A rival to Google's search engine launched on Monday.
Developed and run by the husband-and-wife team of Stanford professor Tom Costello and former Google search architect Anna Patterson, Cuil is pitched as being bigger, faster and better than Google's flagship search engine in pretty much every way.
The most important difference between Cuil and Google is its ranking system. Rather than assigning priority to pages based on inbound links as Google does — a system Google calls 'Pagerank' — Cuil analyses the content of web pages to divine their relevance to a search query.
Asked by ZDNet.co.uk sister site CNET News.com whether Cuil was therefore a 'semantic search engine' such as PowerSet (recently sold to Microsoft), Costello said Cuil's search was "contextual". "We're trying to understand the real world, not the web," he said.
This means Cuil's results are automatically categorised. When a user searches for a common name, for example, Cuil will return a result page where results for different individuals with that name are groups under tabs. It will also break out sub-topics related to each name — a search for 'Harry' results in different tabs for "Harry Potter" and "Prince Harry of Wales". The Harry Potter tab then presents further sub-links devoted to actors, Gryffindor dorm-mates and so on. "We have a strong ontological commitment," Costello said, meaning that parsing search results into readable chunks is a very big part of the Cuil value proposition.
The service also displays images from web results whenever possible. Another potential advantage of the context-based search is that it may allow Cuil searches to be more respectful of user privacy. Unlike Google, which simply has to track every single click to refine its index, Cuil's context-based search does not. In practice, the distinction may be moot because Cuil will need to track clicks to see if their results are working for people, but it could serve as a marketable distinction.
Context-based indexing also presents a juicier target for search spammers, but Costello said: "That's a success problem."
It is regarding the size of Cuil's web index, however, where the company makes its boldest claim. Costello said the engine was launching with 120 billion pages indexed, well over the 40 billion he said Google had.
Costello also claimed Cuil's web crawler was three times faster than Google's, although it was not clear whether he meant that 'per search computer' or for the entire system. Compared with Google's globe-spanning data network of datacentres — some of which are even set up near dams so they can tap hydro power more efficiently — Cuil's two relatively tiny datacentres, hosting fewer than 2,000 PCs in total, will have to run fast to outpace Google's crawlers.
Cuil launched on Monday without a 'beta' tag — Costello said it would be good enough to use from Day One — but the site is not as complete as Google. While Google has had failures in extending its brand, such as with Froogle or Google Base, its collection of services that are affiliated with its mainstream search product, such as Google Maps, Image Search and Desktop Search, can make switching away from Google difficult for users. Costello said he realises that Cuil needs to layer in additional services, but added that the company had to start somewhere.
Talkback
Just gave it a try and I have to say, its the slowest search engine I have ever tried. Searches that return instantly with Google are taking 10 to 15 seconds on Cuil. If they can't sort out the poor performance, and work out a better way to display the results, then its a non-starter for me.