Former employees of Google and IBM have recently launched a rival search engine called Cuil (pronounced 'cool'). Results are returned in a
'magazine' layout, complete with graphics and snippets of relevant text that allow users to quickly focus on articles of interest.
Cuil also tries to differentiate itself from the competition through its privacy policy - it claims to record 'nothing that is personally identifiable about your searches', which includes search-history, ip-addresses, cookies, names, etc.
Cuil also claim to have more pages indexed than any other search engine (including Google). However, the relevance of this point is a questionable for several reasons - search engines live or die by the quality of the results they return, not by the quantity of data stored in their databases. Quite whether having a huge database helps of hinders the quality of search results is open to debate.
Another interesting feature of Cuil is the lack of 'sponsored-links' in the search results - so far the site appears to be advert-free. However, it will be interesting to see if advertising is introduced at a later date, as this is one of Google's primary revenue generators (in the form of Google Adwords).
Some may also question the need for another search engine altogether - in reality there are relatively few in mainstream use today. Clearly Google hold a significant market share (78.35%) with Yahoo (11.78%), Microsoft (5.22%) and Ask (1.19%) tending to be the main other competitors*.
It's worth knowing that several other well-known search engines actually use one of the four main competitors to generate their results - for example, AltaVista uses Yahoo, while Lycos returns results generated by Ask.
There are many other search engines, but these tend to focus in particular areas - for example Baidu is the most popular search engine in the Peoples Republic of China.
So how does Cuil perform when compared to, say, Google?
Not bad, but it depends what you're searching for. Our experiments seem to suggest that Cuil is best suited to 'research' queries (think wikipedia or academia etc) rather than to 'purchase' queries (think amazon, ebay etc).
In particular Cuil seems to excel if you need relevant literature discussing entities or concepts in depth - this is to be expected as Cuil specifically base their results on 'content' and 'relevance', rather than just popularity.
However, the Cuil servers are noticeably slower to respond than other search engines - and occasionally time-out altogether. This is a potential problem, but it appears to be due to a higher-than-expected level of interest; hopefully these are simply teething issues and we should expect responsiveness to improve over the coming weeks and months.
Cuil is certainly worth a look - it may not be a Google killer, but it could become a viable alternative and is likely to influence how other search-engines present their results in future.
* figures taken from hitslink
