Birgitte SB wrote:
I think there was recently a thread about the press about the paper A Gene Wiki for Community Annotation of Gene Function [1]. I was reading it today and found it interesting in respect to views generally expressed on this mailing list against bot created articles. Personally I can't see why this sort of work described here should be required to be done by hand (as is the case where some wikipedias don't allow this sort of bot creation). Especially when analysis found that after the bot created stubs for all genes in the authorative database that were missing from Wikipedia, "approximately 50% of all edits to gene pages were made on the newly created pages."
PLoS Biology is a recognized journal for biology research, but not for wiki research. Their statements about the usefulness in wikis of bot-generated stubs are not backed up by verifiable evidence.
For example, they don't define what a "stub" is, and how the usefulness varies with that definition. The stub shown as example in the article (fig. 1) is far longer and more well-written than what one usually has to confront when criticizing stub articles in Wikipedia.
Their statistic that 50% of edits landed in new articles doesn't indicate quality or usefulness. It only says that carpet bombing might sometimes hit a target.
Their work is interesting biology. But for wiki research, this paper is merely of anecdotal interest. Maybe they are writing a separate article focused on wikis? Are the authors coming to Wikimania?
There is also interesting argument is made about how the existence of a complete network (even if, as in this case, partially consisting of bot-created stubs) leads to more efficient browing of the entire subject area.
Yes, if your task is to create a navigational user interface, then a wiki might be a useful tool. But that doesn't imply that this is a good method for creating a free encyclopedia.