) leaves the question open. One vendor will sell me a system with KDE. Another vendor will sell me a system with Gnome.

Why isn't Wikipedia like that? Instead of "edit wars" where wikipedia editors compete for the favor of King Wales, why don't we have one competitive market for articles and a second competitive market for encyclopedias (that is, aggregations of articles)?

Wikipedia could have been created as a set of interoperability standards: a way in which anyone could create an article project that would be compatible (more or less) with all Wikipedia aggregates. Jimmy Wales could have created the analog of unix architecture for encyclopedias. It would have been an interesting experiment.

Instead, Jimmy carved himself a kingdom. And with One Laptop Per Child, His Kingdom is about to become an empire. This can't be good.

-t