Summary

  • The contemporaneous minute records consensus on progressing AFPUB-2010-v4-005 to Last Call. It does not record final adoption at AFRINIC-13, and later events confirm that substantial process remained.
  • The result can be reconstructed with unusual precision as an amendment ledger: broaden exhaustion-phase scope beyond the Final /8; rename the two sub-phases; clarify that the /22 maximum applied per allocation rather than once per organisation; retain the /27 minimum despite a routability concern; then have the author incorporate the instructed changes for later review.
  • The archive dates that incorporation to 25 January 2011, two months after the meeting. The sealed record contains no character-level redline from the 18 November text to the January draft, so it supports a bounded procedural mandate but cannot prove exact textual fidelity.
  • A same-meeting comparator matters. Another IPv4 proposal had rough support yet did not advance because its meeting text differed from the version posted before the one-week deadline. AFRINIC-13 therefore treated document identity as outcome-relevant, which makes the post-meeting amendment custody of Soft Landing an audit question rather than a drafting footnote.
  • The legitimate role at issue was private coordination of unallocated inventory and service eligibility. AFRINIC’s useful ledger function did not make it a sovereign, regulator, adjudicator or owner, and meeting participation did not confer authority over absent operators, resource holders, users, states or Africa as a whole.

The verb that defines the event

The most important word in the 25 November 2010 record is not “consensus.” It is “progressing.” The interim co-chairs determined that there was consensus on progressing the Soft Landing proposal to Last Call. The object of agreement was thus a procedural movement, not a completed instrument. The proposal had a name, AFPUB-2010-v4-005; it had been presented in the meeting; it had a recent publication history; and it was accompanied by a set of directions arising from the room. But the meeting did not end with a claim that every line had become final policy.

That grammatical precision is institutional precision. “Adopt” would describe completion. “Progress to Last Call” describes the opening of another review stage. The process text itself makes Last Call consequential: after chairs determine rough consensus at a public meeting, they initiate a Last Call of at least two weeks and then assess the feedback from both the meeting and Last Call before deciding whether consensus has been achieved. On that design, Last Call is not ceremonial applause for a settled result. It is part of the evidentiary basis for a later determination.

The archive uses a looser retrospective shorthand, saying that the proposal “finds consensus” at AFRINIC-13. Read alone, that phrase can suggest a completed communal verdict. Read against the contemporaneous minute, it must be narrowed. The minute states the act more exactly, and it records that the room also agreed changes which were still to be incorporated. Archive shorthand is useful for locating the episode; it is not a licence to enlarge the episode.

This distinction also survives the later chronology. The archive records further revisions, a no-consensus conclusion after a May 2011 Last Call, renewed meeting consensus at AFRINIC-14 subject to further changes, another Last Call, a final consensus declaration in November 2011 and Board ratification on 21 November 2011. Those later steps did not retroactively alter the decision made at AFRINIC-13. They show why that decision was intermediate. Nearly a year separated the November 2010 meeting from ratification, and the path between them was neither automatic nor empty.

The correct description is therefore neither that nothing happened nor that a policy was adopted. Something important happened: a private policy forum authorized the next stage for an identified draft under bounded amendment instructions. The distinction matters precisely because intermediate acts can redirect capital and expectations even before they become operational rules. A procedural green light shapes which text receives institutional momentum, which objections must fight uphill, and which assumptions become the baseline for the next conversation. That is significant power, but it is not sovereignty and it is not finality.

The room, the record and the absent denominator

The official minutes place the Public Policy Meeting on 24 and 25 November 2010 at the Protea Hotel Balalaika in Sandton, Johannesburg, South Africa. They name Alan Barrett and S. Moonesamy as the interim PDWG discussion leaders and say that proposal discussions followed AFPUB-2010-GEN-005. The Soft Landing author was absent, while Alain Aina presented AFPUB-2010-v4-005. These details identify the institutional setting and the people recorded as performing particular functions. They do not establish a legislature, an electorate or a public mandate.

The minute gives no attendance denominator for the Soft Landing session. It gives no eligible-participant denominator, no counts of support, opposition or abstention, no remote-participation denominator, no mailing-list denominator, no transcript and no issue-by-issue explanation of the method used to assess consensus. That absence does not prove that the chairs assessed the room badly. It also does not justify inventing a majority, unanimity, representative mandate or quantified balance. “Rough consensus” is the process’s recorded method; the sealed evidence does not permit it to be translated into arithmetic.

Participation has value without becoming jurisdiction. People in the room supplied technical knowledge, operational concerns and competing scarcity judgments. Graham Beneke supported the proposal while questioning the routability of a /27 or longer prefix. Alan Barrett, speaking as an individual, supported it while suggesting changes concerning phase names, retention of /27 and scope beyond the Final /8. Dr Viv Padayatchy asked whether exhaustion-phase applicants must already hold IPv6, and Alain Aina answered that a member without IPv6 resources could apply for them simultaneously.

Fiona Asonga and William Stucke supported the proposal as presented and also supported extending its exhaustion rules to address space received outside the Final /8.

These recorded interventions reveal what entered the discussion. They do not reveal who was absent, what absent operators would have said, or how many resource holders shared each position. A meeting room is not a demos. Technical participation can improve coordination, expose defects and identify implementable compromises, but attendance cannot manufacture authority over those who did not authorize the forum to speak for them. Africa is a region named in the service context, not a single political principal that a private meeting can embody.

The limitation is not an argument for ignoring the discussion. It is an argument for describing its force honestly. Participants helped a ledger coordinator decide how to administer service eligibility and unallocated inventory within private arrangements. Their input may be technically persuasive. The resulting coordination may be useful and widely relied upon. None of that turns AFRINIC into government, gives it ownership of operators’ networks, or converts a chair’s consensus assessment into law.

Official records establish what the institution said, wrote and did. They are strong evidence that particular questions were raised, that the co-chairs announced a particular outcome, and that a later draft appeared in the archive. They are not independent proof that the announced outcome represented every affected party, that each participant understood the same proposition, or that the text would produce the economic effects anticipated in the room. Those broader claims require reasoning bounded by the record, not institutional vocabulary treated as self-authenticating.

Reconstructing the exact consensus object

The best way to understand AFRINIC-13 is to treat the result as a composite object with five parts. First came an identified draft: AFPUB-2010-v4-005 as presented on 25 November, following archive-recorded author updates on 7 and 18 November and an AFRINIC online posting on 8 November. Second came the procedural act: progress to Last Call. Third came three affirmative amendment instructions. Fourth came one recorded non-change, the retention of the /27 minimum. Fifth came custody: the author was expected to incorporate the meeting modifications, after which the revised instrument would be exposed to the next review stage.

This object was more specific than a vote for a theme such as “prepare for IPv4 exhaustion.” It was also less complete than consensus on a character-perfect final document. The room was not merely endorsing scarcity management in the abstract, because it resolved questions about scope, naming and repeat allocations. Yet it could not have been assenting to the exact January wording, because that online draft did not appear until 25 January 2011. The legitimate mandate sits between those extremes: advance this proposal, changed in these named ways and not changed in that named way, for substantive Last Call review.

That middle category is easy to lose because institutional histories prefer compact milestones. A list saying “consensus at AFRINIC-13” is tidy. An amendment ledger is not. But the ledger is what preserves the actual allocation of authority. The room had authority within the private process to supply technical feedback and accept a bounded direction. The interim co-chairs had the procedural role of assessing rough consensus. The author retained mechanical custody of the text. AFRINIC published the proposal and its history. Last Call participants would test the revised result.

Later process, not the November meeting alone, would lead to eventual ratification.

Each link is narrower than the whole. The author’s power to incorporate agreed changes was not a blank cheque to redesign the proposal. The chairs’ determination to progress did not allow them to skip Last Call. AFRINIC’s publication role did not make its archive entry a sovereign act. Meeting support did not bind every operator. Eventual ratification did not prove that the November room had agreed to later wording. Keeping those functions separate is the difference between traceable coordination and retrospective mandate inflation.

The power chain begins outside the meeting with the IANA exhaustion trajectory and AFRINIC’s available inventory. It then passes through the proposal author’s draft custody, AFRINIC’s administrative publication, input from the list and meeting, the interim co-chairs’ limited process assessment, the author’s incorporation of specified amendments, and Last Call’s testing of the revised instrument. Eventual Board ratification occurred only after later stages. At the far end, resource applicants and network operators—not the meeting record—would carry routing, investment, continuity and customer consequences.

The chain explains why the most defensible reading is procedural rather than mystical. “The community” did not become a single author. Different people performed different tasks against a scarce resource background. Consensus compressed their discussion into a direction for the next stage. That compression becomes legitimate only when the object remains identifiable and every handoff preserves the recorded limits.

The first amendment widened the inventory in scope

The first agreed change was substantive. Exhaustion-phase rules would apply to all IPv4 address space available to AFRINIC during the phase, not only address space from the Final /8. Fiona Asonga and William Stucke are recorded as supporting coverage of space that AFRINIC might receive after exhaustion outside that particular block. Alan Barrett, speaking individually, also suggested scope beyond the Final /8. The room’s consensus converted that concern into an amendment instruction.

This was not a cosmetic repair. A rule framed only around the Final /8 could leave later-received IPv4 inventory outside the exhaustion regime. Broadening the scope changed which addresses would be administered under the scarcity rules. The amount of any such future inventory is not given in the sealed record, and the article cannot infer it. The economic direction is nevertheless clear: a definition of the covered pool determines which stock is subject to phase restrictions and allocation ceilings.

Scope clauses often look clerical because they operate through nouns rather than commands. Here the noun phrase answered a capital-allocation question. If scarce inventory acquired from another source sat outside the exhaustion phase, applicants might face different eligibility or size treatment depending on the administrative origin of otherwise usable address space. If all available space fell inside the phase, the restriction would follow the scarcity condition rather than the provenance of a block. The meeting chose the latter direction.

That choice illustrates why a private bookkeeper’s drafting discipline matters. AFRINIC did not own the networks that would use the resources, and it did not gain sovereign authority by defining the pool. It did have a narrow coordination task: maintain a unique ledger and administer unallocated inventory predictably within its arrangements. A scope ambiguity inside that task could create operational differences between applicants. Clarifying it was therefore legitimate coordination, provided the wording faithfully tracked the meeting instruction and remained open to the next process check.

The archive’s modification history says version 11 expanded the exhaustion scope beyond the Final /8 in response to AFRINIC-13 comments. That supports incorporation at the level of described purpose. The sealed set does not include a character-level redline from the 18 November draft to the January text, so it cannot establish that every word was the only possible implementation or that no other wording changed. The evidence proves the amendment direction and the archive’s account of its response; it stops short of textual forensics.

The second amendment repaired the phase names

The second agreed change renamed the two sub-phases “Exhaustion Phase 1” and “Exhaustion Phase 2.” Compared with scope or allocation frequency, naming can seem minor. But phase names are control surfaces. They determine how operators, applicants, staff and later reviewers refer to different rule states. A label that is ambiguous, misleading or unstable can detach an operational trigger from the obligations meant to follow it.

The archive’s modification history identifies the phase-name change as one of the version 11 responses to AFRINIC-13 comments. The significance lies less in literary style than in identity. If a later discussion says “Phase 2,” everyone must be able to know which condition and rules that label denotes. Naming supports traceability across the proposal, the archive and subsequent implementation. It reduces the chance that a policy debate shifts objects because different participants use different shorthand.

This amendment also helps locate the boundary between principle and text. Participants may have understood the two-stage model before the names were settled. Agreeing to rename the phases did not require them to rehearse every rule in each phase. Yet the fact that the room instructed a name change confirms that it was not simply approving the posted document without qualification. The text still had an assigned transformation to undergo.

Calling the change “editorial” would not eliminate the need for custody. Editorial changes can alter cross-references, defined terms and the clarity of triggers. The sealed record does not report such an error, and none should be invented. The narrower point is that even faithful renaming creates a new textual version. The author’s task was to produce that version; the archive’s task was to publish it; Last Call’s task was to test the resulting instrument rather than a remembered oral consensus.

The third amendment rejected a lifetime single-block ceiling

The third agreed clarification concerned the /22 maximum. The room resolved that the maximum applied independently to each allocation, with no limit on how many qualifying allocations the same organisation could receive. A /22 contains 1,024 IPv4 addresses. That arithmetic describes block size; it does not describe how many applicants existed, how much demand they had, or how long the pool would last.

Alan Barrett put the scarcity choice plainly: should an organisation that had used 90 percent of its first /22 remain eligible for a new /22, or should each organisation be limited to a single /22? Mark Elkins argued that a lifetime single-/22 ceiling would leave the remaining pool unused. The agreed clarification selected repeat eligibility under the relevant qualification, not a one-block lifetime ration.

This decision changes the incentive structure of scarcity. A per-allocation cap restrains the size of each increment while permitting organisations with continuing, qualifying need to return. A lifetime cap spreads access differently by barring repeat allocations, but it may also strand inventory if eligible first-time demand does not consume the pool. The minute records the choice and one argument for it. It does not supply demand projections, pool-depletion models, applicant counts or distributional outcomes, so none can be asserted.

The wording “per allocation” is economically dense. Under a repeat-allocation rule, verification of qualification becomes more important because the same organisation can make successive claims on the pool. The 90 percent use question anchors the recorded discussion, but the available record does not provide a complete implementation specification beyond the clarification itself. What can be said is that the meeting rejected an interpretation under which “maximum /22” meant “one /22 ever.”

Again, AFRINIC’s role must stay narrow. Administering allocation size and repeat eligibility for unallocated inventory is coordination through service terms and ledger practices. It is not confiscation of resources already held, political rationing by a sovereign, or ownership of an operator’s deployed capital. Scarcity makes the administrative choice economically important; it does not transmute the administrator into a state.

The distinction also disciplines later summaries. Saying that AFRINIC-13 accepted a /22 maximum without saying “per allocation” would omit the feature that decides whether repeat claims are possible. Saying that it authorized unlimited addresses would be equally wrong: each allocation remained capped, and eligibility still mattered. The accurate object is a constrained repeat-allocation structure, not a lifetime ceiling and not an unrestricted claim.

The /27 remained, and that non-change belongs in the ledger

The meeting heard a routability concern about a /27 or longer prefix. A /27 contains 32 addresses. Graham Beneke supported the proposal but questioned whether such a small prefix would be routable. The presentation described the /27 minimum as serving IPv6 transition mechanisms rather than ordinary global routing. The minute also records views that future technology might route small blocks or that some allocations might not require global routability.

The room did not require a change to the /27 minimum. That is not a missing fourth amendment. It is a recorded non-change, distinct from the three agreed changes. Preserving that distinction prevents a concern from being mistaken for an instruction and prevents silence after debate from being mistaken for an unresolved drafting omission.

The operational issue was real without having a single universal answer in the record. Routability is not identical to allocation eligibility. A small block might serve a transition mechanism or a use that did not require ordinary global routing. It might also confront filtering practices that made reachability uncertain. The meeting chose to retain the minimum despite the concern, but the sealed record does not establish a guarantee that every /27 would be globally reachable, nor does it establish that every such block would be unusable.

This is where technical specificity protects institutional modesty. A registry can enter an allocation in a ledger. It cannot command the global routing system to propagate the prefix. Operators make routing choices; networks bear reachability and continuity consequences. Treating allocation as a promise of universal routing would enlarge clerical coordination into a power the registry did not possess.

The non-change therefore defines both text and limit. Textually, the author was not instructed to remove or enlarge the /27 minimum. Institutionally, retaining it did not make AFRINIC guarantor of routing acceptance. Economically, applicants considering transition mechanisms would have to account for the difference between receiving a block and achieving the desired network treatment. The minute records the debate, not a technological warranty.

Two months of amendment custody

The archive dates the relevant sequence with enough precision to expose the custody gap. An author update went to the RPD list on 7 November 2010. AFRINIC posted the updated proposal online on 8 November. Another author update went to the list on 18 November. On 25 November, the meeting recorded consensus to progress the proposal, three agreed changes or clarifications, and no change on the /27 issue. The archive then records an updated online copy on 25 January 2011 incorporating the meeting modifications under the new reference AFPUB-2010-v4-005-draft-01.

The two-month interval is evidence of elapsed custody, not evidence of misconduct. The archive does not explain why incorporation took until 25 January. The sealed record does not show that the author or chairs acted improperly, altered the text in bad faith, or exceeded the instruction. It also does not supply a hash or full sealed copy of the exact 18 November proposal, and it provides no character-level redline between the meeting draft and the January version.

Those unknowns matter because the agreement was composite. Faithful implementation required the new draft to broaden scope, rename the phases, express the /22 maximum per allocation with repeat eligibility, and retain the /27 minimum. A redline could have made verification cheap. Without one in the sealed set, the archive’s modification history supports two of the described changes and the chronology supports the existence of a later incorporation, but a reader cannot audit every textual delta from these sources alone.

The appropriate conclusion is calibrated. The January posting is evidence that the author performed an incorporation step. The minute is evidence of the instruction the room gave. The lack of a sealed redline is an audit limitation. It is not proof that the January draft diverged materially. Nor can a later history label substitute for direct version comparison where exact language matters.

Amendment custody is a control problem because it separates social agreement from operative wording. The room produces a bounded instruction; a custodian produces a document. If the custodian has discretion to choose among formulations, that discretion should remain visible. If the choice is mechanical, a redline can demonstrate it. If a later review stage is substantive, participants can test whether the new text embodies the instruction. Each mechanism reduces the risk that “consensus” becomes a fog covering who actually wrote the operative rule.

The author’s absence from the meeting makes the handoff especially legible, though it supports no accusation. Alain Aina presented the proposal, the room discussed it, the co-chairs stated the outcome, and the author later posted the updated copy. The institutional record should therefore be read as a sequence of distinct acts by distinct people, not as if one collective mind continuously controlled every sentence.

The same meeting proved that stable text could decide an outcome

AFRINIC-13 supplies its own comparator. A different proposal, AFPUB-2010-v4-006, had rough support but did not progress because the text used at the meeting differed from the text posted before the deadline. The minutes invoked the rule in AFPUB-2010-GEN-005 that prohibited change within one week of a meeting so that participants could consider a stable version.

This comparator does not prove that Soft Landing violated the stable-draft rule. The archive records an update on 18 November for a meeting outcome on 25 November, and the available record makes no finding of a timing breach. The comparator instead proves a narrower and more important point: the process treated document identity as capable of defeating procedural progress even where rough support existed. Text was not incidental to consensus.

That internal standard sharpens the meaning of the Soft Landing result. The room could validly agree on an identified stable draft plus named amendments, but it could not honestly collapse the draft, oral instructions and later January wording into one timeless object. At the moment of the announcement, the posted proposal and the amendment ledger together formed the mandate. After incorporation, a new textual artefact existed and required the promised Last Call scrutiny.

The comparator also rebuts two opposite simplifications. One says no progress can ever occur until final typesetting is complete. AFRINIC-13’s Soft Landing decision shows how a process can advance a draft with explicit amendment custody. The other says wording does not matter if participants support the general policy direction. AFPUB-2010-v4-006 shows that AFRINIC’s own meeting rejected that view. Useful coordination lives between paralysis and vagueness: stable enough to identify the object, flexible enough to incorporate bounded decisions, and open enough for a later check.

The one-week rule protects more than reading time. It protects the mapping between support and text. Without a stable object, a statement of support can attach to one version while the procedure advances another. That risk is particularly acute when changes affect inventory scope or repeat eligibility. The rule cannot itself guarantee broad representation or wise policy, but it can prevent version drift from masquerading as consensus.

For Soft Landing, the correct audit question is therefore not, “Did the room endorse exhaustion planning?” It is, “Did the January draft faithfully implement the three agreed changes while preserving the recorded non-change, and was that result genuinely tested at Last Call?” The first half cannot be resolved character by character from the sealed material. The second half is shown to be nontrivial by the later no-consensus conclusion after the May 2011 Last Call.

A chronology that refuses retrospective compression

The immediate story begins before Johannesburg but does not require a retelling of the proposal’s entire development. A prior Last Call ended with objections on 14 July 2010, returning the proposal to list discussion. The author posted an updated copy on 7 November, AFRINIC placed it online on 8 November, and another update went to the list on 18 November. Those dates establish the pre-meeting draft sequence relevant to document identity.

On 25 November, the public meeting recorded the bounded result: progress to Last Call, three changes or clarifications, and retention of the /27 minimum. On 25 January 2011, the author posted the updated draft said to incorporate the meeting modifications. That interval is the central transition in this article because it is where agreed direction became new text.

The subsequent dates guard against mistaking the transition for completion. A later Last Call ended with a no-consensus conclusion on 28 May 2011. Draft 04 gained consensus at AFRINIC-14 on 10 June subject to further changes. Final consensus and Board ratification came only later, with ratification recorded on 21 November 2011. The later record shows continuing contestability and revision; it does not allow AFRINIC-14 or November ratification to be imported backward into the meaning of AFRINIC-13.

Likewise, the July objection history should not swallow the November event. It explains why another review mattered, but the present subject is not the full thirty-four-month life of Soft Landing or every earlier no-consensus result. The evidence boundary keeps attention on the exact institutional act at AFRINIC-13: what the meeting advanced, what it changed, what it left unchanged and who held the text until the next published version.

A disciplined chronology is an antidote to legitimacy by hindsight. Once a policy is eventually ratified, earlier steps can be retold as inevitable stages of one community will. The May no-consensus conclusion demonstrates that inevitability would be false even as process history. More fundamentally, each step had its own object and evidentiary basis. A later outcome cannot confer on an earlier room powers or agreement it did not possess at the time.

Scarcity turned syntax into capital allocation

Soft Landing addressed an exhaustion trajectory and a finite pool. Under abundance, a phrase about covered inventory or maximum allocation size might remain an administrative detail. Under scarcity, the same phrase determines who can obtain how much, under which trigger, and whether an organisation can return. These are capital-allocation effects even when implemented through a private registry ledger.

The first amendment determined the stock governed by exhaustion rules. The third determined the unit and repeatability of allocations. The /27 non-change preserved a smallest-block rule whose operational utility could depend on routing practice or transition use. The phase names helped attach rules to the correct stage. The four ledger entries therefore touched scope, timing, size, recurrence and usability—the mechanics through which scarcity reaches network planning.

Operators bear the downstream costs. They plan address use, deploy transition mechanisms, obtain equipment and services, manage routing exposure, serve customers and preserve continuity. A meeting participant can argue about a /22 or /27; an operator must make the resulting block work. A registry records allocation and applies eligibility rules; it does not absorb all capital or customer consequences if assumptions prove wrong.

This asymmetry establishes a narrow duty. The registry must keep the ledger unique, legible and predictable. Predictability includes version integrity: applicants should be able to identify the rule, the phase and the allocation interpretation that will govern a request. It includes amendment traceability: a change affecting repeat allocations should not disappear inside vague claims of consensus. It includes reasons and review: technical concerns such as routability should remain visible even when they do not change the text. And it includes exit from mistake: Last Call must be able to expose defects rather than merely certify momentum.

Scarcity does not create a mandate for broader control. Economic importance is sometimes used rhetorically to turn a technical coordinator into a regulator: because addresses are valuable, the body keeping the allocation ledger is said to govern the resource in a political sense. The logic runs backward. Greater economic consequence increases the need for institutional modesty, transparent custody and operator choice. It does not give a private body sovereignty over number resources or ownership of deployed networks.

The same point constrains community rhetoric. Participants can contribute evidence about conservation, transition and distribution. They cannot turn their presence into authority over absent resource holders. Consensus can coordinate use of unallocated inventory within voluntary arrangements; it cannot settle political ownership of address space, command routing, punish operators, or represent a continent. The meeting’s legitimate achievement was smaller and more useful: it specified a next procedural step and a bounded amendment package.

The strongest defence of the meeting’s approach

The best contrary case is practical. A policy-development meeting need not wait for final typesetting before making progress. If an identifiable draft is on the table, participants can agree on explicit amendments, entrust faithful incorporation to the author and use Last Call to verify the result. Requiring another full meeting for every precise wording change could let editorial mechanics paralyse useful coordination during an approaching scarcity problem.

AFRINIC-13 contains facts that support this defence. The minute names AFPUB-2010-v4-005. It records the three changes rather than issuing a featureless endorsement. It records that the /27 concern did not require a change. It sends the work to Last Call rather than presenting the room’s act as Board adoption. The archive later identifies a revised draft and attributes named modifications to the meeting comments. On that reading, the process did not pretend the text was finished; it openly used an amendment-and-review method.

The same-meeting stable-text comparator can strengthen rather than destroy the defence. It shows that the forum was capable of refusing progress when the meeting object differed from the timely posted text. Soft Landing, by contrast, could be understood as a stable draft followed by amendments formed through the meeting itself. A rule against last-minute pre-meeting substitution need not prohibit participants from deciding changes during the meeting, provided those changes are captured and the revised instrument receives later scrutiny.

This defence gets the institutional scale right only if it remains narrow. It justifies an efficient method for coordinating a private service; it does not create public authority. It justifies authorial incorporation of named changes; it does not authorize free revision. It justifies movement to Last Call; it does not make Last Call a formality. It justifies reliance on rough consensus within the process; it does not prove representation of all affected operators.

Nor does the two-month gap alone defeat the defence. Elapsed time may heighten audit interest, but the archive gives no explanation for the interval and no evidence of wrongdoing. A faithful draft can take time. What the gap does is expose why documentary controls matter: memories fade, participants may differ about nuance, and the later text becomes harder to verify without a redline. Delay increases the value of traceability; it does not supply a verdict.

The contrary case therefore survives, but conditionally. It is persuasive if every amendment remains identifiable, the new version can be compared with the mandate, the custodian’s discretion is bounded, and Last Call is prepared to reject the result. The later no-consensus conclusion is evidence that later review was not necessarily automatic. The defence cannot erase the central thesis because its own legitimacy depends on maintaining the distinction between procedural progress and final agreement.

The private ledger is not a sovereign chamber

The correct institutional frame begins with the work actually performed. AFRINIC coordinates number-resource records, publishes proposal history and applies service rules to inventory within private arrangements. Those functions can be important, technically demanding and economically consequential. They remain bookkeeping and coordination functions. Importance is not sovereignty.

This matters in interpreting words such as “policy,” “community” and “consensus.” A private policy can organize service eligibility without becoming public law. A community can supply expertise without becoming an electorate. Consensus can direct the next stage without manufacturing jurisdiction. A Board can ratify an internal policy without acquiring ownership of the resources recorded or the networks built by operators.

The distinction protects the Internet’s operational reality. Routing arises from distributed operator decisions, not from a registry decree. Capital is committed by network builders, not by meeting chairs. Customers experience service continuity at the edge, not in an archive entry. The registry contributes coordination by keeping records unique and predictable, but its value depends on serving that distributed system rather than claiming power over it.

In scarcity, the temptation to inflate the role grows. If the remaining pool is limited, allocating it appears to decide winners and losers. Yet the remedy for consequential discretion is not to invent sovereign legitimacy. It is to constrain the discretion: identify the inventory, publish stable drafts, record objections, specify amendments, disclose custody, provide review, and preserve the ability to challenge or exit. Authority should become more traceable as consequences rise.

AFRINIC-13 can be defended on precisely these modest terms. The meeting supplied a bounded instruction about unallocated inventory. The co-chairs announced a limited process outcome. The author incorporated changes. Another review stage followed. None of those acts required the fiction that the room spoke for every African network or owned the address space. Indeed, the narrower account is stronger because it matches the evidence.

The danger lies in later compression. “The proposal found consensus” may be harmless archive shorthand until it is used to claim that a final instrument enjoyed a unified mandate. “The community decided” may summarize a discussion until it obscures missing denominators and distinct actors. “AFRINIC policy” may describe a service rule until it is treated as sovereign regulation. Precision is not semantic fussiness here; it is the boundary that keeps useful coordination from becoming an unsupported claim of rule.

What the sealed record can and cannot bear

The evidence strongly supports five conclusions. AFRINIC-13 advanced an identified proposal rather than adopting final policy text. The meeting result can be reconstructed as three changes, one non-change, an authorial incorporation step and later review. The handling of AFPUB-2010-v4-006 proves that the same meeting considered stable document identity important enough to affect progress. The scope and allocation clauses carried economic consequences because they governed scarce inventory. And the minute’s phrase “progressing to Last Call” is more exact than the archive’s broader shorthand.

The evidence does not support a quantified majority. It does not show that all participants agreed, because counts and denominators are absent. It does not establish bad faith, capture, corruption, illegality or motive. It does not show that the author or chairs exceeded the instruction. It does not prove that the January draft materially diverged from the meeting mandate. It does not allow a later no-consensus result or eventual ratification to rewrite the event of 25 November.

Several unknowns remain concrete. The exact 18 November text is not separately sealed here as a complete text or hash. No character-level redline to the January draft is available in the sealed set. The method used to assess consensus, the eligible population and participant counts are unpublished in the minute. The record does not state when Last Call was expected to begin after AFRINIC-13, why incorporation took two months, or whether every participant understood the same distinction among principle, draft and amendment instruction.

These limits do not reduce the analysis to agnosticism. They tell us which level of claim is warranted. We can identify the procedural object and its amendment ledger. We can explain why custody mattered and why the later review was substantive. We can compare the treatment of stable text within the same meeting. We can locate the economic mechanisms. We cannot fill gaps with a story of unanimous continental authorization or, on the other side, a story of secret manipulation.

The resulting conclusion is exact. AFRINIC-13 agreed to advance Soft Landing before the text was finished, but it did not advance an empty idea. It advanced a known draft under three affirmative instructions and one deliberate non-change, with the author holding bounded amendment custody and Last Call reserved as the next test. That is a legitimate form of private coordination when its limits remain visible. It becomes misleading when procedural movement is retold as final adoption, when meeting attendance is inflated into mandate, or when scarce-resource bookkeeping is dressed as sovereignty.