Summary

  • In May 2006, AFRINIC’s Board adopted Resolution 200605.27, ratifying AFPUB-2006-v6-001 and instructing staff to take the necessary implementation action. The proposal governed upstream IPv6 allocations from IANA to Regional Internet Registries: a minimum /12 and enough space for at least 18 months of registration needs. The final “.27” is a resolution sequence identifier, not evidence of a meeting on 27 May; the checked cross-regional record gives 15 May as the Board-adoption date.
  • The regional process record is thinner than the precision of the policy text. AFRINIC’s Cairo material says this proposal had no specific discussion yet realised consensus. A staff notice initially said the community had voted, but Alan Barrett corrected it to no vote and apparent consensus; after an uncertain contrary recollection from Alan Levin, AFRINIC chief executive Adiel Akplogan confirmed that no vote occurred because the policy process used consensus. The proposal still required the separate Board-ratification gate.
  • AFRINIC’s assent was necessary to the all-region common-text procedure, but it was not sufficient to bind IANA, another RIR or any nonparty. NRO, ASO, ICANN and IANA later performed distinct acts before the IANA registry recorded 2c00::/12 for AFRINIC. A useful private coordination act did not create a public mandate, territorial title, or sovereign authority over African networks.
  • The durable lesson is to preserve both sides of the event: objective replenishment rules made top-level IPv6 supply more predictable, while incomplete meeting and Board receipts leave the claimed consensus impossible to measure. Technical necessity can justify accurate, non-duplicative bookkeeping; it cannot turn a registry’s operational leverage into legislative, regulatory or coercive power.

A resolution at the hinge of two systems

Resolution 200605.27 sits at a precise hinge. Below it was AFRINIC’s own policy-development and corporate decision process. Above it was a multi-institution procedure for converting matching regional texts into a global instruction capable of being executed by IANA. The resolution mattered because the upper chain could not honestly treat the proposal as common to every RIR unless AFRINIC completed its regional step. It remained bounded because AFRINIC could complete only that step.

The Board page groups the resolution under May 2006. It says the Board heard a report from the Policy Working Group chair on the status of open policies, resolved to ratify four proposals and instructed staff to take the action necessary to implement them. One of those proposals was AFPUB-2006-v6-001, the proposed global policy for allocating IPv6 address space from IANA to the RIRs. A later ICANN comparison of the regional processes records AFRINIC Board adoption on 15 May 2006. Nothing in the checked record establishes 27 May as the date.

Reading the final digits of “200605.27” as a day would convert a numbering convention into an invented fact.

The bundled form needs similar care. Alongside the IPv6 proposal, the resolution named a temporary resource-assignment proposal, regional ASN eligibility criteria and a four-byte ASN proposal. Those instruments explain why the resolution speaks collectively, but their substantive rules are not part of this inquiry. The old minutes also contain ambiguity in the policy list, and the public page does not show whether directors treated the four items as one indivisible motion or as separate decisions recorded together. Neither the existence of a bundle nor the instruction to staff supplies the missing answer.

This article therefore follows one conversion only: AFPUB-2006-v6-001 from proposal text, through the Cairo consensus characterisation and last call, to AFRINIC’s corporate assent, then far enough along the later chain to identify the legal and operational effect of that assent. It does not retell AFRINIC’s recognition, the transition of regional service, the general history of global policy, or the wider technological argument between IPv4 and IPv6. Those stories would obscure the institutional act under examination.

The scope distinction is basic. The proposed rule concerned allocations from IANA to RIRs. It did not prescribe how AFRINIC would make downstream allocations to members or end users. The global rule stocked a regional administrator; it did not decide every regional distribution choice. Indeed, the text allowed each RIR to retain its own allocation and reservation strategies. Global compatibility at the supply boundary and regional autonomy below that boundary were both designed into the policy.

The common text replaced bargaining with a replenishment formula

The practical achievement was a clearer upstream supply rule. A globally unique pool cannot be managed reliably if each regional registry negotiates incompatible replenishment terms whenever its inventory runs low. The policy specified the unit, the planning horizon, two eligibility tests, calculations for stock and need, an evidentiary route for exceptional forecasts, and public announcement duties. Those elements reduced some room for ad hoc bargaining without pretending that every operational question had been solved.

The minimum allocation unit was a /12. At the policy’s inception, each existing RIR holding less than a /12 of unallocated IPv6 space was to receive a /12. A newly recognised RIR would receive an allocation upon recognition. For scale, a /12 contains 1,048,576 distinct /32-sized units: 2 raised to the power of the difference between 32 and 12. That arithmetic describes address-prefix capacity only. It says nothing by itself about downstream assignments, utilisation, demand, market value, legal ownership or how many networks would benefit.

The supply horizon was at least 18 months of an RIR’s registration needs. Additional space became available when either of two conditions was met. The first was that the RIR’s available space had fallen below 50 per cent of a /12. Half of the 1,048,576 /32-sized units is 524,288 such units, equivalent in prefix arithmetic to a /13. The second test was forward-looking: available space was below the quantity established as necessary for the next nine months. The “or” matters.

One test set a stock threshold; the other allowed demonstrated near-term need to trigger replenishment even where a simple half-block comparison did not capture the operating position.

“Available space” was itself calculated rather than guessed. The rule added currently free addresses to reservations due to expire during the following three months, then subtracted fragmented space. Fragmented space meant blocks within the current stock that were smaller than the RIR’s own minimum allocation size and therefore could not serve the ordinary allocation task in view. Treating unusable remnants as fully available would have overstated workable inventory. Deducting them linked the mathematical stock ledger to the operational shape of allocations.

Ordinary necessary space came from recent conduct: the average number of addresses allocated per month during the preceding six months, multiplied by the number of months in the planning period. For an 18-month horizon, that means 18 times the six-month monthly average, or three times the total issued during those six months. The formula offered repeatability and comparability, but it also made historical demand the default predictor of future need.

The policy consequently provided a special-needs route. An RIR could project a different rate when particular facts made the historic average inadequate, but it had to give a clear and detailed justification. Depending on the basis asserted, the supporting material could include trend data, analysis of the effect of policy changes, or analysis and verifiable references for external factors. IANA was to treat the projected special need over the next 18 months as valid if it lacked elements that clearly called the projection into question.

That wording preserved judgement at the evidentiary boundary: the route was not automatic, but nor did it invite rejection merely because a forecast departed from the recent average.

Once eligibility was established, the additional allocation was to be a single block sufficient for the established 18-month need. IANA, the NRO and the RIRs were to announce allocations and update their sites, while ICANN and the NRO were to establish administrative procedures. The publication duties mattered because a uniqueness system is credible only when its authoritative records converge and changes can be seen. A top-level decision concealed from adjacent registries and operators would defeat much of the reason for having a common process.

One omission was express rather than accidental. The text did not establish IANA service-performance requirements. It left those to suitable agreements between ICANN and the NRO. An RIR could therefore know the eligibility mathematics from this policy yet still lack, in this instrument, a promise about response time, reasons, escalation or remedy. The distinction is operationally important: a rule may define when an applicant qualifies without defining how quickly or accountably the service provider must act.

The design had several economic effects without settling broader IPv6 economics. An 18-month inventory horizon reduced dependence on last-minute negotiation and gave AFRINIC a basis for planning registration-service capacity. The /12 unit and the fragmented-space deduction aimed to preserve administratively usable, aggregatable stock rather than leave a registry dependent on small remnants. Evidence requirements made unusual forecasts more contestable. Regional autonomy avoided global micromanagement of downstream choices, but it also meant operators remained exposed to the quality of those regional choices.

Equal top-level blocks did not establish equal end-user access, deployment, price or regional benefit. The policy was a supply mechanism, not proof of an IPv6 transition success or an answer to continuing IPv4 reliance.

Consensus without a ballot, and without a denominator

The proposal’s documentary history is notably less exact than its formulas. ICANN’s later background table places introduction into the AFRINIC process on 1 August 2005. AFRINIC’s current policy page carries the older reference “afpol-glbipv6200508” and identifies Alan Barrett as author, although the page misspells his surname and displays a details date of 17 May 2005 that is inconsistent with the reference, the reported introduction and the later sequence. The best checked introduction date is 1 August 2005. The stray May date is not an adoption or implementation date.

The exact day of the Cairo event should not be silently harmonised either. The current AFRINIC policy history says consensus was reached on 3 December 2005. Yet a mailing-list message on 5 December still described the proposal as awaiting discussion in Cairo. The published meeting minutes are dated 14 December. A February last-call notice describes the meeting as having occurred on 12–13 December, while another AFRINIC event page describes 13–14 December. These sources establish a Cairo process in December 2005; taken together, they do not justify an unqualified exact day.

The substantive record is spare. The AFRINIC-3 minutes include IPv6 allocations from IANA to RIRs among proposals said to have received consensus. They say the proposals would return to the mailing list for 15 days before final Board approval. The later proposal-specific last-call summary is blunter: there was no discussion, yet the policy realised consensus. The checked material provides no proposal-specific speaker list, argument, objection, support count, show of hands, ballot or attendance denominator. It does not explain how consensus was assessed when there had been no discussion of this proposal.

That does not prove there was opposition, manipulation or an invalid result. It establishes an evidence boundary. Consensus is the process characterisation preserved by AFRINIC; its breadth and reasoning cannot be reconstructed from the available record. An open room is not a denominator. Silence is not a count. A label applied after a meeting is evidence of what the process custodian concluded, not independent evidence that every affected operator had spoken, consented or appointed the participants as representatives.

The February mailing-list exchange sharpens this distinction. On 9 February 2006, AFRINIC staff opened a 15-day last call on four proposals. The notice initially said the community had consented and voted, while its proposal-specific line said there had been no discussion and that consensus was realised. Two days later Alan Barrett, the proposal author and a participant, corrected the description: there had been no vote, although consensus appeared to exist.

On 13 February, Alan Levin offered a contrary memory cautiously rather than definitively. He thought there might have been a vote before the discussions and recalled only two voting members being present. Adiel Akplogan, AFRINIC’s chief executive, then answered that there had been no vote and that policy was not a voting matter under the applicable policy-development process; adoption rested on consensus. The fairest account preserves all three epistemic positions. Barrett issued a contemporaneous correction. Levin recalled uncertainty and a strikingly small possible voting group.

Akplogan gave the final first-party confirmation that no vote occurred. None of them supplied a measurable consensus denominator.

The correction matters because later readers could otherwise import the certainty of a ballot into a process that explicitly did not use one. It also matters because the only public 13–0 tally in this story belongs elsewhere: to the ICANN Board’s September decision, four months after AFRINIC’s resolution. It is not an AFRINIC vote and cannot be used to fill the gap in AFRINIC’s Board record.

The last-call close contains another small but instructive conflict. AFRINIC’s current history gives 9–24 February 2006. The opening notice promised expiry after 15 days, which is consistent with 24 February by elapsed-day arithmetic. A staff message sent on 7 March says the last call expired at noon on 25/02/2005 and that the Policy Working Group would forward the proposals to the Board. The year 2005 conflicts with the message date and the whole surrounding sequence; the day also differs by one. The honest record is therefore a 9 February opening and a reported 24 or 25 February close, with an apparent year error in the later notice.

Correcting it invisibly would erase evidence about the quality of the archive.

What the Board did—and what its page cannot show

The regional process did not make the policy final for AFRINIC. The minutes and the closing notice both point to a distinct corporate gate: delivery to the Board for approval. Resolution 200605.27 records that gate. After hearing the Policy Working Group chair’s report on open proposals, the AFRINIC Board ratified the four listed instruments, including AFPUB-2006-v6-001, and told staff to take the necessary implementation action. Within the corporation and the inter-institution procedure, that was a consequential act.

The public receipt is nevertheless incomplete. The Board page does not name the chair in the resolution, publish the chair’s report, or provide a proposal-specific Board paper. It does not list attendance for this act, establish quorum, identify a mover or seconder, reproduce a motion, state whether conflicts were declared, give individual votes, record abstentions or publish a tally. It supplies no item-specific reasons. Nor does it resolve whether the four proposals were an indivisible bundle. A signed record tying each director to the act has not been recovered from the checked material.

Those absences do not warrant inventing a defect. A private corporation may validly approve an internal policy using the method its rules permit, and the surviving resolution is affirmative evidence that AFRINIC recorded approval. The absences do, however, cap what can responsibly be claimed. The record cannot sustain a detailed account of Board deliberation, demonstrate unanimity, or establish that a measured constituency instructed the directors to act. Institutional validity within the corporation and public authorization by everyone affected are different propositions.

AFRINIC’s current history marks the proposal implemented on 17 May 2006, two days after the Board-adoption date in ICANN’s table. The page does not identify the concrete staff action completed that day, its owner, its acceptance test or its evidence. Regional “implementation” also cannot mean that IANA had already begun operating the final global policy: the NRO, ASO and ICANN stages still lay ahead. The label is evidence of AFRINIC’s reported internal state, not a substitute for an execution receipt.

This distinction is especially important because the Board instruction had two faces. Internally, staff could update AFRINIC’s policy posture, participate in the common-text chain and prepare for operation. Externally, neither the instruction nor the implementation label could command IANA, bind another RIR or impose obligations on an operator outside the relevant contractual relationships. AFRINIC’s authority over its own action was not authority over the whole chain.

Necessary assent, separate acts

The institutional logic came from the then-operative 2004 ASO Memorandum of Understanding between ICANN and the NRO. It defined a global number-resource policy as one agreed by all RIRs through their own processes and by ICANN, where implementation required action by IANA or another ICANN-related external body. Its procedure required common text, ratification by each RIR using a method of its choosing, referral by the NRO Executive Council, review by the ASO Address Council and ratification by the ICANN Board.

That design made AFRINIC’s assent necessary but not sufficient. Without a regional ratification, the text could not truthfully advance as one accepted by all RIRs under the stated process. With AFRINIC’s ratification, the proposal still could not skip the other regional dispositions, the NRO referral, the ASO process review, ICANN’s decision or IANA’s operational act. Necessity conveyed procedural leverage, not a sovereign veto over the Internet.

The later sequence is traceable. ICANN’s background report says that all five RIRs had formally adopted the second version by 16 May 2006. The NRO Executive Council submitted the proposed policy to the ASO Address Council on 6 June. The Address Council adopted it on 12 July and its secretary forwarded it to ICANN on 13 July. These were reported coordinating and review acts distinct from AFRINIC’s May ratification.

On 7 September, the ICANN Board ratified the global policy, directed staff to implement it, required timely allocation reporting with the RIRs and requested an evaluation of the effects within three years. Thirteen directors present voted yes, none voted against and none abstained. That clear 13–0 receipt demonstrates what a public tally can establish—and why it must remain attached to ICANN’s September meeting rather than being retrofitted to AFRINIC’s May resolution.

On 3 October 2006, IANA’s global unicast registry recorded 2c00::/12 as allocated to AFRINIC. The same table records /12 allocations to APNIC, ARIN, LACNIC and RIPE NCC on that date. It also preserves an earlier, smaller AFRINIC entry, 2001:4200::/23, dated 1 June 2004. The October entry is the direct operational outcome relevant to the policy chain. It demonstrates that IANA recorded a coordinated top-level allocation after the separate institutional approvals.

It does not demonstrate that Resolution 200605.27 itself allocated 2c00::/12. The checked sources do not include AFRINIC’s request, IANA’s review file, the decision time, the implementation correspondence or a legal theory of title. Sequence supports a bounded causal account: AFRINIC completed a required regional stage; later actors completed their own stages; IANA then made and recorded the allocation. Collapsing those acts into a single AFRINIC decision would misstate both responsibility and control.

Nor is agreement repeated across private institutions equivalent to a public delegation. The NRO Executive Council could coordinate and refer the common text. The ASO Address Council could review the reported regional process and forward or return it. The ICANN Board could ratify within the agreed framework and direct ICANN staff. IANA could execute an allocation and update the top-level registry. Each actor had a real function. Passing the same proposal through them did not manufacture sovereign power that none of the instruments granted.

The operator value was real

The strongest defence of AFRINIC’s decision begins with engineering rather than political metaphor. Unique numbering requires reconciled top-level records. Five unrelated, case-by-case bargains would invite inconsistent standards, strategic delay and opaque discretion. An equal /12 minimum, objective replenishment thresholds, an 18-month planning horizon and published allocation notices gave each region a common frame. Separate regional stages also created opportunities for people closer to operators to identify incompatibilities before the text reached the global level.

On that view, demanding a state-like electorate for every narrow registry-supply rule would confuse administration with legislation and could make dependable inventory impossible. AFRINIC’s Board performed a practical and legitimate internal approval in a private technical system. The formula constrained some discretion. The multi-stage chain enabled review. The October registry entry shows that the system could convert the common rule into coordinated action. These are substantial concessions, not rhetorical preliminaries.

Operators and resource users had an economic interest in that result. Predictable upstream stock reduced the chance that AFRINIC would face a last-minute shortage while processing regional registration demand. A large, usable allocation lowered the administrative danger of depending on fragments. A written special-needs route made exceptional forecasts capable of being examined rather than negotiated entirely out of view. Public announcements helped operators and neighbouring institutions reconcile expectations with the top-level ledger.

The LARUS operator-risk perspective helps explain why provenance matters even when operators never see the IANA request. Registry decisions can become infrastructure risks when records, authority boundaries or continuity arrangements are unclear. That is a present-day analytical lens, not evidence about who spoke in Cairo or what the AFRINIC directors discussed. Applied carefully, it shows why a consequential ledger position calls for better audit, correction and continuity controls. It does not turn registry consequence into a claim of public jurisdiction.

NRS likewise offers a present-day explanation of the IANA-to-RIR hierarchy, the role of large IPv6 units, the fact that authority stops at contractual boundaries and the way global coordination remains constrained by local law. The contemporary instruments remain controlling for the exact 2005–2006 chronology and mechanics. NRS did not write AFPUB-2006-v6-001, call consensus in Cairo, ratify the resolution or make the IANA allocation. It researches, advocates, convenes and represents members who explicitly authorise it; it does not operate a registry, IANA, RPKI, WHOIS or RDAP, elections, appeals, settlement, custody or continuity functions.

The operator case therefore strengthens, rather than weakens, the need for narrow institutional language. Network operators build, route and serve customers; they bear the costs of delay, mistaken records and service interruption. Yet the checked sources do not show that all operators, most operators or any measurable share authorised the Cairo characterisation or Board act. Economic exposure makes them affected principals. It does not make meeting attendees their agents without evidence of appointment.

Coordination without constitutional inflation

Heng Lu’s uniqueness-coordination doctrine supplies the controlling distinction. A registry performs a valuable narrow service by keeping non-duplicative records and coordinating issuance. Accurate bookkeeping can be essential to interoperable networks. But the importance of that service does not make the bookkeeper sovereign. A service region is not territory; an allocation entry is not political title; a successful formula is not a constitution.

The related principal-versus-stakeholder analysis applies directly to the Cairo record. Participation, mailing-list availability and a consensus declaration are meaningful process inputs. Without a defined principal, a response denominator and evidence of authorization, they cannot be expanded into a mandate over absent operators. Otherwise, attendance becomes representation by assertion and repetition of the claim across institutional layers launders a private process into a public power it never received.

BTW’s analysis of a contractable IANA function reinforces the same point from the top of the chain. The numbering function can be specified, audited and operated as a narrow technical service without political ownership of the resource. Its companion method distinguishes delegation, recognition and certification instead of treating every upstream institutional act as the same kind of authority. Here the IANA registry entry proves an allocation record and operational reliance. It does not, without a separate instrument, prove ownership, jurisdiction or a downstream right to coerce.

NRS, Heng Lu, LARUS and BTW are first-class present-day sources in this analysis: respectively, they illuminate the allocation hierarchy and contract boundary, the controlling coordination-versus-authority doctrine, operator-continuity exposure, and the distinction between technical service and political ownership. None is used as evidence that it authored the 2005 common text, called the Cairo consensus, ratified AFRINIC’s 2006 resolution or made the IANA allocation. Those historical propositions rest on the contemporary institutional record.

This is the bounded answer to the strongest defence. The coordination achievement was real. AFRINIC’s internal approval could be valid and necessary within the agreed procedure even without universal operator authorization. But private process validity cannot supply sovereignty. The sparse proposal-specific and Board records cannot bear later claims that the act represented every African network, government, user or resource holder. A working allocation system deserves accurate credit, not constitutional inflation.

AFRINIC’s assent was therefore a useful private coordination act whose scope ended at corporate, procedural, contractual and technical functions. It created no sovereign, legislative, regulatory, police, prosecutorial, judicial, punitive or confiscatory authority. The Board could approve a policy for the corporation, accept responsibilities in the RIR coordination stack and direct staff. It could not legislate for states, prosecute a network operator, adjudicate public rights, punish nonparties, confiscate resources or convert a service relationship into rule over a polity.

For this analysis, AFRINIC was and remains a private uniqueness-ledger administrator and coordinator. Supply and registration rules do not create enforcement power outside contract and law. States and courts retain public-law authority. Operators retain their own economic and legal spheres. Where a contract creates rights and remedies, those rights arise from the contract; where public law applies, authority comes from the competent public institution—not from the mere operational importance of an address registry.

The October 2c00::/12 entry illustrates the distinction cleanly. It increased AFRINIC’s operational leverage because networks depended on accurate administration of a large unique block. That consequence supports stronger recordkeeping, portability, audit and correction. It does not make the prefix AFRINIC’s sovereign property, an allocation to an African political community, or proof that downstream access was equal. Recording uniqueness is an executable act; community ownership and territorial mandate are symbolic claims unless connected to a valid instrument.

Official institutional language must be read with the same discipline. AFRINIC, NRO, ASO, ICANN and IANA records are first-party evidence of their documents, dates, decisions, registry entries and self-descriptions. Claims that a process was bottom-up, fair, non-discriminatory, representative, mandated or historically important remain claims of the institution making them. They may describe sincere design goals. They do not independently prove public legitimacy or authorization by all affected people.

What the record establishes—and leaves open

The evidence is strongest where the act and text are concrete. It establishes the proposal identity and upstream scope; the /12 unit; the 18-month horizon; the two replenishment tests; the stock and need formulas; the evidentiary route for special forecasts; announcement duties; the express performance omission; the reported Cairo consensus; the no-vote correction; the last call; the separate Board ratification; the later global stages; and the IANA registry outcome.

It is weaker on human and corporate accountability. No checked item provides the number of Cairo participants who considered the proposal, a list-subscriber or response denominator, or a proposal-specific support and objection count. No source explains how consensus was tested without proposal-specific discussion. The chair’s report to the Board is unavailable. Attendance, quorum, motion, mover, seconder, conflicts, votes, tally, abstentions and reasons for Resolution 200605.27 are unreported. The record does not show whether the bundle was procedurally indivisible.

The documentary defects must remain visible. The Cairo dates conflict. The last-call close is 24 or 25 February, and the March notice prints 2005 where the surrounding chronology requires 2006. The current policy page contains an inconsistent 17 May 2005 details date. Its 17 May 2006 implementation label lacks an action receipt. Although ICANN reported adoption of a second common version in every region, the checked record contains no cryptographic hash or complete region-by-region redline proving byte identity.

The operational receipt is incomplete too. The AFRINIC request for 2c00::/12, IANA’s review reasoning, decision timing and correspondence have not been recovered. The policy contains no IANA service-performance standard. No source establishes property title or sovereign jurisdiction arising from the allocation, and none establishes authorization by every affected African operator, member, government, user or resource holder.

These are not invitations to fill gaps with suspicion. They are limits on inference. The appropriate conclusion is neither that nothing happened nor that the process created unlimited authority. AFRINIC made a documented corporate assent; the common rule later passed through distinct institutions and produced a recorded allocation. The missing receipts affect how confidently one can describe participation, accountability, implementation and legal effect. They do not erase the technical result, and the technical result does not erase the missing receipts.