Summary

  • AFRINIC-27 considered Draft 6, accepted two narrow amendments and moved the resulting Draft 7 into a fifteen-calendar-day Last Call; the Appeal Committee later accepted that meeting-stage call but rejected the separate conclusion reached after Last Call.
  • The co-chairs’ description of identical petition letters and one substantive multinational-company concern could be a legitimate exercise of rough-consensus judgement, but the record did not expose the participant denominator, unique positions, authorisations, material-objection map or item-by-item dispositions needed to test that judgement.
  • A durable remedy is a reasoned consensus receipt published before a Board recommendation: it should preserve chair discretion while proving what text was assessed, who or what positions appeared, which objections mattered, how each was answered and why the result remained a bounded private allocation rule rather than public law.

L3 — How Draft 7 crossed the Last Call line

Two official conclusions frame the problem. The first came from the co-chairs after the announced Last Call had ended. Their report treated a series of identical petition letters as organised repetition, isolated a concern about adverse effects on multinational companies, recorded a staff answer and recommended that the Board ratify the proposal immediately. The second came from the Appeal Committee. It found considerable opposition during Last Call, concluded that consensus did not exist during or at the end of that period, and held that the recommendation to the Board had been an error.

Both conclusions belonged to AFRINIC’s own private policy machinery. Yet they gave materially different accounts of whether the same review stage had cleared its procedural threshold.

That disagreement cannot be dissolved by saying simply that consensus is subjective. It is true that rough consensus is not a ballot and that chairs must exercise judgement. But a judgement that controls whether a scarcity instrument advances must leave enough evidence for another reader to understand its path. Here, the available official records tell us what each body concluded without supplying a complete, reproducible bridge between the Last Call corpus and the conclusion.

They do not disclose a session or list denominator, a count of unique participants, a complete taxonomy of materially different objections, or an itemised account of why each objection was resolved, repeated, immaterial or still blocking. The absence is important not because counting would automatically settle the matter, but because the later reversal shows that characterisation itself was contested.

The stable object under review must be identified first. On 30 November, at AFRINIC-27, the policy discussion concerned Version 6 of the Soft Landing BIS proposal. It did not concern an already published Version 7. The meeting minutes recorded mostly supportive statements and two refinements accepted by the authors. One removed a specific reference to XLAT translators from section 5.4.7.1. The other added IDN country-code top-level domains to the Core DNS provision in section 5.4.7.5.3. With those changes accepted, the co-chairs decided that the resulting text could proceed to Last Call.

The next day, 1 December, the new text was published as AFPUB-2016-V4-001-DRAFT07. The archived instrument identifies Omo Oaiya, Joe Kimaili and Alain P. Aina as its authors and records its status as under discussion. The Last Call notice announced the two modifications and opened a period of fifteen calendar days, scheduled to end on 16 December. This chronology matters. The face-to-face judgement concerned the Version 6 discussion plus two accepted refinements; the Last Call concerned the resulting Draft 7.

A claim that the meeting had supported the transition does not, by itself, establish what happened during the later and distinct review of the finalised draft.

The Appeal Committee preserved precisely that distinction. Having reviewed the meeting video, it found no opposition at the face-to-face session and agreed with the co-chairs’ conduct, evaluation and declaration at that stage. This was not a finding that every feature of the proposal was beyond dispute, nor a retrospective invalidation of the meeting. It was a conclusion that moving the instrument out of the room and onto the list was procedurally defensible. The Committee’s adverse finding concerned the end of Last Call: it said the list showed considerable opposition and that the proposal should not have been recommended to the Board.

The separation of stages prevents two opposite mistakes. One would be to treat the later successful appeal as proof that the meeting call was itself improper. The official review did not say that. The other would be to treat the favourable meeting call as a mandate carried automatically through Last Call. The process described by the co-chairs on 7 December treated these as separate decisions. Following face-to-face consensus, the list had an opportunity to examine the resulting text; only after consensus at that later stage was the proposal to proceed with a discussion and feedback report to the Board.

If the second stage could add nothing to the first, its announced fifteen-day duration would have been ceremonial.

The substance being tested explains why the distinction had economic weight. Draft 7 proposed to replace section 5.4 of AFRINIC’s Consolidated Policy Manual and govern allocations and assignments from the registry’s final /8, 102/8, during exhaustion. It would have imposed a /24 minimum in both phases. In Phase 1, the maximum would have been a /18; in Phase 2, a /22. Phase 2 would begin when no more than a /11 of non-reserved space from the Final /8 remained. These are not merely different notations. A /18 contains 16,384 IPv4 addresses, a /22 contains 1,024, a /24 contains 256 and a /11 contains 2,097,152.

The change in phase therefore sharply altered how much an applicant might receive.

Other clauses shaped time as well as quantity. Draft 7 used an eight-month planning window and required justified and demonstrated need. It proposed a 90 per cent prior-use test, while exempting a new Local Internet Registry or End User’s first request from that prior-use condition. It also capped what an organisation could receive in any twenty-four-month period at the equivalent of a /18 in Phase 1 and a /22 in Phase 2.

Taken together, the planning horizon, utilisation test and recurring total would affect when an applicant could return, how much growth could be supported from the unallocated pool, and how operators might plan around the prospect of transfers or other alternatives.

The reserve provisions made the control surface broader still. Draft 7 proposed a contiguous /12—1,048,576 addresses—to facilitate IPv6 deployment. Within that reserve route, a qualifying organisation could receive a one-time /24. The text included staff discretion over justification, a six-month condition tied to non-receipt, and a requirement to show that earlier IPv4 space could not meet the need. It also contained the Core DNS provision amended to include IDN ccTLDs.

These details meant that the proposal would govern not only an ordinary ceiling but a distinct route into a large reserved block, with eligibility, timing and staff assessment built into it.

None of that proves the draft was adopted, implemented or applied to an applicant. The official archive’s labels are not enough to establish such an outcome: one current proposal-history presentation uses a Last Call label, while the backup archive records Draft 7 as under discussion. The evidence establishes a proposal, a meeting-stage decision, a Last Call, a recommendation and an appeal result. It does not establish Board ratification or operational use. The Board recommendation was a gate, not the destination.

Keeping that boundary intact is essential to serious institutional analysis, because exaggerating the outcome would obscure the real defect: the evidence for crossing the gate was not publicly reproducible.

The 7 December discussion also requires careful dating. In explaining the process, the co-chairs referred to the eight weeks after publication of Draft 6 and said that period had produced one comment and a +1 concerning restriction of the IPv6 reserve to new members. They also said they had observed objections from people who preferred removing soft-landing restrictions. Those remarks may illuminate the climate before AFRINIC-27, but they do not measure the complete corpus produced during the later fifteen-day Last Call on Draft 7. A pre-meeting eight-week observation and a post-meeting Last Call are different evidence windows.

Conflating them would make the later objections disappear into an earlier baseline.

By 26 December, the co-chairs’ report characterised the Last Call record more narrowly. It said identical petition letters had been rallied, identified one substantive objection concerning effects on multinational companies and recorded a staff response that multinationals could apply through separate entities. It then recommended immediate ratification. The report is evidence of the reasoning the co-chairs chose to disclose. It is not an independently verified census of positions, and the reported staff answer is not accompanied here by a separate assessment of its legal, cost or operational implications.

A practical answer can be relevant without necessarily resolving every version of the underlying concern.

The Appeal Committee’s report supplies the institutional counterweight. It worked from an RPD discussion summary, meeting minutes, Last Call material, the co-chairs’ assessment and the Board recommendation. It did not rule on whether /18 or /22 was the better scarcity ceiling, whether the reserve design was economically optimal, or whether the multinational concern should prevail as policy. Its conclusion was procedural: considerable opposition existed in the Last Call record; consensus was absent during and at the end of that stage; the co-chairs should not have recommended Draft 7 to the Board; and the appeal therefore succeeded.

It corrected a private process without converting itself into a court or the policy forum into a legislature.

The word “considerable” remains qualitative. The Committee did not enumerate every objection or announce a numerical threshold. Meanwhile, the co-chairs’ references to identical petitions and a single substantive concern also leave unresolved how many independent principals stood behind the messages, what authority any writer claimed to represent an organisation, and whether similar wording concealed different operational circumstances. Neither description provides the evidence needed to reproduce its categorisation. One side compresses the record; the other says opposition was considerable.

The resulting contradiction is the heart of the case, not a reason to invent a number that neither report supplies.

The apparent neatness of the two final edits should not distract from the wider instrument. Removing an XLAT reference and adding IDN ccTLDs may have been narrow, but Last Call was on Draft 7 as a whole. Participants could evaluate the full scarcity architecture, including provisions inherited from the prior text. Nor does the fact that the proposal had undergone seven versions over nearly two years mean that all surviving objections had expired. Revision history can demonstrate effort and responsiveness. It cannot predetermine whether the final text earned consensus in a newly opened stage.

Scarcity strengthened the need for this distinction. When a private registry administers a finite, unallocated pool, phase triggers, queue timestamps, request ceilings and reserve eligibility affect operating options. A /18-to-/22 transition is a sixteenfold reduction in the maximum block size. A twenty-four-month total can constrain repeated access. An eight-month planning horizon can shape capital and network decisions. The /11 threshold decides when the tighter phase begins. Those consequences do not turn AFRINIC into a sovereign owner of IPv4 capital.

They do mean that the private judgement advancing the rule should be legible to those whose planning is affected.

The L3 conclusion is therefore deliberately narrow. Draft 7 was an identifiable scarcity instrument. The transition from Draft 6 was visible: two accepted changes, publication on 1 December and a stated fifteen-day review. The meeting-stage call was later upheld. The end-of-Last-Call call was later reversed. What is missing between those facts is not necessarily a ballot result or evidence of wrongdoing.

It is the record that would let an independent reader see how the Last Call’s messages became a set of positions, how those positions became objections, how the objections were answered and why the remaining disagreement did or did not block a Board recommendation.