Summary

  • IANA assigned 43/8 to the WIDE Project in 1991, before APNIC existed. One /11 was later used in IPv6-transition work; the remaining 43/9, 43.128/10 and 43.192/11 stayed historical resources rather than entering the ordinary APNIC free pool.
  • On 25 March 2020 WIDE transferred those three prefixes—14,680,064 addresses—to the Asia Pacific Internet Development Trust under two conditions: sell them through the IPv4 market to networks that still needed space, and use the proceeds for Internet development in the region.
  • The important change was institutional, not numerical. A resource once administered through early research-network relationships acquired a policy status, a corporate trustee, fiduciary duties, a staged tender, transfer records, audited accounts and a beneficiary purpose that could survive its original experiment.

Seven-eighths of a network-era assumption

At first glance, 43/8 looks like abundance preserved by accident. A /8 contains 16,777,216 addresses. Take away one /11, and 14,680,064 remain. In 2020 that remainder appeared in three registry-sized pieces: 43/9, 43.128/10 and 43.192/11. The arithmetic is clean enough to tempt a simple story: an early pioneer received a huge block, used a fraction and eventually gave the rest back.

Almost every verb in that sentence is misleading.

The APIDT record says IANA assigned 43/8 to the WIDE Project in 1991. That was before APNIC had begun operating as the Asia Pacific regional Internet registry. Murai’s own account places the request in a wider experiment. In the late 1980s, he was asking Jon Postel for Class B and Class C space, and for a Class A block, as part of testing how network information centre functions might be delegated for Japan and the region. Address distribution still depended heavily on correspondence, technical trust and the judgement of a small number of people.

The mature chain—global pool, RIR policy, member request, documented need and registered delegation—was still being built.

Nor was WIDE a shell created to hold numbers. Its chronology begins with a Keio campus network in 1982, JUNET in 1984 and a WIDE research group in 1985. By 1989, WIDE records a 9,600-bps connection to NSFNET with NACSIS cooperation and a 64-kbps leased line to the University of Hawaii. The Internet Hall of Fame credits Murai with developing JUNET and founding the WIDE Project; the same sources place him in a collective of universities, engineers and institutions, not above it.

That context matters because 43/8 was a capability reserve before it became a scarce financial asset. Its original logic was regional experimentation: give a research community enough room to build, renumber and learn without requesting a new fragment for every trial. The address block did not carry a trust deed, a market valuation or a plan for disposal. Those were solutions to a later world.

The part that was used, and the part that waited

From 1992, Murai says, WIDE concentrated on IPv6 development and promotion. Some of 43/8 supported Japanese networks that had to renumber during the transition. APNIC allocated one /11 to participants in that work; WIDE retained the rest for other research and development.

This is not evidence that a single /11 was the only useful outcome of the allocation. Address space can provide option value: it lets an experiment proceed without first knowing which branches will survive. But option value is not permanent exemption from review. Once APNIC was operating in 1993, the region had an institution capable of making continuing allocations under community policy. Once IPv6 was in production but IPv4 remained commercially necessary, the unused remainder had acquired a different opportunity cost. Keeping it untouched, routing it, returning it and selling it no longer meant the same thing.

The historical label made the choice less obvious. APIDT’s FAQ is explicit: the space was assigned before APNIC and therefore sat outside ordinary APNIC management policy. It was not available to be returned to APNIC’s unallocated pool under any circumstances. A historical-resource transfer could move it to another account and, in doing so, make it current space subject to APNIC policy. That route is different from surrendering addresses into a free pool for fresh delegations.

It is also why “Murai gave back a /8” is the wrong headline. WIDE was the holder of the historical resource. The used /11 was not part of the transfer. The remaining seven-eighths did not go into a no-cost allocation queue. It went to a trust so that address rights could be transferred through the market and their monetary value redirected toward a defined public purpose.

Two conditions are not an institution

Murai announced two conditions on 25 March 2020. The unused space should be placed on the IPv4 market for organizations that still needed IPv4 addresses. The proceeds should support Internet development in the Asia Pacific. Those conditions join two moments in the Internet’s history: the scarcity value of an old protocol resource and the continuing need to finance skills, critical infrastructure and research.

But a condition spoken by a respected custodian is not yet a durable control. Who receives the addresses? Who holds the money? Which law binds the holder? Can the purpose survive the founder? Who records the sale and separates registry approval from commercial selection?

APIDT supplied a layered answer. A corporate trustee, APIDTT Pty Ltd, was incorporated in Australia. APNIC and WIDE own that company and act as guardians of the trust. The trustee is bound by the deed and by duties to act honestly, carefully and in the interests of the beneficiaries. The trust can place sale proceeds in an investment fund, with returns supporting regional Internet development through the APNIC Foundation and other channels.

The roles should not be blurred. WIDE made the transfer decision as holder of the historical space. APNIC accepted the transfer under its policy framework and maintained the registry function. APIDTT became the legal trustee. APIDT described the beneficiary purpose. The APNIC Foundation could be one spending channel. Naming every layer makes the arrangement sound more cumbersome than a pioneer’s promise. That is the point. Institutional memory is supposed to be more cumbersome than memory in one person’s head.

A market process made the conditions observable

The first tender, released on 25 May 2020, turns the abstract plan into a sequence that can be inspected. Stage 1 offered 43.0.0.0/9, divided into as many as eight /12 blocks. Each /12 contained 1,048,576 addresses. A bidder could seek one or more whole blocks at a price per address, but also had to supply identity information, an APNIC membership identifier where applicable, documents supporting need and executed transfer paperwork.

Those requirements did not make the trustee a regional registry. Commercial selection and registry eligibility remained different decisions. A high bid did not remove the need to satisfy the applicable transfer policy; an APNIC account did not guarantee success in the tender. KPMG and Maddocks Lawyers were engaged to manage or advise on process and probity. The record therefore created several receipts: an offer, an evaluated bid, a payment obligation, a registry transfer and, later, an audited financial position.

APNIC’s resource reporting confirms the unusual scale. Its membership analysis called out the /9, /10 and /11 received by APIDT. Its review of 2020 transfers identifies 43.0.0.0/9 moving in August from APIDT to Alibaba.com Singapore E-Commerce. That is evidence for the first half of 43/8, not a licence to assign the other two blocks to the same buyer. The tender itself was staged; the evidence should remain staged too.

By 30 June 2021, the trust’s audited financial statements said that a majority of the donated address assets had been sold, with the remainder intended for disposal within twelve months. A subsequent-events note says the rest was later disposed of. The accounts recorded the donated addresses at AU$586,215,074, or US$440,716,493, and showed AU$32,951,061 in assets still held for sale at the reporting date. They also showed cash of AU$68,701,690 and equity investments of AU$471,480,810.

These figures describe accounting treatment and a balance sheet, not one neat sale-price total. The donation value is not today’s address price. Cash plus investments is not automatically the gross proceeds. The remaining-asset line is not the price of every unsold prefix. What the report proves is narrower and more useful: by the reporting date, an Internet number resource had been recognized, sold in substantial part and converted into financial assets subject to an audited institutional record.

The stewardship was in the conversion chain

Murai’s career makes it easy to treat 43/8 as a personal morality play: visionary asks for addresses, visionary uses some, visionary donates the rest. Yet the durable achievement lies between those verbs.

An early delegation was legible inside a small technical community because participants knew the people and the purpose. Three decades later, neither acquaintance nor reputation could tell an APNIC member which addresses moved, under what authority, to whom, on what eligibility basis, where the proceeds sat or who owed duties to future beneficiaries. The transfer needed to replace social memory with linked records.

That conversion also preserved something from the original experiment. WIDE did not insist that the same addresses remain reserved forever. It translated “support Internet development in this region” from a technical option into a financial one. The resource changed form—from unallocated historical address space to current transferable prefixes, then to invested capital—while the purpose was written into a different custody mechanism.

This does not prove that every later investment or grant is wise. A trust deed cannot guarantee equitable distribution. An annual audit cannot measure whether training improved a network, and a clean transfer record cannot settle whether address markets are socially desirable. Those are separate evaluations. The narrower lesson is that stewardship can include retiring one form of control, provided the successor form exposes its authority, conditions and beneficiaries.

Sources