Remember AltaVista and Lycos? In the late 1990s, these websites were up there with Yahoo! in the popularity ranks. In 1999, Lycos was the most visited site in the world, before the internet bubble burst and Google emerged as the all-conquering search king.

Could the same happen to Google, Yahoo or Microsoft Live – which together control 90 per cent of the US search market?

Probably not, if Cuil.com – a new Silicon Valley-based search engine launched last week – is anything to go by.

Cuil – an ancient Gaelic term for 'knowledge', and pronounced 'cool' – was founded by a husband-and-wife team who seem intent on knocking Google off the number-one spot. Its president Anna Patterson was the architect of Google's search technology, experience that adds some credibility to her company's claim that Cuil's index – a whopping 120 billion pages – is three times bigger than Google's.

The start-up does have some advantages over Google. Cuil promises not to retain records of searches, something that Google does for up to two years. It also categorises results: for example, if you search for 'Dubai', then it gives you handy subject tabs that include 'Buildings And Structures In Dubai' and 'Airports In The United Arab Emirates'.

But that's just about all that's 'cool' about Cuil. The site has been criticised for its seemingly random results; a more serious problem is that Cuil is unsophisticated in filtering out obscene content, something that Google excels in. The newcomer does not have the associated services (such as news or image searches) that its rivals offer, nor does it display advertising, making its future revenue stream uncertain.

However, what Cuil does offer is a glimpse of how search may work in the future. Much has been written about the 'semantic web' – a smarter internet that 'understands' the underlying meaning of web content, and characterising the next generation of online. Cuil is a step in this direction. Its 'web page content analysis' claims to pick up on the concepts behind each query. This allows the site to group together information into subjects so that, taking the example above, it recognises which web pages fall under the category 'Buildings And Structures In Dubai'.

However, the true basis of the 'semantic web' is a new language called RDF, which allows website owners to tag content appropriately. This means that Emaar would tag pages about the Burj Dubai as 'Buildings And Structures In Dubai'. At the moment, sites such as Cuil can only make educated guesses regarding the content of web pages.

Yahoo! recently announced that it is investigating new RDF technology. For it must realise that fickle Web 2.0 users are always looking for something, well, 'cooler'.