The most anticipated launch of the year so far was the Wikia Search Alpha on Monday (although with the PhD calling I had to force myself to ignore it until today). Whilst I welcome the idea of a transparent search engine, and hate the dominance of Google in the English speaking world (such a monopoly is not in the public interest), I am unsure how well the wikia search project will work, or even if I want it to succeed.
Search engine transparency appeals to me both as a regular search engine user, and as a webometrician. I broadly agree with Wales’s awfully worded sentiment that “sunlight is the best disinfectant”, and access to crawling and ranking data could allow webometricians to have access to larger crawls of the web without compromising our understanding of where the data comes from. However, I am not a big fan of human-edited directory approach.
Wikia, like Mahola before, have decided to mix-up what we used to label directories and search engines, and create human-edited search engines, where for certain searches the search engine results include a human edited article on the topic. Whilst Mahalo pays people to create these articles, Wikia will allow any user to add to an article, in the true wiki spirit. But whilst you may be willing to contribute to a wiki article for the good of the community, would you be willing to continue doing this when the money starts to roll into Jimmy Wales’s pockets? Wikia Search is a commercial venture.
Whilst human edited search engines may have their place in world of search, they appeal to the generalist rather than the specialist. Pages for the iPhone and Heroes will quickly appear, but I can’t imagine people will still be jumping on-board with enthusiasm if they starting seeing Wales get rich, and such a search engine needs mass participation for it to be of any use for those interested in more than the latest Paris Hilton gossip.
In a perfect world an individual would have the opportunity to personalise a search engine’s ranking system to their own requirements, preferably for each specific query, varying the effects of different features according to the sort of data that was required, e.g., links, keywords, anchor text, domains. However, such personalisation is still a distant dream.