Friday, August 11, 2006

Wikipedia and science

It never ceases to amaze me how people just don't get Wikipedia. Not just any old people, but those who you would expect to grok the concept often have the most difficulty with it.

Case in point: economist Stephen J. Dubner of Freakonomics has written a blog post entitled "More Welcome Ridicule For Wikipedia". A tad hostile, don't you think?

First he damns it with faint praise:

Wikipedia is generally fun, sometimes useful, often entertaining. What it isn’t is very dependable, for the very reason that makes it fun: it is an encyclopedia whose content is generated by random contributors.

Then he links to a "a better job of ridiculing Wikipedia than we could ever dream."

Hardly the first time Dubner has attacked Wikipedia. What's prompted this new attack?

Comedian Stephen Colbert ran amuck on American television, attacking and defacing Wikipedia entries. Hardy har har.

Of course, neither Dubner nor Colbert mention that the system worked: Colbert's vandalism was detected rapidly and reversed.

I'm forced to wonder what it is about Wikipedia that's eating Dubner. The Wikipedia system has striking similarities to the way free markets work. Anyone can go into business as a supplier of information in Wikipedia, just as anyone can go into business as a supplier of goods and services in a free market. In both, these suppliers vary greatly in quality; in both, there is no centralised "expert" who decides what is good and what is bad; in both, there are problems with cheats and vandals. One shouldn't discount the free market and institute central planning because of problems with, say, polluters. One creates systems that control or limit the problems. Free markets have laws which punish polluters, or "sin taxes" which make polluting industries pay for their anti-social activities. Wikipedia has revision control.

As an economist, no doubt Dubner could talk for hours about the emergent properties of free markets, and how the "random contributors" to the market create dependable supply and demand of goods and services without any central planner "expert" who decides what suppliers should supply or where investors should put their money. So why is he so resistant to the idea that "random contributors" to Wikipedia can lead to emergent properties of reliability, accuracy and accountability?

Had Stephen Colbert run a coffee shop and, for a lark, started putting powerful laxatives in the food and drink he served to customers, would Dubner claim this was "welcome ridicule for capitalism"? No of course not -- he'd surely point out that market economies don't collapse because of the actions of a few malicious or stupid sellers, and that buyers would rapidly learn that Colbert was providing tainted goods and they'd avoid his store, sending him broke. The free market's feedback loops are slow: likely it would take weeks or months before Colbert's dodgy store was shut down. Compare that to the Wikipedia system, which locked the vandalised articles within twenty minutes. (Of course, not all vandalism is detected that quickly. The higher the profile of the article, the more likely vandalism will be detected quickly; but the lower the profile, the less likely anyone will bother vandalising it.)

But there is another similarity to Wikipedia's open content that is even stronger than the free market. That is science. Like Wikipedia, science operates by consensus and peer review. There are experts in science, but it your peers' respect that proclaims you "expert", not merely your paper qualifications -- there are many Ph.D.s but they are not all held in the same regard, and some of the greatest scientists never held formal qualifications. Likewise, not all Wikipedia editors are equal. There is no central committee that decides scientific truth, instead there is an open market in ideas, just like in Wikipedia.

Science is transparent: much of the scientific consensus emerges from public debates in journals which are open for anyone to subscribe to. But while science also includes private debates behind the closed doors of offices and laboratories, Wikipedia is completely open: virtually everything is debated in public, with the debates recorded in the Talk pages and the editing history of the articles recorded for anyone to see. Wikipedia is even more transparent than science. For those who wish to take the time to dig under the surface, Wikipedia gives the reader the means to judge the reliability of every sentence.

You can't say the same thing about Encyclopædia Britannica. You're reduced to a take-it-or-leave-it attitude towards the articles: you either trust everything in the encyclopædia, or you trust nothing. You either trust Britannica's status as an expert source, or you don't. You can't easily determine whether this particular editor or that writer has an axe to grind, you simply have to trust that they don't. (Britannica has a good reputation for a reason, so that trust is probably not misplaced -- but you're still taking them on trust.)

Like science, Wikipedia emphasises how do you know rather than who are you -- and I suspect that's what is eating Dubner. The old authoritarian impulse at work perhaps? "Obey my authori-tay!" Both science and Wikipedia demand references, and reliable ones at that, and give no credence to people based only on "qualifications". Both operate through peer review rather than executive fiat. That's hard for some people to deal with, even some scientists.

There's a further similarity between Wikipedia and science that is amusing. In the early days of the Royal Society, there were many amateurs -- actually they were all amateurs -- and meetings often became extremely heated as people passionately argued at each other for their own unsupported ideas. Sounds just like Wikipedia during "edit wars". Nevertheless, from this unpromising start, the scientific demand for evidence rather than "expert authority" won out. In fact, a close look at the history of science shows that periods of reliance on authority have resulted in a reduction in reliability and accuracy, and for much the same reason as centralised planning usually leads to inefficient allocation of resources.

Free markets work because of, not in spite of, lots of ordinary business people. Science works because of, not in spite of, lots of ordinary scientists who aren't neccesarily leading experts in their fields. And likewise, Wikipedia doesn't depend on experts, although experts are always welcome for the depth of their knowledge and their understanding of the need for evidence and references. Wikipedia doesn't need experts because of who they are, but because of what they know. Anybody who knows what an expert knows, even if it is only a tiny little bit of what an expert knows, can contribute and help the incremental emergence of knowledge.

No comments: