Summary

  • RIPE's Address Policy Working Group adopted a written chair selection procedure in April 2015. One co-chair would step down each year, other candidates could volunteer in advance, and the remaining co-chair would judge whether the group had reached consensus on the successor. A secret ballot was the fallback if consensus failed.
  • The first rotation selected Gert Doering, a chair since 2003. He volunteered again and the group reconfirmed him at RIPE 70 without comment. The procedure created an opening for a challenger without forcing a ballot in a case where the meeting recorded no contest.
  • Remote participants could join the consensus discussion. The ballot fallback and a no-confidence ballot were limited to people physically present. That difference is a concrete boundary on participation, even though no ballot was needed for Doering's reconfirmation.

The quickest resignation in Amsterdam

Gert Doering walked into the Address Policy Working Group session at RIPE 70 in May 2015 as one of its two chairs. The minutes show him welcoming the audience and introducing co-chair Sander Steffann. The next item recorded the group reconfirming Doering as co-chair without comment. He had stepped down under the group's freshly adopted rotation procedure, volunteered to continue, and returned to the same seat. His own verdict, recorded in the RIPE 70 minutes, was that stepping down and straight back up felt a bit ridiculous. He also said it mattered to give people in the community an opportunity to step up.

The episode invites a joke about an office revolving door with a very short corridor. The corridor is the point. A chair who simply carries on because everyone is used to seeing him there has a different claim to the job from a chair who announces a vacancy, invites others to stand and then continues with the group's assent. The person can remain. The route into office becomes visible.

That difference mattered especially for Doering. RIPE NCC's biography says he became an Address Policy Working Group chair in 2003. By 2015, he had held the role for roughly twelve years. Longevity can mean accumulated trust and a deep memory of arguments. It can also make a volunteer role appear hereditary, however well the volunteer does it. The 2015 procedure kept that experience in play and gave others a public, scheduled chance to ask for the seat.

The RIPE 70 record describes reconfirmation without comment. The minutes' sparse wording is useful here. The new selection path was used for the first time, and the incumbent stayed. The human comedy is already in the record.

The seat had an incumbent. The vacancy had to be real.

Working-group chairs often acquire authority by doing the work. They read the messages that everyone else skims, sit through the meetings, recognize the people who have been arguing for years and know where an objection first appeared. None of that experience automatically answers a different question: who invited them to keep doing it? The Address Policy Working Group’s new procedure answered with a recurring vacancy. Once a year, one chair would step down. The other would remain, so the group kept continuity while opening one chair to a new candidate. Rotation made the incumbent’s role contestable at a date everyone could see.

That arrangement is more interesting than a term limit dressed in gentler clothes. A term limit removes the familiar person when the clock expires. RIPE’s procedure let Doering run again. Its force came from giving others the same announced doorway. The distinction is small on a calendar and large in a room. If a long-serving chair is still the best volunteer, participants can say so. If another person wants the responsibility, that person no longer needs to challenge a chair’s character to ask for the seat. The vacancy exists because the procedure says it exists.

The March 2015 proposal required at least a month’s notice on the working-group mailing list before the RIPE meeting. Candidates could introduce themselves before the session. That month gave a potential challenger time to decide whether to volunteer, explain their experience and hear what the community expected from a chair. A candidacy sprung at the end of a conference agenda would have offered far less room for scrutiny. The list notice mattered because it created preparation time outside the hotel, where many working-group participants spent the meeting week without being physically present at all.

The first use followed that logic. After the procedure’s last call closed, the April archive announced that Doering would be the chair to step down. He had already put his name forward again. Other candidates were invited to do likewise. By the time the meeting reached its chair-selection item, the group knew both the incumbent’s intention and its opportunity to choose someone else. The minutes then recorded no comment on his reconfirmation. That is a thin sentence, but the prior notice makes it meaningful. Silence at a surprise appointment says little.

Silence after an open call, a named candidate and a scheduled session tells a clearer story about this particular choice.

The other chair’s role was equally deliberate. Steffann remained in office while Doering’s seat opened. Under the written procedure, the remaining chair would judge whether the group had reached consensus on the new chair. If the group had no chair at all, the RIPE Chair would make that judgment. There was always a named person responsible for declaring the result. This did not remove the interpretive character of consensus; it located that interpretation in an identifiable office. If people later disputed a selection, they would know which decision to inspect and who had made the call.

The chair who judges the room

The chair of an address policy working group performs more than meeting housekeeping. Proposals about Internet number resources draw support, objections, amendments and people who have strong operational reasons to care. Someone has to assess whether the discussion has reached the level of support required to move forward. In the RIPE process, the chair's interpretation of rough consensus is a practical control point. It shapes when a proposal can progress and how objections are handled. A microphone queue is only the visible part of that job.

Doering explained the other part at RIPE 65 in 2012. Speaking during a policy discussion, he reminded attendees that decisions were not made at the meeting. The co-chairs listened to the room to understand which direction a proposal might take; the discussion had to reach the mailing list, where the consensus work happened. The distinction put an important brake on the excitement of a conference session. People who could not travel still had a channel, and a lively room could not be treated as the whole working group.

At RIPE 67 in 2013, the practical weight of that judgment was on display. During a separate debate over a policy proposal, Doering told the group that the current text had attracted substantial mailing-list support. Some participants still objected; the chairs found no way to absorb those objections while keeping the proposal's core. He said they would apply rough consensus and move ahead. That policy judgment shows why selection of the judge deserved attention: the job includes deciding when broad support can carry a proposal past remaining disagreement.

Later in the same RIPE 67 session, the discussion turned to how chairs themselves were chosen. Steffann asked whether the working group wanted a procedure for appointment and replacement. A suggestion of three-year terms and elections had been circulating. Others worried about creating unnecessary bureaucracy. The room was debating a governance choice with a very ordinary human shape: it valued experienced chairs, yet wanted a way to test whether the confidence placed in them still held.

Jan Zorz, according to the minutes, linked periodic re-election to a chair's credibility when making a difficult call. Lu Heng said he was happy with the chairs and supported elections, in part because a route into the position could draw newcomers into the policy process. Hans Petter Holen suggested overlapping terms so both chairs would not be replaced at once. Steffann wanted a process simple enough to avoid regular elections with voting slips. Those contributions differed on mechanics. They shared a recognition that a chair's standing depends on a route through which the group can renew or replace that person.

The minutes record an argument for rules while the participants still valued their chairs. They wanted a way to renew trust before a bad situation forced the question. A chair who judges everyone else's consensus needs a visible route into the chair himself. The 2013 exchange made that abstract problem sound like a practical meeting agenda.

A rough-consensus judge takes the witness stand

The 2013 conversation had bite because Doering had just shown what a chair’s judgment can mean. During the policy discussion at RIPE 67, he explained that a proposal had attracted broad support on the list despite continuing objections. The chairs could not find a change that would satisfy those objections while preserving the proposal’s central design. He said they would proceed by rough consensus. The phrase does not mean that every hand is raised or every critic falls silent. It requires a chair to decide whether the objections are sufficient to stop a proposal that otherwise has substantial support.

That call can affect how address policy develops across the RIPE region.

The participants discussing chair selection were therefore asking who should hold a demanding kind of discretion. Zorz’s point about periodic re-election and credibility went straight to it: when a chair later makes a contested call, people need a reason to accept the chair’s standing even if they dislike the outcome. Lu Heng’s case for elections added an entry problem. If the same capable people remain indefinitely by familiarity, newcomers may struggle to see a path into the work.

Holen’s staggered-term idea addressed a different risk: replacing both chairs together could discard useful memory at the very moment a group needs someone to keep its process moving. Steffann’s preference for simplicity recognized that volunteers can bury a working group under elaborate electoral machinery.

These were not four versions of the same complaint. They concerned legitimacy, access, continuity and administrative cost. The annual alternating step-down became a practical bargain among them. It offered a scheduled test of legitimacy without automatic exile, an advertised opening without constant campaigns, and continuity through the other chair. The trade still left hard questions about who can participate when consensus fails, but it converted an inherited chairmanship into a role the group could periodically revisit.

The mailing list is central to that bargain. At RIPE 65, Doering had reminded participants that the policy decision would not be made from the mood of one conference room. The meeting let chairs hear questions and assess the direction of debate; the list remained the place for the wider policy process. At RIPE 67, he relied on that list support in a difficult policy call. For chair selection, the group likewise used the list to announce vacancies and candidates, then used the meeting session to decide who would take the office. The participation channels were related, but the written rules assigned them different jobs.

Treating “the community” as a single audience would hide those assignments.

This is why the story cannot be told as a joke about Doering personally leaving and coming back. It is about the conditions under which the person who says “the group has reached rough consensus” gets to keep that authority. Doering was experienced enough to make the question awkward: the room could value his judgment and still insist on a route for somebody else. The published minutes show that combination more clearly than a dramatic leadership fight would have. They show people designing accountability while the incumbent remained welcome.

The year the joke became a rule

The issue was still on the agenda at RIPE 69 in November 2014. Doering addressed a rumor that he was about to leave the chair and said he was staying. He then gave Steffann the floor for the selection discussion. Randy Bush asked the group to stop making the matter too complicated and get on with it. Steffann said he would prepare new wording for the mailing list. Doering had expected more comment from the room because the list had already seen discussion. The minutes show a problem working its way from an awkward question into text.

The version Steffann circulated on 21 March 2015 was concrete. The group would aim for two chairs. Once a year, one would step down, and the chairs would alternate. The impending vacancy would be announced on the working group mailing list at least a month before a RIPE meeting. Anyone could volunteer, including a former chair, by introducing themselves on that list before the meeting. That is a surprisingly useful safeguard for a humble administrative detail. A seat cannot be fairly challenged if potential candidates learn it is vacant only when the session chair reaches that line of the agenda.

The selection itself would happen at the working group session. People attending in person or remotely could take part in determining a consensus on the new chair. The remaining co-chair would decide whether that consensus had been reached. If the group had no chair left, the RIPE Chair would perform that judgment. The design split the work of standing for the seat, participating in the discussion and judging the result among identifiable people. It also gave the process a way to function if both seats were empty.

There was an explicit fallback. If consensus failed, a secret ballot would be held at the session. Everyone physically present could vote. RIPE NCC staff would count votes under a specified proportional method. The rule thus treated an online participant as part of the consensus discussion but excluded that participant from a ballot. That is a real access line. It would matter most precisely when the group could not agree by the method open to remote participants. The record does not show this fallback being used in Doering's 2015 reconfirmation; it shows the line drawn in advance.

The draft also supplied a no-confidence route. A significant dissatisfied group could request a ballot on the mailing list at least a week before a RIPE meeting. A simple majority of those physically present at the session would remove the chair immediately. The text offered an exit as well as an entrance. It did not leave dissatisfied participants to hope that the annual rotation would arrive soon enough.

The meeting room had a second door. It was narrower.

The secret-ballot fallback deserves attention precisely because it was never used in this first case. A remote participant could join the consensus discussion about a chair. If the group failed to find consensus and moved to a ballot at the session, that person could not vote unless physically present. The published procedure states the distinction without softening it. RIPE NCC staff would count the votes under a single transferable vote method. The arrangement made the final contested choice a smaller electorate than the earlier deliberation.

There are practical reasons people write meeting ballots this way. A room ballot has an obvious venue, a count and staff who can supervise it. But the access consequence is equally plain. The person able to debate a candidate from afar loses that final ballot if consensus breaks down. The asymmetry becomes most consequential at the exact moment the group is divided. In an uncontested reconfirmation, as with Doering in 2015, it stays in the appendix of the story. In a contested choice, it could decide whose preference is counted.

The rule announced that boundary in advance, so a participant could assess the process without waiting for a disputed election.

The no-confidence path used the same physical boundary for a different purpose. A significant number of dissatisfied participants could request a vote on the list at least a week before a RIPE meeting, after discussion had failed to resolve their concern. The meeting would then hold a secret ballot. A simple majority of those physically present could remove the chair immediately. The annual step-down offered an ordinary opportunity to replace a chair; no confidence supplied a faster route when trust had collapsed. One chooses a successor for an opening. The other removes an office holder. Their thresholds and timing reflect the difference.

The draft also considered the unpleasant edge case of a working group with no chair. It allowed the group to choose an interim chair at a session, with candidates nominated there, and required that interim holder to step down at the next session while allowing them to stand again. This little contingency shows the authors thinking beyond the picturesque first rotation. A working group cannot suspend policy discussion because its officers vanish. It needs a temporary keeper of the procedure and a prompt path back to an ordinary selection.

The RIPE Chair’s substitute consensus role covered the same kind of gap from another angle: when no remaining working-group chair can judge, another identified officer must do it.

The detailed fallback machinery also exposes the main design choice. Consensus was the preferred method, including voices from the room and from a distance. Ballots were instruments for breakdown, constrained by who was at the meeting. The rules did not promise that all participants would have identical power in every phase. They specified where power changed hands. For an Internet community that prizes open participation, that specificity matters. A remote contributor can read the text and know exactly when their access to the decision stops. The door was open through discussion; the last corridor had a physical turnstile.

Those are narrow procedural powers. RIPE NCC staff would count a ballot if one occurred. The RIPE Chair's substitute role applied if the working group had no chair to judge consensus. The working group chose its own chairs within the procedure. Each role identifies who was responsible for a particular step.

April's paperwork, May's anticlimax

The April 2015 mailing-list archive records the final turn. On 7 April, after a last-call period with no further comments, the chairs declared consensus for the chair-selection process. The rule had gone through the very list whose role Doering had defended in policy work. Adoption of the selection procedure was a separate act from the later decision to keep Doering in office.

The chairs then had to decide whose term would be tested first. Steffann announced that, after what he jokingly described as long deliberation by random-number generator, Doering would step down at RIPE 70 in Amsterdam. Doering had already volunteered to stand again. The same message invited anyone else interested to put a name forward on the mailing list before the meeting. This was the moment the procedure acquired a person, a date and a genuine opening. An incumbent's renewed candidacy was allowed. So was another person's candidacy.

At the May meeting, the group reconfirmed Doering without comment. The minutes record an orderly first use of the rule: the working group opened the seat and kept the chair it wanted. For a volunteer institution, that is an advantage. A scheduled vacancy can test confidence without manufacturing a crisis.

The chronology matters. In 2013, the working group argued over whether chair selection needed a clearer method. In 2014, Doering was still fielding rumours about his own departure while Steffann promised another draft. In March 2015, the group had text precise enough to specify notice periods, remote participation, vote counting and removal. April brought the last-call closure, consensus on the procedure and the identity of the first chair to rotate. May supplied the first use.

A retrospective that jumps straight from “new rule” to “same chair” misses two years in which volunteers converted an uncomfortable question into something a meeting secretary could administer.

It is also why the random-number-generator joke works as more than decoration. Choosing whose turn came first could have looked like a judgment on one of the two chairs. The April announcement treated it as a starting-order problem and kept the focus on opening a seat under the same rule for both. Doering’s candidacy made the first use especially clean: the group could see that stepping down was a procedural event, while reappointment remained an available outcome. If another candidate had appeared, the procedure had a route for that person too. The minutes record the path that actually occurred: reconfirmation without comment.

The process gave working-group participants a defined means to fill a working-group office. The group held a chair-selection discussion at a RIPE meeting and made one role challengeable through a known process. Its precision is the value of the 2015 change.

What changed when the chair stayed

Before the rotation rule, a person watching from outside could see familiar chairs at familiar meetings and perhaps infer that the group was content. After it, the evidence of continued trust had a recurring public form. The vacancy had to be announced. Candidates had a way to appear. The meeting had a way to test consensus. If that failed, the text said exactly how the ballot would proceed. Dissatisfaction severe enough to trigger no confidence also had a route with notice and a stated threshold.

The rule's own design shows the limits of an open room. Remote participants could help form consensus, yet only people at the venue could vote in the fallback. A person excluded from that ballot could still argue on the mailing list before a meeting and join the consensus discussion remotely. Those channels did not erase the difference in voting access. Anyone assessing the group's procedure should keep both facts in view, especially when a contested chair choice is imagined. The 2015 reconfirmation itself reached no such contest.

There is another question the text answers elegantly: who judges consensus on the consensus judge? The co-chair who is not stepping down does. That choice avoids asking a candidate to certify their own return. It also puts a great deal of interpretive responsibility on the remaining chair. The rule identifies the responsible person, which is the first condition for challenging a decision if the group thinks the judgment is wrong. The alternative, an unspoken continuation of office, offers no comparable record of where the call was made.

He had spent years in the chair, knew why judging rough consensus could draw disagreement, helped bring the selection question before the group and then let the first annual test fall on his own seat. The 2013 minutes show the demand for credibility coming from several voices. The 2014 minutes show the wording still being argued through. The March text and April archive show the resulting procedure. The May minutes show its unglamorous first result: the same person, now through an announced opening.

That is why his little joke still works. A revolving door is silly if it moves a person out and back for ceremony alone. Here the short trip created a space in which someone else could have entered. The chair stayed. The door acquired hinges.