On Friday at the O’Reilly Open Source Convention, Wikia, the Web 2.0 community builder behind Wikipedia, gave the world an update on its progress toward building a new search platform based on open-source software and human collaboration.
Wikia executives said that by combining Grub with the power of a wiki to form social consensus, the Wikia search project has taken the next major step toward a future in which search is open and transparent.
Full article at http://www.newsfactor.com/story.xhtml?story_id=13200C4Q51SO&nl=2; additional story at http://www.linuxinsider.com/edpick/58567.html
At first blush, this sounds like a great idea! But, where human input is allowed–especially when from anonymous sources–one must watch for agenda-driven results, rather than relevance-driven results. We see this already with Google-bombing, Digg-ing, and even on Wikipedia itself. Part of the death knell of DMOZ was ego-driven editors keeping rivals from the listings. The directory was wildly incomplete, and its utility was limited and eventually surpassed by Google’s
For politicians, their entries in Wikipedia are a constant tug-of-war between their supporters (or their staff) and those who oppose the politician. Usually, the truth is somewhere in the middle, but depending on the moment you read Wikipedia, are you getting the full story?
User reviews have also become targets of “black-hat” techniques. Recently here in Pittsburgh, someone rated six flower shops with almost identical comments (now removed) and 1-2 stars. There was obvious malicious intent, possibly from a competitor. The reviewer could not show she had ever been a customer from any of the shops she reviewed, which is why the reviews were promptly removed when pointed out to the review service.
Anonymous human collaboration has opened up an entire new industry known as “reputation management”. The mere existance of such an industry puts the lie to the idea that harnessing human collaboration is superior to cold mathematics. I’d much rather have the scatter from mathematics than a single train of thought provided by the most active linkers/diggers/editors.