Monday, January 07, 2008

Wikia search: a bizarre Google killer

Jimbo Wales has just released his
Wikia Search
that is supposed to be a Google killer, combining the advantages of Google-like search engines with the virtues of Wikipedia.

So far I would bet the ranch against the success of this product. For example, you want to know something about Edward Witten. So I suppose that you must first click at "Science & Nature". In the right upper corner, you may type "Edward Witten".

The first hit is the Uncyclopedia page of your humble correspondent, a semi-witty page full of typos. The second page is another Uncyclopedia web page about the Protocols of the Elders of Zion where Witten appears together with Adolf Hitler, Jesus H. Christ, Uri Geller, and Henry Kissinger.

OK, maybe I used a totally wrong website. So I eventually found a page with a more Googlish design,
If you try to search for "Edward Witten" here, the first hit is an unexpected 14-line Spanish biography of Witten, the second page in the list is a home page of your humble correspondent's Czech translation of "The Elegant Universe" by Brian Greene, and the following three hits are dedicated to random programs at the IAS, KITP, and Oxford.

At this moment, the search engine is clearly unusable but the question is whether it can be refined to become a serious competitor of the conventional search engines. I don't quite understand how the human members of the social network behind Wikia are supposed to influence the hits. But frankly speaking, even the very fact that the algorithm is not clear indicates that it won't become a method that will use millions of people's work to improve the results.

Without this work, it is hard to imagine how a superior outcome could ever be achieved. But once again, the success of Wikipedia was unexpected for many people, too. We may also be surprised in this case. Nevertheless, at this moment, I feel that Jimmy Wales has completely misunderstood the real power and essence of the objective ranking algorithms behind Google and why these algorithms are better than recommendations of a few random biased individuals who are simply not enough to keep the most relevant pages at the top of the respective lists.

If the volunteers are expected to modify the rank of individual pages, I am afraid that the hits would eventually be contaminated by organized cliques of morons. For example, about half a million of imbeciles visit every day. An article on that Nazi hate site would be enough to convince Wikia members who also read to promote a bad page about a right-wing guy or a flattering page about a left-wing jerk. These undesirable tendencies are partially suppressed at Wikipedia because the editors are kind of responsible for their edits. It is just hard to see how can one enforce any verification of responsibility of editors for their ranking of tens of millions of pages.

Don't get me wrong: human editors could improve the hits, if they were smarter than the impersonal mechanisms behind Google. But the average internet users are clearly much more stupid than the Google's impersonal mechanisms and it is the average users who would eventually control the hits. The comparison is analogous to the comparison between direct and representative democracy. The representative democracy also contains the virtues of competition that makes it possible for the elected officials to be much better than an average citizen. Direct democracy - and the Wikia Search - kind of directly puts the average person's opinions at the top which might be too little to make a new search engine useful.

No comments:

Post a Comment