Published online 11 January 2008 | Nature | doi:10.1038/news.2008.437

News: Briefing

Wiki search engine launched

Can a 'low quality' search engine get better through the power of the people?

Come together: more people should help make wiki search better. Maybe.GETTY

When Wikipedia founder Jimmy Wales launched his new search engine on Monday, reception proved frosty to say the least. As <a href="http://search.wikia.com/wiki/Search_Wikia">Wikia Search’s homepage</a> admits, "What you see here is our first alpha release. We are aware that the quality of the search results is low." Others have variously declared it an "inexcusable waste of time" and downright "crappy" in their reviews (see <a href="http://www.techcrunch.com/2008/01/06/wikia-search-is-a-complete-letdown/">Techcrunch</a> and <a href="http://searchengineland.com/080107-131756.php">SearchEngineLand</a> respectively). Nature News finds out just how a wiki search engine is meant to work, and whether it can realistically get better — or even good enough to take on Google.

What is the point of Wikia Search?

It's as much a political statement as anything else, according to founder Wales. “Search is a fundamental part of the infrastructure of the internet, and it should be open, democratic and participatory. That was really the initial concept [behind Wikia],” he says. Of course he hopes to make money out of it too — eventually.

Google has come to dominate the websearch market since its launch in 1998. The vast majority of people searching the internet today do so via Google; most of the others opt for Yahoo or MSN. None of these are completely open about how their systems work, says Wales: they don’t explain their algorithms and they make editorial decisions behind closed doors.

Although reception of the actual product has been hostile, the principles behind it have received a warmer welcome. "It’s important there are open-source tools because the web is so big now," says Mark Levene, an expert on web searching at the School of Computer Science and Information Systems at Birkbeck, part of the University of London. "Even five years ago I could probably have crawled the web on the resources here in my school," he adds. Now this is simply not possible, he says.

How does it work?

Wikia Search is based on a similar concept to Wales’s Wikipedia: users will control the content. The backbone of the system is the open source Nutch search engine. On top of this, users will be able to comment on the search results. Eventually, Wales hopes these comments will allow the searches to improve to a level that mean the 'low quality' statement on the homepage can be removed.

But how exactly user data will be used isn't yet known. "What we’re going to do first is to make those kinds of suggestion publicly available … and basically see what that data looks like," says Wales. "It’s very open ended," he adds. "That’s really something for the community to decide upon."

Has this been done before?

Another search engine, called Mahalo, uses teams of people to recommend and select which searches are presented to users. However, unlike Wikia Search, the people conducting this work are paid and the reasons behind their decisions are not in the public domain.

How good is it now?

A search for "Jimmy Wales" on 11 January returned, in addition to his blog, a link to a news story that isn’t there any more and a 'Bibliography of Association Football in Wales'. By contrast, Google gave his blog, his entry on Wikipedia, and a host of news stories that are indeed still available.

Users have heaped vitriol on the results. "We knew and expected that people were going to say things like that," says Wales. "Whatever. I don’t really care."

Will it get better?

Getting a sizeable community behind Wikia Search will be key to improving the quality of the results. "We will only really be able to tell if it’s going to work if in time people’s contributions make it worth considering it as one of your alternatives for searches," says Levene.

Wikipedia similarly relies on a large user base, but seems to have acquired enough to prove a valued resource to many users.

Remind me: how does Google work?

Google is based on a system called PageRank, which works on the idea that the number of links towards a site is a measure of its value. The details of the original PageRank system have been published, as have various improvements and updates to their sites. (Though if you search for "how does Google work" on Google, you get their April Fools' Day page about <a href="http://www.google.com/technology/pigeonrank.html">'PigeonRank'</a>, which is amusing but not very helpful.)

Will Wikia ever be better than Google?

Wales insists that beating Google is not part of the plan; making a viable, transparent alternative to it is. "I doubt if we will replace Google," he says. "What we’re saying is we would be thrilled if we had 5% of the search market. That would be a fabulous success."

"In terms of really getting to good quality search I think we’re looking at at least two years of work," he admits.

What's next for Wiki?

"I haven’t really thought beyond this project, which is absorbing all my time," says Wales. We'll have to wait and see. 

Commenting is now closed.