There's been an interesting discussion, summarized in a Ross Mayfield post on Many2Many about Wikipedia and authoritative knowledge.

Ross proposes some research:

I’ve been quitely[sic] forming a group of journalism schools, media centers and experts to engage in the Wemedia Project, which begins with a formal Wikipedia Article fact checking excercise and publishing findings. The USC Annenberg Center has already announced their support and next month we will begin the collaborative research process within a Socialtext Workspace. Without getting into defining truth, you can separate issue of fact, value or policy. The approach is to apply a formal fact checking process to a sample of articles to gain a baseline measure of factual accuracy and explore issues of reputation.

Good discussion follows the article too.