Summary

  • EARN’s August 1989 paper compared country contributions calculated from gross national product, sent traffic, and traffic in both directions. It described the numbers as illustrative and omitted server compensation.
  • At its October Board meeting, EARN approved its 1990 budget and country contributions, but the traffic-based proposal for 1991 had no universal support. The Board asked the Executive to return with a definitive model that included server-compensation figures.

The title of EARN’s August 1989 paper sounds like a decision: “Country contributions based on traffic.” Its status line says otherwise. The paper was a draft, and the Board of Directors later recorded that the proposed model for 1991 did not have universal approval. EARN’s October 1989 vote adopted a 1990 budget and country contributions; it did not settle the next year’s formula. The distinction matters because a spreadsheet that allocates costs can look like a rule before participants have agreed to use it.

EARN connected academic and research computing across Europe and linked the European network to BITNET in the United States. IBM had helped fund international circuits and technical staffing, with support planned through 1987 and a gradual withdrawal completed in spring 1990, according to EARN’s historical account. The Association already had a scheme for participating countries to share operating costs. By 1989, its officers were building a more explicit financial model as the network’s recurring costs and funding responsibilities changed.

An Executive paper dated 19 June 1989 described the model as a spreadsheet for tracking historical budgets and audited accounts, the current year’s budget and actual income and spending, and draft budgets for later years. It separated expenditure into volume-independent and volume-dependent parts. The contribution calculation also accounted for other income, prior-year surplus, and working capital. The proposal said that working capital would become a carried-forward surplus, building a contingency fund in a controlled way.

This was an attempt to make the budget legible and adjustable, not simply to divide one telephone bill by a traffic counter.

Paul Bryant’s 17 August paper then tested traffic as one basis for country contributions. It used January-to-May 1989 traffic records and compared projected 1990 and 1991 contributions under three approaches: gross national product, traffic sent, and traffic sent plus received. The paper said the traffic data contained anomalies but were accurate enough for further decisions. It also said the comparisons were illustrative and made no allowance for server compensation because relevant figures were unavailable.

The problem was not just whether a country sent more records. Network traffic could reflect the location of shared research facilities as well as the demand of a national community. Bryant singled out Switzerland: CERN’s traffic made its totals anomalously high. He noted other international facilities, including DESY, IIL, and RAL, as similar cases. A country-based meter could therefore charge the country hosting a major research hub for traffic generated by users and institutions beyond that country. At the same time, a gross-national-product measure answered a different question: capacity to pay rather than measured use.

The paper observed a poor correlation between the two.

The Board’s minutes from its meeting in Lisbon on 19–20 October 1989 preserve the boundary between a budget decision and a formula decision. Directors approved the proposed 1990 budget and country contributions by 13 votes to one, with two abstentions and one no-vote. For 1991, they noted that the traffic model was not universally approved and asked the Executive to prepare a definitive proposal for the next meeting, using the model with server-compensation figures added. The surviving minutes do not say which final formula was later adopted.

That is why the traffic meter was not yet the bill. A usable accounting model needed more than a reliable count. It had to decide what counted as network use, where a shared server’s contribution belonged, and how to treat a facility whose international role distorted a country total. The record documents the test and the disagreement, but not the eventual settlement. It would be inaccurate to present traffic-based charging as an adopted EARN rule on this evidence alone.

Sources

Primary records: EARN Executive, “Proposal to adopt an EARN financial model,” EXEC80 89, 19 June 1989; Paul Bryant, “Country contributions based on traffic,” EXEC113 89, 17 August 1989; minutes of EARN Board of Directors meeting 22, Lisbon, 19–20 October 1989. Context: EARN Association history and EARN network history. The archive says its collection is incomplete; the minutes leave the eventual 1991 contribution formula unresolved.