Summary

  • REUNA and Unired reached the international Internet through separate 56 kbps links in 1992, but a 1993 paper says the networks had no local connection.
  • The same account links the split to cost and control disputes: per-megabyte billing proved difficult to meter, and traffic between nearby Santiago computers travelled through the United States and back.

A computer in Santiago could be only a few kilometres from another and still send its traffic abroad before it came home. That is the sharpest detail in a 1993 paper about Chile’s first Internet years. The authors, Ricardo Baeza-Yates, José M. Piquer and Patricio V. Poblete, described two independently operated 56 kbps international links and no local connection between their networks. They said that traffic between nearby machines therefore crossed the United States, with delays that made interactive communication almost unusable. It was an account by participants, not a packet trace; its precision matters as much as its limits. The contemporaneous paper

The dates describe different milestones. REUNA’s institutional timeline says the project began in 1990 with support from CONICYT and CRUCH universities, and that the REUNA corporation was formed in 1994. CONICYT dates REUNA’s Internet connection to 2 January 1992 and says it funded the first year. The 1993 paper, written from inside the network’s early operation, says Chile had two international links from January and that both REUNA and Unired appeared in MERIT’s routing tables in February, after a delay the authors attributed to an NSFNET “one-country-one-link” policy. A project, one network’s first connection, two international paths and a later corporation are not one “first Internet” event. REUNA’s timeline · CONICYT’s infrastructure history

REUNA was planned as a national TCP/IP backbone for universities. But the participating institutions could not agree on how the network should be controlled or how operating costs should be shared. Some left to build a separate link, first as Internet-Chile and later as Unired. The split was not simply a dispute about which cable or satellite to buy. The international link was expensive—the paper estimates about US$7,000 a month during planning—and each university was wary of paying for traffic generated by someone else.

The authors say government funding would cover initial investment, while institutions and any future customer revenue would have to sustain operations.

The proposed bill made that anxiety concrete. REUNA’s member institutions would share most costs in proportion to their budgets, while incoming and outgoing international traffic would be charged per megabyte. Users objected when they saw what email and FTP might cost. The campaign pushed the policy back into discussion, but the technical equipment could not produce the detailed accounting the rule required. REUNA moved to agreed traffic brackets, with separate day and night thresholds.

The authors report that Unired took a different path after user criticism: it dropped international-traffic charges for academic and nonprofit customers, while continuing to charge commercial clients.

Those fee changes did not create a local route between the two networks. In the paper’s May 1993 account, they still had no direct local connection. REUNA had earlier barred its subscribers from connecting to Unired; it later relaxed that rule to allow third-party connections, but added a charge for each one. The authors argued that the networks’ separate and complicated international-cost schemes made a direct connection difficult to negotiate. That is their explanation, not proof that a tariff alone caused the split.

What they do document is the operational consequence: nearby traffic went through the United States, and the REUNA link was congested.

There was one revealing exception to the rivalry. The same paper says the primary .CL name server stayed at the University of Chile’s computer-science department, with secondary servers administered by REUNA and Unired. The department described itself as a neutral technical intermediary, even though it belonged to the University of Chile, one of REUNA’s founders. A shared naming service could keep working across the divide without making the networks themselves locally reachable. The distinction is easy to miss: a common name space did not amount to a common route.

The record is therefore more useful than a single “first connection” date. CONICYT’s account identifies REUNA’s 2 January link; the contemporaneous researchers describe the arrival of two separate international paths and the friction of running them. Neither establishes that all Chilean universities, companies or households were online in 1992. In the live portion of REUNA that the paper describes, three universities, one research institute and CONICYT were connected; the authors say the network’s broader national deployment was still ahead.

Chile’s early network problem was not a shortage of international links alone. It was the absence of an agreed way to connect and account for traffic between domestic networks. A price rule designed to make scarce capacity fair could not be separated from the equipment’s ability to measure use, the institutions’ right to set access conditions, or the route users actually received. The packet that went abroad and back made that institutional boundary visible.