Summary

  • CyberTwiga reported a 9.6 kbps IP connection over X.25 via London, operating from 25 July 1996; a separate 64 kbps University of Dar es Salaam link was still being tested.
  • An NSRC traceroute reached twiga.com on 26 August. That proves a route to one host at one moment, not nationwide access or continuous service.

On 27 August 1996, Bill Sangiwa sent a short operational update into the Network Startup Resource Center’s archive. CyberTwiga, he wrote, had a working 9.6 kbps IP connection over an X.25 link via London. It had been operational since 25 July. The company described itself as Tanzania’s first commercial IP service provider. In the same message, Sangiwa added a useful distinction: the University of Dar es Salaam was then testing a separate 64 kbps link.

The numbers can tempt a neat origin story: one country, one first connection, one speed. The documents describe something more specific. There was a working path for CyberTwiga, a different university link still under test, and an operator reporting both to an international networking community. “First” here belongs to a dated claim about an international IP link, not to every earlier form of electronic communication in Tanzania. Email had circulated through store-and-forward systems years before a live IP route.

A day before Sangiwa’s note, NSRC staff published a traceroute to twiga.com. The recorded path left a US network, crossed several US providers, reached the UK academic network and continued to the destination address. It is an unusually concrete artifact: a packet path, hop by hop, ending at a Tanzanian company’s host. But a traceroute is a narrow witness. It shows that this probe reached that host on 26 August; it does not reveal what ordinary customers could retrieve, how often the link stayed up, or how much traffic it could sustain.

A later account by Michael Jensen, CyberTwiga’s co-founder, describes the commercial service as dial-up email carried over a 9.6 kbps SITA link and recalls a monthly cost above US$4,000. That recollection does not line up neatly with the contemporary message’s description of X.25. The sources may describe different layers or use different shorthand, but the available record does not settle the relationship. The careful conclusion is not to choose the more vivid version; it is to keep both attributed. The contemporary message is strongest for what the operator reported in August 1996.

Jensen’s later account adds a participant’s view of service limits and cost.

The distinction matters because international reach was not the same thing as a national network. A connection could make global email or other IP services possible for a small set of users while leaving access dependent on local telephone lines, dial-up points, institutional networks, and the price of international capacity. The university’s 64 kbps test suggests a separate path being explored; an NSRC provider record updated in 1998 later describes a full 64 kbps IP link through London. That later record should not be backdated to the earlier test.

Tanzania’s 1996 milestone is therefore best read at the scale the evidence supports: an operational international IP route, a company willing to provide a service over it, a university testing another link, and a trace that made one destination visible from abroad. The route was real. Its existence did not itself authorize a claim that the whole country had become connected.

Sources