Summary

  • RFC 1396 recounts the 1992 POISED working group's response to an IAB/IESG process crisis. The November compromise kept the existing bodies but put standards-track processing, including final advancement, with the IESG.
  • The IETF had no formal membership electorate for a direct leadership ballot. The proposal used seven eligible community volunteers on a nominating committee, selected by lot if more than seven volunteered; IAB and ISOC retained separate ratification roles. RFC 1601 later put an annual version of that committee into the 1994 IAB charter.

An architecture argument exposed an appointment problem

In 1991, Internet engineers were worried about address exhaustion and routing tables that kept growing. The ROAD working group recommended a short-term course but left the long-term plan unsettled. The IESG sent further recommendations to the IAB. After the June 1992 INET meeting in Kobe, the IAB raised additional ideas, including elements of CLNP. Many readers heard a decision. IAB participants, in Steve Crocker's account, thought they were offering advice and widening the discussion.

That gap mattered because the IAB and IESG were no longer a small circle whose intentions everyone could read from the room. Working groups, area directorates and the steering group had grown into layers. By the July IETF meeting in Cambridge, the protocol choice remained open, while the organization's authority had become the sharper dispute. Who made a standards decision? Who selected the people with that power? Crocker's January 1993 RFC 1396 is a participant's account of how those questions became a working group called POISED.

The immediate reform proposal was bold. Crocker and Carl Malamud suggested replacing both IAB and IESG with two boards, one for technical management and one for process oversight. At the November 1992 IETF meeting in Washington, members of the existing bodies offered changes that preserved their structure. The decisive concession put all standards-track processing, including the final decision to advance a specification, in IESG hands. Crocker reported that the IAB soon cleared its queue of standards actions and the IESG began handling them alone. The question of who sat on these bodies then became impossible to dodge.

A vote needed voters

A formal election would have made selection look straightforward. It would also have required a rule saying who could vote. The IETF had no formal membership requirement and made its technical decisions through consensus rather than ballots. Inventing an electorate in the middle of an appointment dispute would have been its own constitutional decision. POISED chose a nominating committee instead.

Seven community volunteers would hold its voting seats. If more than seven volunteered, the seats would be assigned by lot. To volunteer, a person had to have attended two IETF meetings under the 1992 proposal. Representatives from the IAB and IESG and a chair supplied by ISOC would participate without votes. The seven did not become Internet governors by chance; they selected nominees for posts in the IAB and IESG. IESG nominees went to the IAB for ratification, and IAB nominees went to ISOC. The mechanism put randomness at the gate to the nomination process, with existing institutions still holding the final appointment checks.

There was another gate beneath the lottery. Two meetings were a familiarity test, but travel and attendance also determined who could enter the volunteer pool. RFC 1396 describes the intent: committee members should know the community and standards work. The rule converted participation in meetings into eligibility for choosing leaders. It was a practical answer to a voter-list problem and a visible limit on who could take part in the draw.

POISED sketched staggered two-year terms and a recall route as well. Complaints would first reach an ISOC ombudsman after other oversight avenues were exhausted. Following a cooling-off period, a volunteer recall committee selected at random could remove a member by a two-thirds vote. The process aimed to give the community a way to challenge incumbents without turning every technical disagreement into a popularity contest.

The trustees kept a hand on the brake

At the December 1992 ISOC trustees' meeting, the group sought temporary approval for selections while a new charter was written. The trustees supported the direction but raised a concrete liability question: what responsibility might ISOC bear for future IESG decisions? Their approval let the process move forward; it did not settle the long-term charter in December. RFC 1396 still looked ahead to the first round of selections.

The later paper trail matters. RFC 1601, dated March 1994, replaced the 1992 IAB charter. It documented an annual seven-person nominating committee chosen randomly from volunteers who had participated in two IETF meetings within two years, with a non-voting chair and four non-voting liaisons. It also specified thirteen voting IAB members and ISOC trustee approval for the twelve full IAB seats. The charter turned the POISED compromise into written institutional machinery, while appointments and day-to-day decisions remained separate acts.

RFC 1396 is memorable because the apparently comic lottery answered a serious problem. Open participation did not produce a ready-made electorate. Consensus did not explain who appointed leaders. The committee, ratification chain and shifted standards authority supplied distinct answers. The Internet's standards process kept its open door, then installed a very specific hinge on the room where its leaders were chosen.

Sources