Summary

  • Draft 2 replaced a Chair-and-Vice-Chair arrangement with two chairs, clarified when a two-week appeal window began, required three participating supporters, set a four-week maximum for review and named a three-seat internal committee.
  • Those changes improved redundancy and predictability, but the original decision, the appeal appointments, urgent Board variance and eventual implementation still sat within an interlocking private institutional chain.
  • AFRINIC’s useful role is narrow technical coordination and accurate bookkeeping; open participation supplies evidence, not a sovereign mandate over operators, resources or capital.
  • A sounder design would preserve open deliberation while adding a decision ledger, silence-neutral consensus, direct standing for materially affected principals, conflict-screened review, reasoned remedies and implementation only after ratification and complete operational notice.

L3 — The redraft that put two hands on the same controls

The most revealing way to read Draft 2 is as a redline, not as a free-standing constitution and certainly not as an enacted code. AFRINIC’s revision record identifies it as AFPUB-2017-GEN-002-DRAFT-02, dated 9 November 2017 and still under discussion. Its significance lies in what moved between Draft 1 and Draft 2: who sat at the front of the working group, where operational rules were written, how emergency Board action was described, what event opened an appeal window, how reviewers were chosen and when implementation was meant to follow. Each alteration changed the handling of private procedural discretion. None created public power.

Draft 1 had proposed a recognisable hierarchy. One Chair led the Policy Development Working Group, assisted by one Vice-Chair. Assistance is not the same thing as a co-equal office. Even where a vice-chair can substitute, the title signals a primary holder of the agenda, the room and the institutional voice. Draft 2 replaced that pairing with two chairs. Its companion guidelines assigned the pair the practical work of coordinating discussion, preparing meeting agendas, controlling the floor and assessing rough consensus. They were also connected to the movement of a proposal through its phases.

If one or both became unavailable, the proposed arrangements contemplated substitution with staff assistance.

That change matters because chairing is not merely ceremonial in a rough-consensus system. The chairs decide when discussion is mature enough for an assessment, distinguish a substantial objection from repetition or obstruction, and state whether the group has crossed a phase boundary. A second chair can check memory, share workload and make it harder for one person’s inattention to define the record. Two people can compare interpretations of a room, challenge each other before declaring consensus and provide continuity when one is absent. In a geographically dispersed technical community, those are not cosmetic benefits.

But the redraft put two hands on the same controls; it did not create an appeal from one control room to another. Both chairs were assigned to administer the same process and make the final rough-consensus assessment. A show of hands remained a gauge, not a vote. The open working group could expose the chairs to more evidence, but the chairs still had to interpret the significance of that evidence. Pairing may reduce individual error. It can also diffuse responsibility if their division of labour, disagreements and individual reasoning are not recorded. The important comparison is therefore not “one person bad, two people good”.

It is one concentrated office versus a paired office whose internal checks need to be visible.

Draft 2 also divided the proposed rulebook. The main Policy Development Process and a separate Working Group guidelines-and-procedures document were presented as companion instruments. There is a benign administrative case for such a split. A concise process can state the stable stages, while a more readily maintained companion can explain meeting practice, agenda management and participation. Readers looking for operational detail may find a focused guide easier to use. Editors can avoid loading every facilitation convention into the constitutional text.

The split also creates a control question: which document governs when their language or revision timing diverges? The proposed core process depended on chairs whose most practical powers were elaborated in the companion. A participant trying to understand how an objection would be heard therefore needed to reconstruct one procedure across two texts. AFRINIC’s staff assessment of 21 November took the view that the documents should be merged for consistency with the Consolidated Policy Manual. That concern was not an objection to clarity as such.

It was a warning that clarity can be lost when the power-conferring rule and the operational instruction sit apart.

The appeal redline was more concrete. Draft 1 used “public knowledge of the decision” as the event from which a two-week appeal period ran. That phrase leaves a claimant with a threshold uncertainty before the merits can even be considered. Did public knowledge begin when a chair spoke in the meeting? When minutes appeared? When a mailing-list message summarised the outcome? When a particular participant actually saw it? A deadline without a defined notice event can become a discretion over timeliness, because the reviewer must decide when knowledge should be attributed to the appellant.

Draft 2 tied the period instead to chair actions or decisions publicly announced on the RPD list or at a Public Policy Meeting. That is a genuine improvement. It identifies institutional events that can be recorded, dated and checked. An operator should not have to prove an ambient state of collective awareness in order to show that an appeal is in time. A public announcement also puts pressure on the decision-maker to state that something appealable has happened, rather than allowing a consequential procedural choice to emerge through implication.

The improvement was incomplete because a defined trigger still requires a defined record. A spoken announcement at a meeting and a written announcement on a list are not identical forms of notice. The text needs to say which event controls if they occur on different dates, what minimum content makes an announcement sufficient and how a correction affects the clock. If an act changes phase, closes an objection or declares consensus, the notice should identify the act, the reasons, the material considered and the date by which review may be requested. Draft 2 made the clock more legible; it did not by itself specify a complete notice.

Access to the appeal was also structured. A formal appellant needed three participating working-group supporters. The word “supporters” must be handled carefully. Their role was to help a request enter the review channel, not to cast three votes proving the complaint correct. A supporter could reasonably believe that an objection deserved examination without endorsing every fact or remedy advanced by the appellant. Treating the threshold as a vote on the merits would confuse a docket filter with adjudication.

There is a plausible reason for the filter. Review consumes time, and a process open to anyone may attract repetitive, tactical or insubstantial filings. Requiring modest evidence that a concern is shared can keep the committee focused on disputes with some working-group salience. Three is a low number in the abstract. It can encourage an appellant to formulate the alleged process failure clearly enough that others will attach their names.

Yet the number can be too high in the case that matters most: a solitary but materially affected principal who identifies a real procedural error before others understand its effect. Participation levels vary by time zone, language, employer support and familiarity with the list. The people most exposed to an implementation burden may not be the people most able to recruit co-signers within two weeks. Silence from the rest of the list cannot reliably distinguish indifference, lack of capacity, fear of social cost, incomplete information or consent.

A supporter threshold is therefore defensible as one route to general-interest review, but dangerous as a veto over the direct standing of a person or operator who can show material effect.

Draft 2 gave the appeal a destination and a timetable. The proposed Appeal Committee had three members. One would come through a Board selection, one through a Council of Elders selection, and the immediate past PDWG co-chair would supply the third institutional channel, subject to fallback arrangements. The Council of Elders nominee would chair the committee. The committee would seek consensus and, failing that, decide by majority. It had a maximum of four weeks to report. It could annul a chair decision where the Policy Development Process had not been followed, and its decision would be final.

Compared with an unnamed or improvised reviewer, those specifications increase predictability. A claimant can see the size of the panel, the origins of its members, its preferred decision mode, its fallback voting rule, its deadline and at least one available remedy. A four-week maximum limits procedural drift. Finality can prevent a private technical process from becoming an endless ladder of internal appeals. Annulment is a meaningful remedy because it can return the process to the point before the defective decision rather than asking the same chairs merely to reconsider informally.

Composition, however, is not independence. A Board designee comes through an organ that retains ratification and urgent-variance functions. A Council of Elders designee may bring experience, but the title and selection channel do not establish external judicial independence. The immediate past co-chair brings direct knowledge of the process, but also belongs to the office whose successors are under review. Expertise and familiarity can help a panel understand context. They can also reproduce the assumptions of the original decision chain. Draft 2 named three internal seats; it did not turn them into a court.

The missing safeguards become sharper because the committee’s decision was final. A deadline and an endpoint answer how long the process lasts. They do not answer what evidence forms the record, whether all parties see it, how conflicts are disclosed, when a member must recuse, what happens if a seat cannot be filled, whether the challenged decision is paused, what reasons must be published or how the remedy should match the defect. “Final” can be valuable only after those questions have credible answers. Otherwise it may close the door with impressive punctuality while leaving the reason for closure obscure.

Draft 2 also changed the treatment of urgent Board variance. Draft 1 had included an explanation, a two-week working-group review and machinery for non-binding comment around the proposed emergency route. Draft 2 removed those consultative restrictions and left the Board route stated through references to bylaw sections 11.4 and 11.5. One defence is institutional competence: a policy proposal should not casually restrict directors where corporate duties or genuine operational necessity require action. Removing duplicative or conflicting constraints can keep the proposal within its proper place.

The other side of that defence is accountability. Urgency is the point at which a private chokepoint is most capable of bypassing ordinary deliberation. If consultation cannot occur first, the need for stated necessity, narrow duration and retrospective explanation becomes greater, not smaller. Cross-referencing external bylaw provisions also makes the proposed process harder to understand as a self-contained instrument, particularly if those provisions change independently. The 21 November staff assessment accordingly said the external bylaw references should be removed.

Removing the proposal’s restrictions did not make the Board a sovereign lawmaker; it widened a private corporate discretion whose limits still needed to be expressed.

Implementation supplied the final sequencing problem. Draft 2 added wording under which the implementation date should be less than six months after the end of Last Call unless a waiver was requested. That formula tried to answer a practical question often neglected in procedural design: when should an accepted rule become operational? A maximum interval can encourage planning and stop an agreed change from languishing indefinitely.

But Last Call is a deliberative milestone, not ratification. A proposal emerging from Last Call has not thereby completed every corporate step needed for a valid change to registry services. Starting the implementation measure from that earlier moment can create ambiguity about whether preparation, notice or operational change is expected before the Board has ratified the proposal. AFRINIC staff’s 21 November assessment said the clock should run from Board ratification instead. The concern is logical even without evidence of any premature deployment: sequence should follow authority within the corporation.

First the working group completes its process; then the Board decides whether to ratify; then staff issues complete operational notice and implements on a stated date.

Nothing in the record establishes that Draft 2 was adopted, ratified, implemented or applied. That limit is central, not a footnote. On 30 November, the AFRINIC-27 meeting record captured the authors’ presentation of the two-chair change, the document split, the revised appeal and Board provisions, as well as staff concerns about implementation and bylaw references. The co-chairs found no consensus and returned the proposal to the mailing list. That was a disposition within discussion, not later invalidation of an operative law. The article’s proper object is the proposed distribution of control at that moment.

Seen together, the redline produced a mixed institutional result. It improved redundancy at the front of the process, made the appeal trigger more knowable, put a ceiling on review time and disclosed the panel’s internal appointment channels. At the same time, it separated related rules, made urgent Board variance less constrained, placed review inside an interlocking circle and anchored implementation to the wrong procedural event. The AFRINIC-27 outcome prevented those propositions from being mistaken for settled procedure.

It also leaves a useful question for analysis: when does making a complaint process more definite make dissent genuinely reviewable, and when does it merely process dissent more efficiently inside the same private chain?