But it is presently showing signs of strain, as millions of new users pour in from outside the academic computing and science communities. If you try to retrieve files from the archives at the Massachusetts Institute of Technology any time when anyone's awake in the USA, you'll probably get the message "there are too many people connected from your region." (Please do not try this now.) So how should MIT fund increased capacity, or how should London's Imperial College or the University of Cologne pay for a full-service regional archive?

Any mention of charging for services produces a welter of panicked messages. Often, in this febrile infosphere of instant response, people confuse charges for services with charges for communications; the more thoughtful see the latter as the biggest threat. It is precisely because use of the net bears a fixed cost, many say, that it has been able to develop so rapidly and imaginatively. Equity of access is also a worry: if the net is as powerful as its proponents believe, what happens if a global underclass is excluded?

"We believe in the pressing need for global democracy, not a global supermarket," declared the February 94 New Delhi Symposium on New Technologies. The net "represents the most important new effort to expand democracy into a wide range of human endeavors," the Washington, DC-based Taxpayer Asset Project says in its call to resist metered use of the US infobahn. Last week [ca. 20/Jun/94] Senator Daniel Inouye introduced a Bill to the US Congress, supported by the lobbying group Computer Professionals for Social Responsibility, calling for a "public lane" on the infobahn. It would reserve up to 20% of the capacity of new advanced telecommunications networks, such as 500-TV-channel cable into homes, for "non-commercial educational and informational services and civic discourse."

The TAP says that usage pricing could extinguish list-servers, which work like 18th-century corresponding societies but rather faster, re-distributing messages to hundreds or thousands of subscribers. The New Zealand net has begun charging for outgoing international traffic; the problem is that on the net it's hard to identify an "originator" for a transaction. You can send a very short mail message requesting a very large file from a server in New Zealand. File-server operators there at present merely post messages asking for discretion.

The US National Science Foundation is on the point of signing contracts to provide four infobahn "Network Access Points". At least two of the contractors (Pac Bell and Sprint) are phone companies. As regular net-commentator Carl Kadie says, "I think the biggest threat to unmetered Internet access can be expressed in one phrase: 'Free International Long-Distance Telephone'. It is now possible with software to have an almost real-time voice conversation over the Internet. This has the potential of putting a huge amount of new traffic on the net and of competing with the phone companies that own much of the net."

Here we enter the twilight world of the economics of telecommunications. Almost all costs are fixed. Once cable or fibre is laid, switches are installed and support staff are hired, the usage-sensitive costs are minuscule. Fixed-cost access makes sense.

Richard Cox, principal telecomms consultant for Mandarin Technology in South Glamorgan, agrees that it's largely a matter of social policy that I pay some [UKpound]1500 a year for telephone service while my grandmother, who's never made an international call and won't speak long-distance for more than a minute, pays a tenth of that. We cost the world's phone companies about the same. He estimates that around 20% of the cost of telephone service is the cost of keeping records, issuing bills and collecting payment. (BT say the proportion is "commercially confidential".)

A subscriber to the British CIX bulletin board system, who prefers to remain anonymous, recalls a 1981 meeting with a manager of the old Prestel on-line service, who "told us amazed information providers that the average Prestel user bill at that time was [UKpound]1.50 per month, while the cost of sending out the bill was [UKpound]8.00. When it was pointed out that the best thing they could do was probably to stop sending the bills, he just didn't seem to understand."

Things get even more complicated on the net. One reason it's so cheap to operate is that it's "connectionless": when you log your computer on to a machine in Nevada, the end-to-end link is illusory. Information travels in "packets", which effectively find their own way to their destination, possibly by several different routes in the course of a minute. So record-keeping of heroic proportions would be required to issue bills.

So how can use be rationed? Economists Jeffrey MacKie-Mason and Hal Varian at the University of Michigan propose a "congestion pricing" scheme. When your computer sent off a packet, it'd attach a note of how much you were prepared to pay for it to arrive promptly -- say 0.002 US cent for 200 bytes. Automatic real-time "auctions" at switching centres would determine the lowest "bid" at which packets would be passed; every packet which got through would pay that lowest price, and others would be delayed. When there was no congestion, all packets would travel free.

The neat thing about this is that it's the commercial services which Murdoch and friends want to put on the net -- interactive TV and so on -- which demand real-time communication. Mailing lists and text discussions can tolerate delays. In this sense, the "netizens" really would become hitch-hikers on the infobahn, squeezing into the gaps in the coming flood of commerce.

However, as Steve Wolff, Director of the NSF Division of Networking and Communications, points out: "Congestion pricing is technically challenging... two or three orders of magnitude [i.e. 100 to 1000 times] more work than switches do at present." While TAP's concerns "are valid," he says, "their fears are at present quite groundless... Since it's clear that users want flat-rate pricing -- and I should make it clear that I like it too -- any [company] that didn't offer it would go out of business."

And author Bruce Sterling comments pithily that "You want an example of a communication system that doesn't charge for transport? The English language. Think of the Internet as a language rather than a machine and most of your questions become irrelevant. I think the Internet is tougher than you give it credit for. From now on the struggle will not be over mechanical control of the means of information, but over spin- control of the zeitgeist."

home

Written:
June 26 1994

An edited and doubtless thus improved version of this article appeared in the Guardian OnLine section.

This version is © copyright 1996 Mike Holderness; moral rights are asserted.