Summary

  • Simple PDP Update Draft 1 proposed a narrower and faster route through part of AFRINIC’s policy-development process, but it never became policy: it was submitted on 23 October 2018 and later withdrawn.
  • Its strongest practical case was inclusion. It required at least eight weeks of mailing-list discussion, retained one meeting and a stable-text period, and allowed the main consensus determination after the meeting instead of making presence in the room decisive.
  • Its central trade-off was control. By ranking the nature of objections above their number and limiting Last Call mainly to editorial points or exceptional new technical objections, the design placed greater interpretive weight on the co-chairs.
  • Draft 1 did not delete Board approval, implementation, appeal, chair recall or emergency variation. Those downstream clauses sat outside its redline, even though the proposed changes could have affected the decision later sent into them.
  • A well-run private registry process can support accurate coordination, but neither participation nor rough consensus supplies a public mandate. A better design would keep mandatory rules narrow, publish reasons, preserve meaningful review and leave wider operational choices voluntary.

The redline that moved the decision

The most revealing way to read the Simple PDP Update is not as a catalogue of aspirations but as a set of edits. Draft 1, formally identified as AFPUB-2018-GEN-002-DRAFT01, version 1.0, was submitted by Jordi Palet Martinez on 23 October 2018. It proposed amendments to selected provisions in section 3.0 of AFRINIC’s then-current Consolidated Policy Manual.

Its chosen instrument was subtraction: less lead time before a meeting, less time before an agenda became due, less time before minutes had to appear, less reliance on the meeting as the instant at which consensus was declared, and fewer kinds of objection that could unsettle an already reached conclusion during Last Call.

That description needs an immediate legal and institutional qualifier. Draft 1 was a proposal under discussion. AFRINIC’s own record now labels it withdrawn. Nothing in the published history establishes that its clauses were adopted, ratified by the Board, implemented, used in a live decision, appealed or enforced. The verbs in this analysis are therefore “proposed,” “would have” and “could have,” not “did.” A later Draft 2 was announced on 27 November 2018, but that was a different instrument. Two days later, the author withdrew the proposal, explaining the step as an effort to restart cooperative discussion.

Draft 2’s language, and the still later PDP Bis work, cannot be smuggled backward into the Draft 1 redline.

The first proposed addition concerned the meaning of consensus. The then-current manual described consensus as general agreement without sustained opposition and entrusted the chairs with deciding whether objections had been adequately addressed. Draft 1 sought to make that logic more explicit. Consensus, in its formulation, was not a vote or a headcount. Critical technical objections had to be resolved. Low participation did not by itself prove that consensus was absent, while disagreement unsupported by an explanation was not enough to defeat it.

The quality and nature of an objection mattered more than the number of people associated with it.

There is a sound engineering intuition here. A protocol defect does not become harmless because only one engineer notices it, and an unsupported “no” does not become a technical argument because many people repeat it. RFC 7282’s account of rough consensus similarly directs attention to issues rather than vote totals and does not promise that every objection must be accommodated. Draft 1 tried to stop a technical forum from degenerating into a show of hands.

The proposal’s difficult move came in specifying who would decide that an argument had run its course: the chairs could disregard additional support for an objection they considered already addressed or refuted. That rule filtered repetition, but it also converted the classification “answered” into a decision gate.

The second proposed addition defined Last Call. Draft 1 treated it as a brief final opportunity to comment after a consensus determination, with particular value for people who had not previously taken part. But it did not reopen the entire discussion. The proposed acceptable material was largely editorial comment or an exceptional, newly discovered and substantiated technical objection. Subjective objections were excluded from the proposed test. Last Call would remain, but its function would narrow: it would check whether a previous conclusion survived rather than supply the forum in which the first conclusion was formed.

That shift becomes clearer when the timetable is reconstructed. Under the then-current section 3.4.1, a proposal had to reach the Resource Policy Discussion mailing list at least four weeks before a Public Policy Meeting. Draft 1 proposed cutting that meeting-specific submission interval to two weeks. Taken alone, that edit sounds like a simple acceleration. It was not the whole design. The proposal also required at least eight weeks of mailing-list discussion and consideration at at least one Public Policy Meeting before consensus could be declared.

It retained the one-meeting condition rather than abolishing meetings, preserved the six-month expiry mechanism, and left the author’s ability to withdraw intact.

The combination matters more than either number in isolation. A two-week pre-meeting deadline could admit a text later in the conference cycle. The eight-week mailing-list floor meant that the decision could not validly be compressed into those same two weeks. If a proposal had not accumulated the required list discussion by the meeting, the chair decision would need to wait. Draft 1 was thus not simply a “faster meeting” rule. It was an attempt to redistribute evidentiary weight across asynchronous discussion and an in-person forum while retaining both.

Section 3.4.2 contained the next cluster of edits. The then-current manual required the meeting agenda at least two weeks beforehand, a stable text at least one week beforehand, a chair decision at the meeting, and minutes within three weeks. Draft 1 proposed an agenda at least one week before the meeting and minutes within two weeks afterward. It retained the one-week stable-text safeguard. Most significantly, it moved the consensus determination away from the meeting itself: once the required list discussion and meeting had occurred, the chairs could determine consensus within a maximum two-week post-meeting window.

This was the redline’s pivot. The room would no longer be the necessary instant of judgment. Comments made across time zones could remain relevant during a defined post-meeting interval, and the chairs would have a short period in which to assess the record rather than make an immediate declaration before the room dispersed. In the benign reading, that is a practical correction to physical-presence bias. In the skeptical reading, the same move makes a later, less observable act of interpretation more important.

A decision announced after the meeting can be more considered, but only if its reasons and the materials assessed are visible enough to be tested.

Draft 1’s treatment of section 3.4.3 completed the sequence. It proposed a two-week Last Call followed by a chair confirmation, within one week of Last Call ending, as to whether the already determined consensus had been maintained. The then-current arrangement described Last Call as a period during which meeting feedback was evaluated before the chairs decided whether consensus had been achieved. Draft 1 would have placed the initial decision earlier and made Last Call a maintenance test. It did not eliminate Last Call. It changed the baseline from “has consensus been achieved?” to “has the consensus already found survived?”

Baselines influence institutional behavior. Once an outcome is named consensus, a later participant is no longer merely contributing to an open question. The participant must fit an objection through a narrower aperture and dislodge an existing characterization. Editorial observations remain welcome. A newly discovered technical problem may qualify if it is exceptional and substantiated. A concern characterized as subjective does not. This can protect a process from endless repetition, strategic delay and late objections that add no information.

It can also make the chairs’ earlier description of an objection—technical, resolved, repetitive or subjective—more consequential than the surface timetable suggests.

The redline’s discipline is easiest to see as a sequence. First, the mailing list must carry discussion for at least eight weeks. Second, the proposal must be considered at at least one Public Policy Meeting, with the text stable for the preceding week. Third, the chairs receive up to two weeks after the qualifying discussion and meeting to decide whether consensus exists. Fourth, a two-week Last Call tests that result under narrower conditions. Fifth, the chairs have one week to say whether the result has been maintained.

The design shortened some notices and reporting deadlines, but it did not abolish deliberation, the meeting, stable text, Last Call, expiry or voluntary withdrawal.

Equally important is what the redline did not touch. Draft 1 displayed replacement language for sections 3.1, 3.4.1, 3.4.2 and 3.4.3. It displayed no replacement language for section 3.4.4 on Board approval, section 3.4.5 on implementation, section 3.5 on conflict resolution, internal appeal and chair recall, or section 3.6 on varying the process in emergencies. In the then-current manual, the later sequence included a recommendation to the Board after working-group consensus, Board ratification, implementation provisions, an internal appeal supported by three participants, chair recall, and an emergency-variation provision.

Those mechanisms remained outside Draft 1’s proposed edits.

“Outside the redline” does not mean “irrelevant to institutional effect.” If the proposed consensus rule changed which objections were treated as critical, when a concern was treated as answered, or whether a prior finding survived Last Call, it could change the decision later presented to the Board or challenged through internal appeal. An upstream filter determines what reaches a downstream safeguard. But that is different from textual abolition. Draft 1 did not remove appeal, Board ratification, implementation or emergency variation. Any account saying it did confuses an indirect effect on inputs with a deletion of the receiving machinery.

The official status also places a hard limit on claims about impact. We can compare clocks, identify allocations of judgment and test likely incentives. We cannot report that the new timetable reduced delay, saved administrative cost or broadened actual participation, because the published record does not establish implementation. It does not supply a complete denominator for who participated, who was absent, or whom any participant was authorized to represent. Nor does it show how co-chairs would have classified a disputed objection under these words in practice.

Draft 1 is evidence of a proposed institutional design, not a natural experiment.

That distinction makes the proposal more, not less, analytically useful. Operative rules often gather anecdotes, personalities and later disputes that obscure their architecture. A withdrawn redline can be inspected as a clean allocation of burdens. Draft 1 asked participants to explain objections; asked chairs to distinguish critical technical concerns from other disagreement; shifted the main decision beyond the meeting; asked late objectors to bring editorial points or exceptional new technical evidence; and placed the confirmation of maintained consensus on a clock.

The question is not whether those allocations secretly caused later events. It is whether they would have reduced exclusion without making classification itself the scarce source of power.

The answer begins with the proposal’s strongest case. A person unable to cross a border, obtain a visa, pay for a flight, leave work, or join at a convenient local hour should not have their written technical contribution discounted merely because others assembled in a room. The stated problem behind Draft 1 was precisely that mailing-list participants who could not attend a Public Policy Meeting might be disadvantaged when chairs assessed rough consensus at the meeting. Moving the decision into a bounded period after both list and meeting discussion recognized that participation can be real without being simultaneous.

The draft also tried to make time legible. Eight weeks of list discussion offered a minimum window. One stable week protected participants from arriving at a meeting to discover that the target had just moved. A maximum two-week decision window prevented indefinite chair deliberation. Two weeks of Last Call and one week for confirmation made the closing sequence knowable. Minutes due in two weeks rather than three could place the meeting record in participants’ hands earlier. None of these features guaranteed efficiency, but together they described a more scheduled administrative service.

Finally, rejecting headcounts can protect minority expertise. The technical safety of shared numbering and routing systems cannot be determined by applause. One substantiated objection may reveal an interoperability or continuity problem that a large majority missed. Conversely, the bare accumulation of names behind an already answered point need not force endless reopening. Draft 1’s emphasis on reasons rather than numbers was therefore not inherently exclusionary. Its fairness depended on whether the institution could distinguish a reasoned filter from a conclusory label—and show its work.