Summary

  • SL-Update Draft 1, published on 28 October 2018, proposed a limited amendment to CPM 5.4.7.2. Publication did not adopt, ratify, activate or implement the proposal.
  • The draft changed the reserve-release trigger and removed the Board’s discretion within that clause. If the reserved /12 was still unused after the remaining available space had been allocated, the reserve would return automatically to the pool.
  • Draft 1 attached that returned /12 to the existing Phase 2 allocation conditions. It did not numerically change the Phase 2 minimum of /24 or maximum of /22 per allocation or assignment.
  • The /12 equals 1,048,576 addresses and can be divided arithmetically into 1,024 /22 blocks, but that is only a theoretical block count—not evidence of recipients, approvals, actual release or realised economic effects.
  • Constraining a private registry’s discretion can matter greatly for predictable access to a finite resource. It does not turn AFRINIC, its Board, or an amorphous “community” into a sovereign or public authority.

L3 — The one-paragraph switch: trigger, reserve and inherited ceiling

The cleanest way to understand SL-Update Draft 1 is to put the two decision structures side by side. The then-current clause was a fallback for exhaustion. When AFRINIC could no longer satisfy requests from the final /8 or other available address space, its Board could decide whether to move all or part of any available address space into the exhaustion pool. In making that decision, the Board could consider demand and other factors; it could also determine the rules by which the replenished pool would be allocated or assigned.

The clause therefore left three linked choices open at the critical moment: whether a reserve should be released, how much should be released, and which distribution conditions should apply.

Draft 1 proposed a narrower, pre-committed mechanism. Its subject was the reserved /12. If that /12 remained unused at the point when the remaining available space had been allocated, the reserve would return to the AFRINIC pool. It would then be distributed under Phase 2 conditions. The proposal did not redraw the whole Soft Landing system. It did not replace all of its phases, reserves or request rules. It replaced the decision structure in CPM 5.4.7.2, one paragraph whose practical role was to determine what happened to a particular reserve after the ordinary stock ran out.

That redline changed four connected features. First, it changed the trigger. The old trigger began with an operational inability to meet further requests and then opened a decision. The proposed trigger was more exact: remaining available space had been allocated, while the reserved /12 was still unused. Second, it changed the decision owner. The Board would no longer decide within this clause whether a release should occur. The text itself would supply the instruction. Third, it changed the character of the decision from discretionary to automatic. If the conditions were satisfied, the draft said what should happen next.

Fourth, it changed how the distribution rule was selected. Instead of leaving the Board to set future rules for the replenished pool, the proposal incorporated the Phase 2 conditions already found elsewhere in the policy.

The word “incorporated” is essential. It prevents the article from making the most tempting factual error in this record. The Phase 2 envelope at the time already prescribed a minimum request size of /24 and a maximum of /22 per allocation or assignment. Draft 1 changed the reserve trigger and attached the existing Phase 2 envelope to the returned space; it did not numerically change the /24 minimum or the /22 maximum. The proposal’s significance cannot be measured by pretending it invented a new ceiling. It took an existing ceiling and made it govern a newly specified path by which the reserved /12 might re-enter the pool.

This distinction is more than drafting etiquette. A numerical amendment and an incorporation by reference solve different problems. A numerical amendment asks how much one approved allocation or assignment may contain. Incorporation asks which already-defined rule set applies when a particular stock becomes available. Draft 1 answered the second question. Any account that says it reduced, raised or established the /22 maximum assigns to the proposal a change that its text did not make.

It also obscures the real institutional choice: the authors were trying to prevent a later Board from deciding both whether to release the reserve and what conditions to attach to it.

The reserve’s scale helps explain why such a short clause deserved scrutiny. IPv4 prefix arithmetic gives a /12 a total of 1,048,576 addresses. A /22 contains 1,024 addresses. On paper, therefore, a /12 can be divided into 1,024 blocks of /22. That calculation is transparent and useful as a measure of the stock governed by the clause. It is not a forecast. It does not show that the /12 was released; it does not show that 1,024 requests were approved; and it certainly does not identify 1,024 organisations.

The policy did not limit every organisation to a single request, and real administration need not produce a neat sequence of equal maximum-sized blocks.

The arithmetic also says nothing by itself about utilisation, eligibility, timing, fragmentation or the size mix of approved requests. A qualified requester could seek an allocation or assignment within the Phase 2 range, subject to the applicable evaluation. Some requests could be smaller than /22. One organisation could not simply be assumed to correspond to one block. Nor can a theoretical partition establish the number of networks that would benefit, the period over which distribution would occur, or whether the reserve would ever be returned in practice.

The million-address scale reveals the maximum stock at issue; it does not manufacture transaction evidence that the source record lacks.

The publication record provides a firm identity for this proposal. AFRINIC identified it as SL-Update, version 1.0, Draft 1, submitted on 28 October 2018 by Gregoire O. Ehoumi, Arnaud A. A. Amelina and Adeola Alain Patrick Aina. The public proposal identifier appeared as AFPUB-2018-V4-001-DRAFT01 and AFPUB-2018-v4-001-DRAFT-01; the two forms identify the same proposal, not separate drafts. Its stated target was CPM 5.4.7.2. The revision history contained the October entry, and the 2018 annual record treated the proposal as under discussion. These facts keep the subject bounded: the event under analysis is publication of a draft amendment, not a later revision and not a general retelling of the Soft Landing policy.

Publication status matters because institutional processes generate several dates that are easy to collapse into one. On 28 October 2018 there was a proposal available for consideration. There was no consensus finding, Board ratification or implementation on that date. At AFRINIC-29 on 29 November, the proposal was presented, but the co-chairs concluded that it needed more discussion and returned it to the mailing list. That was a continuation of deliberation, not approval. Treating the November meeting as adoption would be as inaccurate as treating the October publication as an operative rule.

The next recorded step came on 19 June 2019 at AFRINIC-30. The minutes said there had been no intervening input or refinement to the proposal since the previous meeting. The same version was presented, and a co-chair declared consensus and moved it to Last Call. This is relevant later-lifecycle evidence because it shows the draft eventually crossed a process threshold without a recorded textual refinement. It cannot be projected backwards onto October 2018. The publication event remains a proposal event even though the same text later advanced.

Ratification followed in August 2019 through Board Resolution 201908.513. The official records do not give a perfectly unified exact meeting date: the Board minutes are dated 23 August, while AFRINIC’s public notification says the relevant meeting occurred on 26 August. Nothing in the present analysis requires choosing one over the other, so the responsible statement is that ratification occurred in August 2019 while preserving that 23/26 August documentary discrepancy. AFRINIC then announced on 6 February 2020 that the policy had been implemented in CPM 1.4.

That announcement belongs to 2020, not to the publication, presentation or Last Call stages.

This chronology imposes a hard boundary on the verbs available to an analyst. AFRINIC published or received the draft in October 2018. Meeting participants discussed it in November. A co-chair recorded consensus and moved it to Last Call in June 2019. The Board ratified it in August. AFRINIC announced implementation the following February. “Proposed,” “discussed,” “advanced,” “ratified” and “implemented” are not interchangeable descriptions. Each one identifies a different institutional act with a different evidentiary record.

The staff assessment also belongs within that discipline. It can illuminate how AFRINIC staff read the operational scope of the amendment and whether the assessment recorded legal or implementation objections. It cannot make the proposal effective, and the absence of a recorded objection is not proof of public legitimacy or measurable success. An internal assessment is evidence about an institution’s evaluation of its own policy proposal. It is not an independent warrant for enlarging that institution’s status.

The narrow textual record supports a strong conclusion without exaggeration. Draft 1 proposed to replace a discretionary contingency with a rule-bound contingency. It identified the stock, specified the release condition and selected the applicable distribution regime in advance. Those changes could affect expectations about access to a substantial pool. Yet the draft neither created the Phase 2 numerical endpoints nor demonstrated any later distribution. Its constitutional interest lies in the placement of discretion: a decision formerly left to the Board would be settled by prior text if the stated condition arose.

That is also why this episode should not be merged with broader proposals that pursued different changes to Soft Landing. Other drafts elsewhere in the policy history considered wider redesigns, different maxima, named reserves or replacement strategies. Those are separate policy events. Importing their features here would convert a precise account of CPM 5.4.7.2 into an inaccurate history of everything AFRINIC debated during scarcity. The evidentiary advantage of this proposal is its surgical nature. One can see exactly what the authors wanted to remove, exactly what they wanted to insert, and exactly what they left untouched.

The operative text therefore yields a concise test for every claim made about it. Does the claim concern the unused reserved /12? Does it distinguish allocation of the remaining ordinary space from return of the reserve? Does it recognise that Phase 2’s /24-to-/22 range already existed? Does it preserve the difference between a publication and a later effective policy? Does it avoid treating theoretical CIDR arithmetic as an observed allocation ledger? If an account fails any of those tests, it is not merely imprecise; it changes the institutional event being described.

At L3, then, the finding is exact. Draft 1 was a proposal to alter the sequence by which an unused reserve could become distributable. It did so by replacing a Board-controlled fallback with an automatic textual trigger and by importing the existing Phase 2 conditions. The stock was large enough for the choice to matter. The documentary record is clear enough to separate the proposal from its later lifecycle. What remains unknown is equally important: whether the reserve was ultimately released in practice, who received any part of it, how many transactions followed, and what effects those transactions produced.