Summary

  • PDP Bis Draft 1 tried to repair genuine ambiguity in AFRINIC’s policy process, but placed the powers to classify participation, control phases, determine consensus and vary procedure largely in one Chair, with ratification and appeal remaining within a Board-linked institutional chain.
  • Its safeguards—open participation, stated phases, objection categories, a Last Call, reasons for Board rejection and an appeal route—made policy formation more legible, yet its competing definitions of consensus and its treatment of Last Call silence left the decisive evidential standard unstable.
  • A legitimate private registry process should remain thin: publish denominators, evidence, unresolved objections and reasons; separate facilitation from review; treat silence as neutral; confine ordinary policy to technical coordination; and require direct notice and stronger authorisation when existing rights or running networks are exposed.

L3 — The small constitution hidden inside Draft 1

The title “Policy Development Process Bis” understated the reach of the proposal. AFPUB-2017-GEN-002-DRAFT-01, version 1.0, was submitted on 28 April 2017 by Komi Abel Elitcha, Arnaud A. A. AMELINA, Honest Ornella GANKPA and Alain P. AINA. AFRINIC recorded it as under discussion and described it as an amendment to Article 3.0 of its Consolidated Policy Manual. Those details establish publication of a proposed text, not its adoption.

They matter because the fairest reading starts with what the instrument was: an attempt to specify how a private registry’s policy working group would receive proposals, organise debate, make a finding of consensus and pass the result into corporate implementation.

The proposal’s diagnosis was not fanciful. It said the existing process did not define working-group leadership, consensus, proposal adoption or Board variance with enough precision. Ambiguity at that level is costly. A rule about allocations or transfers can be examined clause by clause, but the procedure that decides whether any such rule has passed governs the whole future policy inventory. If its offices are uncertain, every later result inherits a question about who could move it, who could stop it, what counted as agreement and what happened after an objection.

Draft 1 therefore deserves to be read as institutional design rather than mere administrative housekeeping.

Its core office was a single PDWG Chair, supported by one Vice-Chair. The Chair would coordinate the Policy Development Working Group, guide proposals through the prescribed phases and gauge consensus. The Vice-Chair would assist and substitute where functions were delegated or the Chair was unavailable. On its face, that allocation promised accountability: a participant could identify the person responsible for managing the record, rather than watching two indistinct offices pass responsibility between them. Yet the same clarity concentrated several different kinds of power.

The person facilitating discussion would also classify some contributions, control important transitions and make the final consensus determination. Draft 1 named an office clearly without always separating the office’s roles.

The leadership-selection provisions contained a revealing tension of their own. The text spoke of the community electing a Chair and Vice-Chair for alternating two-year terms. It also assigned NomCom a role in appointing them through a process that culminated in a show-of-hands election at a Public Policy Meeting. Candidates had to be active on the policy mailing list and to have attended at least two AFRINIC Public Policy Meetings. Only candidates physically present at the election meeting would enter the vote.

Each condition had an intelligible purpose: list activity might demonstrate familiarity with the work, prior meeting attendance might show practical experience, and physical presence might permit scrutiny. Together, however, they narrowed eligibility to people able to absorb the time and travel costs of institutional participation. The procedure called the resulting choice a community election while defining the electable community through prior access to the forum.

That did not make the election meaningless. It made its denominator important. A show of hands among those in the room can choose an internal facilitator. It cannot by itself demonstrate what all number-resource holders, operators, customers or other affected principals authorised. Draft 1 partly recognised the distinction when it said a show of hands could gauge support but was not a vote in the policy-consensus sense. Yet it relied on a show of hands for leadership selection. The same physical gesture therefore carried different institutional meanings depending on context.

A well-written constitution should say precisely which population is making which decision and what authority that decision supplies.

The Chair’s control began before consensus. Draft 1 allowed input to be rejected or deferred under classifications including Old, Minor, Timing and Scope. Those categories can be useful. Repetition can drown a working list; a minor editorial point need not halt a proposal; an intervention can arrive too late for the immediate phase; and a submission may genuinely concern a different policy. Without some moderation, persistence or volume can substitute for relevance. The difficulty lies in making the classifier the same officer who guides the proposal and ultimately declares whether objections have been answered.

A scope judgement can shape the evidential record on which the later consensus judgement rests.

The participation power was not merely editorial. Following the stated warning process, the Chair could suspend a participant’s posting privileges for as long as 30 days, subject to appeal. A bounded suspension with warning and review is more disciplined than unexplained exclusion. But in a process whose evidence is produced on the list, temporary removal can affect both speech and the apparent distribution of views. The key question is not whether moderation is ever permissible.

It is whether reasons, examples, duration, notice and review are sufficiently visible that a disagreement about substance cannot quietly become a finding about conduct. The proposal created labels and a remedy, but it left considerable initial control with the officer who also held the consensus pen.

Draft 1 tried to give substance to consensus by distinguishing minor from major objections. It described rough consensus in terms of objections having been resolved or duly considered, with the proposal’s benefits judged to outweigh its disadvantages. That is more demanding than counting applause. It invites a reasoned record: what was the objection, what evidence supported it, how did the proposal change, and why did remaining costs not defeat the benefit? In technical coordination, that method can be better than a bare majority because an objection from one operator may reveal a failure affecting everyone’s routing or registry data.

At the same time, the text said consensus is achieved when everyone consents to a decision acceptable to all participants. That formula sits uneasily beside rough consensus and a Chair’s weighing of benefits against disadvantages. If everyone consents, unresolved major objections should be absent. If rough consensus can exist after an objection is considered but not accepted, the decision is not literally unanimous. If the Chair makes the final determination, “acceptable to all” may describe an aspiration rather than a test. Draft 1 brought the vocabulary of agreement into the open, but it did not settle which of these standards controlled.

The contradiction was consequential because the Chair received the final call. A decision-maker can responsibly find rough consensus without unanimity, but only if the record exposes the bridge from participation to conclusion. It should identify how many distinct participants engaged, which affected interests were represented, what objections remained, whether the evidence answered them and why the proposal stayed within the forum’s remit. Draft 1’s language offered concepts for such a record but did not make the denominator and reasons the decisive constitutional artefact.

The phrase “the community agrees” could therefore carry more weight than the underlying evidence warranted.

The four-phase structure was Draft 1’s most visible attempt to turn debate into an orderly sequence. Adoption, Discussion, Review and Concluding were not just labels; they gave a proposal a path and placed the Chair at each gate. During Adoption, for no more than two weeks, the Chair screened such matters as clarity, relevance, duplication and whether the proposal was ready to enter the working process. This was a rational intake stage. A public forum benefits when an idea arrives with a defined problem and text that participants can actually test.

But intake discretion should remain a test of form and technical scope, not an early judgement about whether an inconvenient proposal deserves a constituency.

The initial Discussion Phase was to last at least four weeks. Here the mailing list and public meeting could do their best work: expose operational assumptions, compare effects, identify implementation requirements and force authors to answer objections. Draft 1’s decision to give discussion a minimum rather than an immediate deadline acknowledged that distributed technical expertise takes time to surface. Yet elapsed time is not evidence of adequate notice or representative participation. Four weeks on a specialist list can be ample for regular insiders and invisible to a network that does not know a proposal will affect it.

The Review Phase could last up to four weeks. In that stage, the Chair would summarise the discussion, assess the state of objections and determine whether rough consensus had formed. A review window can prevent endless recirculation of familiar points. It can also turn the Chair’s summary into the authoritative version of a more complicated record. Accuracy therefore depends on a right to correct the summary and on an explicit account of minority evidence. A summary that says an objection was “addressed” may mean it was answered, accommodated, rejected with reasons or merely mentioned. Each meaning has a different bearing on consensus.

The Concluding Phase contained a Last Call of at least two weeks. Last Call is valuable when it operates as a final error-detection interval: participants can identify a drafting change that reopened an issue, or point out that the recorded reasons omitted an operational consequence. Draft 1 went further by expressly treating the absence of community feedback at that stage as consensus. That move confused the state of the message queue with an affirmative institutional fact. No feedback can mean satisfaction.

It can also mean that earlier participants have nothing to add, that affected parties never received notice, that the cost of re-entering the discussion is too high, or that operators are occupied with running networks. Silence shows that no new comment arrived. It does not reveal why.

Nor can Last Call silence repair an earlier weakness in the denominator. Suppose ten people debated a proposal, two raised objections, the Chair recorded those objections as considered, and nobody wrote during the final fortnight. The final silence does not establish the consent of every participant, still less of thousands of absent resource holders or networks. It says only that the Last Call did not produce another message. Draft 1’s treatment of silence supplied a convenient closure rule, but described that closure in the language of consensus.

A constitution for technical coordination should distinguish administrative finality from evidential agreement.

Once the Chair found rough consensus, Draft 1 sent the proposal to the AFRINIC Board. The Board’s stated function was to check the process and ratify or reject. A rejection required published reasons; ratification activated staff implementation. This was a significant safeguard. It made a negative corporate decision explainable and recognised that a policy cannot become an operational registry instruction without a responsible company checking feasibility and legal fit. It also revealed the private character of the chain. Board ratification was corporate review, not public-law enactment.

The Board could confirm that the prescribed procedure had been followed and that staff could implement the result. It could not turn the working group into a regional legislature.

Draft 1 also provided an appeal route. A formal appeal needed support from three participants in the discussion and had to be filed within two weeks after “public knowledge” of the decision. It would go to a Board-appointed Appeal Committee, which could direct that a Chair decision be annulled where the PDP had not been followed. The presence of a defined appeal was better than leaving an aggrieved participant to petition informally. The threshold discouraged purely solitary procedural manoeuvres, while the power of annulment could force a defective step to be taken again.

But each design choice narrowed the remedy. Requiring three participating supporters could exclude an affected party that discovered the issue late or whose objection was unique but technically decisive. A two-week period can promote finality, yet a clock triggered by “public knowledge” is unstable: public to whom, through which channel, and proved by what event? The committee’s appointment by the Board placed review within the same institutional family as ratification. Its remit, as described, concerned failure to follow the PDP; it was not an independent tribunal with general power to decide rights.

Annulment could correct a procedural decision, but it did not itself resolve whether the underlying policy exceeded the registry’s legitimate technical role.

The structure linked appeal to other Board powers. Draft 1 involved the Board in recall arrangements and some appeal functions, even as it reserved the Board a ratification role. Institutional involvement is not automatically improper: the company’s governing body cannot be absent from corporate legality, staffing and implementation. The concern is accumulated dependence. If one institutional chain selects or controls reviewers, ratifies outcomes and retains exceptional powers, procedural independence needs more than a differently named committee.

Appointment criteria, protected tenure, conflicts rules, published reasons and a clear bar on ex parte direction would matter more than the label “appeal”.

Emergency variance made the allocation still sharper. Draft 1 gave the Chair a route to compress or vary the ordinary process, subject to a combined minimum of four weeks for Discussion, Review and Concluding. It also described an urgent-policy route for the Board with at least two weeks of working-group review, while making the comments non-binding. Emergencies can make ordinary timetables impractical. A newly exposed technical vulnerability may require coordination faster than a leisurely sequence allows.

Yet an exception capable of shortening participation should require an objective trigger, a stated necessity, the least departure possible, a fixed expiry and after-action review.

The two routes raised different problems. The Chair’s minimum four-week combined period preserved some time for participant scrutiny, but added variance power to an office already controlling classification, phase transitions and consensus. The Board’s route gave the working group at least two weeks but described its comments as non-binding, placing the exceptional decision more directly within corporate control. Neither mechanism should be read as proof of public emergency authority. At most, a private coordinator can adjust its internal process to protect narrow technical operations within its contractual and corporate scope.

AFRINIC’s staff and legal assessment of 19 May 2017 confirms that these were live drafting questions, not merely later objections projected backwards. The assessment asked for a public definition of rough consensus, clearer phases and editing, an implementation timeline and sufficient time for staff assessment. AFRINIC’s legal comments challenged the proposed limit on urgent Board action and identified the “public knowledge” appeal trigger as unclear. These comments are valuable contemporaneous evidence because they show the institution itself recognised ambiguity in the instrument.

They are not independent adjudication, and they do not tell us that Draft 1 was later used.

The revision record reinforces the evidential boundary. The parent record lists Draft 2 on 9 November 2017, Draft 3 on 28 April 2018 and Draft 4 on 29 October 2018 after the first version. That sequence demonstrates revision, not what any later clause meant in April 2017. Draft 1 must stand on its own language. The archived official pages identified the proposal as under discussion; the fact of later drafts cannot be converted into adoption, implementation, a consensus result or proof that the first architecture worked.

The proper reconstruction is therefore mixed but precise. Draft 1 published offices, categories, stages, minimum periods, reasons and a remedy where the earlier regime was said to be unclear. It also placed facilitation, classification, discipline, phase control, consensus judgement and one exceptional route around the Chair, then placed corporate ratification, Board-linked appeal and another exceptional route inside AFRINIC. Its constitutional problem was not secrecy alone.

It was the possibility that a more visible procedure could still make one private chain’s account of participation look like authority over people who did not authorise it.