Summary

  • ICANN Resolution 05.25 proclaimed AfriNIC a fully approved and recognized Regional Internet Registry for IP address registration and cognate services in the Africa service region. Read in context, it certified a bounded coordination function: maintaining accurate records, protecting uniqueness, operating reverse-DNS and directory interfaces, and completing the transfer of regional registry work.
  • The resolution did not transfer sovereignty, ownership, territorial title, police or punitive power, general regulatory authority, or immunity from ordinary law. Its application record, peer endorsement and ICP-2 evaluation support administrative competence within registry service; they do not turn private institutional reliance into public-law jurisdiction.

A consequential sentence, and no more than a sentence

The legal imagination is often tempted by short institutional texts. A sentence acquires a long history; the history acquires ceremonial language; practical dependence accumulates; and eventually the institution is described as though the original sentence had granted every power it later found useful. ICANN Board Resolution 05.25 rewards the opposite discipline. Its significance lies in reading what it says closely, identifying the evidence on which it rests, and refusing to fill its silence with a theory of government.

The operative action taken on Friday, 8 April 2005 was specific. The Board “proclaims AfriNIC to be a fully approved and recognized Regional Internet Registry” to provide IP address registration and other services for the Africa service region. Those words changed AfriNIC's institutional status within the system of Internet number-resource coordination. They made it the accepted regional interface for the described registry work. They did not enact a constitution for Africa's networks.

That distinction is not semantic fussiness. A registry record can become indispensable without becoming law. Operators, customers and counterparties may rely on it for accurate holder information, contacts, reverse-DNS administration, allocation history and the avoidance of duplicate registration. The more useful and widely accepted the record becomes, the greater the operator's practical influence. Yet reliance is not consent to unlimited power, and influence is not sovereignty. The office recognized by 05.25 was the office of the regional bookkeeper and coordinator.

The historical tracking page reinforces the need to stay with the enacted words. It expressly cautions that its summaries, explanatory material and recorded implementation actions carry no official authority and do not modify or override Board resolutions. That page records no implementation actions for 05.25 and says the resolution did not address funding. These entries help locate the document, but neither supplies a hidden grant. The operative text and the evidence named by its recitals remain the proper objects of interpretation.

There was also a separate Resolution 05.26. It offered congratulations and expressed institutional values. That provision should not be poured into 05.25 as though praise were delegated power. The two numbers identify two different textual acts: one recognized the registry function; the other spoke ceremonially. Only that separation is necessary here. The reach of 05.25 must be established from its own sentence and its own incorporated record.

Six phrases that define the recognized function

Start with the verb, proclaims. It is an act of institutional recognition within an existing process. ICANN was declaring that an applicant had met the conditions for a recognized place in the RIR coordination structure. The verb does real work: after the declaration, AfriNIC no longer occupied only a provisional or transitional position. But proclamation is not creation ex nihilo. The application said operational work had already moved, staff and facilities already existed, and AfriNIC was already performing RIR functions pending official recognition. The Board accepted that operating body into the recognized arrangement; it did not create a state or vest a private company with powers that ICANN itself did not possess as a public sovereign.

Next comes fully approved and recognized. “Fully” distinguished the status from the provisional recognition granted on 30 September 2004. It signalled that the staged evaluation had reached its endpoint: the transition had been completed to the assessors' satisfaction, the application had been updated, peers had delivered a favourable assessment, and the President had reviewed the application and IANA report against ICP-2. “Approved” and “recognized” therefore speak to institutional qualification. They answer, “Is this organisation accepted as the RIR for the stated service region?” They do not answer, “Who owns all number resources associated with Africa?” or “Who may legislate for every operator connected to the region?”

The noun phrase Regional Internet Registry supplies the nature of the office. A registry is a system and organisation for maintaining authoritative coordination records. Its necessary function is to prevent the same unique identifier from being registered incompatibly, to preserve accurate information about assignments and contacts, and to provide continuity in the shared administrative view. A later technical description, RFC 7020, usefully clarifies this enduring architecture: allocation-record accuracy supports uniqueness, while operational routing decisions lie outside the registry system. That later document does not enlarge the 2005 grant; it helps expose the difference between administering a registry and commanding the networks whose operators consult it.

The phrase IP address registration names the central service. Registration is neither ownership nor confiscation. A record of which resource is associated with which holder, contact, control state or security assertion can be important evidence. It can coordinate conduct among many independent networks. But the bookkeeper does not acquire the assets entered in the book merely because the book is widely relied upon. Nor does an error, alteration or dispute in the ledger automatically rewrite the operational or proprietary world outside it. Recognition made AfriNIC an accepted keeper of a regional number-resource record; it did not transform the record into the source of every relevant right.

Then there are other services. Read without context, the words might appear elastic. Read with the recitals, application and criterion-by-criterion evaluation, they refer to services of the same RIR kind. The transition record identifies resource-record administration, maintenance of the reverse-mapping in-addr.arpa service, WHOIS and other customer interfaces. It also describes the coordination required to transfer service from incumbent RIRs. These are natural companions to address registration. The application even supplied a concrete boundary: AfriNIC said it would not become involved in forward domain-name assignment or administration. “Other services” therefore cannot reasonably operate as a blank cheque for unrelated commercial regulation, control of routing, regulation of speech, or supervision of every transaction in which a number resource has economic significance.

Finally, for the Africa service region describes coverage. A service region tells users which registry interface is organised to serve which set of customers and records. It is an administrative geography. It is not territorial title. Africa did not become AfriNIC's property; African states did not dissolve their legal authority into the registry; and people or companies did not become political subjects of a private membership body because their networks used addresses registered through it. Geography makes coordinated service practicable. It does not manufacture sovereignty.

Together, these phrases describe a substantial but bounded office. AfriNIC could operate the regional ledger, administer registration, coordinate uniqueness, maintain cognate interfaces and exercise the judgment necessary to deliver those services. It could not rule merely because it could record.

The recitals form a chain of evidence, not a chain of title

Resolution 05.25 did not present recognition as an unexplained favour. Its recitals identify an evidentiary sequence. ICANN had procedures and standards for the establishment of new RIRs. AfriNIC had submitted an application and a transition plan. The incumbent registries and the Number Resource Organization supported the transition. The Board granted provisional recognition in September 2004 on conditions later described in terms of successfully completing the transition and supplementing or finalising the application.

The transition was then represented as complete; an updated application was available; the NRO chairman gave a favourable assessment; and the ICANN President reviewed the application and the IANA evaluation against ICP-2 before recommending final recognition.

Each link has a distinct function. The procedures established what would be tested. The application supplied the applicant's account of its organisation and capability. The transition put operating claims into practice. Peer support indicated that the incumbent registries were prepared to coordinate the handover. Provisional recognition created a conditional intermediate status. The updated materials and IANA report supported a final readiness conclusion. The President brought that record to the Board, and the Board adopted the recognition sentence.

None of these steps is a deed to address space. None is a treaty by African states. None converts peer approval into a popular mandate. Horizontal endorsement by other private or technical institutions can be strong evidence that an applicant will interoperate with them; it cannot confer police power. Likewise, an evaluation may establish compliance with institutional criteria while remaining incapable of delegating sovereign authority. The recitals explain why the Board trusted AfriNIC to perform the function. They do not silently add powers absent from the operative sentence.

This is also why four events must remain analytically separate. AfriNIC's incorporation gave it a private legal form and home. Operational handover transferred practical registry work. ICANN recognition established status within the coordination arrangement. Later NRO membership placed AfriNIC in another institutional relationship. The fact that these events occurred in sequence does not make them one indivisible act, and a power cannot be backdated from one instrument into another. The 8 April resolution recognized what it said it recognized.

ICP-2 tested an institution capable of serving, not a government entitled to rule

The incorporated standard, ICP-2, set ten establishment criteria. The proposed service region had to be of sufficient scale. The applicant needed broad support among local Internet registries. It needed a bottom-up policy structure and a commitment to neutrality and impartiality. It had to show technical expertise; adherence to global policies concerning conservation, aggregation and registration; an activity plan; a funding model; reliable recordkeeping; and protection of confidentiality.

These are demanding institutional tests. They go well beyond asking whether a database server can be switched on. A useful RIR must have competent people, durable financing, procedures for allocating unique resources, processes for participants, accurate records and safeguards for sensitive information. Some judgment is unavoidable. Conservation and aggregation cannot be administered without criteria; fraud and security cannot be managed without service conditions; conflicting requests cannot be resolved without a process.

Recognising those necessities is fully consistent with treating AfriNIC as a real operating institution rather than a passive spreadsheet.

But the ten criteria remain criteria for establishing an RIR. They do not describe the conferral of sovereignty, governmental jurisdiction, immunity, territorial title or ownership of address space. They do not confer legislative, prosecutorial, confiscatory or police power. They do not grant a general jurisdiction over commerce. Bottom-up policy is an organisational method, not a constitution binding every absent principal. Neutrality is a duty of administration, not a licence to govern. A funding model shows that services may be sustained; it does not create a taxing authority.

The standard's recordkeeping criterion is especially revealing. The registry must keep accurate records because unique number resources must not be registered inconsistently and because operators need dependable coordination information. Confidentiality protects information entrusted to that service. Both criteria describe obligations of stewardship. They point inward, toward how the keeper should maintain the shared record, rather than outward, toward power over networks' capital, routes, customers or speech.

Conformity on the date of evaluation also has a temporal limit. It supports the conclusion that the conditions were assessed as met then. It does not certify every future act as legitimate. Nor does a historical recognition renew itself into an expanded mandate whenever an institution adds a new policy. Later claims must still be tied to the function necessary for uniqueness, accurate records, proof, security and continuity. Historical approval cannot substitute for an express and competent source of any broader authority.

What the application showed—and what its numbers do not show

The final application described AfriNIC as a registered regional, not-for-profit, non-governmental, membership-based organisation. It said the entity had been incorporated in Mauritius in February 2004. An October 2004 ICANN letter was more specific, describing AfriNIC Ltd as a company limited by guarantee incorporated under Mauritius's Companies Act 2001. This corporate form matters. The organisation seeking recognition already had an ordinary legal identity under the law of a state. Resolution 05.25 did not displace that law, create immunity from courts or contracts, or convert the company into an international public authority.

The application also described a deliberately distributed operation. Incorporation, administration, business and legal functions were located in Mauritius; registration services and engineering in South Africa; replication and disaster recovery in Egypt; and training coordination in Ghana. Four locations demonstrate regional planning and resilience. They do not establish four seats of public power or territorial command. The map was an operational design for a small institution serving a large geography.

Its staffing snapshot makes that scale tangible. The application named four people: chief executive Adiel Akplogan; chief financial officer Harish Gowrinsunkur; registration service and engineering lead Ernest Byaruhanga; and administrative and business assistant Nooriah Woozeer. It planned to hire two additional technical staff. The names and roles are evidence of a team assembled to operate a registry. They also discipline inflated retrospective descriptions. The body being recognized was a developing membership company with four named staff and two planned technical hires, not a continent-wide executive government.

The funding account points in the same direction. According to the application, the NRO registries committed a total of US$100,000, contingent on specified transition activities. Membership fees were arranged in five categories linked to the quantity and kind of resources held by a local Internet registry. Those facts were evidence of a viable funding model under ICP-2. They were not a grant of taxation or a proprietary claim over the resources used to calculate service fees. And the resolution itself did not address funding, a point the historical tracker records without converting it into a substantive interpretation.

Support figures require even greater care because large numbers can be made to resemble democratic authorisation. The application reported that more than 68 organisations signed a founding membership charter in May 2000 and that it had about 40 formal letters of support. Those are two different measures: signatories to a charter and letters in support. Neither denominator represents all network operators, all resource holders, all governments or all inhabitants of Africa.

The application separately reported about 120 registered participants in the first public policy process, including representatives of more than 30 African local Internet registries. It later described more than 85 participants from 30 countries at the May 2004 Dakar meeting. Again, the categories are different. About 120 registered participants are not the same measure as more than 85 attendees. More than 30 LIR representatives are not the same denominator as 30 countries. These statistics may demonstrate meaningful engagement, regional reach and organisational readiness.

Collapsing them into a single claim of “African community consent” would destroy what the figures actually say.

The same caution applies to the proposed representation of six African subregions in the Board structure. Representation design can support institutional legitimacy within a membership organisation. It cannot make private participants into a legislature for nonparticipants. Voluntary consensus can govern arrangements among those who have joined, contracted or otherwise authorised representation. It does not create law for every operator merely because the registry's service geography is continental.

Operational readiness before the proclamation

The application claimed that AfriNIC was already performing all RIR functions pending formal recognition. It said operational services transferred to it included resource-record administration, in-addr.arpa maintenance, WHOIS and other customer interfaces. Until recognition, incumbent RIRs provided a second opinion on resource-allocation decisions. This arrangement explains what final status changed: it ended a transitional qualification around a service already being delivered and confirmed AfriNIC as the accepted regional registry interface.

That history blocks two opposite errors. The first is to trivialise recognition as empty symbolism. It was not empty. Moving from conditional transition and second-opinion arrangements to full recognition affected how peers and operators coordinated with AfriNIC. A single accepted interface can lower transaction costs, reduce ambiguity and improve continuity. The second error is to treat recognition as the source of every underlying capability or right. Operational handover preceded the final proclamation. The service did not appear by magic on 8 April, and sovereignty did not emerge from the end of a technical transition.

The express exclusion of forward domain-name assignment is an unusually useful boundary marker. An institution applying to provide number-resource registry services acknowledged that even a nearby Internet naming function lay outside its role. If the applicant itself distinguished cognate but separate administrative domains, the phrase “other services” cannot sensibly be read to reach any activity an RIR might later wish to influence. Context narrows the residual phrase to RIR work shown in the contemporaneous record.

The provisional stage confirms that the recognition process evaluated readiness. In July 2004, the Board acknowledged AfriNIC's willingness to begin the process and noted its progress toward ICP-2; it did not confer full recognition then. Provisional recognition followed in September. The October letter treated that status as a milestone on the road to a final position. By February 2005, the application placed the second transition phase and full service handover in the relevant period, and the IANA report records that on 21 February the NRO chairman and AfriNIC CEO jointly communicated the favourable peer assessment.

This is a record of institutional maturation, not territorial investiture.

The IANA evaluation: a certificate of criteria met

The IANA report bears a printed date of 6 April 2005, two days before Resolution 05.25. Its surviving filename contains “05aug2005,” but that filename must not be used to invent an August date. The date printed in the report controls; the reason for the mismatch is unknown. Small documentary anomalies matter because disciplined interpretation depends on resisting convenient inference, whether about a filename or about institutional power.

The report evaluated the final application criterion by criterion. It concluded that AfriNIC met all ten ICP-2 conditions, that the transition was virtually complete, and that the necessary requirements for approval and recognition had been met. This was the expert evaluation placed in the recognition chain. It supports a strong conclusion about assessed readiness on 6 April: the organisation, in the evaluator's judgment, had the service-region support, processes, expertise, policy adherence, plan, finances, records and confidentiality arrangements required of an RIR.

That conclusion deserves its full weight. A narrow reading does not mean a hostile reading or a denial that AfriNIC qualified. The operating record supported recognition. Incumbent RIR collaboration, the NRO chairman's favourable assessment, the transition of interfaces and the distributed resilience plan all contributed to a reasonable institutional finding that the Africa service region could be served through AfriNIC.

It deserves no more than its full weight. The IANA report evaluated conformity with an RIR-establishment standard; it did not conduct a constitutional convention. Its praise of competence and policy processes cannot enlarge the Board's sentence. A favourable recommendation proves the recommendation was given and relied upon. It is not an independent source of sovereign mandate. Even unanimous confidence among peer institutions could not manufacture public-law power that none was competent to confer.

Nor does the report establish that every statement in the application was independently audited in every respect. The published evaluation is evidence of the assessor's review and conclusion. Applicant statements about participation, support, staffing and operations should remain attributed to the application where appropriate. The record supports recognition without requiring a reader to pretend that every historical claim has a separate external verification.

What 05.25 did not grant

The negative space is unusually clear because the operative sentence is so concise. It contains no language of sovereignty or governmental jurisdiction. There is no legislative power, no police or prosecutorial authority, no punitive or confiscatory grant. There is no immunity from Mauritius company law, courts, contractual liability or independent review. There is no territorial title over Africa and no ownership or reversionary interest in IP addresses, autonomous system numbers, networks, operators or customers.

The resolution does not convert internal policy into law for people who did not consent to the organisation's rules or authorise someone to represent them. It does not confer a general power to regulate resource leasing, pricing, transfers, customer geography, routing choices, speech, morality or business structure. It does not license the registry to treat dependence on its database as submission to any condition it might later formulate. It does not make “other services” an expandable reservoir from which unrelated powers can be drawn.

Any account that says recognition granted AfriNIC authority over Africa's address space reverses the resolution's terms. The text recognized registry service within an administrative geography; it did not grant control of address space as territory or property. Replacing “service region” with “authority over” is not a harmless paraphrase. It adds the very mandate that the instrument omits.

That does not leave AfriNIC unable to administer a registry. It may apply service rules needed to protect uniqueness, verify claims, keep records accurate, control fraud, maintain security and preserve continuity. It may make bounded allocation judgments and require information necessary to deliver a reliable service. Contracts and ordinary law can make legitimate service conditions enforceable between relevant parties. The line is functional: the power must serve the registry office rather than mimic a sovereign's control over capital or conduct.

Consider an accurate allocation ledger. Refusing to register two incompatible holders for the same unique resource is intrinsic to the function. Requiring authenticated contact information can be intrinsic to accurate records and security. Maintaining reverse-DNS delegations is cognate registry administration. By contrast, asserting ownership of an address because it appears in the ledger confuses record with asset. Threatening registry continuity merely to punish unrelated commercial behaviour converts an administrative chokepoint into a sanction.

Treating service geography as jurisdiction over any transaction connected to Africa turns coverage into territory. The first set of acts keeps the book; the second attempts to rule through it.

The distinction also protects the integrity of the record. If entries can be changed for purposes unrelated to uniqueness, accuracy, fraud, security or continuity, the ledger stops functioning as neutral coordination evidence and begins functioning as an instrument of leverage. Operators then have reason to discount it, duplicate it, litigate around it or seek alternative forms of proof. Mandate expansion can therefore undermine the very shared trust on which a registry's practical influence depends.

The strongest case for a broader institutional reading

The strongest contrary case should not be caricatured. An RIR cannot operate as a mute filing cabinet. ICP-2 expected bottom-up policy development, neutrality, global policy adherence, conservation, an activity plan and technical competence. The application described allocation decisions, member control and capacity-building activity. The IANA report treated policy competence as part of the organisation's work. “Other services” plainly reaches beyond the keystroke that enters an address into a database. Fair, secure and coherent administration requires judgment, policies and conditions that have practical consequences.

From this, an institutional defender can reasonably argue that 05.25 recognized an evolving policy institution, not a static ledger. If AfriNIC must avoid conflicting registrations, steward finite IPv4 records, assess requests, respond to fraud and coordinate with global counterparts, it must exercise discretion. A rule that treats every decision as unauthorised unless quoted in the resolution would make the recognised service impossible to deliver. Practical recognition carries the powers necessarily incidental to the function.

The bounded answer accepts the operational premise and rejects the sovereign conclusion. Administrative discretion exists inside the service. A private bank, exchange, carrier, standards body or membership association may make consequential decisions under contracts and ordinary law without becoming a legislature, court or government. The question is not whether AfriNIC may decide anything. It is whether a claimed decision is tied to the narrow registry purposes that justify its role, or whether the institution is using dependence on the registry to regulate something else.

Necessary discretion remains anchored to uniqueness, accurate records, proof, security, fraud control and continuity. It remains subject to the organisation's private legal form, its contracts, ordinary courts and appropriate review. It cannot become territorial title, confiscation, punishment for unrelated conduct, general commercial regulation or immunity. No accumulation of adjectives—regional, recognised, bottom-up, community-based—crosses that boundary. The more consequential the decision, the more important it is to identify the exact service necessity rather than rely on status alone.

This answer also respects the realities of scarce IPv4. Scarcity makes allocation and transfer records economically important, but it does not vest the registry with ownership of the capital reflected in them. IPv6 abundance narratives do not dissolve existing IPv4 reliance, and they do not justify wider institutional control over IPv4 interests. A shared ledger can coordinate scarce resources without claiming the resources as its own.

Why narrow recognition matters to running networks

The practical value of 05.25 was coordination status. Network operators gained an accepted institutional interface for the Africa service region. Counterparties could look to one registry for records and contacts. Reverse-DNS and customer interfaces could be administered through a regional operator. The completed transition reduced the cost and uncertainty of relying on incumbent registries' second opinions. These are material benefits, not mere clerical niceties.

Accurate and continuous records assist troubleshooting, security assertions, transfers, financing, customer continuity and planning. The registry may not own the economic interests at stake, but its performance can affect them. LARUS's operational analysis places this dependence in concrete terms: governance choices at a registry can quietly disturb live infrastructure. BTW's chronology likewise makes the sequencing important: incorporation, operational handover, recognition and later NRO membership were distinct institutional events.

The operator's real-world influence comes from the services networks rely upon, not from an imaginary crown bestowed by recognition.

Heng Lu's running-code primacy fixes the controlling discipline: historical recognition must remain subordinate to the deterministic needs of uniqueness, interoperability, proof, security and continuity. Geography and community procedure cannot manufacture ownership. A registry may coordinate, record and protect the coherence of the shared ledger, but it may not turn that useful service into control over the capital or conduct of the operators whose systems give the ledger relevance.

NRS's accounts of ICANN's recognition role and RIR/NRO coordination history are first-class evidence for understanding that institutional arrangement. Its proper boundary is equally clear. NRS advocates, researches, convenes and represents members who have expressly authorised it. It does not operate the registry, RPKI, WHOIS or RDAP; it does not run courts, appeals, elections, settlement, custody or continuity. Treating advocacy as operational control would repeat the same category error this resolution requires us to avoid.

Narrow interpretation protects both registry and operator. It preserves the value of a trusted shared record while limiting the hold-up risk created by dependence on one regional interface. If service status is mistaken for punitive authority, operators face uncertainty over whether their resource records can be used as leverage in disputes unrelated to registry integrity. That uncertainty can increase the cost of capital, complicate financing and transfers, and make customers less confident in continuity. Registry overhead, delay and discretion are not abstract.

Their costs fall disproportionately on operators with fewer alternatives and less capacity to absorb disruption.

A service-region monopoly can also create lock-in even without formal sovereignty. When one institution is the accepted interface and there is no practical portability, the cost of resisting an overbroad demand may be loss of continuity rather than an ordinary change of supplier. This is why functional limits cannot depend solely on goodwill. The discipline follows from the architecture: the registry exists for accurate coordination among running networks; it must not turn the networks' need for a coherent record into authority over their unrelated conduct.

What remains unknown

The record closes some questions decisively and leaves others open. It establishes the date and text of 05.25, its separation from 05.26, the evidentiary chain in the recitals, the ten ICP-2 criteria, the described transition functions, AfriNIC's private corporate form, the IANA evaluation's printed date and conclusion, and the absence of sovereign or punitive language.

It does not provide a comprehensive catalogue of everything the Board understood by “other services.” The safe reading is limited to cognate RIR services evidenced in the application and evaluation, not an invented exhaustive list. It does not independently prove every support letter, participant count or operational claim beyond the published IANA assessment. It does not reconcile about 120 registered participants with more than 85 attendees because the figures describe different groups. It does not explain why the IANA report's filename appears to encode an August date while the document itself is dated 6 April.

Nor does 05.25 define the duration of recognition, replacement, portability, continuity upon institutional failure, derecognition or successor arrangements. It does not resolve the legal nature or ownership status of number resources. It creates no explicit remedy, appeal, liability or review regime for later adverse action. The absence of those provisions should not be filled with either an assumption of perpetual institutional entitlement or an assumption that recognition was meaningless. It is simply an unresolved design space around a clearly identified function.

The source record also contains no signed instrument by which any state delegated sovereignty to AfriNIC. Such a transfer cannot be inferred from community support, technical readiness or ICANN recognition. And nothing in this close reading decides the validity of any later board, court dispute, policy or resource action. Those matters would require their own evidence and their own legal analysis. Resolution 05.25 supplies neither a blanket defence nor a blanket condemnation of future conduct.

What it supplies is a precise baseline. AfriNIC was recognized to serve as the regional number-resource registry. Its authority is administrative and relational: it keeps and coordinates the record through the services for which it was accepted. Its legitimacy rises when it protects uniqueness, accuracy, proof, security and continuity; it does not expand merely because dependence on the service gives it leverage. Recognition names the useful office. The sentence does not build a throne.