Photo: Flickr user lifeontheedge

Thursday, October 27, 2005

Another deeply critical Register article: Why Wikipedia isn't like Linux. It isn't like linux, in the way programming isn't like writing: humans are better at reading text than computers are at reading code, so writing can be wrong in ways that code can't (code and writing can both be misguided, of course). But orders of magnitude more people know how to write than know how to code -- wikipedia has the advantage of scale.

This is also an excuse to link to my Wikipedia/Linux comparison.

Tuesday, October 25, 2005

In the mailing list, Daniel P. B. Smith has an interesting take on the Guardian's panel-of-experts wikipedia analysis.

Crappy prose isn't the main "quality" problem.

The quality problem in our articles isn't "crappy prose." Sure, a sentence like

"Implicit order means a system has hidden information which is not apparent based solely on direct observation."

makes me wince.

But improving the quality of an article is more than just rewriting crappy prose. Or checking facts. Or adding references. Or fixing typos. Or removing inaccuracies.

As the -paedia root implies, a good encyclopedia article should teach. That means it should go beyond Gradgrind, facts, facts, facts. It should convey not data, not information, but knowledge. It should integrate and synthesize; it should be comprehensible to a wide range of readers, for example by having a progressive structure that gives the basics quickly without sacrificing detail later. It should be well balanced, giving suitably proportioned weight to all aspects of its subject. It should be analytical.

It takes a _lot_ of work to do that.

If you look at the Guardian criticisms, Mike Barnes complains that some of the writing in an article is "unhelpful." He thinks encyclopedia articles should be helpful.

Alexandra Shulman complains that an article "inaccurate and unclear" and that "every value judgment it makes is wrong." She thinks encyclopedia articles should be not only accurate but clear, and that they _should present value judgements--sound ones._

Mark Kurlansky complains of some factual details. No comment on the prose.

Anthony Julius complains that an article is "purely factual and not in any way analytical." He thinks encyclopedia articles should be analytical.

Claire Tomalin complains of minor accuracies, complains more about major _omissions_, and by the failure of the article to comment on the literary merit of Pepy's diary. She thinks an encyclopedia should provide balanced coverage of a topic. Like Shulman, she thinks it should present sound value judgements.

Derek Barker indeed complains of prose style.

Robert McHenry complains that an article shows "no understanding of the cultural and historical contexts involved. In other words, it is a school essay, sketchy and poorly balanced."

Most complain of inaccuracies, but characterize them as minor.

Only one complains of bad prose.

The others complain of articles that fail to be helpful; clear; present sound value judgements; be analytical; and provide balanced coverage of their subject.

Monday, October 24, 2005

My Doppelgänger

Matthew White publishes WikiWatch, a weblog that, like this one, is about Wikipedia. But his focuses on faulty articles (it's a good resource if you're a wikipedian looking for things to fix) and sharp criticism.

I mostly disagree with him, of course (see yesterday's post. Well, pretty much any post). Here's his FAQ.