SURVEY   TELECOMMUNICATIONS From circuits to packets Click here to go to part 1 “Down with distance”     Will the Internet steal the telephone business?   Cannibalistic urges But companies that are already big carriers of international traffic may also be forced into cannibalism. Mr Mashinsky of Arbinet, in his poky office over the disco, has built software that allows him to help a big American long-distance carrier, keen to pick up international calls in Japan but prevented from doing so by Japanese regulatory restraints. The American carrier installs a switch in Japan; Arbinet installs software between the caller and the switch; and when a customer dials a local access number, the software finds inexpensive ways to transmit the call. It strips out fax calls and sends them over the Internet, and treats voice calls as requests for a call-back service. This allows a big foreign carrier to compete directly with Japan’s international carriers before the regulations change to make that easy; and to fill its vast number of unused minutes. As for Mr Mashinsky, he has grander ambitions. He aims to persuade most of the seven big carriers, which between them carry 80% of the world’s international traffic, to allow him to install his software at their switches. That done, he plans to match a database that records each caller’s requirement—for especially high security, say, or an especially low price—with another set of data that will record from moment to moment where there is spare capacity on the world’s networks, and at what price. Add in the support of a financial intermediary to ensure that the capacity-provider gets paid, and the result will be an automated trading floor, aeons away from the laborious paraphernalia of accounting rates and settlement payments. A deluge of data Behind such dreams is not merely the spread of the Internet, but an explosive growth in demand for capacity to carry packet-switched data of all sorts. The figures are staggering. “The volume of our American Internet business is doubling every three to six months,” says Colin Williams, head of WorldCom’s international operations. “In Europe, we expect to see even faster growth, though from a smaller base. At present, total telecommunications traffic is roughly 80% voice to 20% data. That will be reversed in the next six years or so.” As companies learn to use Internet-type networks, their demand for capacity soars, mopping it up as fast as it becomes available. The sheer speed of this change creates new problems. While some of the big carriers seem to have lots of spare cable capacity (keeping it empty helps to prop up prices), newer cables fill up as soon as they are laid. Mr Williams remembers that he “could have bought cable factories in spring 1996 for practically nothing.” Now there is a serious shortage of cable capacity as new long-distance carriers scramble to lease what they can. WorldCom’s policy, of having as much capacity under construction at any one time as it has in use, owes more to the computer world than to the telephone business. Clearly within a decade the vast bulk of communications traffic will be data. Bob Collett, who helped to build Sprint’s Internet activities and is now head of data services at Teleglobe, Canada’s main international carrier, argues that the communications network will eventually be packet-switched, allowing all kinds of multimedia to be carried around the world. Nor will the growth stop there. “During the latter half of the next decade,” says MCI’s Mr Cerf, “there will be a new driver: billions of devices attached to the Internet.” As a result, the voice call that is now the mainstay of the telephone business may one day become a small, specialist activity, perhaps thrown in for nothing along with other services. But to get there from here is not straightforward. Before becoming connected to this brave new world, people first need to find a way around the main roadblock: that bastion of the telephone monopoly, the local loop.  © Copyright 1997 The Economist Newspaper Limited. All Rights Reserved