Summary

  • Autonomica’s 2006–07 laboratory test found expected behavior when test IDN top-level labels were added to a simulated root and queried through selected DNS software.
  • ICANN’s own account says the experiment was neither a live-root trial nor an end-user usability test; its result could inform deployment, not approve a name or settle who should choose it.

A top-level domain is only a short label, but adding one to the root changes what the public DNS can refer people toward. In 2006, ICANN’s test plan separated two questions that are easy to collapse: could DNS machinery carry internationalized labels through a root-like environment, and would applications and interfaces make those names usable to people? The first question went to a laboratory. The second was described as a later phase, not a result that the first test could claim.

Tina Dam was close to the program, not the author of the experiment. ICANN’s archive describes her as Senior Director for IDNs, responsible for developing and managing IDN projects including the country-code Fast Track. The 2007 announcement names her as the contact for the test; the report itself is by Lars-Johan Liman of Autonomica AB. When ICANN announced the Fast Track in 2009, Dam described years of drafts, tests and volunteer work. That is evidence of a sustained program role, not a claim that she personally designed or ran every test. (ICANN biography; 2007 test announcement; 2009 launch announcement)

The test was concrete. After engaging Autonomica in October 2006, ICANN published a draft plan for comment. Autonomica built a closed environment in Stockholm with two root name servers, a top-level-domain server, iterative resolvers and a query generator. The root zone was based on a copy of the active zone with test delegations added. The two root-server implementations were BIND 9.3.2 and NSD 2.3.5; the resolver mix included several BIND releases and Microsoft DNS from Windows 2000 and 2003. The team ran ordinary DNS queries through the path and looked for bad answers or unexpected delay. The report records expected results in the tested cases and no unexpected delays. (ICANN’s December 2006 test plan; Autonomica’s test report)

That result was useful because it reduced a specific uncertainty: the selected software handled the added delegations in the simulated setup. It was not a proof about every resolver, device, browser, language table or registration workflow. The report explicitly limited its coverage to the implementations it could test. ICANN’s March 2007 announcement draws the larger boundary: the test did not use a live root and did not include the end-user perspective. The earlier plan had described application and user-interface work as a separate phase. A clean server result therefore could not stand in for a person typing, seeing or following an internationalized name. (ICANN’s test announcement)

Nor could the lab choose a name. ICANN’s 2009 Fast Track required more than technical compatibility: its announcement referred to government and community support and a stability evaluation. The current process still separates String Evaluation, which considers a proposed label as a representation of an eligible country or territory, from String Delegation, the later request to operate an approved string and add it to the root. A test can inform one part of a decision chain without becoming the decision itself. (Fast Track request system)

The protocol layer adds another boundary. IDNA distinguishes a Unicode U-label from its ASCII-compatible A-label; the two are related forms of a label, not a guarantee that every user interface will display or accept it consistently. The later IDNA2008 documents specify registration and lookup behavior, while leaving some registrar-side processes outside their scope. Tina Dam’s public record also places her among ICANN staff supporting a multi-constituency revision of the IDN Guidelines in 2010. Taken together, these records show why “the root test passed” is an incomplete summary: technical compatibility, permitted strings, delegation, and actual user experience belong to different checks. (RFC 5890; RFC 5891; 2010 IDN Guidelines draft)

Sources