Summary

  • UNZANET built an email community before Zambia had a direct IP circuit: scheduled store-and-forward messages connected the University of Zambia, NGOs and health institutions.
  • ZAMNET converted that demand into a service plan, but its first year depended on World Bank support, a costly leased line and subscriptions expected to carry salaries and future operations.
  • The November 1994 link did not settle the access question. Public fee-paying service began in March 1995, and within months its line and server were already under pressure.

A network could arrive by the clock

The first recorded UNZANET message left the University of Zambia Computer Centre on 30 September 1991, addressed to Baobab, an African-interest network in Washington. The campus hub existed because an IDRC-funded regional project supplied a microcomputer and modem. International messages first moved through the Association for Progressive Communications network in London. In November, Rhodes University began a more direct route: computers called one another three times a day, carrying mail toward the Internet and back. The South African Universities Network paid those calls.

This was not interactive Internet access. A person wrote a message, a local system stored it, and a scheduled call moved a batch onward. The wait was part of the service. So was the network of people who could send something at all. The University’s 1996 account says the system grew from academic use into an email network that served NGOs, health institutions and development organizations. Its January 1995 update put the footprint above 270 email points. These are operator-reported counts, not a census of Zambian Internet users.

That distinction matters. An email point could make a useful connection without giving its user a live route to the Web or a remote database. Store-and-forward mail traded immediacy for a cheaper operating arrangement. In a country where an international telephone call was expensive, a message carried through shared infrastructure could serve organizations that could not justify their own direct links.

Demand became a financing question

By the middle of 1993, the Computer Centre’s team wanted a direct link. Its first proposal to outside donors found no direct support. At the start of 1994, the University chose a different structure: a private campus company, ZAMNET Communication Systems Ltd., would build the connection and sell Internet access. The World Bank then agreed to fund 80 percent of the new company’s first-year operating costs. The University remained the controlling shareholder.

The proposed budget was US$150,000. It covered the leased circuit and upstream subscription, service and training equipment, technical advice, and three staff positions. The plan expected customers to pay the salaries of one administrator and two technical staff from the start. The intended public service also mixed different kinds of users. Organizations that needed higher-volume access could pay for leased lines; city users with suitable telephone lines could dial in interactively; and users in rural areas could continue with Fidonet or perhaps UUCP where telephone quality made live sessions difficult.

The arrangement was not a simple move from “free” to “paid.” The old mail system had made a broad, noncommercial user base possible; the new company was meant to turn some of that demand into recurring revenue. Robinson’s own account says commercial customers could partly subsidize NGOs and health institutions because the Computer Centre had reason to pool staff and equipment instead of running separate systems. That is an intended cross-subsidy, not an audited balance sheet. The plan set a target of self-sufficiency after the World Bank’s first-year support ended. The sources here do not prove that the target was met.

The long circuit ended at a short, difficult line

In April 1994, ZAMNET applied for a four-wire, 9,600-baud leased circuit between Lusaka and its chosen provider in Cape Town. The Cape Town company agreed to carry Internet traffic without charge for the time being. That concession did not remove the cost of the leased circuit inside Zambia and across the international route; an operator update in January 1995 put the annual line charge at about US$60,000.

The planned July commissioning slipped. Technical trouble on the satellite section delayed testing until October, and the team discovered that a modem chosen from its Fidonet experience handled a two-wire connection, while the leased circuit was four-wire. Two more suitable modems cost US$1,100 each. Then the signal had to cross the final kilometres from Lusaka’s exchange to the University. Robinson’s retrospective says loss on that stretch exceeded the loss over the much longer route to Cape Town. On 22 November, the line finally carried a signal to and from the Internet.

A message posted that day from South Africa recorded several-second ping times to ZAMNET’s router; it captures one route and moment, not a national performance measure.

Even that date needs care. The 1996 chapter describes full Internet access in December 1994, while the November archive records the leased line up and the router responding. In January 1995, Robinson said interactive service was still limited to the University Computer Centre. The international mail for more than 270 points was now gated locally and carried over the leased circuit. A working IP link, an institutional service and a public commercial offer were different milestones.

Opening the service made its capacity visible

ZAMNET opened to fee-paying customers at the beginning of March 1995. By June, Robinson reported nearly 200 accounts and more than 300 mailboxes. He also reported a line still described as 9,600 baud but running synchronously at 14,400 between Lusaka and Internet Africa in Cape Town. About 100 Fidonet users remained; their host exchanged mail with the Internet server five times a day. The new service had not made the older mode irrelevant.

The same June update reported that traffic was growing faster than expected. The link was congested at some times, the mail server lacked memory and disk capacity, and plans to upgrade both were being moved forward. Most accounts still used email. These figures show early take-up and a capacity problem at once. They do not establish national reach, profitability or lasting financial independence.

The account’s author helped build and operate ZAMNET, so his explanation of demand and commercial design is valuable but interested. The contemporaneous January and June updates help pin down what had actually been launched; a November ping provides an independent network-side timestamp; and a World Bank retrospective repeats the earlier explanation that subscriptions and shared users were meant to make the arrangement viable. None supplies audited post-grant accounts.

The most defensible conclusion is narrower: a shared mail network offered a plausible starting customer base, while the cost and reliability of an interactive Internet service still depended on who financed the line, who could pay, and whether the local operating system could grow with demand.

Sources

The interpretation uses Heng Lu’s Note 64 on voluntary adoption and Note 65 on running-code primacy as editorial lenses; neither is historical evidence about Zambia: https://heng.lu/minimum-initial-specification-localized-future-decision-voluntary-adoption-internet-coordination-system/ and https://heng.lu/running-code-primary-the-patch-needed-to-preserve-the-internet-original-design/.