At Google, Anna Patterson designed TeraGoogle, a system that is able to index a large number of documents, while Russell Power worked on web ranking and automatic spam detection.
"Cuil's goal is to solve the two great problems of search: how to index the whole Internet - not just part of it - and how to analyze and sort out its pages so you get relevant results." Cuil thinks that today's search engines can't index all the information that is available on the web (more than one trillion pages, according to Google). Even Google admits that it's selective: "many [web pages] are similar to each other, or represent auto-generated content that isn't very useful to searchers".
Regarding ranking, Cuil combines metrics that measure popularity with information about the context of each web page. "Cuil prefers to find all the pages with your keyword or phrase and then analyze the rest of the content on those pages. During this analysis we discover that your keywords have different meanings in different contexts. Once we've established the context of the pages, we're in a much better position to help you in your search."
The most striking new idea is the way search results are formatted. Instead of the ten blue links displayed linearly, Cuil makes better use of the space by using columns. The search engine also shows thumbnails next to some of the results, but they don't always represent images included in the adjacent web page. Another interesting idea is the explorative category section that shows related Wikipedia categories and topics. Cuil has an excellent auto-complete feature and it displays a list of related searches using an design pattern that suggests exploration.
It's probably not fair to compare Cuil with Google, but when Google was launched, users could see substantially better results. Cuil returns results that are either similar to Google's results or substantially worse. In some cases, the site doesn't return any result for your queries, probably because of the huge traffic from the launch day.
Cuil has problems with relevancy, spam, robots.txt (the site indexes albums from Picasa Web) and the number of search results for almost every query is smaller than the number of Google results. This is especially obvious for queries that return a small number of results:
[louis monier altavista research labs]:
- Google: 609 results
- Cuil: 8 results
[teragoogle]
- Google: 634 results
- Cuil: 42 results
All in all, Cuil is the best search engine launched this year, but it doesn't offer convincing reasons to switch from Google. If Cuil focuses on developing technologies that allow faster indexing of web pages, it's probably the perfect match for existing search engines with less frequently updated indexes like Live Search or Ask.com.
Related:
Cuil launches - can this search start-up really best Google?
Ex-Google engineers debut 'Cuil' way to search
TechCrunch's coverage
Cuil shows us how not to launch a search engine