Summary

  • On 27 April 2005, unnamed members at AFRINIC-2 raised the need for a policy to recover unused resources in the AFRINIC region. The meeting report says the issue would be addressed by the policy working group. It records no identified recovery proposal, definition of “unused”, vote, consensus, affected resource, recovery order or timetable.
  • Referral was the safer institutional act. It kept an undefined concern outside immediate staff discretion and placed it within a process that required open mailing-list discussion, face-to-face consensus, a last call and Board ratification. Yet that process could make an internal policy; it could not manufacture sovereign authority.
  • AFRINIC is a private ledger operator, bookkeeper and technical coordinator. It can protect uniqueness, maintain accurate and contactable records, accept voluntary returns, and correct demonstrable duplication or fraud. It has no sovereign, legislative, police, punitive or confiscatory power over resource holders.
  • A defensible correction regime must distinguish abandonment from non-routing, low utilisation, temporary idleness, leasing or changed use. Before any adverse action, it needs a stated authority, objective evidence, notice, a meaningful cure opportunity, independent review, continuity protection, a hold against conflicting reassignment and a proportionate remedy.

What happened in Maputo, and what did not

A sentence that marked a boundary

The crucial act at AFRINIC-2 was not the recovery of an address block. It was not the adoption of a recovery policy. It was not even the presentation, at least in the surviving official account, of an identified recovery proposal. It was a referral.

AFRINIC’s second public policy and member meeting took place on 26 and 27 April 2005 at the Telecomunicacoes de Mocambique conference and training centre in Maputo. The official report says more than 100 participants attended. The timing gave the gathering unusual institutional weight. ICANN had formally recognised AFRINIC as the fifth regional Internet registry on 8 April. On 26 April, AFRINIC joined the Number Resource Organization. On the morning of 27 April, APNIC, ARIN and RIPE NCC transferred registration and accounting documentation and DNS administration for the African responsibilities they had previously served.

That context can make the language of institutional arrival sound grander than the powers actually transferred. The report described a transfer of formal authority and the completion of custodial and registry processes. Those phrases identify a handoff of technical and administrative functions: who would keep regional records, coordinate number resources and administer associated services. They did not turn AFRINIC into a government. Recognition by ICANN, membership in the NRO and cooperation among registries did not confer legislative, police, punitive or confiscatory authority.

The objects transferred were registry functions and records, not sovereignty over the people and networks recorded in them.

The policy day on 27 April included material that the report identifies more concretely. Itemised policy work concerned IPv4 end-user assignments, temporary assignments and experiments, and changes to criteria for ASN assignments. The same report also records ordinary organisational decisions, including decisions on budget, fees and Board elections. Against those more definite entries, the treatment of unused resources is strikingly spare. Some members raised the need for a policy to recover unused resources in the AFRINIC region. The issue, the report says, would be addressed by the policy working group.

Every noun and verb matters. “Some members” leaves the speakers unnamed. “Need” records a concern, not a finding. “A policy” describes something still to be developed, not an instrument already before the room. “Would be addressed” sends the matter elsewhere; it does not approve a substantive answer. The policy working group was a forum for deliberation, not an enforcement unit receiving an instruction to recover anything.

The safest reading is therefore also the most exact: the meeting surfaced a concern and referred it. Nothing in the record allows that sentence to be upgraded into proposal, consensus, adoption, implementation or recovery.

The zero-decision record

What the report omits is not a minor archival inconvenience. The omissions define the legal and operational limits of the event.

No speaker is named. No holder is identified. No prefix, ASN or quantity is specified. There is no definition of “unused”. There is no account of what evidence might distinguish abandonment from low utilisation, a delayed deployment, a private network, an unannounced route, a changed business purpose or a temporary operational state. The report does not say whether the concern was about encouraging voluntary return, correcting fraudulent records, ending registration services, compelling transfer, deregistering a holder or reallocating resources to someone else.

Nor does it record the procedural events that would convert discussion into policy under AFRINIC’s own process. There is no stable proposal identifier, author, text or revision history for unused-resource recovery. There is no declaration of consensus. There is no vote. There is no last call, Board ratification, recovery order or implementation timetable. The number of recorded recovery decisions at AFRINIC-2 is zero.

This distinction is especially important because the report itself says that some new policies and amendments had been proposed only about two weeks before the meeting. AFRINIC’s transitional policy development process required at least 30 days of discussion on the open policy-working-group mailing list before face-to-face consideration and consequent consensus. The report’s itemised proposals therefore faced a timing problem of their own. The unused-resource concern appears after those items and is described differently again. It would be an error to borrow procedural status from identified proposals and confer it on the later concern.

The absence of a named proposal is not proof that nobody ever wrote or discussed one elsewhere. It is a boundary on what can be said about this meeting. The available record does not establish such a proposal. It also does not establish what happened after the referral. Whether the issue later received an identifier, entered a mailing-list thread, reached consensus, survived last call, received ratification or led to an actual recovery remains open. Later events cannot be read backwards into a sentence written on 27 April 2005.

That discipline protects both sides of the institutional question. It prevents opponents of recovery from claiming that the meeting rejected a developed rule; the record says no such thing. It also prevents supporters of recovery from treating the referral as an early mandate. The meeting neither approved nor rejected recovery. It preserved the question for a process capable, at least in form, of defining it.

Why referral was substantive restraint

Procedure is sometimes dismissed as delay around a result that everyone assumes must eventually arrive. Here, delay was part of the protection. The adjective “unused” was doing too much work. Before it could justify consequences, someone had to turn it into testable categories, state the authority relied upon, expose the evidence standard to objection and explain what would happen to a holder and its customers.

The 2004 transitional process provided a sequence. Anyone could propose policy. The text would be published to an open mailing list. It required at least 30 days of discussion. Face-to-face discussion could then seek consensus, understood as general agreement rather than a majority vote. A 15-day last call followed, and the Board had a ratification and adoption role.

Each stage offered a distinct protection. A written proposal fixed the claim so that participants could not agree to different private meanings. Open-list discussion gave affected operators time to test edge cases. The 30-day minimum prevented a consequential rule from arriving too late for serious response. Face-to-face discussion could expose points of common ground and unresolved objection. Last call created a final opportunity to identify a material defect after the room’s apparent agreement. Board ratification created internal accountability for the completed process.

Referral thus made it harder for a vague concern to become an immediate administrative command. Staff could not properly take the Maputo sentence as an instruction to investigate a particular holder. The meeting chair could not properly translate it into consensus. The Board could not properly ratify words that had never been specified. A concern had to become a proposal before it could become policy, and policy had to pass the sequence before it could be internally adopted.

But procedural completion would answer only one question: had AFRINIC followed its internal method for making policy? It would not answer whether the resulting action was within the registry’s legitimate authority. A perfectly documented private process cannot create a power its participants do not possess. Open attendance is not universal representation. Consensus among those present is not legislation. Board ratification is not a court judgment. Referral was necessary restraint, not the first step in an inevitable march towards a sovereign recovery mandate.

Stewardship at the point of handoff

The strongest case for raising the concern should be taken seriously. AFRINIC was taking over coordination of a finite pool allocated through documented-need practices. A new registry could not maintain a credible ledger if explicit abandonment, duplicate registration or proven fraud had to remain forever untouched. Genuinely returned resources could become available again. Fraudulent records could impair both accuracy and fairness. A technical coordinator needs some ability to reconcile its ledger with demonstrable reality.

That is why the referral was better than either extreme. An instant recovery decision would have been procedurally indefensible. Refusing even to examine the problem would have treated registry accuracy as optional. Sending the concern to the policy working group created space to ask which situations were genuine record-integrity defects and which were merely operational choices disliked by others.

The correct distinction begins with the registry’s invariant. Number coordination needs uniqueness: the same resource must not be represented as validly controlled by incompatible claimants. It needs accurate records, reliable contacts, security metadata and a means to isolate disputes. A registry can accept an unambiguous voluntary return. It can correct a demonstrable duplicate. It can act on proven fraud affecting its own records. It can record explicit abandonment or a binding legal determination. Those acts align the ledger with independently established reality.

The registry crosses the boundary when it treats its ledger as the source of reality. If AFRINIC alone defines a resource as unused, treats that classification as forfeiture, removes the recorded holder and reallocates the resource, it is no longer merely correcting a record. It is using control of the record to impose a change on the network and the holder. That is coercive power. The official handoff in Maputo did not supply it, and the unused-resource referral did not silently create it.