Summary

  • Resolution 201111.138 records the AFRINIC Board's internal ratification of AFPUB-2010-v4-005-draft-05 after a 1,048-day documentary path. The resolution proves what the private company said it adopted; it does not by itself prove representation, legal mandate, consent, implementation or public-law authority.
  • The ratified mechanism divided exhaustion into two phases, reduced maximum request sizes, shortened planning to eight months, imposed a 90% prior-use test on existing applicants, attached a regional-use condition and reserved a /12. These were administrative rules for access to AFRINIC's unallocated inventory, not commands backed by sovereignty.
  • The most revealing entry in the policy's modification history removed IPv6 adoption plans and deployment as conditions for receiving IPv4 because members' technology choices were outside AFRINIC's purview. That correction draws the proper institutional line: a registry can coordinate unique numbers and ration its remaining pool, but it cannot govern an operator's architecture.

On 21 November 2011, the AFRINIC Board published a compact institutional claim. Based on reports submitted by the Policy Development Process Working Group co-chairs, it said, the Board resolved by consensus to ratify AFPUB-2010-v4-005-draft-05, the IPv4 Soft Landing Policy. The wording has the tidy finality of a closing entry in a ledger. Yet the object behind that sentence was anything but simple. It had travelled for 1,048 days—34 calendar months and 14 days—from the archive's first-post date of 7 January 2009 to the Board's stated ratification date.

The useful question is not whether three years of discussion sounds long enough to confer legitimacy. Duration is not authority, and repetition is not representation. The useful question is what, exactly, the final sentence purported to adopt. The answer lies in the clauses and modification history attached to the policy, not in the ceremonial force of the word “ratify”. Together they show an institution trying to administer a finite stock through smaller parcels, tighter utilisation tests and a shorter planning horizon.

They also preserve an unusually clear admission of institutional restraint: requirements concerning members' adoption plans and deployment of IPv6 were removed because their technology choices lay outside AFRINIC's purview.

That admission is the centre of the record. It separates two activities that are often blurred whenever scarcity creates pressure for broader control. One is registry coordination: keeping unique-number records, managing an unallocated pool, publishing transparent procedures and applying intelligible terms to requests for that pool. The other is technology governance: deciding which protocol an operator should deploy, on what timetable, and as the price of receiving a different resource. The first can belong to a private coordinator.

The second belongs to the operator, subject to its contracts and applicable law—not to a regional registry merely because the registry maintains a ledger.

The sentence and the missing file around it

Resolution 201111.138 identifies a precise ratification object: AFPUB-2010-v4-005-draft-05. AFRINIC's policy page, in turn, labels AFPUB-2010-v4-005 as ratified, dates it 29 July 2011 and names Douglas Onyango as author. The archive says draft-05 was posted to the website and the Resource Policy Discussion mailing list on 5 August. Those are two different published fields. The page date and posting-history date should not be collapsed into a single event, nor silently harmonised. Their coexistence is a small example of why documentary identity matters.

The resolution also says the Board acted on reports submitted by the PDPWG co-chairs. But the material available with the resolution does not reproduce, date or link those reports. Nor does the published sentence provide a Board notice, agenda, attendance list, quorum record, proposer, seconder, individual votes, abstentions, conflicts, legal advice, signatures, implementation order or effective time. “Resolves by consensus” is left undefined. It might describe unanimity, an absence of formal objection, a procedural convention or some other voting state. Without the underlying record, the phrase cannot choose among those meanings.

These absences do not prove misconduct or invalidate the decision. Missing evidence remains missing evidence. But they limit the proposition that the resolution can support. It is strong evidence of AFRINIC's official claim that its Board internally ratified a named draft. It is weak evidence for assertions the sentence does not contain: that every affected operator was represented, that members delegated public authority, that legal jurisdiction existed, that implementation followed, or that any particular clause was later enforced lawfully.

This distinction is elementary but consequential. A company's Board can close its own process. It cannot turn itself into a legislature through the vocabulary of ratification. A working group's declared consensus can inform a private administrative decision. It cannot turn a mailing list into a polity or manufacture sovereign consent from participants' silence. The archive therefore supplies an authority chain only as far as the documents carry it: co-chair reports are said to have reached the Board; the Board is said to have ratified draft-05. Beyond that point, authority must be shown rather than assumed.

The exact-instrument method associated with NRS is useful here because it refuses to treat a title as a substitute for a chain of authority. Which instrument was adopted? By whom? Under which private powers? On what record? With what limits? Applied to Resolution 201111.138, that discipline does not require hostility to common rules. It requires the rule-maker's claim to remain proportionate to the evidence. The instrument can govern AFRINIC's handling of its own unallocated inventory within the relationships that law and contract permit.

It does not make AFRINIC a government, regulator, police force, prosecutor, punishment authority, confiscator or adjudicator.

What draft-05 actually rationed

The policy cast a wide administrative net over the pool it described. Its exhaustion regime was written to apply to Local Internet Registries and End Users, and to all IPv4 space allocated, assigned or otherwise managed by AFRINIC during and after the transition, whether or not that space formed part of the final /8. That phrasing expanded the policy beyond a single inherited block. It was a statement of the policy's intended private scope; it was not a grant of sovereignty over the networks using those numbers.

The transition mechanism began when an otherwise valid request could not be fulfilled without drawing on the final /8, or when fulfilling it would empty the pool outside that final /8. Pending applications would then be handled under the new policy, and AFRINIC would announce exhaustion. The final text divided what followed into two neutral phases rather than clothing the stages in moral or political labels.

Phase 1 retained a /24 minimum for an End User and a /22 minimum for an LIR, while cutting the maximum request from /10 to /13. It continued until no more than a /11 of non-reserved space remained in the final /8. Phase 2 then set /24 as the minimum and /22 as the maximum for each allocation or assignment. Prefix notation can obscure the practical move: the policy reduced the maximum parcel available per approved request as scarcity intensified. It did not create new IPv4 supply. It distributed access to a finite remaining stock through smaller administrative lots.

Nor did the policy set an explicit limit on the number of times an organisation could request more IPv4 space during the exhaustion period. Repeat applications remained possible, but every request still had to satisfy the other published criteria. That combination—smaller maximum parcels without a fixed request-count ceiling—made the utilisation test and planning horizon important. It was not a simple one-block-per-organisation ration.

The planning period fell from twelve months to eight. In administrative terms, a shorter horizon can spread scarce inventory by asking applicants to justify a nearer-term need. In operational terms, it can also compel more frequent forecasting and applications, shifting transaction costs and timing risk to members. The text offered fairness as a rationale, but the archive does not prove that eight months was economically optimal. A design choice can be intelligible without being self-validating.

For existing applicants, the policy required at least 90% use of all prior allocations or assignments before another request. A genuinely new LIR or End User with no previous allocation or assignment was exempt on its first request. The threshold plainly affected growth timing, spare-capacity strategy and the cost of demand forecasting. Yet the public material considered here does not settle every measurement method, evidence demand or contested edge case. The figure is an exact criterion, not a complete adjudication manual.

The policy also said AFRINIC resources were for the AFRINIC service region and that any use outside the region should be solely in support of connectivity back to it. During the September 2011 last call, the co-chair notice described the movement away from an earlier 10% cap toward the connectivity-back wording. The discarded percentage was not part of the final ratified rule. The final clause reached towards topology and customer geography, but recording its text is not the same as conceding that a registry possessed regulatory, police or confiscatory power over network use.

Finally, section 3.9 withheld a /12 from the final /8 for unforeseen future uses. It also said that, if AFRINIC could no longer meet further requests, the Board could at its discretion replenish the exhaustion pool with address space then available to AFRINIC, taking demand and other factors into account and acting in what the text called the community's best interest. The /12 was an inventory reservation, not proof of a later use, benefit or lawful disposition. The replenishment language described a discretionary control surface, not a later decision.

The mechanism therefore combined exact numbers with open texture. The /13 maximum, /11 trigger, /22 Phase 2 maximum, /24 minimum, eight-month horizon and 90% test could be reproduced from the text. “Other factors” and “best interest of the community” could not. Scarcity magnifies the economic value of each discretionary decision. It therefore increases, rather than diminishes, the need for criteria, custody records, conflict controls and review. A benevolent phrase is not its own audit trail.

The line the policy learned to draw

The modification history records that Version 1 removed IPv6 adoption plans and deployment as requirements for receiving IPv4 space because members' technology choices were outside AFRINIC's purview. That sentence is more than an editing note. It identifies a category error and corrects it.

AFRINIC had a plausible coordinating interest in the depletion of an unallocated pool it administered. Scarcity can support common, transparent rules on request size, sequence, documentation and utilisation. It does not give the custodian a proprietary claim over an operator's entire technical roadmap. Requiring an IPv6 adoption plan or deployment as the administrative price of access to IPv4 would have leveraged control of one registry service to pressure a separate architectural decision.

The correction did not declare IPv6 unnecessary, unsuccessful, fraudulent or technically inferior. It made no universal engineering judgment. Operators could have compelling commercial, resilience or growth reasons to deploy IPv6. The point was institutional: whether and when they did so was their technology choice. Heng Lu's analysis of the “IPv6 escape from scarcity” narrative reinforces precisely this boundary. A registry cannot convert a forecast about address scarcity into an entitlement to command protocol adoption.

The final introduction still spoke of maintaining IPv4 networks while deploying IPv6 as part of a transition rationale. There is no contradiction once purpose and condition are kept separate. A document can explain that IPv6 deployment forms part of the environment motivating a policy while declining to make deployment an eligibility test. Background language describes why the authors thought rationing mattered. A compulsory criterion determines whether an applicant receives a resource. Confusing the two would reinsert, by interpretation, the very mandate the modification history removed.

That is why the deletion carries more institutional weight than the Board's compact claim of ratification. The resolution says that AFRINIC finished a private procedure. The modification record says where AFRINIC's competence stopped. The first is an assertion of internal action. The second is an admission of limit—and institutional limits are most credible when the institution recognises them in its own operative history.