Summary

  • Resolution 202006.557 was an internal corporate authorisation: AFRINIC’s Board authorised its chief executive to sign the NRO Internet Number Registry System Joint Project Agreement Addendum on the company’s behalf. It was not itself the five-party agreement and does not establish that the four duties were implemented or measured. The signed attachment is the separate act that records AFRINIC, APNIC, ARIN, LACNIC and RIPE NCC accepting the text, with visible dates from 10 to 12 August 2020.
  • The addendum makes four shared obligations legible. Each RIR undertook not to act in a way that would violate registry uniqueness, to take effective measures promoting accuracy, to publish registry entries for timely global Internet operations, and to cooperate in providing consistent and effective global registry services. Those verbs create a common contractual floor, but the public instrument package does not attach the measurements, incident rules, data controls or external remedies that would let an operator follow every duty from promise to evidence.
  • Section 10 of the base NRO memorandum was extended so an RIR could seek arbitration over disputes concerning those four obligations. That is a private remedy among eligible RIR parties, under the base memorandum’s ICC-rules framework; it is not a public court and it gives no contractual standing in this instrument to an operator, resource holder, customer, NRS or the public. It strengthens peer accountability without creating regulatory or sovereign power.
  • The base memorandum supplies essential context: commitments and resource decisions require unanimity; costs are shared equally unless a specifically approved alternative applies; amendments require a signed writing; assignment is restricted; and the arrangement disclaims partnership and agency. Yet the examined public package does not identify an INRS budget, a service catalogue, authoritative-data and correction controls, or a withdrawal and successor schedule. Those matters are unverified here, not proved nonexistent elsewhere.
  • The continuity test is therefore institutional but not statist. AFRINIC and the other RIRs are private technical bookkeepers and coordinators. Their useful commitments can preserve shared records and services, but neither criticality nor cooperation gives them sovereign, legislative, regulatory, police, confiscatory or public-law adjudicative authority. Continuity belongs to the ledger, authorised changes, security chain, running networks and downstream users; a sound arrangement must make the administrator replaceable without making the service disposable.

L3 - Five signatures, four duties and one private remedy

The 2020 instrument begins with a practical coordination problem rather than a constitutional one. Five separately organised regional registries each maintain a regional registry, while the entries must fit into a globally unique and non-overlapping whole. A mistake or incompatible act need not stay regional merely because the responsible administrator is regional. The seams matter: one participant’s treatment of uniqueness, accuracy, publication or service consistency can affect the usefulness of the shared system. A written inter-RIR commitment is therefore a sensible way to state a minimum across organisational boundaries.

That is why the legal sequence matters. AFRINIC’s archived Board register records Resolution 202006.557 as authorising its chief executive to sign the NRO INRS JPAA for African Network Information Centre Ltd. Authorisation is the first link, not the entire chain. It answers a corporate-capacity question: who was empowered, inside AFRINIC, to place the company’s assent on the proposed instrument? The record does not state the Board vote, deliberation, meeting date, conditions or supporting paper. It also does not substitute for the attachment bearing the parties’ signatures.

The signed attachment supplies the next link. It names AFRINIC, APNIC, ARIN, LACNIC and RIPE NCC. The visible dates span 10 to 12 August 2020. AFRINIC’s block names its chief executive, Eddy Mabano Kayihura, and bears a visible date of 10 August. The other visible RIR dates follow on 11 and 12 August. Those marks establish the documented assent of the five named organisations to the text. They do not, by their own force, prove how a duty was later assigned internally, how performance was measured, whether an exception arose, or whether a remedy was ever needed.

Keeping those acts separate prevents a common analytical shortcut. A board resolution may authorise a signature. A signature may bind a party to contractual language. Implementation then requires owners, procedures, systems and decisions. Measurement requires evidence against a defined standard. Breach requires facts about non-performance and the applicable obligation. Remedy requires standing, a forum and an enforceable result. Transition requires yet another set of people, artefacts and rehearsed steps. The examined sources establish authorisation and the signed common text.

They do not prove a breach, outage, inaccurate record, disputed cost, invoked arbitration or customer loss.

The instrument describes the Internet Number Registry System as a joint cooperative project. It says each RIR independently operates its regional registry while the organisations cooperate globally. It also recites that they operate the system “on behalf of” the global network-operator community. That final phrase must remain attributed to the contracting parties. It is their institutional description of the project, not evidence that every operator delegated power, that the public granted a mandate, or that the organisations acquired ownership or sovereign authority.

This distinction does not diminish the real coordination work. The addendum describes entries concerning globally unique identifier values and non-overlapping ranges. It associates an entry with a specific party issued an exclusive right of use, and it identifies uniqueness and identification of the registered holder as critical functions. Those are the instrument’s descriptions of the registry. They do not settle every question of ownership, routing, regulation or public-law title. The agreement can be useful as a private contract without being inflated into a constitution for the Internet.

The first operative verb is restraint: each RIR agreed not to take action that would violate INRS uniqueness. This is narrower and more exact than a claim that an RIR guarantees every possible result. It identifies a boundary on each participant’s conduct. The shared registry depends on values and ranges not colliding, so a party promises not to introduce action inconsistent with that property. To follow the promise operationally, a contract map would still need to identify which proposed changes are checked, against which authoritative state, by whom, with what evidence and through which exception process.

The public addendum states the duty; it does not supply that implementation map.

The second verb is active but qualified: each RIR agreed to take effective measures to promote accuracy. “Take” requires action, “effective” invites some test of practical sufficiency, and “promote” is not the same as warranting an error-free registry at every instant. That language can support disciplined peer accountability, but only if the surrounding system makes evidence visible. A useful implementation record would distinguish input validation, authorised correction, detection, reconciliation and restoration. It would show how accuracy is assessed without pretending that the examined sources establish an actual error or failure.

The third verb is publish. The parties agreed to publish INRS entries publicly for timely global Internet operations. Publication makes the registry useful beyond an internal back office, while “timely” connects the entry to operational need. Yet a publication promise alone does not specify an availability target, an update interval, an incident-notice period, a correction deadline, a privacy boundary or an external customer remedy. Those controls could have existed in procedures outside the examined public package. The proper finding is not that they did not exist.

It is that the public contract chain inspected here does not show where they lived.

The fourth verb is cooperate. Each RIR agreed to cooperate in providing consistent and effective global Internet Number Registry Services. Cooperation recognises that no regional administrator can make the global result alone. “Consistent” points towards compatible treatment across regional seams; “effective” asks whether the combined service actually supports its purpose. But cooperation is not a metric. To become auditable, it needs a service inventory, ownership, availability and timeliness measures, incident escalation, cure records, cross-RIR tests and evidence that exceptions have been handled.

Again, these are questions raised by the obligation, not allegations that any party failed it.

Together, the four verbs form a useful architecture: avoid harmful inconsistency, act for accuracy, publish operationally useful entries, and cooperate on the global service. They move the arrangement beyond a vague statement of collegial intent. At the same time, they stay at the level of common commitments. The public text does not connect every verb to a named control, metric, budget line, evidence repository or transition procedure. The value of the addendum is therefore twofold. It supplies a minimum floor, and it gives readers a precise set of duties against which to ask where the implementation detail is recorded.

The dispute provision adds another concrete link. The addendum changes section 10 of the base NRO memorandum so that RIRs can seek arbitration over disputes concerning the four INRS obligations. The base clause provides for arbitration under ICC rules in Bermuda, unless the parties unanimously agree another location. This is an inter-RIR enforcement route. It may deter peer breach or provide a forum when a disagreement cannot be resolved through cooperation.

Its limits are as important as its existence. The base memorandum creates rights only for the RIR signatories and admitted additional RIR members. Operators, resource holders, customers, NRS and the public do not acquire standing under that clause merely because registry service affects them. Arbitration under the agreement is not a public court, regulatory hearing or universal customer-redress scheme. The arbitrator’s role does not make the NRO a government, and the parties’ ability to enforce a private promise does not give an RIR legislative, police, prosecutorial, punitive, confiscatory or public-law adjudicative power.

There is a legitimate design choice in keeping the forum inter partes. A shared project agreement can focus on obligations that only the participating registries owe one another, without absorbing every downstream dispute. But that choice leaves a separate continuity question: if the service impact reaches an operator that lacks standing under the clause, where is that operator’s lawful route to timely protection or remedy? The examined evidence does not answer that question, and it does not establish that such a route was ever needed. A transparent contract map should simply identify whether the answer lies elsewhere.

The base memorandum also prevents the coordination arrangement from being mistaken for a new public institution. It says the arrangement creates no partnership, agency, association or franchise. It restricts transfer or assignment of interests, rights or obligations without every signatory’s prior written consent. It disclaims founding-RIR responsibility for financial commitments made contrary to the agreement. These clauses preserve the separate legal identities and bounded commitments of the participants. They reinforce the point that cooperation is being organised through private agreement, not through a supranational grant of authority.

Amendment control makes the chain traceable. The base memorandum permits amendment only through a writing signed by all RIRs, and a superseding writing must identify with particularity the provisions it replaces. The 2020 attachment does exactly one clearly relevant thing to the dispute architecture: it expressly changes section 10 for disagreements over the four duties. That specificity allows a reader to follow the extension from the later instrument back to the base clause. It does not justify assuming amendments, schedules or powers that the examined writings do not show.

Unanimity operates similarly. The base memorandum requires unanimous RIR approval for NRO commitments and allows its Executive Council to commit RIR resources only unanimously. That rule can protect each organisation against being bound to a collective commitment it did not accept. It can also make responsibility legible when the parties do agree: the decision is not merely attributed to an undefined “community” or to an autonomous NRO centre. Each RIR’s assent matters. Still, unanimity is a decision rule, not evidence that implementation was complete or that affected outsiders consented.

The resulting picture is deliberately modest. The June record establishes AFRINIC’s internal permission to sign. The August attachment records five private organisations accepting four shared duties. The addendum extends a pre-existing private arbitration clause for disputes about those duties. The base memorandum supplies decision, amendment, cost and liability context. None of those links transfers number resources, confers registry ownership, creates public power or proves universal representation.

Their strength lies in the narrower proposition: regional bookkeepers can make enforceable promises to one another about a shared technical service.

That proposition is enough to matter. It turns a cooperative narrative into obligations with identified parties and a forum available to those parties. But a signature chain becomes a continuity chain only when a reader can proceed further: from each promise to the funded service, from the service to measured evidence, from the evidence to an authorised correction or remedy, and from the incumbent administrator to a workable successor. The next layer is therefore not a demand for grander authority. It is a demand for the ordinary legibility expected of critical private infrastructure.