Summary

  • AFPUB-2026-GEN-001-DRAFT01 is under discussion. AFRINIC-37 records that its authors did not ask for consensus and that the proposal returned to the policy mailing list.
  • The canonical draft uses a sixty-day trigger, while the official meeting minutes say ten days. The public record inspected for this article does not reconcile the difference.
  • Fifteen supporters do not, by themselves, ratify anything. The proposed chain also requires registered contacts of distinct AFRINIC members from at least three sub-regions and a new staff analysis finding no new issue.
  • The public impact assessment is not absent: it contains detailed operational and legal material. Yet its interpretation and clarity sections still contain drafting instructions, leaving the most consequential transition insufficiently explained.
  • A Petition-to-Implementation Ledger could preserve continuity while making Board receipt, supporter qualification, staff analysis, legal review, decision, challenge and rollback separately verifiable.

A continuity problem worth solving

The most persuasive case for Draft01 is not theoretical. AFRINIC has already experienced a period in which the absence of a functioning Board quorum interrupted public policy meetings, appointments and policy implementation. Networks did not stop needing registry services during that interval. A policy process that depends on one corporate organ must therefore answer a practical question: what happens when that organ is temporarily unable to perform its part?

That question deserves more than a reflexive defence of the status quo. A Board-ratification step can protect institutional responsibility, but it can also become a permanent veto if incapacity has no bounded alternative. Draft01 tries to prevent that outcome. It keeps the ordinary route in which co-chairs recommend a consensus proposal to the Board, then adds a petition path for cases in which the Board cannot ratify or refuses without reasons for a stated period.

The design contains real friction. A single person cannot simply declare a policy adopted. The petition must attract at least fifteen PDWG participants who are registered contacts of different AFRINIC members. Those supporters must span at least three AFRINIC sub-regions. Staff must then conduct a new analysis. Only if that analysis finds no new issue does the draft say the proposal must be considered ratified and implemented by staff.

Those conditions matter. They distinguish the draft from a bare mailing-list vote and acknowledge that continuity needs membership diversity, geographic distribution and a fresh operational check. The authors also presented the text as unfinished. The AFRINIC-37 record says they were not seeking consensus, that the questions were complex and that the proposal would return to RPD. This is therefore the moment to repair the mechanism, not to report it as a completed transfer of authority.

What the draft actually changes

The current Consolidated Policy Manual separates several states that are easy to blur in public argument. A proposal is discussed. Co-chairs assess consensus. A successful draft is recommended to the Board. The Board ratifies it. Adoption is announced. Procedures are prepared and published. Implementation follows. Each verb describes a different event and should have its own evidence.

Draft01 retains much of that sequence but creates an alternative after the Board stage. Its canonical text describes a proposal already approved by the PDWG that has not been ratified because the Board is unable to do so or refuses without reasons for sixty days. A PDWG member may initiate a petition. When the defined supporter threshold is met, a new staff analysis follows. If no new issue is found, the draft says the proposal must be considered ratified and implemented by staff.

The phrase “must be considered ratified” is the hinge. It turns evidence about participation and analysis into an institutional decision. Yet the draft does not identify the complete set of receipts needed to prove that the hinge may lawfully and operationally move. It does not say what event constitutes delivery to the Board, which timestamp starts the clock, how Board inability differs from delay, what counts as a refusal, or what makes reasons adequate. Without those definitions, two honest administrators could calculate different deadlines from the same history.

The supporter test also needs a time model. “PDWG member” is not self-defining. “Registered contact” can change. A company may have more than one contact, a person may be associated with more than one account, and an account’s sub-region may depend on a maintained classification. The draft does not specify the snapshot date, deduplication rule, identity-verification method, privacy boundary, withdrawal rule or treatment of later changes. A published total of fifteen would not show that all conditions had been met.

The staff-analysis stage is equally consequential. The draft calls for a new analysis and asks whether it finds a new issue. It does not define the baseline against which “new” is measured, who writes the analysis, who checks it for conflicts or completeness, whether the legal section must be resolved, what evidence is public, or who signs the final finding. If the analysis is both the main safeguard and the event that unlocks deemed ratification, its version and review history cannot remain implicit.

Ten days in the minutes, sixty in the proposal

The official record contains a material version-control conflict. The canonical proposal uses sixty days. Author explanations in the RPD archive also discuss sixty days. The AFRINIC-37 minutes, however, summarise the alternative as activating after ten days. No inspected source reconciles those figures.

This cannot be handled as a typographical curiosity. A fifty-day difference changes the practical balance between continuity and Board deliberation. It affects when supporters may organise, when staff must prepare, when implementation exposure begins and how much time exists for a Board to recover quorum or give reasons. It also affects whether an administrator can prove that a petition was premature.

The proper conclusion is narrow: the public record contains two official numbers, and the canonical proposal text is the current proposal text. The minutes are not silently promoted into an amendment; nor should they be dismissed without correction. A future revision should state the chosen period, identify the version it replaces, explain the purpose of the duration and specify the authoritative clock receipt.

The conflict is also a useful stress test. If a proposal cannot reliably carry its own waiting period from meeting presentation into the public decision record, its later deemed-ratification mechanism needs stronger version binding. Every petition should cite the exact proposal hash, the exact policy state, the exact Board-delivery receipt and the exact rule version used to calculate the deadline.

An impact assessment that is substantial and unfinished

The public impact-assessment page should not be described as empty. It contains operational findings covering registry systems and organisational functions. It addresses finance. More importantly, it publishes a detailed legal analysis of the relationship among the PDP, the Consolidated Policy Manual and AFRINIC’s Bylaws.

That legal section takes a restrictive view of the proposed transfer. It treats the PDP and manual as subordinate to the Bylaws, identifies Board direction and statutory responsibilities, and argues that a recommendation to the Board calls for judgment rather than automatic endorsement. It also says a refusal should rest on reasonable, objective and clearly articulated grounds. The position raises a serious design question: if the Bylaws allocate a function to the Board, can policy text alone create a staff-operated substitute?

That is AFRINIC staff’s institutional legal analysis, not a judgment of a Mauritian court and not a final statement of law. It must be reported with that limit. But the draft cannot simply ignore it, because the legal issue sits directly between the no-new-issue finding and implementation.

At the same time, two central parts of the published assessment remain visibly unfinished. The interpretation section contains “Publish in public IA” and instructions to provide a concise explanation. The clarity section likewise contains drafting instructions to list ambiguities and map them to policy or procedural touchpoints. These are not completed public findings.

The result is uneven rather than absent assessment. Detailed legal objections are public; core interpretation and clarity work is not. That asymmetry matters because the missing material should answer the exact operational questions on which a petition would turn: eligibility, timing, evidence, authorship, review and the meaning of “new issue.” Publishing those answers would allow supporters, opponents, staff and the Board to disagree about a defined mechanism rather than about different imagined versions of it.

The ledger the proposal is missing

Heng Lu argues that continuity of the verifiable registry function should not be confused with continuity of every gatekeeping power. That distinction offers a constructive way to assess Draft01. The goal is not to preserve one actor’s veto. It is to preserve the ledger, services and legitimate decision path while ensuring that any temporary substitute authority remains constrained and auditable.

A Petition-to-Implementation Ledger would begin before the petition. It would record the proposal identifier and content hash, the co-chair consensus decision and its evidence, delivery to the Board, the receiving actor and timestamp, and the policy and Bylaw versions then in force. It would identify whether the claimed trigger is inability or refusal. For inability, it would record the condition and the evidence that it persists. For refusal, it would preserve the decision and reasons, or the evidence that no reasons were supplied.

The clock would be a derived value, not an assertion. Its start event, timezone, duration, pauses and expiry would be machine-readable and human-readable. A correction—such as a ten-day reference being replaced by sixty days—would create a new version rather than overwrite history.

The petition receipt would identify its initiator and the rule under which that person is eligible. Support should be recorded without publishing unnecessary personal data. Each validation could attest that, at a defined snapshot, the supporter was a PDWG participant and a registered contact of a particular distinct member, while a privacy-preserving public digest proves that the count and geographic distribution passed. Withdrawals, duplicates, challenged records and later changes would remain visible as state transitions.

Next would come the staff analysis. Its mandate, evidence cutoff, authors, reviewers, conflicts, operational tests, legal treatment and final version should be recorded. “No new issue” should refer to a published test: new relative to which earlier assessment, material by which criterion, and decided by whom? An independent review should confirm that the analysis addressed the issues raised in consultation, not merely that a document exists.

The deemed decision would then have its own receipt. It would cite the validated petition, final analysis, legal-resolution record and authorising provision. It would identify the accountable signer and explain why every prerequisite is satisfied. Implementation would be a later, separate event with a change plan, safety checks, effective date, communications and a rollback path.

Finally, the ledger would preserve challenge. A member or Board able to show an identity error, clock error, unresolved legal conflict or operational risk should have a defined route to request review or pause. The standard, decision maker, time limit and effect on implementation should be known before a dispute occurs. Continuity without challenge can become self-validating power; challenge without a bounded timetable can recreate the paralysis the proposal seeks to solve.

Private authority, public consequences

AFRINIC is a private, membership-based Internet registry incorporated in Mauritius. Its Board, PDWG, co-chairs, Secretariat and staff occupy private corporate, policy-process, employment and contractual roles. Their decisions can have serious operational and economic consequences, but they are not legislation, regulation or adjudication.

Nothing in a fifteen-person petition creates sovereign, legislative, regulatory, judicial, police, prosecutorial, punitive or confiscatory power. It does not confer ownership of Africa’s number resources, a continent-wide public mandate or self-validating authority over disputed facts. Saying so does not diminish the importance of registry governance. It locates the source and limit of authority accurately.

That boundary reinforces the case for receipts. A private institution that administers scarce identifiers earns legitimacy through a clear mandate, consistent process, verifiable records and reviewable discretion. It cannot rely on the symbolism of “the community” to cure missing evidence about which members acted, under which rule and after which institutional event.

The same boundary protects participants. Disagreement with Draft01 is not evidence of hostility to continuity. Support for a fallback is not evidence of an attempt to seize public power. The inspected sources establish no unlawful act, bad faith, actual petition, deemed ratification, implementation or harm. They establish a proposal under discussion and a chance to improve it before any of those states can occur.