Summary

  • The appeal filed on 1 October 2020 tested whether the co-chairs’ non-consensus conclusion on AFPUB-2020-GEN-001-DRAFT01 could be reviewed through an inspectable record; it did not prove that the proposed dashboard should have advanced.
  • Qualitative rough consensus can weigh reasons rather than votes, but reviewability requires a fixed decision object, distinct objection families, relevant staff and operational evidence, the proponent’s answers, the chairs’ treatment of those answers and a reasoned disposition.
  • The Appeal Committee examined the available meeting, chat and mailing-list material, found opposition and unresolved concerns, and unanimously let non-consensus stand. An upheld decision can still be meaningfully reviewable.
  • AFRINIC remained a private technical bookkeeper and coordinator throughout. Neither the dashboard proposal, the chairs’ conclusion nor the internal appeal created sovereign or other public coercive authority.

L3 — Seven objections enter the review record

A dashboard between a record and a consequence

By the open microphone at the AFRINIC-32 online public-policy meeting on 17 September 2020, the proposal called Policy Compliance Dashboard had put a deceptively simple interface at the centre of a difficult institutional question. It proposed a per-member dashboard, automated compliance reviews and notifications, and staff alerts associated with categories of registration-services-agreement and policy compliance. Its text also described an escalating three-month sequence that could culminate in nameserver removal, registry-record removal and resource recovery, while treating unauthorised transfers, non-payment and fraud separately.

The dashboard was therefore not merely a display. Classification could become the first step in a chain of operational action.

That possibility matters to an operator or number-resource holder even though the record establishes no implementation and no completed consequence. Stable, accurate registry records are part of the environment in which networks plan, invest and maintain continuity. A private classification that is difficult to understand or correct can increase the time spent tracing the basis of a notice, reconciling records, obtaining advice, assessing contractual exposure and preparing for interruption. Automation can improve consistency, but it can also repeat a mistaken or contestable judgment at speed.

The difference between a reviewable assessment and an opaque label is therefore practical: one exposes a route by which an error can be found and corrected; the other allows a disputed judgment to harden into routine administration.

The co-chairs announced at that meeting that significant objections remained and that the proposal had not achieved rough consensus. The conclusion was confirmed on the policy discussion list on 21 September. Jordi Palet filed an appeal on 1 October. His filing challenged both the general handling of the policy-development process and the treatment of seven families of objection recorded by the co-chairs. The decisive issue was not whether his preferred outcome should replace theirs.

It was whether a reviewer could identify what was decided, which objections mattered, what answers had been offered and why the remaining concerns justified non-consensus.

That distinction makes the episode narrower, and more useful, than a general dispute about dashboards or rough consensus. The proposed mechanism placed private classification upstream of notifications and possible service consequences. The chairs controlled whether objections to that design were answered or unresolved. The Appeal Committee controlled whether the chairs’ process and conclusion could be tested on a bounded record. Each control point involved judgment. None was a public-law act, but the quality of the reasons still affected the legitimacy and operational reliability of private coordination.

The decision object and its chronology

The object under review must first be fixed. AFRINIC’s proposal record identifies AFPUB-2020-GEN-001-DRAFT01, version 1.0, as submitted on 17 February 2020 by Jordi Palet Martinez, Frank Habicht, Mark Elkins and Haitham El Nakhal. It was later marked expired. The proposal page establishes what that draft said and how AFRINIC recorded its status. It does not establish adoption, implementation, contractual validity, actual enforcement or harm.

The public staff assessment in August 2020 is important because it shows that the administrative surface was not yet self-explanatory. Staff sought clarification about the scope of compliance, the categories and thresholds the system would use, what would be disclosed publicly, how quarantine would work, how penalties would be handled, and what people, systems and other resources would be required. These were not peripheral presentation questions. They reached the meanings that would drive classification, the visibility of the resulting label, the handling of exceptional cases and the feasibility of operating the mechanism.

The 17 September meeting conclusion and the 21 September list confirmation then supplied the challenged decision. The 1 October filing supplied the challenge. On 15 October the Appeal Committee held its first meeting, confirmed that the appeal complied with its Terms of Reference and began arranging its review. On 12 November it published a timeline and sought meeting recordings, video, chat and relevant messages from the policy discussion list. It met repeatedly in January and February 2021. On 18 February 2021, a quorate committee unanimously reported that there had been no consensus during the public-policy meeting, so the appeal failed.

This chronology does not turn the February report into a new event detached from the appeal. It is the disposition of the same chain that began with the chairs’ non-consensus conclusion and the October filing. Nor does committee unanimity mean that all participants agreed, that the proposal had no merit or that every contested proposition was settled. It tells us the result reached by the internal review body on the record it considered.

The discipline of fixing dates, text and decision is not clerical fussiness. Without it, a later clarification can be mistaken for an answer that existed when the chair decided; a proposal’s aspiration can be mistaken for an operating system; or the reviewer can unconsciously assess a better design than the one actually before the meeting. Reviewability begins by refusing those substitutions. The relevant question is what the co-chairs could reasonably conclude about Draft 1 from the discussion and evidence available at the time, followed by whether the committee could test that process from the preserved materials.

What the seven families actually represented

The appeal filing arranged the dispute around seven recorded families of objection and gave the appellant’s replies. That structure was valuable because it made disagreement more inspectable than a simple statement that opposition existed. Yet the filing was partisan advocacy. It proves that Palet made the allegations and supplied those responses; it does not by itself prove that the chairs erred, that the replies were complete or that every objection was resolved.

One objection family treated the proposal as redundant because staff already performed compliance work. Palet answered that the draft sought a uniform, public process instead of ad hoc private practice. This exchange exposed a real design tension. Existing administrative activity could support the case for consistency, but it could also raise the question of what additional work the proposal required and how a new system would relate to what staff already did. A reply that identifies a benefit does not automatically answer feasibility or duplication.

Other objections asked whether existing guidance already covered the subject, whether the proposed mechanism could bind or enforce, and whether ignorance of obligations justified creating a new system. These concerns operated at different levels. One questioned necessity, another the source and reach of private consequences, and another the behavioural premise behind the proposal. Compressing them into a single pro-or-anti-dashboard sentiment would have hidden their different relationships to the draft.

Further objections concerned the absence of a comprehensive Board framework for actions and severity, the resources required compared with existing systems, and whether the administrative work should remain with staff. Again, these were not interchangeable. A severity framework concerns the connection between classification and consequence. A resource objection concerns feasibility and opportunity cost. A role objection asks which institutional actor should administer the work. A qualitative consensus assessment must be able to recognise that several speakers may be describing distinct defects even if all end by opposing the same proposal.

The inverse is also true. Repetition does not become a new substantive reason merely because another message restates it. A participant cannot acquire a unilateral veto by repeating an objection after it has been answered. Chairs must distinguish a genuinely distinct reason, new evidence or an unanswered clause from advocacy that adds no substance. That is why counting objectors or messages would not resolve the appeal. The relevant unit is the reason and its relationship to the proposal text, not the number of times it appears.

Palet also named participants whose mailing-list posts he relied upon and said they would be copied in support. That could help a reviewer locate material. It did not make those participants authorised principals for every holder, operator or state in the region. Participation in a private policy process is evidence of participation, not a transfer of public sovereignty. The same caution applies on the other side: opposition expressed in the process is neither a statutory command nor proof of universal regional will.

What the committee could see

AFRINIC describes its appeal process as a review of co-chair decision-making rather than an opportunity to decide, according to a fresh policy preference, whether a proposal ought to have reached consensus. That division is institutionally sensible. A committee examining a challenged conclusion should be able to ask whether the chairs identified the relevant proposal and decision, considered material objections and reached a reasoned result. It should not quietly reconvene the policy room, rewrite the draft or manufacture consensus after the meeting.

The committee’s effort to obtain recordings, meeting video, chat and list messages therefore mattered. It watched the public-policy-meeting discussion and the videos of the co-chair conclusion. It considered the material available to it rather than treating the appellant’s filing as the whole record. The report invoked the policy-development rule assigning the co-chairs the task of determining during the public-policy meeting whether rough consensus had been achieved. That rule established the location of the original judgment, while the preserved material allowed the committee to examine how it had been exercised.

The report found significant opposition. It treated the operational or redundancy concern as unresolved because staff had already automated some compliance activity without a policy requirement. It also identified a fairness concern: a staged or partial dashboard operating alongside existing practice could produce two systems and unequal opportunity. Staff recommendations and requests for clarification remained unresolved in the committee’s analysis. The report further considered that a compliance dashboard required a complete examination of the compliance areas in the policy manual and their associated consequences.

Those findings show more than a conclusion, but less than an exhaustive clause-by-clause account of everything every participant said. They disclose why the committee believed material concerns persisted. They do not permit the stronger claim that every one of the seven appellant replies was rejected on every ground, or that every objection was independently held valid. They also do not establish that any proposed consequence was lawful, applied or technically executed. The committee report proves its review steps, reasoning and disposition; it does not become a judgment of a court.

On that limited basis, the decision was reviewable enough to be upheld. The committee reconstructed the relevant record, identified unresolved operational, fairness and clarification concerns, and unanimously concluded that consensus had not existed at the meeting. The appeal therefore failed. Review did not require reversal. Indeed, an institution in which review automatically produces a different policy outcome would merely move the original discretion to another room. The value of review lies in forcing the first judgment to encounter the record and survive stated reasons.

Evidence without inflation

The source classes set firm boundaries around the analysis. The proposal record establishes the draft’s wording, submission details, staff questions and status. It cannot prove that the proposed dashboard became operational. The appeal establishes what Palet alleged, how he framed the seven objection families and what answers he offered. It cannot make his answers neutral findings. The appeal register and committee report establish the internal process, materials considered, reasons stated and final disposition. They cannot create judicial authority or certify every claim made during the debate.

No affected member is established here. There is no proved nameserver removal, registry-record removal, resource recovery, contract termination, outage or quantified loss. The escalation described in the draft is still relevant because it explains why accuracy and review would matter if such a system were adopted, but prospective mechanism must not be narrated as completed history. The honest formulation is conditional: when an automated private classification sits upstream of consequential registry action, incomplete reasons can increase correction and continuity risk.

That evidential restraint strengthens rather than weakens the central conclusion. The appeal need not show an actual outage to demonstrate the institutional value of a reasons record. Review is meant to catch errors before a classification travels further through an administrative system. Nor must the proposal be condemned in order to demand reviewability. A useful administrative idea can still contain unresolved definitions, role questions or fairness concerns. The record can recognise potential benefit and incomplete design at the same time.

The bounded account is therefore clear. Draft 1 proposed a standardised compliance interface with automation and escalating private consequences. Staff raised questions about its definitions, disclosure, quarantine, penalties and resources. The chairs concluded that significant objections remained. The appellant challenged how those objections had been handled. The committee examined a broader set of meeting and list materials, identified unresolved concerns and upheld non-consensus.

What the episode tested was not the inherent virtue or vice of a dashboard, but whether qualitative discretion could be made answerable to an inspectable institutional record.