Wikipedia has always been a thorn in the side of scholars because no one knows who has written the information or who has edited it. That makes it hard to judge the value of the information, without know the credentials of the writer.
An academic, Dr. Robert Hoffman, has suggested that a new uber-wiki, where all knowledge is signed by its author (including edits) and where it is rated by readers would be more creditable. You can read about it in a brief note in The Register or in a scholarly article in Nature Genetics (ref: Nature Genetics 40, 1047 - 1051 (2008), doi:10.1038/ng.f.217).
However, I think his idea may (in the nature of how things happen on the web) have already been overcome by events since Google's Knols seem to be exactly what he's describing -- or very close to it. If you haven't checked them out, do. The best way to check out a reference work, of course, is to look at an area you know very well and to see how well they do. I checked out the articles on Diabetes (which I have and know a great deal about as a consequence); the articles are amazingly detailed and if there is an error I couldn't find it.
Of course, any new venture into building a knowledge data base will run into the fact that Wikipedia has a huge number of articles and building a new set of articles will take a long time, but having a second (and signed) resource sounds like a great idea. I'm thinking about writing a couple of knols myself.