Summary

  • In May 2006, the AFRINIC Board used Resolution 200605.27 to ratify AFPUB-2006-GEN-002 and instruct staff to implement it. The instrument covered temporary unicast IPv4 addresses, unicast IPv6 addresses and AS numbers, normally for one month or less, on a renewable one-month “lease basis”. An applicant had to make the activity publicly knowable, give detailed identity and use information, state when the activity ended and say when the resources would be returned.
  • The one-month form was a useful private coordination service, not a licence or title. It could make a short-lived network’s identity, need, size and end visible while conserving unique identifiers. But no complete public grant, renewal and return ledger was recovered, and the policy did not specify decision times, renewal ceilings, reasons, review, reminders, return receipts, closeout tests or a response to non-return.
  • The strongest case for the policy is practical: conferences, exhibitions, conventions and comparable disclosed activities can need globally unique numbering without needing it indefinitely. The strongest answer to its gaps is therefore better bookkeeping—a precise lifecycle record, bounded reasons and auditable closure—not an enlargement of AFRINIC’s power.
  • AFRINIC was and is, for this analysis, a private bookkeeper and coordinator. It is not a sovereign, legislature, regulator, police force, prosecutor, judge, punisher, confiscator, owner or public licensor. Expiry of a temporary registry entry is not cancellation of a public licence, forfeiture of property, a penalty or a command backed by the state.

A resolution with a very small centre

Resolution 200605.27 contained four ratifications. Three concerned subjects that must be kept outside this account: a global IPv6 policy, ordinary ASN eligibility and a four-byte ASN transition. The only act examined here is the fourth, the Board’s ratification of the proposed policy for temporary resource assignments and its instruction to staff to take the steps necessary to implement it. Keeping that boundary matters. A bundled resolution can invite a broad story about number policy, but AFPUB-2006-GEN-002 was a much narrower instrument.

It addressed a temporary need, the evidence offered for that need, the size and duration of the registry entry, the possibility of renewal and the expected return.

The policy page calls the instrument the “Policy for Temporary Assignments/Allocations”, gives it the reference AFPUB-2006-GEN-002 and preserves an older reference, afpol-tmpal200504. It attributes the text to Aina Alain and marks the status as implemented. Within the document, the relevant resources are expressly defined as unicast IPv4 addresses, unicast IPv6 addresses and AS numbers. Nothing in the available record extends this particular instrument to multicast space, domain names, certificates, route objects or any other asset. Nor does the presence of AS numbers invite the ordinary ASN eligibility question into this article.

Here an AS number is simply one of three possible resource classes in a temporary request.

The policy’s introduction described organisations that might need resources for a fixed period, “usually one month or less”. It named exhibitions, conferences and conventions as examples. Those examples make the intended problem easy to see without creating a closed statutory list. A disclosed experiment or another comparable time-bounded activity could fit the stated logic; the text does not say that every event qualifies, and it does not provide an exhaustive taxonomy of acceptable events. Its organising idea was impermanence supported by evidence.

That modest idea is analytically valuable. An event network may need numbering during preparation, live operation and dismantling. A temporary record can identify who is responsible while the network exists and can mark the expectation that the entry will close. It can also avoid turning every brief need into an indefinite reservation. None of this requires a theory that AFRINIC owns the numbers or grants permission for lawful network activity. It requires a serviceable ledger: who asked, for what, for how long, for which exact resource and with what plan to return it.

The policy itself used the phrase “lease basis” for the one-month arrangement. That phrase should neither be erased nor overloaded. It describes the temporal form adopted by the instrument: a one-month, return-oriented registration arrangement that could be renewed on a further application. It is not proof of a government licence, a land lease, a final conclusion about property, or a wider jurisdiction over network markets. Calling the record a lease did not make AFRINIC a landlord. Calling the end of the record an expiry did not turn that end into a public sanction.

The chronology the surviving page actually supports

The path to the May 2006 Board act can be reconstructed only to the level preserved in AFRINIC’s own records. The policy page’s history says the proposal was first posted to the RPD mailing list on 17 April 2005. The original posting is not in the available evidence, so that date is a retrospective entry on the policy page rather than a conclusion drawn from the message itself.

Ten days later, on 27 April 2005, the AFRINIC-2 meeting report recorded discussion of temporary assignments. The report summarised several central features that survived into the instrument: a publicly documented activity, return after the activity, a one-month lease basis, renewal with information and sizing according to the submitted plan. It also preserved objections and design suggestions. Participants asked whether one month was too short, discussed whether repeated announcement, withdrawal and later reuse might “taint” a prefix, proposed using a single reserved block, asked for routing plans and debated how fees should work.

The report preserved contrary views on the supposed taint rather than resolving the issue. These were discussion points, not clauses merely waiting to be transcribed. The report also noted a process concern: the policies had been proposed only about two weeks earlier, while the process then described required at least 30 days of mailing-list discussion before meeting consensus.

The surviving policy page displays “Date: 17 May 2005”. That field presents a genuine anomaly. The same page’s history places the first posting in April 2005, while official records place meeting consensus in December 2005, last call in February 2006 and Board ratification in May 2006. The page does not explain whether 17 May 2005 refers to a draft, a metadata event or something else. It cannot responsibly be treated as the ratification date, and its implemented status does not transform that unexplained field into proof of implementation on that day.

On 14 December 2005, the AFRINIC-3 report and minutes page listed temporary address assignments among five proposals said to have achieved meeting consensus. They also said the proposals would return to the policy list for a further 15 days before Board approval. That is evidence of a reported procedural result. It is not Board ratification, still less evidence that meeting participation conferred public authority over people who did not participate.

The policy page then records a 15-day last-call period beginning on 9 February and ending on 24 February 2006. The available evidence does not contain the complete last-call message set, the objections submitted during it or their disposition. The page establishes the dates it reports, but it cannot establish details that are absent from the record.

Finally, the Board page places Resolution 200605.27 in May 2006. It says the Board heard a status report from a Policy Working Group chair, ratified the proposed temporary-resource policy with the other three proposals and instructed staff to implement it. The page supplies no exact meeting day, attendance list, individual votes, indication of whether the four proposals were voted separately, copy of the chair’s report, signed standalone instrument or detailed implementation specification. It would therefore be wrong to infer unanimity, an exact date or a richer implementation order.

At an unspecified later date, the temporary policy appeared as section 9 of AFRINIC’s Consolidated Policy Manual. The manual preserved its core mechanics: public documentation of the activity, a one-month assignment, renewal on necessary information, plan-based sizing, specified application fields and a possible administrative fee. Manual inclusion proves that these requirements appeared in the published policy corpus. It does not prove uninterrupted practice, complete compliance or the outcome of any application.

What counted as a knowable temporary need

AFPUB-2006-GEN-002 made disclosure the first practical filter. The temporary activity was to be publicly documented and available, preferably on a website. “Publicly accessible” was defined in demanding terms: the material had to be openly available free of charge and free of disclosure constraints. The text further said AFRINIC would not recognise an activity under the policy if it could not be publicly disclosed.

“Recognise” is an administrative word inside a private service instrument. It describes whether AFRINIC would accept the activity as a basis for this temporary registry arrangement. It does not mean that a private or undisclosed event was illegal, that its packets were invalid, that its organiser lost title to anything, or that the organisation had violated a public command. The policy supplied no criminal rule, civil prohibition or general licence requirement. Its consequence was local to the stated service: an activity that could not be disclosed did not fit this route to a temporary entry.

There is nevertheless a real trade-off in that disclosure design. A public activity page could deter an invented request, give carriers and peers context and make the claimed dates testable. It could also exclude a legitimate confidential experiment, security-sensitive trial or private technical exercise. The available instrument gave no alternative route for such a case, no trusted confidential disclosure process and no enumerated exception. That silence should be described as a design limit, not repaired by inventing an exception and not converted into a judgement that all confidential activity was undeserving.

The public duty also has two different possible objects, and the text clearly settles only one of them. It expressly required the applicant to publish or otherwise make available information about the activity. It did not clearly require AFRINIC to publish a historical register of every accepted temporary assignment. A later RPD exchange in December 2018 exposed precisely this distinction during debate over a clarification proposal. Participants considered whether information such as the requester, purpose and prefix size should be published by AFRINIC and discussed retention of historical data.

They also debated a proposed IPv4 cap and whether 15, 20 or 30 days was operationally suitable. That later exchange demonstrates that the two publication questions were not identical. It does not retroactively amend the 2006 text, prove adoption of the clarification or establish what AFRINIC actually published for each case.

The policy’s eligibility envelope was therefore both intelligible and incomplete. It had a fixed-duration purpose, illustrative event types and a strict public-evidence rule. It did not state whether an applicant had to be an existing AFRINIC member, what standing it needed, which class of organisation could apply or who was authorised to sign. Those missing matters cannot be filled from a different policy. The ordinary rules for obtaining an ASN, for example, belong to another subject even though an AS number could be requested temporarily here.

From application form to registry state

The information sought from an applicant was extensive enough to sketch a useful record, though not enough to complete its lifecycle. The applicant was to provide its legal organisation name and country of registration; postal and physical addresses; telephone and fax numbers; and a mandatory website. It was also to identify the website describing the temporary activity. Where relevant, the application could point to similar previous activities and provide links from other relevant sites. The policy invited other information that AFRINIC should know, without defining the boundary of that open-ended category.

Operationally, the applicant had to give the date the activity would end and a plan for using the resources. For address space, it had to state the subnet size and the purpose of that subnet. It had to identify any AS numbers required and request reverse delegation if needed. It also had to give an intended return date. After reviewing this information, AFRINIC would require the appropriate resource-request form. The size of the temporary resource was to be determined from the requesting entity’s plan. The original available instrument states no universal prefix-size cap and no scoring formula.

These fields support several legitimate checks. Legal identity and contact information make responsibility traceable. The public activity record ties the request to an observable purpose. A use plan links requested quantity to intended operation. The activity-end and return dates make impermanence explicit. Any AS-number or reverse-delegation need can be recorded alongside the address request rather than inferred after issue. The later form can capture resource-specific detail once the preliminary account has been reviewed.

But “review” remains a thin verb in the surviving text. The policy does not state who reviews, what evidence is sufficient, what conflicts must be disclosed, whether comparable requests receive a common test or how long a decision should take. It requires no published reason for acceptance or refusal, no correction period for a curable omission, no standard exception route and no independent appeal. It does not say how confidential operational details should be protected or how long the evidence should be retained. Those omissions do not prove arbitrary decisions; they make arbitrariness difficult to exclude from the documentary record.

The policy’s open invitation to supply anything else AFRINIC “should know” is especially revealing. A narrowly used question could allow staff to resolve a technical ambiguity. Without a bounded purpose, reason log or retention rule, however, it could widen the information demand from verification into discretionary judgement. Under a bookkeeping approach, every supplementary request should connect to a defined registry need: identity, uniqueness, technical feasibility, contactability, term, closure or cost. Questions about the organiser’s lawful business model or broader worthiness would require authority the instrument does not supply.

The application evidence also separates the applicant’s responsibilities from the registry’s. The requesting organisation defines the activity, discloses it, supplies identity and contact information, states the plan, identifies the requested resources and dates, seeks renewal if required and returns the resource when the activity or experiment ends. Its event technical team builds, routes, monitors and tears down the network, bearing carrier, compatibility, renumbering and route-history consequences.

AFRINIC staff review the submission, request the relevant form, derive size from the plan, maintain the registry entry and may charge an administrative fee if necessary. Staff do not thereby become licensing officers or regulators of the event.

The optional fee belongs within the same narrow service logic. Recovering direct administrative cost can be defensible when a short-lived request requires review, registration, reverse-delegation work and closure. Yet the policy gives no amount, formula, waiver, invoice time or published cost basis. The available record proves neither that any applicant paid a fee nor that AFRINIC waived one. The defensible principle is cost recovery tied to an actual service; the missing evidence is how, or whether, that discretion operated.

One instrument, at least three clocks

The apparent simplicity of “one month” dissolves once the temporary network is treated as an operation rather than a date label. The policy contains at least three clocks. First, the introduction says the organisational need will usually last one month or less. That is a description of the expected need. Second, the operative rule says resources are assigned on a lease basis for one month. That is the stated term of the registry arrangement. Third, the applicant must give both the activity-end date and an intended return date, and must understand that the resources will be returned when the activity or experiment ends.

Those are fact-specific lifecycle dates.

The text does not say which clock prevails when they diverge. An organiser may need numbering before a conference opens in order to configure equipment, establish carrier sessions and test the network. Teardown may continue after the final session. The event could end early, or continue because of a delay. The intended return date could fall before or after the nominal one-month endpoint. A renewal application might be pending when the first month expires. Each possibility changes the operational state, yet the published instrument does not publish a priority rule.

That gap was visible in the AFRINIC-2 discussion. Some participants thought one month might be too short. The concern was not adopted as a longer term, and the final instrument retained one month with possible renewal. It was, however, a sensible warning about setup and teardown. A hard non-renewable month would make turnover clear but could force renumbering or interruption during a genuine extension. Automatic rollover would reduce continuity risk but could turn a supposedly temporary entry into an indefinite one without refreshed evidence. Renewal on application is a reasonable middle form only if timing and state are legible.

The actual renewal clause is spare. Renewal is possible on application to AFRINIC while providing the necessary information. It does not say renewal is automatic, and it proves no entitlement to renewal. It gives no advance-notice period, list of required refreshed fields, test for continuing need, decision deadline, default state while a decision is pending, maximum number of renewals or maximum total duration. It provides no reason requirement and no appeal path. Even the length of a renewed term is left to inference from the one-month form rather than separately specified.

For an operator, those omissions can become concrete continuity risks. Apply too early and the final event plan may still be incomplete; apply too late and a decision may not arrive before the recorded end. Assume continuity and the organiser may build around a resource whose status is about to become uncertain. Assume a hard stop and it may spend money on avoidable renumbering. LARUS’s contemporary public framing of renewal and continuity is useful only at this level: it shows why time at a registry interface can be economically material to an operator. It is not evidence of a 2006 decision, and its service model is not AFPUB-2006-GEN-002.

A sound record would preserve rather than collapse these clocks. It would record application time, issue time, activation time, activity start, activity end, intended return, the one-month maximum date, renewal-request time, decision time and actual return. A reasonable design could treat the earliest applicable end as the closeout trigger unless a documented renewal or justified teardown interval changed it. That is a recommended bookkeeping control, not a clause found in the available policy.

Return was promised; closure was not designed

Return is the instrument’s natural endpoint. The applicant had to demonstrate an understanding that the resources would be returned when the activity or experiment ended and had to state an intended return date. The short term and the explicit return condition are central to the strongest conservation case for the policy. They distinguish a temporary use record from an indefinite reservation.

Yet an intended return is not an actual return, and an actual return is not a closed record. The policy does not publish a return method, an acknowledgment, a registry-update deadline, a reminder sequence or a receipt that the holder can retain. It states no test for route withdrawal, reverse-delegation cleanup or the closing of related security records. It contains no quarantine or reuse-readiness procedure, no rule for preserving the audit trail and no aggregate reporting requirement. Most importantly, it does not say what happens when return is late or disputed.

That final silence must remain a silence. It is not permission to import a general theory of revocation, reclamation or punishment. If the event has ended but routing continues, the immediate bookkeeping questions are factual: did the organisation request renewal, did AFRINIC decide it, was the route actually withdrawn, is the record stale, and what notice or correction step is documented? A registry can flag an inconsistent entry, seek confirmation, preserve evidence and correct its own book. Any dispute over legal use belongs to contract or competent public law, not to powers invented from a missing paragraph.

The AFRINIC-2 report adds one operational concern without resolving it. Participants debated whether repeated withdrawal and later injection could leave a prefix with an undesirable routing history. Some suggested a dedicated reserved block; others disputed that the alleged taint was real. The available evidence supplies no measurement, no incident and no harmed later user. It therefore supports neither a claim that reuse was harmless nor a claim that it caused damage. It supports a design question: before reuse, what observable checks should be logged, and how should uncertainty be communicated to the next operator?

A complete closeout would join four facts that the policy left apart: the activity actually ended; the operator ceased using the resource; AFRINIC updated the registry state; and the resource was ready for subsequent coordination. Keeping those facts separate prevents a date from masquerading as an outcome. It also protects the former holder. Without a receipt and closure timestamp, later activity associated with the resource could be wrongly attributed to an organisation that had already returned it.

The lifecycle ledger that should have existed

The available evidence permits a clear state model even though it does not prove AFRINIC used one. At the first state, “purpose disclosed”, the applicant should supply the public activity record, purpose, dates and a capture timestamp showing what evidence was available. At “applicant identified”, legal identity, jurisdiction, addresses and contacts should be joined to proof that the submitter is authorised. Most identity fields were express in the policy; submitter authority was not.

At “need and size proposed”, the record should identify the resource class, requested subnet or AS number, intended use, plan version, routing considerations and any reverse-delegation requirement. The policy expressly sought the plan, size, purpose, AS-number and reverse-delegation information, but it did not require version control or a detailed routing plan. Versioning matters because a decision based on one plan should not silently appear to have approved a later one.

At “reviewed”, AFRINIC should record who reviewed the evidence, what was checked, which questions were asked, what decision was made, why and when. The instrument expressly contemplated review and a subsequent request form, but not the reviewer, criteria, reason or service time. At “issued and active”, the registry entry should hold the exact resource, the organisation and responsible contact, issue and activation times, the one-month endpoint, the activity-end date and intended return. No complete public issue register was recovered.

At “renewal requested”, the organisation should record when it applied, why the need continued, which public evidence was refreshed, what changed in the plan and what new end and return dates were proposed. At “renewed or refused”, the registry should record the decision, evidence basis, reviewer, time, new endpoint or refusal reason and a challenge route. The policy expressly made renewal possible; it did not define these controls.

At “activity ended or term expired”, the operator should record the actual activity end, network teardown and route withdrawal, while the registry should issue or record the closeout instruction. At “returned and record closed”, the organisation should record actual return and AFRINIC should acknowledge it, close the current state without erasing history, record any reverse-delegation cleanup and mark reuse readiness. The policy expressed return but not this procedure. Finally, at “fee accounted”, any charge should have an amount, cost basis, invoice and receipt, with an equally explicit no-fee or waiver decision.

The instrument only supplied permission to charge an administrative fee if necessary.

This state model is deliberately about records, not authority. It makes the registry’s private service more reliable while limiting discretion. Each state answers a practical question for the applicant, the operator, AFRINIC staff, carriers, auditors and later users. None decides whether the event is politically legitimate, whether the organiser deserves to operate or who owns Internet numbers.

Nor does the model require publication of sensitive operational detail. The applicant’s public activity evidence can be archived or captured while confidential topology and personal contact information remain protected. A privacy-aware public view could report aggregate request, acceptance, refusal, renewal, return and fee statistics, alongside limited status information sufficient to distinguish active, pending renewal, expired, returned and closed entries. The precise disclosure design was absent from the 2006 text; the principle is to make decisions auditable without needlessly exposing operators.

The strongest case for the policy

The best defence of AFPUB-2006-GEN-002 begins with the network problem, not institutional prestige. A registry charged with keeping identifiers unique cannot coordinate temporary use on an organiser’s unsupported assertion. A conference, exhibition, convention or experiment may genuinely need globally unique numbering, perhaps because a venue’s provider cannot supply the required operational independence. Duplicate or improvised use can create confusion. An anonymous, open-ended reservation can waste resources and leave peers unsure who is responsible.

Against those risks, the instrument assembled a proportionate-looking package. The organisation disclosed its legal identity and contacts. The activity had to be publicly documented. The plan tied resource size to need rather than to an unexamined maximum. The one-month term made temporary intent visible. The activity-end and intended-return dates gave the record an expected closing point. Renewal required a new application with necessary information rather than silent rollover. A reasonable administrative fee could recover the direct cost of review and registry work.

This is not merely bureaucratic neatness. A bounded entry can help an event build, test, operate and dismantle a network without pretending that a short-lived need is permanent. It can tell carriers and peers whom to contact. It can make duplicate coordination less likely. A refreshed renewal application can catch changed dates, contacts or technical plans. A recorded return can protect the next user and the former user by clarifying the handover in responsibility.

The finite-pool argument is strongest when expressed as conservation rather than ownership. The registry need not own a resource to avoid recording incompatible simultaneous use. It need not license an event to verify that the stated activity exists. It need not regulate a market to ask how much numbering a particular temporary network plans to use. It need not punish the organiser to record that the agreed temporary state has ended. These are thin coordination functions, and the policy’s one-month form makes their legitimate purpose unusually visible.

The contrary case also gets something important right about human judgement. Event networks differ. A fixed formula cannot capture every topology, preparation period or reverse-delegation need. Some staff review is unavoidable. The answer is not to pretend that judgement can be eliminated; it is to bind judgement to stated registry purposes, recorded evidence, reasons, comparable treatment and reviewable timing. The policy supplied the inputs for such a system but not the full accountability structure.

Where coordination stops

The same features that support the policy also mark its authority boundary. NRS offers a contemporary explanation of AFRINIC as the body through which networks obtain and have Internet number resources allocated and recorded. That description helps identify a practical function. It does not establish sovereignty or prove the 2006 act. NRS advocates, researches, convenes and represents explicitly authorised members. It does not operate the registry, RPKI, WHOIS or RDAP, appeals, settlement, elections, custody or continuity, and it is not evidence of an individual temporary request.

Heng Lu’s controlling doctrine supplies the interpretive discipline. The “Bill of Rights of Uniqueness Coordination” permits the preservation of uniqueness and accurate records but rejects the conversion of recordkeeping, geography, scarcity or registration into rule, ownership or punishment. The bookkeeper analysis warns that a ledger serves operational reality; it does not create the legitimacy of the network it describes. “The Policy Mirror” distinguishes authoritative registry publication from operational and legal reality and argues for bounded functions, records, reasons and audit trails rather than confiscatory consequences.

Applied here, that doctrine preserves everything useful in the policy. AFRINIC can ask who seeks a temporary entry, verify the disclosed activity, assess the plan for a requested size, record a one-month term, process a renewal request, update the state on return and recover a reasonable service cost. It can correct an inaccurate ledger and maintain an audit history. But practical influence over a registry entry is not public authority over the operator. A private board can direct company staff within lawful corporate and service scope; a working group can discuss and report consensus.

Neither act creates authority over nonparties or turns process participation into regional ownership.

BTW’s contemporary analysis of a number registry as a settlement utility reinforces the same distinction. A settlement utility answers narrow questions reliably: which identifier is recorded to whom, in what state and with what history. It becomes a gatekeeper when those ledger questions expand into unbounded judgements about who deserves to operate. A mandate firewall is therefore not hostility to coordination; it is the condition for trusting a shared book.

LARUS adds an operator perspective on continuity and renewal risk. Its relevance is the economic cost of uncertain time at a registry interface: configuration, carrier coordination and live service do not pause merely because an administrative endpoint is ambiguous. That perspective does not make LARUS evidence of the policy’s drafting, any Board vote or any grant. It also does not transform the 2006 “lease basis” into a theory about other arrangements. The comparison ends at continuity risk.

Official AFRINIC sources occupy a different evidentiary role. They prove what the Board page, policy text, meeting reports, list archive and manual say: words, dates, reported discussions, procedural steps and the corporate act. They do not prove a community mandate, regional ownership, stewardship entitlement, public-law jurisdiction or legitimacy beyond AFRINIC’s private corporate and contractual scope. Meeting consensus is a process fact, not sovereignty by headcount.

The boundary must be explicit because the policy’s gaps otherwise tempt an institutional shortcut. Undefined recognition criteria can be mistaken for permission to judge deservingness. An unexplained renewal delay can be mistaken for a power to decide whether an event may continue. A return expectation can be mistaken for title. An expired record can be mistaken for a penalty. Each move confuses the map with the territory. The appropriate response to an incomplete ledger is a better ledger, not a larger ruler.

AFRINIC was and is a private bookkeeper and coordinator in this analysis. It has no sovereign, legislative, regulatory, police, prosecutorial, judicial, punitive or confiscatory authority. It does not become an owner or public licensor by recording, renewing or closing a temporary assignment. The record is not a licence, title deed, penalty or public command. Its expiry is an administrative state change within a narrow service, not a judgement on the legality of the network.

What the evidence cannot answer

The available record is strong on the published instrument and weak on outcomes. It does not reveal the exact day or attendance of the May 2006 Board meeting, individual votes, whether the four proposals were voted separately, the chair’s status report or a signed standalone resolution. It does not contain the original April 2005 mailing-list post or the full February 2006 last-call record. The meaning of the policy page’s 17 May 2005 metadata field remains unexplained.

More consequentially, no complete public historical register of applications, approvals, refusals, issued resources, renewals, returns, fees or closure states was recovered. There is no individual application in the available evidence. The article therefore cannot say how many temporary assignments existed, whether any request was refused, whether any entry was renewed, whether any fee was charged, whether any resource was returned late or whether reuse harmed a later operator. Current access to the policy and manual cannot prove uninterrupted implementation or present practice.

The text itself leaves organisational standing and signatory authority unspecified. It does not identify reviewers or conflict controls, set service deadlines, require reasons, offer correction or appeal, establish confidentiality and retention rules, reconcile its several clocks or define renewal lead time and ceiling. It does not specify a return acknowledgment, record-close deadline, routing-withdrawal test, reverse-DNS cleanup, quarantine or reuse check. It gives no procedure for late or disputed return. The route-taint issue was contested and unmeasured. The 2018 clarification exchange was discussion, not adoption.

These unknowns are not a rhetorical weakness to be covered with confidence. They define the difference between a policy text and an accountable service. The 2006 instrument described what applicants should disclose and the broad form of a temporary assignment. Without lifecycle outcomes, it is impossible to test whether equivalent cases were handled equivalently, whether renewals were timely, whether fees tracked costs or whether returned resources reached a clean closed state.

The evidence also sets a legal and institutional negative boundary. No available source proves that the temporary record was a sovereign licence, title, punishment, confiscation or public-law command. No missing clause can supply those powers. The only warranted conclusion is narrower and more useful: Resolution 200605.27 ratified a one-month, renewable coordination instrument whose front-end application record was considerably more developed than its back-end lifecycle account.