Summary

  • At RIPE 93, Remco van Mook said the engaged core of the Address Policy Working Group had fallen sharply and referred to “24 people right now.” His measure excludes RIPE NCC staff and working-group chairs and counts people posting at least five times in a year.
  • He proposed a references-only policy index as the first move. The room drew a firm line between collecting existing rules and changing their meaning; no policy manual was adopted at the meeting.

Twenty-four people can fill a dining room. They can also become the people left arguing over the rules that govern number resources across a much larger region. At RIPE 93, Remco van Mook gave the Address Policy Working Group that number while pitching a cleaner policy manual. He had just described his count of the engaged mailing-list core: people with at least five posts in a calendar year, excluding RIPE NCC staff and working-group chairs. His account put that core at about 60 in 2006, around 110 at its 2015 high, and roughly a quarter of that peak today. The published talk abstract says the population sustaining consensus has shrunk by a factor of three over a decade. These are the speaker’s counts. The underlying year-by-year table is absent from the published transcript.

The smaller room matters because the rulebook is already scattered. RIPE’s policy page lists current documents by subject. Van Mook said separate texts leave definitions and corner cases that the community has not resolved. RIPE NCC staff then have to apply those texts to live cases. His joke was that the registry reads tea leaves. The joke lands because an applicant receives an operational answer, not a seminar on why the wording is uncertain.

His proposed first step is deliberately plain: an index with chapters for IPv4, IPv6, AS numbers, transfers, special-purpose distribution, contractual relations and adjacent policies. It would point to the existing numbered documents. It would contain zero new policy. A reader seeking the IPv4 rule would find the reference to RIPE 826, then read the actual rule there. Later stages could move text verbatim, chapter by chapter. Altering definitions would require its own policy discussion.

That sequence exposed a real fight in the room. One participant warned that even a “clarification” can disturb settled interpretation; a cleaner sentence may silently move the boundary for a holder. The Database Working Group insisted that collation and correction be separate steps. Others welcomed a map that would let a newcomer find the rules before learning the oral history of every exception. The working-group chair reminded the room that chairs can help turn an imperfect idea into a proposal. A process that requires a perfect first draft from newcomers quietly selects for people who already know the kitchen.

Van Mook gave two examples of cleanup moving at different speeds. A misspelled database status could be corrected once someone pointed to it. An ambiguous policy status needed the proposal process, participants and consensus. The gap between those routes is where RIPE NCC’s practical discretion grows. The registry must answer a request while the community is still deciding what it once meant.

The proposed index would make that gap visible. It would not settle a dispute, and van Mook did not ask the room to approve new policy on 29 October. The immediate test is concrete: publish a references-only map, show the source document and version behind every chapter, and mark conflicting definitions without quietly resolving them. Then put each proposed fix through the process that gives it authority. Twenty-four regulars can start that work. They should not have to be the only people able to understand the result.