Summary

  • The 2020 appeal concerned the co-chairs’ finding that AFPUB-2019-GEN-006-DRAFT02 lacked consensus; it did not establish that Draft 2 should have advanced.
  • A negative rough-consensus decision is consequential because it withholds the next procedural step. It therefore needs reasons as much as a positive decision does, even though rough consensus must not be reduced to a vote.
  • A reviewable record would freeze the exact draft, map material objections to its text and evidence, distinguish independent affected interests from repeated participation, explain each disposition and preserve a bounded path to affirmance or remand.
  • AFRINIC’s appeal machinery can correct its own private policy process. It cannot command networks, decide ownership or acquire sovereign, regulatory, police, prosecutorial, punitive, confiscatory or public-law adjudicative power.

L3 — The negative decision under appeal

A machine-readable signal, a human stopping point

The procedural paradox was unusually sharp. The proposal at issue concerned an AS0 Route Origin Authorisation, a machine-readable statement associated with routing security. Yet the immediate question in 2020 was not one a router could calculate. It was whether human co-chairs, listening to technical argument and exercising qualitative judgement, had enough basis to say that Draft 2 had not achieved rough consensus. Their answer stopped the proposal’s progress at that point. The appeal then asked, in effect, whether the reasons behind that stopping point were available in a form another body could examine.

That distinction matters from the first sentence. The exact object was AFPUB-2019-GEN-006-DRAFT02, identified in the record as the second draft and dated 30 July 2020. AFRINIC’s co-chairs assessed that this version lacked consensus, and that assessment was appealed in 2020. The appeal was not against a declaration that consensus existed. Nor was it a challenge to a deployed routing action, a particular AS0 object or an instruction imposed upon a network. It was a challenge to a negative procedural result inside a private policy-development process.

A little technical context explains why participants would care without deciding the appeal for them. An AS0 ROA can serve as a machine-readable disavowal signal. Route-origin validation can classify announcements into technical states, and operators may use those results in their own routing policies. Neither the signal nor the classification pulls a route out of the global routing system by itself. An operator still decides what to do locally. Likewise, a consensus call inside AFRINIC does not order a network to filter anything.

The proposal’s security purpose made its treatment important, but it did not transform a coordination discussion into public legislation.

The conflict was therefore more precise than a contest between advocates and opponents of routing security. On one side stood the co-chairs’ legitimate discretion to say that material objections remained and convergence had not been reached. On the other stood the appellant’s demand for a decision whose basis could be checked. A proposal can be stopped correctly. It can also be stopped opaquely. The value of review lies in distinguishing those conditions without presuming that the reviewers should substitute their own policy preference.

A negative finding still changes the field

“No consensus” can sound like an absence: nothing agreed, nothing done, nothing to explain. Procedurally, however, it is a decision with effects. The specified draft does not move to its next step. Its authors and participants must decide whether to revise, gather more evidence, repeat arguments, abandon the effort or challenge the call. Operators trying to understand the direction of a routing-security proposal are left to distinguish a temporary pause from a terminal obstacle. Time already spent on technical analysis and participation may have to be spent again.

The decision is not equivalent to adoption, and its consequences should not be exaggerated. The available evidence does not establish that Draft 2 was implemented, that any AS0 object was issued under it or that a route was rejected. It identifies no affected prefix, outage, holder loss or measured security benefit. Yet the absence of proved operational harm does not make the procedural consequence imaginary. Uncertainty itself has a cost. If a bare announcement reveals no route to closure, participants cannot tell whether a targeted amendment would answer the concern or whether another round would simply reproduce the same impasse.

This is where LARUS contributes something the meeting room cannot supply on its own. Its operator-centred account stresses that registry uncertainty can travel into planning and continuity because networks and their customers, rather than policy participants as such, carry the obligations of operating infrastructure. That insight does not prove that this appeal caused an outage or a financial loss. It clarifies the transmission channel. An unclear institutional status can consume engineering attention, delay planning choices and make it harder to know which assumptions are safe.

The economic stake is reduced uncertainty and reduced repeat-process cost, not an invented claim about traffic that the evidence does not show.

A reasoned negative decision can therefore be useful even to those who favour restraint. It can tell authors that one objection remains decisive, tell operators that the proposal has not matured and tell future participants what evidence would change the assessment. It can also reveal that the co-chairs were right to stop. Reasons are not a concession to the losing side; they are the means by which discretion earns confidence.

What the appeal proves — and what it leaves open

The existence of the appeal proves contest. It shows that the co-chairs’ non-consensus assessment did not pass without challenge. Beyond that, the evidential boundary is firm. The available record does not establish the filing date, the identity of the appellant or supporters, the exact grounds, the precise objection text or the exact language of the co-chairs’ announcement. It does not establish who sat on the committee for this matter, when it met, how anyone voted, whether there were separate views, when a report appeared or what disposition was reached.

Those gaps prevent a verdict about which side won. They also prevent a reconstruction in which later events are made to explain earlier reasoning. The proposal series continued after Draft 2, but later text and later status cannot be projected backwards to fill the 2020 record. The relevant question is whether the Draft 2 assessment was reviewable when made, on the objections, evidence and participation then before the decision-makers. A later formulation might respond to an earlier difficulty; it might reflect a different bargain; or it might do neither. Without the missing record, none of those possibilities can be asserted as the answer.

AFRINIC had published Version 2 of its Policy Development Appeal Committee Terms of Reference in November 2017. Those terms described an internal process in which the committee assessed facts and ruled on an appellant’s grounds, using the existing mailing-list and meeting record to consider compliance and its judgement of consensus or lack of consensus. They contemplated remand, reversal or affirmance and required findings accompanied by rationale. This gives a useful model of reasoned, record-based review.

It does not establish which version or clause governed every step of the Draft 2 matter, or that each contemplated power was used in a particular way.

The modesty of that claim is a strength. It allows analysis of what a fair internal review requires without inventing the result of this one. A reviewer could ask whether the correct draft was examined, whether the material objections were actually identified, whether responses were considered and whether the conclusion followed intelligibly from the record. Those are questions of reviewability. They are not permission to declare the proposal wise, to manufacture consensus or to announce an operational effect that has not been shown.

The strongest case for stopping

The benign defence of the co-chairs deserves to be heard in full. Rough consensus is qualitative. It is not a referendum in which a visible majority automatically carries the proposal. A security objective, however attractive in the abstract, does not entitle text to advance while important technical objections remain. Chairs need room to weigh the quality of reasons, to distinguish an answered point from an unresolved risk and to stop a draft whose uncertainties have not been responsibly settled.

The same defence rejects an automatic veto for persistence. If one person repeats an objection ten times, the objection has not acquired ten independent principals behind it. If a familiar circle dominates a mailing list, message volume does not establish the breadth of affected support or opposition. Conversely, silence does not equal assent. A chair who merely counts messages would abandon the judgement that rough consensus requires and would invite tactical repetition. The policy process would become a contest in endurance rather than an examination of reasons.

An appeal also has to remain bounded. It should not become a second policy forum in which the committee rewrites the draft, reopens every technical question or rewards any disappointed participant with another attempt to force adoption. If the original record supports a careful finding that significant objections remained, affirmance respects the co-chairs’ role. The appeal’s mere existence cannot prove that the appellant’s position was sound.

All of this can be accepted without accepting opacity. Indeed, it is because chairs exercise judgement rather than arithmetic that they must reveal the path of that judgement. A count explains itself poorly but at least can be reproduced. A qualitative conclusion needs a more informative account: which concern mattered, which text it reached, what support it had, what answer was offered and why the concern still blocked convergence. The burden is explanation, not surrender of discretion.

The first layer of the appeal’s significance is thus simple. The proposal did not need automatic passage, and the appellant did not gain a right to victory. But a negative finding was not an empty non-event. It withheld movement from a defined text. To review that exercise of private procedural power, a reviewer needed more than the words “no consensus”.