Summary

  • On 26 October 2014, Karmann Olumomo published AFPUB-2014-GEN-005-DRAFT-01, the AFRINIC Service Guidelines proposal, with the status Under Discussion. It described member complaints of extremely long waits for subsequent allocations and requests for non-technical information such as customer data and marketing channels.
  • Clause 3.1 proposed a decision within no more than 20% of the period for which resources were approved. For the draft's 12-month needs example, that is 2.4 months, not two months. Its proposed interim response after further delay is grammatically ambiguous and supplies no safe basis for inventing a percentage, size or entitlement.
  • Clauses 3.2 through 3.5 paired the clock with standard baseline information requirements, limits on storing or requesting marketing and business information, decisions based on current policy, and separation of routine allocation work from Board-level appeal. Together, these could have made a private service more predictable without creating public jurisdiction.
  • The page establishes publication, wording, date, author and proposal status. It does not establish adoption, implementation, contractual incorporation, measured compliance, member consensus, an actual appeal or legal force.
  • AFRINIC's proper character is that of a private technical bookkeeper, coordinator and service operator. It has no sovereignty, public regulatory jurisdiction, police authority, punishment power, confiscation power or final adjudication authority. Precisely because its mandate is narrow, it can be held to strong service obligations that constrain how it processes members' requests.

One publication, and a deliberately narrow question

On 26 October 2014, the publication of the AFRINIC Service Guidelines Draft 1 put an unusually practical proposition into view. The reference was AFPUB-2014-GEN-005-DRAFT-01; the named author was Karmann Olumomo; the status was Under Discussion. The proposal said that some members had experienced extremely long waits when seeking additional allocations. It also said they had been asked for non-technical material, including customer data and information about marketing channels. In response, it proposed a time formula, standardised baseline requirements, limits on business-information collection, policy-bounded decision-making and a separation between ordinary processing and Board appeal.

Those opening facts set both the importance and the limit of the document. The draft was trying to discipline a service. Its principal move was to turn an open-ended administrative interaction into something that could be timed, specified, explained and escalated. Yet the page did not turn the proposal into an adopted rule. It did not show that AFRINIC used the clock, incorporated the terms into membership contracts, issued an interim allocation, or accepted the data limits. The status Under Discussion must carry its ordinary weight: this was a proposal being discussed, not a proven operating commitment.

The right question, therefore, is not whether a single page created a new source of public power. It plainly did not. The useful question is how the seven proposed clauses could have constrained AFRINIC in the role it actually occupied: a private technical bookkeeper, coordinator and service operator handling number-resource requests. A service operator controls an important administrative surface. It receives applications, sets the sequence of questions, decides when a file is complete, records pauses, communicates a result and routes an escalation. That control can matter greatly to a network waiting to plan growth.

But operational importance does not become sovereignty merely because the records are important.

The distinction is the centre of the draft's significance. A private registry can make strong promises about time, evidence, privacy, reasons and review. Members can rely on those promises through contract, membership arrangements, audit, reputation, exit where realistic, independent dispute resolution and ordinary courts where legal rights are implicated. None of those mechanisms makes the registry a legislature or a public-law regulator. They make it a more accountable provider of a narrow, consequential service.

The complaint was a service allegation, not a decided case

The problem statement deserves close reading because it describes the friction the clauses were designed to reduce. The proposal said that some members had faced extremely long waits for additional-allocation requests. It also referred to demands for non-technical information, identifying customer data and marketing channels as examples. Clause 3.3 later named marketing budget as well. These assertions explain why a clock and an evidentiary boundary might have seemed necessary. They do not prove that any particular request was mishandled.

No member is named in the available page. No request file shows when an application arrived, when questions were sent, whether the applicant replied promptly, what technical verification was necessary, how much address space was sought or what decision followed. There is no measured sample of waiting times and no demonstrated commercial loss. It would therefore be wrong to transform the proposal's account into a finding that AFRINIC breached a duty, acted in bad faith or harmed a specified operator. The document identifies a class of service risk; it does not adjudicate a historical dispute.

That limitation does not empty the complaint of analytical value. Service rules are often designed before every underlying incident can be reconstructed publicly. If requesters cannot tell which questions are standard, when the response period began, why a file stopped moving or who can review an impasse, the procedure itself creates avoidable uncertainty. Even a technically justified delay becomes difficult to distinguish from drift when the service supplies no visible clock mechanics.

Likewise, a legitimate need for network evidence can be difficult to distinguish from a fishing expedition when the operator has not published the categories of information it ordinarily requires.

The draft's stated purpose was to improve service quality and transparency in AFRINIC's number-resource services by documenting AFRINIC's roles and responsibilities. A stated purpose is not an accomplished effect. But it tells us what kind of instrument its author proposed: one aimed at the provider's conduct toward requesters. It was not a charter of territorial jurisdiction. It did not purport to authorize coercion against the world. It sought to put boundaries around a member-facing process.

Clause 3.1: a clock tied to the planning horizon

Clause 3.1 supplied the proposal's most memorable device. AFRINIC would decide a subsequent-allocation request under AFRINIC policy and conclude it within no more than 20% of the total period for which AFRINIC approved the resources. The example used a 12-month needs period. Twenty per cent of 12 months is 2.4 months. The decimal matters. Rounding the proposal to a two-month universal service level would alter the text twice: it would change the arithmetic and erase the fact that the formula was tied to the approved needs period rather than stated as one fixed duration for every case.

The formula attempted to connect administrative speed to the operational horizon being assessed. If the registry evaluates how much a network needs for a defined future period, a decision that consumes a large portion of that same period can defeat the practical value of the evaluation. A 12-month plan is not static while a file waits. Capacity forecasts age. Equipment purchases, customer commitments and routing plans may have to be adjusted. The member bears those continuity and planning costs even though the registry controls much of the administrative sequence.

This is why the clock should be understood as a service obligation rather than a grant of authority. It tells the provider when it should finish work that is already within its limited role. It does not expand the subjects over which AFRINIC may rule. It does not make the allocation policy a statute, turn database recognition into sovereign title or authorize sanctions beyond the service relationship. The clock faces inward, toward the operator's performance.

Yet the formula is incomplete on its own terms. It does not identify the precise event that begins the 20% period. Is it the first submission, a formally acknowledged submission or the point at which a baseline package is complete? It does not say whether the clock pauses while the member supplies additional technical material. It does not identify permissible pause reasons, how quickly a question must be raised, whether multiple rounds of questions are allowed or what event restarts time. Nor does it explain how a change in the ultimately approved needs period affects a clock calculated by reference to that period.

Those are not cosmetic gaps. A deadline without start, pause and restart rules can produce only the appearance of measurability. The operator can effectively control the clock by leaving completeness undefined, while the requester can claim delay despite withholding necessary material. A workable version would timestamp receipt, specify the baseline requirements for each request class, identify when the file is complete, and require every pause to name a particular missing item and the party expected to act. It would restart automatically when the item arrived, subject to a published verification window.

That fuller design would be demanding without being reckless. It would give staff room to ask case-specific technical questions, investigate identity or fraud concerns and protect the uniqueness of registry records. At the same time, it would prevent a general appeal to complexity from becoming indefinite administrative discretion. The promise would be measured against recorded events rather than impressions.

The ambiguous interim response after delay

Clause 3.1 did more than state a completion formula. It then proposed an allocation-related response for each additional month of delay until a decision was made. That sentence appears to aim at continuity: if the registry exceeds its own response horizon, the requester should not bear the full operational cost while waiting. But the grammar is too uncertain to establish the remedy's calculation.

The wording does not safely reveal a percentage, an allocation size, a method for deriving the amount, a universal entitlement or the conditions for release. It does not say how an interim response would interact with the final merits decision. It does not explain whether the relief would be reversible, how uniqueness would be protected, what technical threshold would remain mandatory or what would happen if the final decision approved less than the interim amount. No actual request, grant, denial or payment under the sentence is established.

It is tempting to repair such uncertainty by supplying a plausible number. That would make the clause look more complete than it was. The honest reading is narrower: the draft expressed a continuity-remedy idea but did not specify it well enough to support a precise entitlement. The gap is especially important because an allocation-related response can affect records relied upon by many networks. Any provisional mechanism would have to preserve uniqueness, avoid prejudging the final decision and remain capable of orderly reversal or adjustment.

A bounded version could require an escalation before the deadline, followed by narrowly tailored continuity relief only when baseline technical conditions were already satisfied and the delay was attributable to the service operator. Relief could be capped, recorded separately, time-limited and expressly without prejudice to the final assessment. It could never mean that elapsed time alone compels staff to disregard technical integrity. This is an analytical model for making the proposal safe, not a claim about what the sentence actually guaranteed.

The distinction between the draft's text and a better implementation is essential. The text shows an intuition: delay should have consequences for the party controlling the service process, and the waiting member may need continuity support. It does not supply the mechanics. Recognising that intuition while preserving the ambiguity is more faithful than either dismissing the remedy as meaningless or presenting it as a fully formed allocation right.

Clause 3.2: publish the ordinary evidentiary baseline

Clause 3.2 proposed that AFRINIC publish standardised baseline information requirements for each typical allocation type. This provision addresses the evidentiary side of delay. A clock can be undermined if the requested inputs keep changing; a checklist can be oppressive if it is treated as a substitute for judgment. The clause's useful middle path is to publish what normally makes an application ready for assessment while leaving room for specific follow-up questions justified by the case.

The proposal did not contain the lists themselves. It did not say that every request must present identical evidence or that unusual network designs should be forced into an unsuitable template. It is therefore best read as a transparency rule for ordinary intake, not a command for mechanical approvals. Members would know the baseline before applying. Staff would have a shared starting point. Additional questions could then be identified as additional, rather than quietly becoming new general requirements one request at a time.

Standardisation matters because information demands are one of the registry's strongest practical controls over a request. A provider can extend a process simply by asking successive questions, even without issuing a formal refusal. If the expected baseline is unknown, the requester cannot plan its submission, estimate the review period or challenge an irrelevant demand. Publishing request categories, required fields, accepted evidence types and common reasons for follow-up would reduce that asymmetry.

But baseline publication must not become a promise that all complete-looking files are equivalent. Subsequent-allocation requests can differ in size, utilisation history, topology, addressing plan, evidence quality and fraud risk. Staff may need to verify whether a claim maps to technical need and whether records would remain accurate. The service discipline lies in requiring the question to be specific, relevant and recorded—not in prohibiting judgment.

The best operational connection between clauses 3.1 and 3.2 is a defined complete-intake event. The service clock should begin either when the published baseline is supplied or when AFRINIC identifies, within a short initial window, the precise missing baseline items. If a later question becomes necessary, the provider should state the technical or legal reason, record whether the clock pauses and identify the event that restarts it. That structure protects careful verification while preventing moving evidentiary demands from swallowing the deadline.

Clause 3.3: necessary technical evidence, not commercial curiosity

Clause 3.3 proposed that AFRINIC neither request nor store non-technical marketing or business information. It specifically mentioned customer data, marketing channels and marketing budget. The proposal thus paired a collection limit with a storage limit. That pairing is significant: restraint at intake is incomplete if irrelevant material already collected can be retained indefinitely, while a storage rule alone does not stop the pressure created by asking for it.

The clause was not a complete privacy code. It supplied no retention period, deletion method, access rule, audit procedure, breach response or mechanism for correcting information. It also did not resolve every difficult boundary between technical and non-technical material. A customer forecast might be offered to support projected network growth; a contract could contain both relevant capacity commitments and commercially sensitive terms; a financial or identity check might be required for a narrow lawful purpose without establishing technical need. Treating every business fact as forbidden could make accurate verification harder.

Treating every fact that might illuminate demand as technical could make the boundary meaningless.

The practical principle is purpose limitation. AFRINIC should identify what fact must be established for the request, ask for the least intrusive material capable of establishing it, separate technical verification from unrelated commercial assessment and avoid storing fields that do not serve the registry function. Where a document contains both necessary and irrelevant content, redaction or a narrower attestation may be preferable to wholesale disclosure. Access should be limited to personnel who need the information for the identified step, and retention should end when the legitimate service purpose and any applicable record duty end.

This discipline follows from the narrowness of the registry's role. A technical bookkeeper needs enough evidence to keep records accurate and coordinate unique number-resource entries. It does not thereby acquire a general mandate to inspect a member's market strategy, customer relationships or promotional spending. The absence of public regulatory jurisdiction makes clarity more important, not less. A private provider should not use an essential administrative position as a doorway into commercial information unrelated to the service decision.

Data minimisation also improves speed. Unbounded information collection creates more material to review, more questions about inconsistency and more sensitive records to protect. A published baseline focused on relevant technical evidence can reduce both member burden and staff workload. Clauses 3.2 and 3.3 therefore attack the same source of discretionary drift from opposite sides: one says what is ordinarily needed; the other says what should ordinarily remain outside the file.

Clause 3.4: a policy boundary around the decision

Clause 3.4 proposed that allocations be based solely on current policy and not on other factors. In service terms, the provision sought to make the decision criteria knowable. A member should be assessed against the rules governing the request, rather than hidden considerations introduced during processing. That is a constraint on the operator's administrative choice.

The clause does not define every judgment involved in applying the criteria. It does not eliminate the need to verify evidence, examine technical coherence, protect uniqueness, confirm identity or address fraud concerns. Those functions may be necessary to determine whether the policy conditions are actually met. The phrase “not on other factors” should not be stretched into an instruction to ignore falsity, contradiction or technical risk. Its useful force is against unrelated or undisclosed considerations, not against careful verification.

Nor does this clause make current policy public law. An internal policy can guide a private service relationship without acquiring general jurisdiction over outsiders. A policy-bounded decision is different from a statute-backed administrative order. The registry can say, in effect, “these are the criteria we will use in processing this service,” and members can demand consistent application of that promise. It cannot, merely by publishing the criteria, create police power, impose public punishment or become the final judge of legal rights.

A measurable implementation would require a written decision identifying the applicable policy provisions, the material evidence accepted or disputed and a reason code for the outcome. If further verification was decisive, the decision should explain why it was relevant. Such reasons would help a member distinguish a merits disagreement from a procedural delay, and they would give any appeal reviewer a defined record. Transparency would constrain the provider without enlarging its jurisdiction.

Clause 3.5: routine processing below, appeal above

Clause 3.5 proposed that the AFRINIC Board not be directly involved in IP-allocation processing unless an appeal was made. The service-design intuition is separation of roles. First-instance work should be performed through a regular, technically administered process. A different level should review a contested result only when the appeal route is invoked. Keeping the Board out of routine files can protect both staff neutrality and the Board's ability to review without having already directed the result.

The available text does not prove that an appeal system already existed, define how an appeal began or state what the Board could do. It gives no standard of review, deadline, conflict rule, record-access rule or remedy. It establishes no actual Board intervention and no actual appeal. Those unknowns matter because merely naming a higher body does not create independent review.

A credible separation would require much more. The requester should receive a reasoned first-instance decision and a clear filing window. Reviewers should not have directed the original processing. The standard of review should distinguish factual error, departure from current policy, procedural unfairness and new evidence. Conflicts should be disclosed and managed. The appeal should have its own time limit and a defined set of outcomes, such as confirmation, correction or return for reconsideration.

For disputes involving legal rights beyond the association's service process, independent contractual review or an ordinary court should remain available where appropriate.

That last point preserves the constitutional boundary. A Board appeal can be a private review mechanism. It cannot make the Board the final adjudicator of law. Internal review may correct service errors and preserve institutional consistency, but it does not displace courts or confer sovereign authority. The registry remains accountable within a limited private arrangement.

Clause 3.5 also complements the clock. An overdue file should not have to wait for a final refusal before anyone can intervene. A separate escalation channel can address service delay while routine technical assessment continues. That channel should not decide the merits merely because time is running out; it should make sure the file has an owner, pauses have reasons, questions are specific and continuity relief is considered under a published rule. Appeal and escalation are related but different: escalation restores movement, while appeal reviews a decision.

Clauses 3.6 and 3.7: separation provisions bundled into the proposal

The final two clauses move outward from request processing. Clause 3.6 proposed excluding staff and governing personnel of other regional Internet registries from AFRINIC policy development. Clause 3.7 proposed independence between the election of the AFRINIC policy working group chair and the AFRINIC Board nomination committee, while saying that candidates should be free of requirements.

Both provisions concern institutional separation, but their presence should not pull the analysis away from the service-level proposal. Clause 3.6 states a categorical exclusion. The text does not identify any person, prove prior participation, establish improper influence or explain how the category's boundary should be applied. The proposal alone cannot support a claim that another registry captured or interfered with AFRINIC's process. At most, it shows that the author wanted a sharper line between AFRINIC policy development and personnel associated with peer registries.

Clause 3.7 is likewise broad and underspecified. Independence between a chair election and a Board nomination committee can be understood as role separation: the body involved in Board nominations should not control a distinct policy-working-group selection. But “free of requirements” does not reveal a complete eligibility design. Taken literally, the phrase could collide with the ordinary need for a transparent nomination process, basic consent, conflict disclosure and a way to determine the result. No election, candidate, rule defect or outcome is established by the page.

These clauses can be acknowledged as part of the seven-clause package without being turned into a general account of elections or collective legitimacy. Their connection to the main service design is limited but intelligible: institutional roles should not blur in ways that make routine technical work or its oversight less independent. The more immediate and measurable service architecture, however, remains in clauses 3.1 through 3.5—time, evidence, data, decisional criteria and appeal separation.

Strong obligations do not manufacture public authority

The draft's most valuable implication is that accountability does not depend on pretending the registry is a state. AFRINIC can be socially and operationally important while remaining a private institution incorporated and headquartered in Mauritius and serving number-resource functions for Africa and part of the Indian Ocean region. Importance is not jurisdiction. Reliance on its records does not give it sovereignty over networks, property, speech or commerce.

The proper model is the private technical bookkeeper. The registry records and coordinates entries so that number-resource use can be recognised consistently. It can verify submissions, process requests, maintain accurate records, publish service information and enter bounded procedural commitments. It can administer an association or contractual relationship. It cannot legislate for the public, conduct policing, impose punishment, confiscate assets or finally adjudicate legal rights. A service region, database, policy forum or assertion of collective agreement cannot supply powers that were never delegated by public law.

This limitation is not an excuse for weak service. On the contrary, a narrow mandate supports stronger obligations. Members cannot reasonably be told that AFRINIC is too private to owe predictable processing while also being treated as though its internal discretion deserves the deference of a regulator. If the organisation's legitimate control is administrative, that control should be bounded by transparent inputs, clocks, explanations, privacy safeguards and review.

Strong service commitments constrain the operator's own conduct. They say when it will respond, what evidence it will ordinarily request, what it will not collect, which criteria it will apply and how a member may challenge error. They do not authorize new commands to non-consenting actors. The direction of power is decisive: service standards narrow discretion over an existing service relationship; sovereign regulation enlarges coercive jurisdiction over a public subject. Confusing the two would weaken both accountability and institutional modesty.

The distinction also clarifies remedies. A missed service deadline might justify escalation, fee consequences, temporary continuity measures, independent review or a contractual claim, depending on terms actually adopted and applicable law. It would not automatically justify punishment by AFRINIC. An adverse merits decision based on published criteria is not the same thing as confiscation, but neither may a registry recast a punitive measure as routine bookkeeping merely because it controls the database. Purpose, scope, process and external legal authority matter.

Draft 1 did not settle all these questions. It did something narrower: it proposed several restraints whose logic is compatible with private coordination. That is enough to make it worth examining, provided proposal is not confused with implementation and obligation is not confused with sovereignty.