Summary

  • A registry handover carries three separable things — recognition of the registry, the administrative service that answers resource holders, and the registration records themselves — and each is evidenced by a different class of instrument.
  • The published material reviewed here establishes recognition and administration for the successor registry. It does not establish the record layer: no prefix-level transfer schedule, database change history or reverse-DNS delegation record for the early African registrations was obtained in this review.
  • The consequence for an operator is narrow and uncomfortable: a registry answer today describes who currently keeps an entry, not who decided the allocation that the entry records.

Ask a registry about an address block registered before that registry existed and you get a tidy answer in a database field. The answer is true. It is also answering a different question from the one most people think they asked. A whois or RDAP reply tells you who is keeping the record now. It does not tell you who made the original allocation, who authorised its movement, or which of two registry entries should be believed if they ever disagree.

That gap is not a scandal. It is a structural feature of how the internet's number registries were built, and the RIPE NCC to AFRINIC transition is the cleanest place to see it, because the transition happened in public and the record layer did not travel with the same paperwork as the recognition layer. The transition as an entity performs registry functions for internet number resources — which is a description of custody, not of authorship.

Three layers, three kinds of paper

The first layer is allocation at the top level, and it is documented separately from everything else. The framework page for internet number resources describes how addresses are distributed and administered, and treats those as distinct acts. IANA, internet number resources

The second layer is recognition. The parameter record for the regional registry system names the recognised registries and sits apart from any service arrangement a registry might enter with its members. IANA, RIR parameters IANA's top-level IPv4 record is the instrument in which allocation rows live: IANA IPv4 address space registry. The excerpt retained in this review is the registry page itself, not a read of the specific rows, so what each designation form in that record means is a question this review leaves open.

Recognition is also a collective act. The five registries act jointly through the Number Resource Organization, which is where a system-level continuity position would appear rather than in any single registry's notice. NRO

The third layer is the administrative service — the thing a member actually experiences. Here the predecessor registry publishes something unusually clarifying. Its procedure for resource transfers and mergers treats the movement of a resource record as its own transaction, not as a by-product of a change in membership. RIPE NCC, resource transfers and mergers A registry that separates those two events in its own procedure has told you, in its own words, that a member moving and a record moving are different facts requiring different instruments.

That is the whole thesis, and it needs no scandal to matter. Recognition is granted. Service is delivered. Records are handed over, or they are not, and the handover has its own paperwork.

What the transition documents show, and what they do not

The predecessor registry publishes from its own site, and its announcement index is the place where an instrument governing a handover between registries would appear. RIPE NCC RIPE NCC announcements The item that would govern the record-level movement of early African registrations was not obtained in this review, so no operative term from it is quoted, summarised or characterised here.

The successor registry publishes its own notices and member communications, and its site is therefore the primary source for its account of the transition. AFRINIC What was retained in this review is a landing record, not notice text; nothing in this article should be read as quoting AFRINIC's account of the handover.

Corporate administration of the successor registry is a matter for the Supreme Court of Mauritius, and the record classes in which such orders appear are the court's own judgment search and the MauritiusLII index used to cross-check it. Supreme Court of Mauritius, judgment search MauritiusLII No order was retrieved in this review. That means no holding, appointment, date or operative direction is asserted here, and the current corporate administration of the successor registry is not characterised in this article at all.

BTW's earlier reporting on the transition records a dated sequence: interim approval by ICANN on 30 September 2004; operational transition of registry service on 21 February 2005; full recognition on 8 April 2005; formal authority-transfer letters on 27 April 2005; and African memberships and standard service agreements in the predecessor registry recorded as ending on 31 December 2005, with African local internet registries served operationally by the successor while legal membership remained with the predecessor during the overlap.

That sequence is this desk's earlier reporting and was not re-verified against primary instruments in this run. Read it as a chronology to be tested, not as a finding.

The same limit applies to two details this desk has previously reported: two /8 blocks whose registry dates fall in May 1993, administered today by a registry that did not exist on that date; and the top-level IPv4 registry carrying blocks associated with the successor in more than one designation form, including a form expressed as administration rather than as allocation. Neither is treated here as established by the material reviewed, and the formal significance of the differing designation forms is not established at all.

The four items nobody keeps in one place

If you wanted to close the record question properly, you would need a dated, prefix-level transfer schedule naming the early registrations that moved and those excluded; the change history of the affected database objects, including source field, status value and timestamps; the reverse-DNS delegation record for the corresponding in-addr.arpa zones before and after the handover; and the custody record for the resource certificates covering the affected prefixes. One further instrument would settle disputes: any rule that determines which registry's entry governs when the two disagree.

None of those five was obtained in this review. None of them is a press release. All of them are the sort of artefact a registry produces as a matter of internal hygiene and rarely publishes as a package.

What this means if you hold the block

A resource holder who queries a registry gets an answer about the custodian of an entry. That answer is only as authoritative as the provenance chain behind it, and the provenance chain is not the same thing as the response field.

Where an early registration still sits in a predecessor record, a transfer request, a contact update and a reverse-DNS change may route through a registry with which the holder has no membership relationship at all. That is not a malfunction; it is what a layered system looks like from the inside.

Continuity of use, meanwhile, is evidenced by routing history and service records rather than by a registration entry. An operator's own archive — announcements, peering history, invoices, delegation correspondence — is stronger evidence of continuous use than any whois printout will ever be, because it was created contemporaneously by the party that was using the addresses.

Where the two registries' entries diverge, an acquiring network, a lender or an auditor receives two defensible answers and, on the material reviewed here, no rule that ranks them. Divergence is a possibility in this review, not a finding: no prefix-level disagreement is demonstrated. But the absence of a ranking rule is itself the thing to plan around, because it is the kind of gap that is cheap to notice in diligence and expensive to discover in a dispute.

What would change the answer

A dated, prefix-level transfer schedule showing early African registrations re-registered under the successor, with reverse-DNS delegations moved alongside them, would make the successor's record authority unitary and make this article's central distinction much less interesting. Nothing in the material reviewed supplies that schedule.

The distinction earns its keep in the meantime. Recognition of a registry is an act. Administration is a service. Custody of a record is a fact about who holds the paper, and the answer to that question can change without a single new allocation ever being made.