Thursday, December 15, 2005

Nature on Wikipedia

Chris Masse tipped me off that the new Nature contains this article by Jim Giles talking about Wikipedia, and in particular comparing its accuracy to Encyclopedia Britannica. Wikipedia has been in the news recently due to its containing a spoof page alleging a journalist was involved in the assassination of two Kennedys, so some might be surprised that Nature found its overall standard to be not far behind its more illustrious (and expensive) cousin in terms of the number of mistakes identified by reviewers.

The accompanying box (link) "Challenges of being a Wikipedian" mentions William Connolley's involvement in the "climate wars". The fact that Wikipedia works at all suggests to me that most errors are due to ignorance rather than malice, but a handful of trouble-makers can clearly make quite a nuisance of themselves. I'm not really convinced that it's worth the effort but I'm glad someone else is :-)

There's some slightly odd spin in the Nature article, on the number of contributors:
As well as comparing the two encyclopaedias, Nature surveyed more than 1,000 Nature authors and found that although more than 70% had heard of Wikipedia and 17% of those consulted it on a weekly basis, less than 10% help to update it.
A figure in the region of 10% seems like an extremely high contribution rate to me. I would have probably bet on 1% or lower - and how many contribute to usenet or other interactive internet-based media such as blogs, I wonder? Note that this figure represents about 80% of those Nature authors who actually use it regularly (17% of 70%)! In contrast to his tone, I find such a high participation rate to be very encouraging.

The article mentions that the organisers plan to introduce "stable" versions of pages once they reach a high enough quality. In the meantime, here's a small tip for users: when referring to a Wikipedia page, always use a specified version rather than linking to the current editable page (you can get a permanent link to the current version from the "permanent link" in the "toolbox" on the LHS of the page, or use the "history" tab at the top to look though older versions - eg here is a decent page on the Earth Simulator, but this version - although identical at the time of posting - may be changed at any time). That way, at least you know your readers will get to see what you saw, rather than some potentially very different (possibly vandalised) version. Checking the history also allows you to see how controversial the page is. Of course this doesn't ensure that the information is correct, but anyone who bets the farm based on a single unverified web page pretty much deserves what they get anyway.


CapitalistImperialistPig said...

I use wikipedia frequently, but I admit to being pretty unclear on the concept. In particular, what is the virtue of picking some particular version of a page? How does one know that any given version is or is not likely to be better than any other?

In particular, I have no idea what kind of quality control there is. It's pretty good at explanations, so I like it for that, but the whole concept seems reminiscent of the habit my mother-in-law had of asking random strangers in the drugstore for medical advice

James Annan said...


The "quality control" is just the belief that rational debate will win the day. Each page has an associated "talk" page, and there are mechanisms for resolving serious conflicts. The proof of the pudding is that it generally looks reasonable to me, although it's not perfect.

Referring to a fixed page doesn't help if there are serious mistakes you don't know about. But if the page seems reasonable, it means that your readers get the same info rather than possibly some randomly vandalised (or subtly perverted) version. I hope they introduce an official "locked version" system soon, as most pages are IMO good enough. (The idea here is that the current page is frozen and presented as the "standard" one, but editing can continue on a behind-the-scenes version which will replace the main page when considered to be significant improvement.)

Brian said...

The user tip is a great idea - I've wondered about that issue of changes, especially if I want to quote something. People experienced with wikis could always find the latest version themselves if they're curious.

FWIW, I personally created the original wiki entry for Denali National Park, where I used to work:

This is what it looks like now:

An impressive development from my lame beginning.

William M. Connolley said...

The versioning system has been talked about for a while. Its due in soon. Its not really clear at this point (at least to me!) how it will work though. In principle it sounds good though.

James Annan said...

Oh, I've just spotted the "permanent link" listed in teh "toolbox" on the LHS of the page, which gives a permanent link (really!) to the current version. That's easier than clicking through the history.