Summary

  • PDP Bis Draft 4, AFPUB-2017-GEN-002-DRAFT-04, made four targeted changes: authors could no longer be anonymous, PDWG chairs could not act as document editors, chairs received up to two weeks for specified consensus decisions, and a Board-appointed appeal committee replaced Draft 3’s bespoke three-office review panel.
  • The revised map still contained several distinct private gates. Chairs controlled admission, phases and consensus calls; staff assisted and analysed but implemented only after Board ratification; the appeal committee could annul a chair decision for failure to follow the PDP; and the Board retained reasoned remand grounds and a necessary-and-urgent variance route.
  • The benign case is substantial: named authors, separated roles, a visible decision clock, staff impact analysis, published remand reasons and an available internal appeal could make a registry’s policy workflow more accountable and predictable.
  • Better procedure did not turn participation into representation or corporate discretion into jurisdiction. AFRINIC remained a private Mauritius-incorporated technical bookkeeper and uniqueness coordinator, not a legislature, regulator, court or emergency government for African networks.

A fourteen-day decision inside a many-gated chain

PDP Bis Draft 4 was not a constitutional founding. It was a proposed revision to the way AFRINIC organised changes to its own policy manual. The archive identifies it as version 4.0 of AFPUB-2017-GEN-002-DRAFT-04, submitted on 29 October 2018, marked “Under Discussion” and intended to replace section 3.0 of the Consolidated Policy Manual. Its listed authors were Komi Abel Elitcha, Arnaud A. A. AMELINA, Honest Ornella GANKPA and Alain P. AINA. One month later, at AFRINIC-29 on 29 November, the authors described the feedback they had incorporated. The meeting record says the update had attracted no substantial discussion or suggestions and returned it to the mailing list. That disposition is the first boundary around any analysis of the draft: it was still a proposal.

The four recorded alterations were narrow but revealing. First, Draft 4 removed the option for authors to remain anonymous. Second, it prevented PDWG chairs from serving as document editors. Third, section 3.4.3 gave the chairs up to two weeks to decide on a draft presented at a Public Policy Meeting or reaching the end of the adoption phase. Fourth, it replaced Draft 3’s specially composed appeal panel—a director, a member of the Council of Elders and an immediate past PDWG chair—with an appeal committee appointed by the AFRINIC Board.

The meeting account described that last change as alignment with the appeal machinery already used by the then-current PDP.

Each adjustment answered an identifiable administrative problem. A named author can be questioned, credited and held to a stable text. Excluding chairs from document editing reduces the risk that the people managing deliberation also control the language on which they later declare consensus. A decision deadline tells participants when uncertainty should end. Substituting an existing appeal committee for a panel assembled from three offices can make review easier to convene. These are not cosmetic virtues.

In a registry setting, clarity about who holds a document, who decides, and when a decision is due can reduce monitoring expense and strategic ambiguity.

Yet the changes only make sense when placed back inside the larger architecture they revised. Most of that architecture belonged to Draft 3, published on 28 April 2018; Draft 4 did not newly invent it. The inherited sequence gave chairs several points of control. They screened proposals for matters such as clarity, scope and duplication. They managed movement through phases. They assessed rough consensus by considering both meeting discussion and the mailing list. A show of hands was described as a gauge of opinion, not a vote. Under the inherited Last Call mechanism, an absence of feedback could be treated as consensus.

Draft 4’s distinctive addition was the clock: for the specified decision points, chairs had as long as two weeks to announce the result.

This made the chair a strong procedural gatekeeper but not an unreviewable one. The inherited appeal procedure required three supporters who had participated in the working group and imposed a two-week filing window. The reviewing body was to report within four weeks unless that period was waived. Its remedial authority was narrow: it could annul a chair decision when the policy development process had not been followed. It did not receive a general power to decide whether the policy was wise, economically efficient, fair to every operator or lawful under public law.

Nor did it gain a free-standing right to substitute its policy judgment for the chairs’ judgment. Draft 4 changed who staffed that review gate, not the limited character of the gate.

The word “final” in the internal appeal design therefore needs disciplined reading. It can close a route within a private organisational sequence. It cannot erase ordinary legal rights, convert committee members into judges of general jurisdiction or make their conclusion binding on persons who never lawfully submitted a dispute to them. Internal finality is useful: an institution cannot administer a process if every procedural question remains perpetually open. But finality has an object and a perimeter. Here, the object was an internal process challenge to a chair decision. The perimeter was AFRINIC’s proposed policy workflow.

Staff occupied a different place. In the inherited adoption mechanism, chairs or AFRINIC staff could work with initiators to refine a problem statement. Staff also supplied analysis relevant to practical implementation, and the later review stage contemplated further discussion in light of an impact analysis. But staff did not receive a general policy veto or a licence to implement a proposal on its own initiative. Implementation followed Board ratification. The distinction is consequential: expertise can reveal operational cost, ambiguity or technical incompatibility without becoming a source of political authority.

The Board then formed another gate. The inherited ratification clause was framed as a duty to ratify, subject to two classes of exception. The Board could find that the PDP or consensus declaration was materially flawed. It could also consider the proposal harmful to AFRINIC while acting in its fiduciary role. Rejection required published reasons and remand rather than silent burial. Separately, an inherited variance clause provided a route for the Board to adopt a measure it regarded as necessary and urgent, followed by Working Group consideration at the next Public Policy Meeting.

The implementation clause contemplated an implementation date less than six months after ratification unless a waiver was requested.

Those provisions distribute power rather than crown a single sovereign. Chairs could advance a text toward the Board by declaring consensus, but an appeal could annul their decision for process failure. Staff could shape clarity and illuminate feasibility, but not activate implementation before ratification. The Board was described as normally obliged to ratify, but could remand for material procedural defect or perceived harm to the corporation and could use a separate emergency route. The appeal committee itself depended on Board appointment.

When procedural validity and substantive policy diverged, the draft did not give one office an unlimited answer. It placed different questions in different institutional hands.

For an operator, that map translates into cost. A phase decision determines whether a proposal advances or goes back. A consensus declaration starts the next portion of the sequence. A two-week decision window requires participants to monitor a predictable interval, while the two-week appeal window demands a fast response from anyone alleging procedural failure. A Board remand delays implementation and reopens deliberation. An emergency variance can compress the ordinary sequence. Once ratified, staff action turns text into registry administration. None of this proves that Draft 4 caused an outage, resource loss or impairment.

It explains why the allocation of gates matters before harm occurs.

The record also sets hard chronological boundaries. Draft 3 owns the four-phase structure, the rough-consensus vocabulary, silent Last Call, the ratification exceptions, the emergency route, the implementation clock and the original three-office appeal design. Draft 4 owns the four recorded changes and the altered institutional map that resulted. Draft 5 is dated 9 June 2019, and its revision history characterises its changes as corrections of typos and grammar. It is useful as a carrier of the revision history, but its old-and-new table also indicates wording changes.

Where it does so, later wording cannot simply be backdated and quoted as Draft 4 language.

That evidence discipline rules out several tempting stories. There is no basis here to say that Draft 4 was adopted, ratified, implemented, tested in an actual appeal or made operative. The absence of substantial new discussion at AFRINIC-29 is not proof of agreement by affected operators; it is a fact about the recorded discussion. The number of relevant mailing-list participants, meeting attendees, represented principals and valid authorisations is not supplied. The proposal record proves what AFRINIC and participants wrote, proposed and did. It does not, by itself, prove whom they had authority to bind.