If you fancy a challenge, think about making the Web redundant. Think about getting all the programmers and designers who create Web pages in HTML (Hypertext Mark-up Language) to use something different. Think about converting them to Curl.

It sounds like a long shot, but it is not impossible. The information technology industry is used to sweeping changes, such as the move from Microsoft's MS-DOS to Windows, or from proprietary networking systems to internet protocol.

Curl does have a couple of advantages, according to the company's co-founder, chairman and chief executive, Mr Bob Young. It comes out of the Massachusetts Institute of Technology (MIT), where the World Wide Web Consortium (W3C) is based, and it has the backing of Mr Tim Berners-Lee, the Web's inventor and W3C's founder, who is also an adviser to Curl.

The problem with the Web is that it is great for looking at things, but generally poor at doing things. Curl's idea is to provide the functionality of desktop PC applications inside a browser. This becomes possible if you create sites using the Curl content language instead of HTML. The drawback is that to view them, users have to download Surge, which enables the browser to run Curl programs.

Mr Young says: "Users have to get the plug-in, just as they do if you throw some Flash on your site. Surge is about five megabytes to download, which is about the same size as an MP3 file, so it's like downloading a song. So as with Flash, you'll be asked if you want the regular site or the enhanced site. The difference is that with Flash, the site is generally not functionally any better, but with Curl it is."

The intranet is where Curl is being targeted now. The first application has just gone live at Siemens and British Telecom is using it. Curl has opened an office in Britain to sell it to companies. Users can download Surge from Curl's website free, but companies with internal networks can use it today, for a price.

Mr Young says Curl's big advantage is that it brings together the functionality of disparate Web technologies, including HTML, and Java-like programming capabilities. "Everybody thought the Web would get better, but it is integration that is really the key," says Mr Young.

Curl is efficient, because once the user has the plug-in, only small pieces of code need to be sent over the Web. Young argues that this is an advantage, particularly in places such as Britain, "because of the lack of high-speed internet availability".

Curl also makes much better use of the world's computer power. At the moment, Mr Young says, the Web is complicated by the fact that, in the US, the computation for 150 million PCs is being done on 10 to 15 million servers. "It is an unstable situation. It is an enormously inefficient way to handle computation." Push the computation out to the PCs, as Curl does, and you increase the power of the Web "by an order of magnitude overnight.

"The whole idea of the Web being replaced by distributed executables, most people think that's the way the Web is going to go, " says Mr Young. There could be alternatives, of course, such as Sun Microsystems' Java, and Microsoft's emerging .Net strategy. But Java has failed on the desktop and .Net doesn't exist yet. Either way, Mr Young claims Curl isn't out to replace these rival technologies: it complements them. The latter half of the 1990s saw an attempted counter-revolution led by people selling large servers and large databases to run on them - Sun's Mr Scott McNealy and Oracle's Mr Larry Ellison. The need for large hosts has not gone away but the tide has turned and there is more interest in distributed computation and peer-to-peer systems (Napster, Aimster, KaZaA and so on).

Curl may not be the software architecture for the future Web, but it is heading in the right direction.