Asimov once wrote of a universe where intelligent life was intelligence alone, roaming the avenues of pure thought divorced from all physical form. To reach the infosphere with your feet still on this planet though, you need a point of access, a gateway to what could easily become another world in your life. The nature of these gateways, their availability and their owners are among the more contentious aspects of the global information infrastructure.

The task of developing this infrastructure to the point of ubiquity has been compared to the construction of road networks - any place, however remote, must be connected to every other. As this would make little economic sense for private investors, the task had to be one for the government. Present-day governments are poor, so they insist that subsidies should come from business in the form of universal service: providing the same thing at the same prices everywhere.

The rationale for this is noble - equal opportunity to all. Without roads people can't get anywhere, literally and economically, and the same must be true for information highways. But if the info-highway is actually an information ocean, the situation changes dramatically. It is no longer necessary to move from place to place, but simply to dive in.

This is true even in practical terms. On a highway, the route traffic takes is important, because of the distance, while on the Internet routing is a very low-level, technical sort of thing, as data may travel in little pieces on different paths, ignoring geographical distance. Moreover, a highway needs a physical connection between any two points, which affects costs, while the infosphere floats above waiting to be reached through a satellite transceiver from just anywhere.

Nor will the lack of universal service lead to a society of information haves and have-nots. Information is unique in that one doesn't 'have' it. It is created, and a 'have- not' can easily become a 'have'. The possible inequality is in trading information resources, in knowledge opportunities. However, especially in the field of information, people are not equally equipped to take advantage of opportunities, however unequally distributed the latter may be. Instead of demanding universal service, governments should encourage the development of infrastructure where it will be best used, which is in the interests of business, anyway.

Unfortunately even telecom corporations have an incorrect perception of who the best users are. They believe in connecting those who can pay now for services, rather than the many more who would benefit and therefore be able to pay the most, once they exploit currently non-existent opportunities.

Demand-driven markets grow fastest, though companies right now seem to be more interested in shoving interactive television down the eyes of the reluctant and rather bored elite. A demand-driven information market is harder to work with, as prospective buyers (as also the sellers) have to be convinced of the benefits of technology. But the potential markets are huge, if far from the minds of the vendors of 'convergence' technology, as peoples' information (or entertainment) needs are often more basic than their material needs - just look at Bombay slum- dwellers, who despite limited access to clean drinking water, buy colour televisions and VCRs. They only needed to see what TV was good for, first.

An information ocean is cheaper to build than any highway could possibly be. It needs no legislation to exist everywhere, this is in its very nature. Gateways to the infosphere will, governments permitting, sprout spontaneously wherever needed, as cracks on the surface of a frozen sea.