Summary
- APNIC's Tokyo experiment ran from September 1993 to June 1994 and demonstrated real registry demand, but its operating recognition came before a stable corporate person that could clearly hold assets, employ staff, contract or be sued.
- The documentary record contains two different early markers: an allocation table dates the equivalent 202/7 allocation to 10 January 1994, while APNIC's later institutional history dates public recognition or delegation to 1 April 1994. They may record different acts, not a single clean founding moment.
- The later chain moved through APNIC Ltd in Seychelles, a special committee created on 18 May 1996, APNIC Pty Ltd in Australia on 5 February 1998, and a planned 12-month membership migration beginning in March 1999; those steps regularised custody but do not by themselves prove a complete transfer of title, liability or consent.
The first problem was not whether the office worked
The Tokyo pilot should not be judged by asking whether it was useful. It was useful. That is exactly why its legal form matters.
By the early 1990s, address allocation in the Asia-Pacific region had stopped being an informal favour that could be absorbed inside a small technical circle. Networks were multiplying across economies with different laws, different research institutions, different commercial incentives and different relationships to the global Internet. A regional registry could make the system more orderly. It could reduce the delay and distance of relying on a central allocation function. It could let people closer to the operators collect requests, maintain records and explain procedures.
It could also create a new administrative actor before anyone had finished saying what kind of actor it was.
The question raised by the Tokyo experiment is therefore not whether a pilot can be legitimate as a test. A pilot often is the only practical way to build a coordination service. The question is narrower and more institutional: when a pilot receives address space, collects fees, builds records, hires or borrows labour, receives equipment, deals with applicants and becomes the regional point through which future operators must pass, who holds the registry function? Who owns the files? Who is responsible for mistakes? Who can be sued? Who can sign an employment contract? Who can be ordered to correct a decision? Who can bind future members?
APNIC's early history is usually told as a story of regionalisation. That is true but incomplete. It is also a chain-of-custody story. The regional registry did not move from idea to settled corporation in one clean act. It passed through a technical proposal environment, an Asia-Pacific coordination forum, a Tokyo pilot, consequential IANA recognition, a Seychelles company, a special committee under that company, an Australian company and a later membership migration plan. Each stage carried some power. Each stage left a different evidentiary footprint.
The governance problem is created by treating those stages as if they were interchangeable.
For operators, the difference was practical. If an applicant in 1994 thought a request had been wrongly handled, the relevant fact was not only that a registry office existed. It was whether the applicant could identify a legal counterparty. If a staff member was doing registry work, the relevant fact was not only that the work was socially necessary. It was who employed, supervised, insured and paid that person. If a creditor supplied services or equipment, the relevant fact was not only that the regional Internet benefited. It was who owed the debt. If a member paid fees, the relevant fact was not only that membership signalled support.
It was whether membership created enforceable rights in the entity actually controlling the registry.
The Tokyo prototype is important because it separates operational confidence from legal custody. It shows a registry becoming necessary before the law had caught up with the role it was beginning to play.
Recognition came before a settled container
Surviving institutional histories place the APNIC experiment in Tokyo between September 1993 and June 1994. That period was not a decorative preface. It was when the proposed regional registry became an operating service. Requests were processed. Records were made. Coordination with the global allocation function occurred. Participation extended across multiple economies. By the end of the experiment, APNIC's later history reports 27 members in 12 economies.
That figure is valuable, but only for what it actually measures. It is evidence that the experiment had service users and supporters beyond one office or one national community. It is not a denominator for all affected operators in the Asia-Pacific region. It is not proof that every network that would depend on APNIC's records affirmatively authorised the pilot. It is not a vote on corporate form. It is an adoption observation: 27 members, 12 economies, at the end of a defined experiment.
The more consequential evidence concerns delegation. The record does not present one uncontested founding date. One allocation table dates APNIC's equivalent 202/7 allocation to 10 January 1994. APNIC's later retrospective history dates public recognition or delegation to 1 April 1994. Those two dates should not be flattened into a single ceremonial origin. They may describe different documentary acts: an allocation entered on one table, public recognition announced later, operational delegation formalised through another record, or retrospective dating by a history project.
Without the underlying instruments, the honest conclusion is not to choose a winner. The honest conclusion is that operational consequences were attached to the pilot before a durable corporate home is visible in the record.
That distinction matters more than the calendar discrepancy. If the 10 January 1994 entry records the equivalent of a 202/7 allocation, then a resource with regional consequences was associated with APNIC during the pilot period. If the 1 April 1994 date records public recognition or delegation, then the pilot had been acknowledged as a regional registry actor before the end of the experiment. In both readings, recognition preceded the later corporate chain. Neither reading proves title to assets, employment arrangements, insurance cover, tax treatment, member rights or liability.
A block can be delegated to an operating registry function without answering who owns the filing cabinets, who is liable for a wrongful decision or which corporate law governs member disputes.
This is the central APNIC legal-home problem. The administrative fact of recognition did not wait for the legal architecture to become clean. That may have been operationally rational. It was also institutionally expensive. Once a registry is recognised, operators begin to treat it as real. Applicants conform to its procedures. Records acquire evidentiary value. Downstream networks rely on its decisions. The pilot stops being a test only in the laboratory sense. It becomes a public coordination point even if its own legal personality remains provisional.
APCCIRN and APNG supplied a forum, not a finished corporation
The early Asia-Pacific Internet coordination environment was not empty. APCCIRN and related APNG activity supplied discussion, institutional imagination and a regional meeting structure. That matters because APNIC did not emerge from one private company's unilateral decision. It was shaped in a community of research-network entities, operators and technical coordinators responding to a real administrative need. The early case for a regional registry was not frivolous. The region was large. Centralised allocation was increasingly strained. A localised registry could improve service and record quality.
But a forum is not a legal person. A meeting process can create consensus around an experiment. It can identify volunteers. It can document a need. It can recommend a structure. It can give the experiment social authority among those who participate. It cannot automatically hold title to assets unless incorporated or otherwise legally recognised. It cannot automatically employ staff. It cannot automatically accept liability. It cannot automatically bind non-entities merely because their networks are located in the same region.
The distinction is not pedantic. Internet governance often treats technical participation as if it naturally becomes institutional consent. It does not. Participation can be strong evidence of support among entities. It can be weak evidence of consent among the wider affected class. The early APNIC case has both elements: there was real use and regional discussion, but the surviving record does not supply a complete denominator of all affected operators or a formal authorisation process from them. The 27 members in 12 economies show that APNIC was more than a name.
They do not prove that the registry power had a clean mandate from the region as a whole.
The same caution applies to IANA recognition. IANA could recognise a regional registry function for operational purposes. That made APNIC's records consequential. It did not create a regional corporation. It did not identify the employer of every worker. It did not transfer equipment title. It did not create a comprehensive appeals process. It did not settle which law would apply if a member, applicant or creditor challenged the registry. Recognition solved one coordination problem and left several custody problems open.
This does not make the early entities irresponsible. It makes their improvisation visible. They were building an administrative service faster than corporate paperwork could be stabilised. That is a common pattern in infrastructure institutions. The danger comes later, when retrospective histories use continuity of service to imply continuity of authority. Service continuity is evidence that people kept the registry operating. It is not evidence that all rights and liabilities travelled cleanly with the name.
What exactly had to be kept in custody
The custody question becomes clearer if the registry is divided into specific things rather than treated as one symbol.
First, there were records. Address allocation requests, decisions, contact information, membership records, meeting minutes and correspondence had administrative value. They were not just files. They were the proof by which later operators could show what had been allocated, on what basis and by whom. Possession of the records gave practical control. Title to the records, duties of care over the records and obligations to correct them were separate questions.
Second, there was equipment. A pilot office needed terminals, printers, storage, network access, telephones, fax capacity, office space and perhaps donated or borrowed systems. Equipment can be possessed without being owned. It can be used under a host arrangement without being transferred. It can be donated subject to conditions. The surviving public summaries do not supply a full executed inventory showing how every item moved from the Tokyo host environment to later APNIC entities.
Third, there was labour. People processed requests, maintained lists, answered questions, prepared records and coordinated with other institutions. In a pilot, some labour may be contributed by a host, seconded by another organisation, funded through fees or performed under informal arrangements. That may be adequate for a test. It is not the same as a transparent employment chain. The legal question is who employed staff, who could direct them, who insured them and who bore liability for their acts.
Fourth, there were fees and member relationships. Once a registry collects or expects support from members, the issue is no longer only technical. Members may believe they have rights. They may expect voting, notice, access to records, service standards or recourse. The later APNIC by-laws are important precisely because they distinguish old membership from rights under the new arrangement. They show that membership continuity could not simply be assumed.
Fifth, there was the address delegation itself. The delegated resource was not a physical asset sitting in the Tokyo office. It was a recognised administrative authority over entries in the global number system. That is why relocation-cost figures are not a measure of resource value. The 1998 annual report gives two inconsistent 1997 relocation-cost figures, US$18,194 and US$18,914, and reports US$21,182 for 1998. Those figures speak to reported relocation expenditure, not the value of delegated address space or the institutional authority attached to registry decisions.
Sixth, there were liabilities. A wrong allocation, an inconsistent policy application, a lost record, a payment dispute or a staff claim would require a defendant or accountable body. The record used here does not show a contemporaneous legal opinion explaining who could be sued or ordered to correct a registry decision before the later corporate steps. The absence of that record does not prove that no answer existed. It means public claims about clean institutional authority should be made with restraint.
Once APNIC is analysed in these categories, the phrase "the registry moved" becomes too vague. Possession of records may have moved at one time. Staff arrangements may have changed at another. A legal person may have arrived later. Member rights may have been planned for migration still later. Delegation may have been recognised before all of those. A serious governance history has to keep those layers apart.
Seychelles regularised existence but did not erase the pilot problem
APNIC Ltd was incorporated in Seychelles on 30 April 1996. A special committee under that entity was established on 18 May 1996. Those dates are crucial because they show the first visible arrival, in this chain, of a corporate container capable of holding rights and obligations in a way the Tokyo pilot could not obviously do on its own.
Incorporation changes the institutional surface. A company can hold assets. It can contract. It can employ. It can be sued. It can maintain accounts. It can have internal rules. It can act through a board or committee. It can give members a defined legal relationship, depending on its constitution. That is why the 30 April 1996 date matters. It is not merely an administrative milestone. It is the point at which the registry story begins to acquire a recognisable legal body.
But incorporation in 1996 does not retroactively answer what happened in 1993 and 1994. It does not by itself prove who owned the Tokyo pilot's records before incorporation. It does not prove that every pre-existing liability was assumed. It does not prove that every piece of equipment was transferred. It does not prove that all affected operators consented to the new company as their regional registry authority. It does not prove that the IANA-recognised function and the corporate entity were joined by a complete, executed transfer instrument.
That is not an accusation. It is a chain-of-custody limit. A later container can regularise a messy service. It cannot make the earlier mess disappear without records showing how title, liabilities and rights were moved.
The Seychelles choice also deserves careful treatment. The question here is not whether Seychelles law was good or bad for APNIC. The evidence used for this article does not support a full jurisdictional comparison, and the later Brisbane move belongs to a separate inquiry. The narrower point is that a corporate home was selected after the registry had already become operationally consequential. That home introduced legal personality, but it also created a new question: what exactly did APNIC Ltd receive from the prior pilot, and by what instrument?
The special committee created on 18 May 1996 should also be kept distinct from the company itself. A committee can manage, advise or administer under an entity's rules. It is not the same thing as the legal person. If a member had rights, those rights depended on the company's governing documents and the committee's authority within them. If a creditor had a claim, the target would not simply be the phrase "APNIC community." It would depend on the contract, the entity, the committee's authority and the relevant law.
If an applicant sought review, the existence of a committee might provide a practical route for complaint, but that route would not necessarily equal an enforceable appeal unless the governing documents created one.
The 1996 step therefore deserves neither dismissal nor myth. It was a real institutional improvement. It gave the registry a corporate surface. It did not produce, on the public record alone, a perfect retrospective conveyance from the Tokyo experiment.
The Australian company created another surface, not a magic merger
APNIC Pty Ltd was registered in Australia on 5 February 1998 as ACN 081 528 010. The 1998 annual report presents the Australian incorporation as part of a wider transition. The by-laws dated 24 June 1998 distinguish the special committee established under APNIC Ltd on 18 May 1996 from the APNIC special committee established under APNIC Pty Ltd on 24 June 1998. The same materials indicate that old membership did not automatically confer rights under the new arrangement. Existing APNIC Ltd membership rights continued while migration to APNIC Pty Ltd was planned to start in March 1999 and take 12 months.
This sequence is the strongest reason not to write APNIC's institutional history as a single incorporation story. If Australian registration had automatically settled every right, there would be little need to describe a later membership migration. If the old membership automatically became full membership under the new company, the by-law distinction would be unnecessary. If all assets and liabilities were transparently conveyed in one public act, the annual report would not leave the public reader dependent on general transition language rather than a complete transfer schedule.
The Australian company mattered because it created a new legal person under a different jurisdiction. It opened a local-law accountability surface. It could employ staff, hold accounts, contract, take assignment of assets and become the entity through which members related to APNIC. It also created a moment when continuity had to be proven rather than assumed. A new company does not become the old company's rights holder merely because it shares a name, function or staff. It needs assignment, novation, assumption, membership migration or some other legal mechanism, depending on the asset or obligation.
The public record described in the 1998 report does not prove the completion of every transfer. It tells us that a migration was planned. It tells us that the process was expected to begin in March 1999 and take 12 months. It tells us that tax work and staff replacement were part of the transition environment. It reports relocation costs, with the inconsistency in the 1997 figure already noted. It does not give a complete schedule of records, equipment, contracts, liabilities, member rights and delegated authority travelling from the Tokyo pilot to APNIC Ltd and then to APNIC Pty Ltd.
That does not mean the transfers failed. It means the public institutional story is thinner than the operational continuity story. The service continued. Continuity of service is important evidence. It shows that APNIC did not collapse under the weight of its legal reorganisation. It also shows why operators would have accepted the transition pragmatically: the registry kept functioning. But a registry's ability to keep functioning is not identical to a legal audit of custody.
An employee would see the difference quickly. Before a stable employment chain is documented, a worker may not know whether the host organisation, APNIC Ltd, a committee or a later Australian company is the employer for a given period. A creditor would see it too. An invoice for office services, equipment, travel or professional advice has to be issued to someone. A member would see it in voting and notice rights. A right under APNIC Ltd was not automatically a right under APNIC Pty Ltd. An applicant would see it if a registry decision had to be appealed, corrected or challenged.
Operational continuity is comforting until someone needs an enforceable counterparty.
Why the January and April dates should remain uncomfortable
Founding histories prefer clean dates. Institutions like anniversaries. Registries like continuity. Public narratives often select one moment and let it stand for the whole transformation. APNIC's early record resists that habit.
The 10 January 1994 allocation-table date and the 1 April 1994 recognition or delegation date are not a minor footnote. They reveal that different records may be capturing different acts. One may identify an allocation entry. Another may identify public recognition. Another may reflect retrospective institutional presentation. Without the underlying instruments, a responsible analysis should not infer more than the records can bear.
The important governance point is that both dates sit inside the pilot period, before the 1996 Seychelles incorporation and before the 1998 Australian company. That puts operational recognition ahead of corporate settlement whichever marker is used. If January is treated as the relevant allocation marker, APNIC's regional function was already attached to consequential number resources very early in the experiment. If April is treated as the public recognition marker, APNIC was still publicly recognised before the pilot ended and before incorporation.
Either way, the pilot's authority cannot be explained by corporate personality at that time.
This does not invalidate the delegation. It reframes it. IANA could make an operational judgement that a regional registry function was needed and that the APNIC experiment was the appropriate vehicle. That judgement could be sensible. It could be necessary. It could be based on real entities and real demand. But it was not the same as a conveyance of all legal rights. It was an administrative recognition of a function.
That distinction matters because number-resource governance often uses delegation language to carry more legitimacy than it can support. A delegation can authorise operational entries in a global registry system. It does not necessarily prove democratic consent. It does not necessarily identify the entity that owns records. It does not necessarily create member rights. It does not necessarily answer whether the registry can be sued in a convenient forum. It does not necessarily explain how historical liabilities travel when the registry later incorporates.
The January-April discomfort should therefore be preserved. It is a useful guard against retrospective smoothing. It tells later analysts to ask what document did what work. A table entry, a public recognition announcement, a meeting minute, a corporate incorporation, a by-law and a membership migration plan are different instruments. Treating them as one founding event makes the institution easier to celebrate and harder to audit.
The countercase deserves respect
The strongest defence of APNIC's early sequence is not sentimental. It is practical.
The Asia-Pacific region needed a registry service before a perfect corporate design was available. The technical system could not wait for every jurisdictional and membership question to be solved. A pilot is a rational way to test demand, build procedures and learn what staff and records are required. If the experiment had waited for a full legal architecture, operators might have faced slower allocations, weaker regional knowledge and continued dependence on distant administration. The early entities were not building a luxury institution. They were solving a coordination problem under time pressure.
The evidence supports that countercase. The pilot was not a paper-only exercise. It had reported membership across 12 economies. It received recognition or delegation from the global allocation function. It processed real work. Later relocation was reported as low-cost and operationally continuous. The move toward corporate form did not destroy service. The later Australian incorporation created a more conventional accountability surface. The planned membership migration shows an awareness that member rights had to be regularised rather than ignored.
This defence should be given full force because it prevents the custody analysis from becoming an anachronistic demand that a 1993 experiment behave like a mature institution from the start. Early Internet governance often grew through pilots, mailing lists, research networks and voluntary coordination. Formality followed function. In some cases that sequence was the only way anything useful happened.
But the countercase proves usefulness, not completeness. It explains why the pilot existed. It does not prove that the pilot had a clean legal counterparty. It explains why recognition might precede corporate form. It does not prove that recognition supplied title or member consent. It explains why later continuity mattered. It does not prove that every right and liability moved cleanly from host arrangements to APNIC Ltd and then to APNIC Pty Ltd.
The fairest conclusion is therefore mixed. APNIC's early improvisation was operationally defensible. Its later legal record should not be described as if operational defence and legal custody were the same thing.
How different stakeholders would have read the chain
An applicant in the Tokyo period would have encountered APNIC as a registry office, not as a legal theory. The applicant's immediate concern would be whether a request was accepted, delayed, modified or rejected. For ordinary service, the pilot could function. For a dispute, the harder questions would surface. Was there a written appeal right? Against whom would a correction order run? Which body had authority to review staff action? If the applicant was outside Japan, what forum would be practical? The available public history does not supply a full answer.
An employee or worker would see a different version of the same problem. Registry work involves judgement and responsibility. It can expose a worker to pressure from applicants, members, host institutions and global coordinators. A durable employer matters because it defines pay, supervision, confidentiality, insurance, intellectual property, workplace obligations and liability. In a pilot hosted through research-network arrangements, some of those answers may have been supplied by the host or by informal understandings. That may have worked in practice. It is not equivalent to a public employment chain covering the whole pre-1998 period.
A creditor or supplier would ask who could be invoiced. Office space, telecommunications, professional services, printing, travel, systems and legal work all require an obligor. If a pilot uses borrowed capacity, the supplier may be the host itself rather than an external creditor. If a committee approves spending, the supplier still has to know whether the committee binds a company, a host organisation or no one beyond the individuals involved. Incorporation helps because it lets a company become the contracting party. It does not explain all pre-incorporation obligations unless assumption or reimbursement records exist.
A member would ask what membership meant. The reported 27 members in 12 economies at the end of the experiment demonstrate support and use. But membership in a pilot, membership in APNIC Ltd and membership in APNIC Pty Ltd are not automatically the same legal relationship. The 1998 by-law distinction and the planned March 1999 migration make that point directly. Old membership rights continued while migration was planned. That is a sign of care. It is also evidence that a member right had to be moved through procedure, not presumed by institutional memory.
IANA would have seen a regional coordination need. From the global allocation perspective, the key question may have been whether there was a capable regional body to receive and manage delegation. That is an operational test. It is not the same as asking whether every affected operator had authorised the body or whether all internal legal arrangements were mature. Global coordination systems often rely on capability before they demand perfect governance. APNIC's early case fits that pattern.
The public would see almost nothing unless a dispute forced the issue. That is typical of infrastructure governance. Legal ambiguity can remain invisible while service works. It becomes visible when someone asks for correction, damages, voting rights, asset title or an injunction. The absence of a known dispute is not proof that the structure was clean. It may mean the service worked well enough that the unresolved questions stayed latent.
What the 1998 records do and do not prove
The 1998 annual report is one of the most useful documents because it speaks from inside the transition rather than from a distant founding myth. It records APNIC Ltd's Seychelles incorporation, APNIC Pty Ltd's Australian registration on 5 February 1998, the ACN 081 528 010 identifier, the planned membership migration from March 1999 over 12 months, staff replacement, tax work and relocation expenditure. It is therefore strong evidence for dated corporate steps and management concerns.
It is weaker evidence for complete custody. Annual reports are management records, not independent legal audits. They can report that a transition is underway without attaching every assignment, novation, employment contract, insurance schedule, asset register or liability assumption. They can describe continuity without proving every legal path by which continuity was achieved. They can state plans without proving completion. The public reader should treat the report as a key record, not as a substitute for the underlying instruments.
The relocation-cost discrepancy is a small but useful warning. The report gives US$18,194 and US$18,914 for 1997 relocation cost, and US$21,182 for 1998. The difference between the two 1997 figures is not institutionally decisive. It does not undermine the entire report. But it shows why exact numbers should be handled carefully. More importantly, none of these amounts values the delegated number resources. They are reported relocation expenditures. They do not measure the economic or administrative value of being the registry for a region's number-resource records.
The by-laws dated 24 June 1998 are equally important for a different reason. They distinguish the special committee under APNIC Ltd from the APNIC special committee under APNIC Pty Ltd and indicate that old membership did not automatically confer rights under the new arrangement. This is not a mere technicality. It proves that APNIC's own legal transition materials understood that continuity of name and function did not automatically settle membership rights.
That point should discipline the whole history. If old membership needed migration, other rights and obligations may also have needed documented transfer. Some may have been transferred. Some may have been replaced. Some may have expired. Some may have continued under old arrangements for a period. The public record used here does not allow a complete map. The right conclusion is not suspicion for its own sake. It is a ranked custody finding based on what each document can actually prove.
Ranking the custody chain
The APNIC chain is strongest at the level of operational continuity. The Tokyo pilot existed, served users, attracted reported members across multiple economies and received consequential recognition or delegation. The later service continued through corporate changes. If the question is whether APNIC functioned as a registry, the evidence is strong.
The chain is also strong for the existence of later corporate containers. APNIC Ltd was incorporated in Seychelles on 30 April 1996. A special committee under it was established on 18 May 1996. APNIC Pty Ltd was registered in Australia on 5 February 1998 as ACN 081 528 010. A new committee under the Australian company was established on 24 June 1998. These are dated legal facts, not merely narrative claims.
The chain is moderate for planned membership regularisation. The 1998 materials show that existing APNIC Ltd membership rights continued while migration to APNIC Pty Ltd was planned to start in March 1999 and take 12 months. That proves awareness and intended procedure. It does not, from the records considered here, prove completion for every member or show every member's consent.
The chain is thin for pre-1996 title, employment, insurance, creditor liability and appealability. The Tokyo pilot's host arrangements, equipment custody, employment basis, complete asset register and contemporaneous legal advice are not available in the public record used here. The absence of those records does not prove misconduct. It limits what can be claimed.
The chain is weakest for regional consent as a public-law style mandate. The 27 members in 12 economies show adoption by entities. They do not establish a denominator of all affected operators. IANA recognition shows operational acceptance by the global allocation function. It does not prove a regional plebiscite, democratic authorisation or enforceable rights for non-entities. The later corporate steps improve accountability but cannot retroactively create consent where the record does not show it.
This ranking yields a precise conclusion. APNIC's Tokyo prototype was a successful operating experiment before it was a fully regularised legal institution. IANA recognition and service use made the registry consequential. Seychelles incorporation and later Australian registration regularised important parts of the chain. They did not, without the missing transfer and host records, prove a clean retrospective custody path for every asset, liability, member right and accountability claim.
The remedy is archival, not rhetorical
The useful remedy is not to relitigate whether APNIC should have existed. It should have existed. The useful remedy is to publish a custody file adequate to the authority the registry acquired.
That file would include the executed host arrangements for the Tokyo pilot, equipment and records inventories, staff or secondment agreements, insurance documents, fee and bank-account arrangements, committee authorities, legal advice on pre-incorporation liability, assignment or assumption instruments from the pilot environment to APNIC Ltd, transfer schedules from APNIC Ltd to APNIC Pty Ltd, evidence of the planned 12-month membership migration's completion and the instrument by which IANA recognition attached to each legal container. It would not need to disclose private personal data.
It would need to show what moved, when, under whose authority and with what rights preserved.
The measurable transfer test is simple. For each custody category - records, equipment, staff, fees, contracts, liabilities, member rights and delegated registry authority - the archive should identify the holder before transfer, the holder after transfer, the instrument authorising the change, the effective date, the consent or notice mechanism, and the remedy available to an applicant, employee, creditor or member. If a category cannot meet that test, the honest label is not failure. It is undocumented continuity.
The archival test that could change this finding is equally concrete. A complete set of executed host agreements, transfer schedules and membership migration records could upgrade the custody chain from operationally strong but legally thin to legally well documented. A contemporaneous legal opinion identifying who could be sued or ordered to correct a registry decision during the Tokyo period would narrow the accountability gap. A complete record of member migration beginning in March 1999 and finishing within the planned 12 months would strengthen the rights-continuity claim.
A document explaining how the January 1994 allocation marker relates to the April 1994 recognition marker would resolve the date conflict without guesswork.
Until those records are public, the most accurate description is deliberately restrained. APNIC did not begin as a lawless nameplate. It began as a real service inside a regional coordination effort, received operational recognition before incorporation, then searched for legal containers that could carry an authority already in motion. That history is not discrediting. It is instructive. It shows how Internet registry power often forms first as custody of a necessary ledger, and only later as a legal institution capable of explaining who owns the ledger, who answers for it and who has enforceable rights when the ledger matters.