Summary

  • On 17 November 2021, the AFRINIC-34 co-chairs said Publication of Information Draft 2 had not reached consensus because valid concerns remained unaddressed, and they returned it to mailing-list discussion.
  • The proposal would have required AFRINIC, two years after each accepted resource request, to publish a short summary of its justification. Staff said roughly 5,000 historical resources were already beyond that anniversary, while no publication summaries existed and existing members had not agreed to their release.
  • No-consensus prevented unresolved legal, confidentiality, retroactivity, staffing and implementation burdens from becoming an operative private-registry mandate. It did not punish the author, decide legal rights, authorize disclosure, confiscate resources or reject transparency in general.
  • Rough consensus was expressly distinguished from voting. A valid unresolved minority objection could prevent advancement regardless of numerical support, making the quality and treatment of objections more important than a headcount.
  • A stronger successor would publish a minimal decision receipt while protecting the underlying customer evidence, separate old requests from future ones, identify lawful authorization, fund review and correction, and expressly forbid retaliation or business-merit licensing.

The decision in the online meeting

On 17 November 2021, during the first day of the online AFRINIC-34 public policy meeting, the working group reached the decisive point on Publication of Information Draft 2. The proposal had been presented, AFRINIC's impact assessment had been discussed, and participants had raised objections. The co-chairs then recorded that valid concerns remained unaddressed. They did not send the text onward. They returned it to the Resource Policy Discussion mailing list for further discussion.

That procedural sentence did more institutional work than its modest wording suggests. The text before the meeting, AFPUB-2021-GEN-001-DRAFT02, version 2.0, had been submitted by Jordi Palet Martinez on 9 November 2021. It proposed a new article 13.0 in the Consolidated Policy Manual. Two years after every accepted request for number resources, AFRINIC would publish a short summary of the justification for the request, in a venue chosen by AFRINIC. The author distinguished this proposed publication from existing Whois registration and described its object as the overall justification used to test whether the accepted need remained sustained.

By the time of the meeting, however, this was not merely a clean choice between secrecy and openness. Staff reported approximately 6,500 ASN, IPv4 and IPv6 resources issued since 2005, with approximately 5,000 already beyond the proposed two-year point. Those were staff figures, not an audited dataset. Staff also said the underlying justifications lived in request tickets, that AFRINIC did not already maintain the short summaries contemplated by the proposal, and that existing members had not agreed or consented to publication of such summaries.

What might have advanced, then, was a duty to reconstruct a large historical record into new public summaries. That duty arrived without a settled legal basis for the old material, without pre-existing summaries to release, and without a credible resourcing plan for producing them, resolving questions, and handling disagreements. AFRINIC's displayed legal assessment warned of a potential conflict with the Registration Service Agreement's requirement to comply with applicable Mauritian data-protection and privacy law.

It also stated that pre-contractual and agreement-era information was confidential at common law and could not be disclosed without express member consent or a judge's order. Those were attributed views in AFRINIC's assessment, not findings by a court. They mattered as unresolved objections; they did not become judicial truth by appearing on an institutional page.

The immediate significance of no-consensus was therefore narrow but real. It kept a proposed transparency rule from crossing the line between aspiration and operative obligation while its authority, cost and method remained disputed. There was no move to Last Call. The proposal went back to discussion. The record later shows it under discussion until 9 November 2022, when it expired, and archived on 14 November 2022. Submission, assessment, discussion, non-consensus, return to the list, expiry and archive were separate stages. None proves adoption or implementation.

Rough consensus was not a losing vote

The meeting's own process explanation is essential to understanding the decision. The co-chair said rough consensus was not voting. Numerical support did not automatically dispose of a valid minority concern. One unresolved objection could matter even against a majority if it exposed a defect that the proposal had not answered. Co-chairs were expected to identify contentious issues, track open issues and ensure that concerns in staff impact assessments were addressed.

That process rule changes the meaning of “no consensus.” It does not mean that an electorate rejected the proposal in a referendum. It does not establish that every participant opposed disclosure, or that every premise offered for greater transparency was unsound. The available record does not provide a participant denominator or a scored matrix of objections. It records a process judgment: valid concerns remained open, so the working group had not attained the internal state needed to advance the draft.

The objections identified in the minutes were not interchangeable expressions of dislike. They concerned forced disclosure of business information, the possibility of competitive copying or targeting, surveillance, retroactive operation, an unclear statement of the problem, and policy micromanagement of confidential organisational data. Staff added the absence of existing summaries, the need for additional human resources, the prospect of member questions and disputes, and the legal concerns it had identified.

The co-chairs summarized as pending the concern that compelling members to expose business information sat uneasily with AFRINIC's role in managing resources rather than exercising general authority.

The safety value of rough consensus lay in refusing a shortcut from support to command. A headcount could have hidden the fact that the actual work, authority and exposure had not been resolved. Withholding consensus preserved those questions as questions. It prevented procedure from making an unfunded or potentially unlawful requirement appear settled simply because the meeting had reached the item and heard argument.

This did not give every objection a permanent veto over every future proposal. The disposition was a return to discussion, not a declaration that transparency could never be pursued. The test was whether material concerns had been answered well enough for this text, at this stage, to move. It had not. A revised design remained possible.

A safety valve, not a source of force

AFRINIC's institutional character places a second boundary around the decision. It is a private technical bookkeeper, membership service provider and coordinator. It can maintain accurate and auditable registry records, protect the uniqueness of number-resource registration, support contactability and security-relevant coordination, recognize changes in control and preserve operational continuity. It is not a sovereign, legislature, regulator, police force, prosecutor, punisher, confiscator or public-law court. Its service region is not territory, and its policy meeting is not a legislature.

For that reason, consensus itself could not have manufactured a public power that AFRINIC did not possess through contract or law. Equally, the absence of consensus did not produce a new power. The co-chairs did not sanction Martinez for offering the proposal. They did not punish a member, determine anyone's legal rights, issue a judicial ruling on Mauritian law, authorize publication, impose implementation duties or take resources. Nothing in the disposition supports treating it as an act of coercive discipline.

This is the critical distinction between restraint and punishment. Restraint says: the private coordination process has not resolved the conditions for this proposed duty, so it will not advance the duty. Punishment says: the institution may impose an adverse consequence on a speaker or resource holder because of a supposed offence. The first description fits the recorded act. The second does not.

The distinction also protects the transparency case from caricature. Refusing to advance Draft 2 was not a referendum on whether registry decisions should be understandable. Comparable public summaries might reveal artificial justifications, inconsistent criteria, hidden exceptions, drift or favoritism in scarce-resource decisions. That is the strongest benign case for a disclosure mechanism.

A process capable of stopping this draft could also preserve the possibility of designing that mechanism better: prospective rather than casually retroactive, specifically authorized rather than assumed, funded rather than wished into operation, and focused on a public decision receipt rather than the customer file.