Summary

  • Paul Wilson’s documented contribution was membership in a nine-person, five-region team behind prop-069. The public record does not assign him sole authorship or personal control of the process.
  • The first proposal paired local recovery policies with mandatory returns and a needs-based IANA allocation rule. A fifth regional text diverged from the other four, and the proposal was abandoned.
  • The second attempt used an eligibility rule that raised the possibility that one RIR could take the available pool. The third narrowed the global rule to equal redistribution among all five RIRs while leaving the return process outside its scope.
  • The final policy became global only after separate regional adoption paths, NRO and ASO handling, and ICANN Board ratification in May 2012.

The author line is a beginning, not a biography

The surviving documentary fact about Paul Wilson and the first global proposal is unusually concrete. APNIC lists his name, with his affiliation, as one of nine members of the team that developed prop-069. The other eight authors came from AfriNIC, APNIC, ARIN, LACNIC and RIPE NCC. The page marks the proposal abandoned. Its history records an APNIC Policy SIG consensus, member-meeting endorsement, an eight-week final comment period and later APNIC Executive Council action. Those entries establish participation and chronology. They do not tell us which author drafted which paragraph, who persuaded whom, or how much Wilson contributed to the final language. APNIC’s prop-069 record and its version-three text make the distinction visible: the team is named, but individual drafting shares are not.

That is a useful starting point for a profile because the policy itself was built around distributed authority. A proposal carrying the label “global” did not pass through one world legislature. It had to be introduced, discussed and accepted inside five regional Internet registry communities, then move through the Number Resource Organization and Address Supporting Organization structures before the ICANN Board could ratify it. A cross-regional author team could write a common draft. It could not make that draft binding by listing the names of participants from all five regions.

Wilson’s institutional position matters as context, but it should not be substituted for evidence about the policy work. APNIC’s conference biography says he became its Director General in August 1998. APNIC records that he completed the role on 30 June 2024. A director general’s presence in a cross-regional author team does not prove that he personally determined the proposal’s terms or controlled decisions in other regions. The safest account is also the most informative one: Wilson was one APNIC-affiliated member of a team whose document entered several distinct policy systems.

The article therefore follows the policy’s changing design rather than constructing an undocumented story about one person’s private influence. The three proposals show where a common rule touched local recovery policy, where different runout rates created an allocation problem, and how the eventual rule separated the source of returned space from the distribution of that space. Wilson’s role is the human entry point to that record, not a claim that the record belongs to him alone.

The problem appeared when the old source of supply ran out

For years the IANA-to-RIR allocation system could treat the central IPv4 pool as the source from which regional registries received large blocks. That arrangement could not continue unchanged once IANA’s unallocated pool was exhausted. RIRs still expected that some addresses might later return: a holder could give space back, an RIR could recover unused resources under its regional policy, or a legal process could change control. But those possible returns would not arrive as a new large stock under the old allocation rule.

The policy question was not simply whether IPv4 remained scarce. It was where to place recovered space and how to allocate it after the central pool had been depleted. The answer affected the boundary between a regional registry’s authority over its own inventory and the role of IANA as a common allocator. The proposal designers had to define both a way for returned resources to enter a shared pool and a rule for distributing them back to regions.

The distinction matters economically. A regional policy could decide which address blocks were recoverable, how to handle an organization that no longer used them, what evidence established a valid return, and how much administrative effort to spend on the record. A global distribution rule could then decide which registries received the returned supply. If the global rule required a particular kind of return, it would reach backward into local recovery practice. If it ignored how much each region contributed, it could create incentives or perceived inequities in who received the pool.

Prop-069 tried to solve both sides inside one policy. That was its ambition and the point where the legal and economic detail became difficult. A common pool can make returned supply usable beyond the region in which it was recovered. But the common pool also makes local recovery decisions relevant to every region’s future allocation. Whether a registry has a duty to return recovered addresses is not a small bookkeeping detail if the return changes the allocation rights of five regional communities.

The ICANN background report on the third proposal describes the central premise: the IANA free pool had been exhausted, yet returned addresses could still become available. It names the 2009, 2010 and 2011 efforts as the first, second and third proposals for handling that situation. The sequence was not a repeated vote on identical wording. Each iteration moved the boundary between what the global rule had to specify and what remained subject to regional policy.

What prop-069 put into the shared rule

The version of prop-069 that APNIC published in March 2009 began by defining the inputs. It called recovered address space space returned to an RIR because it was unused, voluntarily returned, or reclaimed through legal action or an abuse determination. It excluded a particular category of fee-related reclamation: space reclaimed for nonpayment of contractual fees less than one year before the RIR’s report. That exception shows the authors understood that a record of reclamation could have causes with different meanings. They did not treat every block that reappeared in an inventory as economically or operationally equivalent.

The proposal also defined an RIR’s holdings. Its total was not merely the space already allocated to customers. It included unallocated address space, recovered space not yet returned to IANA, and an adjustment for space committed under the region’s reservation policy and practice. This definition would drive eligibility. Before asking how the pool should be shared, the policy needed to decide which regional inventory counted and which commitments reduced the amount that could be treated as available.

The design then separated activity into two phases. In the first, IANA would establish a recovered IPv4 pool. Each RIR would continue to recover space under its own policies and strategies, but the proposal said it would return recovered space quarterly in aggregated blocks of /24 or larger. No allocation would be made from the pool while this collection phase was under way. This timing created a common reservoir before any region received a new allocation from it.

In the second phase, following ICANN ratification and a declaration that the old IANA free pool was depleted, IANA would begin allocating from the recovered pool. The policy divided time into six-month allocation periods beginning on 1 March and 1 September. At the start of a period, the allocation unit would be one-tenth of the recovered pool, rounded down to a CIDR boundary, with a minimum size of /24. Each RIR could submit one request during the period, and an eligible RIR would receive one unit.

The rule did not promise that a unit would correspond to one continuous prefix. The proposal said IANA would use the smallest possible number of blocks in the available pool to make an allocation. Its examples recognized that an allocation could be assembled from multiple smaller blocks. This is a practical point: a policy that describes quantities but ignores block geometry can create an administrative promise the address inventory cannot fulfill.

Eligibility was based on holdings rather than an open-ended comparative judgment by IANA. An RIR could receive a unit when its total holdings fell below half of the current allocation unit and it had not already received a unit in that period. The proposal called the allocation needs-based. Its “need” test was therefore a threshold built from the defined inventory, not an assessment of population, user demand, economic output, or the merits of a region’s IPv4 policy.

The text also addressed new registries. A newly recognized RIR would be allocated one unit regardless of projected utilization, with the allocation made as soon as a unit became available if it could not be provided at recognition. That clause mattered because a global rule designed around the existing five registries could otherwise make entry by a future region structurally different. The provision was not a demand forecast; it was a defined treatment of a registry at the point of recognition.

Finally, prop-069 required public accounting. Returned and allocated blocks, their dates and the relevant RIR would be recorded in an IANA transaction log; IANA would also maintain a public registry of address dispositions. The proposal did not set performance requirements for IANA’s service. It said those should be handled through appropriate agreements among RIRs and ICANN. The draft thus paired a new allocation mechanism with an information trail, while drawing a line around the operational service contract.

Why the return clause became a regional fault line

A global pool depends on what enters it. Prop-069’s first phase made the RIRs the recovery agents but made quarterly return a mandatory part of the global design. That combination left local organizations with responsibility for identifying recovered space while changing the consequences of recovery beyond their own membership. A region could reclaim a block under its own policy, but the proposal’s rule would direct that block into a shared pool.

That arrangement can be defended as a way to prevent a regional registry from retaining all of the upside when the addresses it recovered might support other regions. It can also be criticized as a global rule controlling a decision whose inputs were produced by local policy and local legal conditions. The proposal’s authors tried to keep regional recovery policies intact: each RIR would choose its own policies and strategies for recovering space. But mandatory return limited the choices after recovery. Local control of one stage did not mean local control over the disposition of the resulting inventory.

The 2011 LACNIC proposal says the original return obligation was a significant concern to the ARIN community. ARIN’s archived policy page supplies the concrete textual history. In its local version, the ARIN Advisory Council changed the clause so an RIR could decide under local policy which recovered space to designate for return. The earlier text directed that recovered space, in blocks of /24 or larger, be returned to IANA at quarterly intervals. The revision made designation optional.

That was not merely a change of verb. It moved a decision right. Under the original, recovery would feed the pool subject to the global requirement. Under the ARIN revision, a region could apply its own policies and strategies, then decide whether to place recovered space into the pool. A regional community concerned about the scope of recovery could accept the purpose of post-exhaustion redistribution while rejecting a universal obligation to contribute every qualifying block.

The difference also changed the economic exposure of the regions. If one policy defines which assets enter a shared pool, participants may worry about how disputes over legacy status, nonpayment, legal claims, use, or administrative records are resolved. The global policy might be formally limited to IANA-to-RIR transfers while still depending on regional judgments about what can be returned. That dependency is why a seemingly narrow “return” clause can carry distributional weight.

The ARIN page records that its version was discussed at ARIN XXIII and that the changed text would have to be reconciled with versions approved elsewhere. The third proposal’s retrospective account says the first proposal reached consensus at four RIRs and board endorsement there, while the fifth region’s text was substantially rewritten to address local concerns. It says the resulting version would have been difficult to reconcile with the versions that had consensus in the other four regions.

That account is important but should be read with source discipline. It appears in a later proposal document written by people advocating a revised design. It is not a transcript of every regional debate or an independent adjudication of motive. ARIN’s version history corroborates a substantive textual difference and explains the local change. Together, these sources support a precise conclusion: a proposal with consensus in four regions did not yield a single version ready for global adoption once the fifth region’s policy process produced different text.

The word “consensus” can obscure that distinction. A consensus within one regional process establishes a decision under that region’s rules. It does not make a different version equivalent to the text accepted elsewhere. Nor does a region’s board endorsement replace consensus in another region. The global project therefore faced a compatibility problem before the allocation formula could be used: the participants needed agreement on the same operative language.

Why a global text was not a global vote

The number of institutions involved did not create a single institutional chain of command. Each RIR had its own policy development process, membership, staff, meeting schedule and way of recording consensus or approval. The proposal team could assemble a document across regions, but it did not have a power to impose that document on a region whose process had not accepted it.

That structure has a cost. Five processes take time, and a document can be edited differently as it passes through each one. It also has a benefit: a common rule that affects regional allocation must be acceptable to the organizations that would implement it. One region’s formal approval cannot be treated as authorization for four others. Requiring separate decisions makes differences visible before a rule is installed in the shared operating system.

The process was layered. The regional proposal stages determined whether each region had an accepted version. The NRO and ASO arrangements provided a cross-regional path for handling candidate global policies. The ICANN Board’s role was ratification at the end of that process, not a substitute for the RIRs’ regional work. That sequence can appear slow beside a conventional hierarchy in which a central authority issues a rule. Its purpose was to create common authorization from several systems rather than assume that the “global” label supplied it.

For Wilson, this process makes the author list informative but bounded. He was an APNIC-affiliated participant in the team that proposed a common design. His membership places him inside one point of cross-regional coordination. It does not tell us how he voted in every forum, whether he backed the local revisions, or what role he played in the later proposals. A profile that moves from name on the author line to personal architect of global policy would claim more than the sources show.

APNIC’s own record helps establish the difference between a regional milestone and a global outcome. It says prop-069 reached consensus at APNIC 27 in February 2009, that suggestions were incorporated into version three, and that the Executive Council endorsed the proposal in May. It also records a final-call process. Those are substantive steps. Yet the page now marks the proposal Abandoned, and its 2011 history says the proposal had been superseded by prop-097. Local progress did not guarantee cross-RIR completion.

The same distinction applies to the 2010 and 2011 texts. A proposal’s acceptance in ARIN, APNIC, LACNIC, RIPE or AfriNIC was one regional outcome. Global policy required the relevant text to move through all five regional processes, then through cross-RIR and ICANN stages. Treating the procedure as one vote would erase the very source of the difficulty: regional communities could accept different versions or reach decisions at different times.

The second attempt changed the trigger

After the first global effort foundered on divergent text, some members of the ARIN community wrote a second proposal, called GPP-IPv4-2010 in ICANN’s background report and prop-086 in APNIC. The APNIC version was authored by seven people, none of whom is listed as an author of prop-069’s cross-RIR team. It was not a Paul Wilson proposal.

Prop-086 addressed the same post-exhaustion setting, but it changed the trigger and return design. It proposed activating a reclamation pool once the first RIR exhausted its inventory after receiving its final /8. An RIR would become eligible after it publicly announced exhaustion, with the draft defining that condition partly by inventory below the equivalent of a /8 and inability to make assignments at a relevant minimum unit. The pool would be divided among the regions eligible in a given period.

The division rule was designed to distribute resources evenly among the eligible RIRs, with leftovers held until CIDR-compatible units could be allocated. That sounds like a safeguard against unequal treatment. The difficulty was that eligibility itself differed by region. If regional runout policies or rates were different, one RIR could become eligible while the others were not. In that period, “equal distribution among eligible RIRs” could mean that one region received all of the pool.

The 2011 LACNIC authors identified that as a risk in their proposal rationale. They argued that a region with a higher allocation rate, or one whose local policies allowed a different path through the final addresses, might claim the pool in the first allocation period. Regions that had reserved some of their last addresses to support a smooth IPv6 transition could be disadvantaged because their inventories would decline more slowly. This was a scenario the authors wanted the policy to avoid, not evidence that a region actually drained a recovered pool.

This objection shows the difference between equal treatment within a selected set and equal access to the selection mechanism. If only one RIR meets the eligibility rule, it is possible to divide the available pool equally among eligible RIRs and still produce a highly concentrated global outcome. The problem is not arithmetic; it is how the rule defines who enters the denominator.

Prop-086 also dealt with transfers and the status of returned space. Its authors sought to make the pool useful for regional allocations without permitting uncontrolled transfer of addresses from the pool in the absence of a coordinated global transfer policy. The text’s restrictions reflect the period’s concern that movement of addresses among regions could create policy asymmetry. But addressing transfers inside the same proposal introduced another set of questions that had not been resolved by the original global return mechanism.

ICANN’s 2011 background report provides the process summary: the second proposal was introduced in 2010, rapidly adopted in ARIN, abandoned in APNIC and withdrawn in RIPE. The report concludes that it appeared unlikely to advance as a global policy. This was not a rejection by a single world body. It was the result of a proposal’s regional trajectory failing to produce the common set of adoptions required for a global outcome.

The third attempt moved the boundary

The third proposal, GPP-IPv4-2011, became prop-097 at APNIC. Its author list is different: Alejandro Acosta, Nicolas Antoniello, S. Moonesamy, Douglas Onyango, Medel Ramirez and Masato Yamanishi. Paul Wilson is not among them. That fact matters for attribution. The later design followed a debate that included prop-069, but the documentary record does not support describing Wilson as its author.

The third proposal made a consequential choice about scope. It governed the process by which IANA would allocate IPv4 resources to RIRs after the central pool was exhausted. The process by which space would be placed into IANA’s recovered pool was explicitly outside its scope. The return of space and the distribution of returned space would be treated as separate policy questions rather than bound together in one global text.

That separation did not answer every political question. It removed the obligation to create a single global rule for how each region recovers addresses. A regional community could keep its own policies for voluntary returns, reclamation or recovery. The common policy then had a narrower task: when addresses had entered the IANA pool, how would they be allocated back to the regions?

The third proposal also changed eligibility from region-by-region runout to simultaneous treatment. ICANN’s background report describes the pool becoming active when one RIR had less than half of its last /8 remaining. If the pool had enough address space, IANA would allocate an IPv4 allocation unit to each RIR. The proposal’s minimum unit was /24. This design meant a single regional trigger could activate the distribution rule for all five RIRs, instead of making each region wait for its own distinct exhaustion announcement.

The difference from prop-086 was material. A trigger based on the first RIR crossing a threshold could bring all regions into the same distribution period. It treated the pool as a common post-exhaustion reserve, not as an inventory awarded only to whichever RIR reached a regional exhaustion test first. The rule could therefore avoid the concentration scenario identified by the 2011 proposal authors.

The redistribution formula was equal across the five regions. ICANN’s ratification notice says the global policy allows returned IPv4 space to enter an IANA-managed pool, from which equal parts will be allocated to all five RIRs in a deterministic manner on a regular schedule after activation. “Equal” referred to the five regional registries. It did not mean equal shares for countries, operators, users or network sizes. Nor did it allocate address space directly to end users.

The third design traded one kind of differentiation for another. Prop-069 had a needs-based unit and eligibility threshold; prop-097 emphasized equal parts among all five RIRs once the trigger condition was met. An equal regional split could limit the risk that one RIR captured a small pool in one period. It could also distribute supply to regions that did not have the highest immediate demand. That was a choice about the level at which fairness would be measured: equality among the five allocation systems, not a demand model inside each system.

The adopted policy therefore did not eliminate scarcity or settle which networks should receive recovered space. It defined a predictable relationship between IANA and the RIRs. Each RIR could still operate its own allocation policy for members and customers. The global formula determined what entered the regions, while the downstream distribution of that space remained local.

What the third design did, and did not, harmonize

Leaving the return process out of scope narrowed the global agreement. The move also preserved a deliberate incompleteness: the final policy described what to do with returned address space without providing a universal method for inducing or compelling returns. That was a way to make progress without requiring all five regions to resolve every difference in how recovery occurred.

This is the central design shift across the three attempts. The first draft linked recovery and allocation. Regional systems would recover address space under local policies, but then had a common obligation to return it. IANA would allocate according to a need threshold and pool unit. The second draft revised the triggers and made returns more voluntary, but its region-specific eligibility raised a concern that one region could receive the pool before others. The third draft separated the return question from the redistribution question, used a common trigger, and divided the pool equally among all regions.

It would be too simple to say that the final proposal “fixed” the previous two. It addressed identified design risks and successfully passed the required process. It also made a choice about what to leave unresolved. A rule that does not define how the pool is replenished cannot guarantee the volume or frequency of returned space. A rule that divides supply equally among regions does not account for differences in population, demand, adoption of IPv6, industrial structure or the sizes of regional member networks. A rule between IANA and RIRs does not decide how individual address holders should be served.

Those limits do not make the policy empty. They clarify its control surface. A common allocator can decide when and in what proportions to distribute a recovered pool. Regional bodies can decide how to recover resources and how to allocate their own inventory under local policies. What can be standardized at the global layer is separated from what each region must continue to decide.

The distinction also helps avoid a misleading story about “global scarcity” as one uniform condition. The same IPv4 space moved through different legal and policy environments. Each RIR had a different inventory profile, members, runout rules and pressures. The policy did not make those differences disappear. It created an interregional rule that could operate despite them.

The tradeoff is visible in the source documents rather than in later ideology. The LACNIC authors explicitly argued that regional policy variation would affect how quickly different regions consumed their remaining addresses. They preferred an equal distribution that could operate across multiple periods and prevent one RIR from receiving the entire pool. They also acknowledged a disadvantage: their new proposal did not define the process by which resources would be returned. A policy text that says what it leaves out is useful evidence about its intended boundary.

Adoption required more than agreement with the formula

The third proposal reached APNIC’s Policy SIG consensus on 24 February 2011. APNIC’s history records an eight-week final call for comments, EC endorsement on 6 May, and eventual submission by the ASO Address Council to ICANN on 13 March 2012. APNIC marks prop-097 as implemented. Each date belongs to a distinct step; none should be substituted for the others.

ICANN’s background report records regional adoption by all five RIRs. It lists the regional processes and the adoption actions: the AfriNIC Board ratified on 21 November 2011, APNIC’s EC endorsed on 6 May 2011, the ARIN Board adopted on 16 December 2011, the LACNIC Board ratified on 19 July 2011, and the RIPE community accepted on 20 October 2011. These steps were not identical because regional institutions did not use one uniform approval mechanism.

After regional adoption, the proposal moved through NRO and ASO handling. The ASO Address Council submitted GPP-IPv4-2011 to the ICANN Board on 13 March 2012. ICANN posted it for public comment from 14 March to 4 April; the ratification notice says no comments were submitted. The Board ratified the global proposal in May 2012. ICANN’s public announcement is dated 15 May, while APNIC’s proposal history records 6 May as the ratification date. The careful summary is that ratification occurred in May, with the exact date depending on whether one is describing the Board action as recorded by APNIC or the publication date of ICANN’s announcement.

This process illustrates why a global policy was not simply a statement approved in a single meeting room. The same principle had to pass five regional communities in forms each could accept. NRO and ASO procedure carried the adopted proposal through a shared organizational path. ICANN’s Board provided the final ratification step. The decision chain was a composition of regional authorization and global ratification.

That composition also gives each stage a different question. A regional forum asks whether a proposal has sufficient local support under that region’s practice. A board or executive council may take a formal institutional action. The ASO Council checks and transmits a candidate global policy through its process. The ICANN Board ratifies the policy at the global level. The fact that a proposal is “global” does not cause one stage to perform the job of another.

The phrase “four RIRs reached consensus” and the later fact “all five RIRs adopted the third proposal” therefore describe different points in the history. The first refers to the first design’s regional outcomes and its text divergence. The second refers to the later design’s completed set of regional adoptions. The difference is not cosmetic. It is the mechanism through which a shared allocation rule obtained legitimacy across institutions with separate policy processes.

The human account should remain inside the evidence

Wilson’s long tenure as APNIC Director General helps explain why his name appears in an APNIC-affiliated cross-RIR group. APNIC’s biographical record also describes earlier work at Pegasus Networks and the International Development Research Centre’s Pan-Asia Networking programme. Those facts offer career context. They do not supply evidence about the relative influence of each author in the policy team or about which regional decisions Wilson advocated.

It is tempting to make one person the center of every system that touches their biography. Here, the temptation would be to turn a name in the author list into the architect of the global policy, or to read the eventual success of prop-097 as Wilson’s personal victory. The source record resists both moves. The first draft was explicitly a cross-RIR team submission; the second and third drafts had different author teams; each regional process reached its own decisions; and the final policy came through all five regions and ICANN.

A profile can still be personal without inventing private agency. The person’s role is the join between career and document. Wilson was not a distant observer: he was named as a member of the 2009 team. But he was also one person within a nine-member group, and the policy’s final form was produced by later authors and decisions. That bounded contribution is more useful than a heroic narrative because it reveals how much of an Internet-wide rule can depend on coordination among people who do not possess a single shared decision authority.

The APNIC record shows that prop-069 was ultimately abandoned because prop-097 had superseded it. This does not mean that the first proposal’s authors failed at their work, nor that the first proposal was adopted by another name. It means the first design did not become global policy and that the later design advanced through a different process and an altered allocation boundary. The most defensible biographical statement is correspondingly narrow: Paul Wilson was one of the named authors of the first effort; he is not listed as an author of the second or third.

A coordination test: common minimum, local choices

One way to assess the sequence is to ask how much the global system needed to specify and which decisions could remain regional. This is an editorial test, not a historical explanation attributed to the participants. A common policy can define the smallest cross-regional rule needed for predictable allocation while leaving local RIRs room to manage recovery, reservations and customer assignments. But the more the common rule dictates the upstream return of addresses, the more it reaches into local policy choices.

Prop-069 tried to make a common pool useful by standardizing both entry and exit: recovered blocks would enter the pool, and IANA would distribute them according to the global needs test. That was coherent as a mechanism, but its mandatory-return clause connected local recovery to global allocation. The ARIN edit exposed the cost of that connection. A region could support the idea of a shared mechanism while objecting to a universal rule about what it must contribute.

Prop-086 tried to soften the return rule but made eligibility depend on each region’s individual runout. That respected differences among local inventories but could make the pool’s distribution contingent on which region crossed its threshold first. The concern was not that regional policies differed; it was that a global pool could respond to that difference by concentrating all of its available supply in one place.

Prop-097 used a smaller shared specification. It left the route by which space entered the recovered pool outside the global allocation policy. Once the pool existed, a common trigger activated distribution to all five RIRs and equal parts were allocated under a predictable schedule. The formula relied less on individual eligibility and more on a rule that each region could calculate from the same shared pool.

The resulting architecture did not abolish regional choice. It placed the global constraint at the IANA-to-RIR boundary and left the recovery and downstream allocation questions to other policies. This is one practical version of localized future decision-making: agree on the minimum shared operation that must be common, then identify which decisions each participant can continue to make under its own process.

“Voluntary adoption” also needs a precise meaning here. The historical record shows regional acceptance through separate processes before ratification. After the global policy was ratified, its operational rules were not optional for each RIR each time the pool was triggered. The voluntary or participatory quality lies in the route to adoption and the ability of regional processes to decide whether to accept the common rule, not in a right to ignore the adopted rule while benefiting from it.

The distinction is relevant beyond IPv4. A global policy that touches assets controlled by different institutions has to say who creates the input, who decides eligibility, how costs and benefits are split, and which authority can make the outcome operative. A common label cannot answer those questions. A narrow rule can sometimes succeed where a broad rule stalls, but only when the narrower scope still provides a service participants value.

What this history says about institutional power

The policy was not a transfer of address blocks directly from one region to another. Each of the three proposals placed IANA between the RIRs as a common pool manager. That design was meant to avoid direct interregional transfers and to make returned space available through a defined allocation process. It also made IANA’s role dependent on rules agreed by the regional systems and ratified through ICANN.

That separation limits one kind of institutional discretion. The final equal-part rule was deterministic: when the trigger occurred and sufficient space was available, the division among five RIRs was specified. The policy did not ask an IANA operator to make an open-ended judgment about which region was more deserving. A log of returns and allocations would provide a public accounting of what moved and when.

But deterministic allocation at the top of the system cannot answer every distributive question below it. If one region gives an equal part to each RIR, the formula stops at that boundary. Each RIR then applies its own policy to its members. The global rule does not decide whether one network’s demand is more urgent, whether a small operator should receive a priority, or how a resource should be transferred between customers. Those downstream choices require their own evidence and authority.

The model also makes the pool’s replenishment an unresolved dependency. The final policy’s allocation formula operates on addresses that have already been returned. If very little space is returned, the formula cannot manufacture new supply. If return policies differ regionally, the input to the pool still reflects those differences. The third proposal did not solve scarcity; it made the distribution of any recovered supply more predictable.

That is why policy design and resource adequacy should not be conflated. A predictable rule can be valuable even when the underlying pool is small. It reduces uncertainty about how returned space will be shared. It does not necessarily produce enough space to satisfy demand, remove the cost of address conservation, or equalize the effect of scarcity across regions.

The process also shows how institutional legitimacy can depend on limiting the level at which authority is claimed. A rule governing allocation from IANA to RIRs can be justified as a common interface among regional registries. A rule that attempted to govern every recovery action, legal claim, member transfer and end-user assignment would have been a much larger undertaking. The third proposal succeeded within a narrower control surface while leaving many consequential questions where the affected systems could decide them.

This is not a claim that the narrow boundary is always optimal. It is a description of the bargain the documents show. The first proposal’s rules were more integrated; its success across four regions was not enough to yield a common text. The later proposal reduced the number of local policy questions included in the global instrument and found a formula the five regions adopted. Agreement became possible once the policy drew a more limited line around the cross-regional mechanism.

The outcome should not be retrofitted onto the first draft

Because prop-097 became the adopted policy, it is easy to narrate prop-069 as a rough draft whose real purpose was to prepare the later outcome. The documents do not prove that simple story. The third proposal’s author list differs. Its scope differs. Its trigger and distribution principle differ. It left return processes out of scope, whereas prop-069 made quarterly return a required part of the global design.

There are continuities: all three attempts addressed the fate of IPv4 addresses that might be returned after exhaustion of the IANA free pool; each considered an IANA-held pool; each sought a predictable path back to RIRs; and each arose from a distributed regional process. Those similarities justify treating the proposals as a sequence. They do not make their policies interchangeable.

Nor should the adoption of the third proposal be used to claim that the first one was “accepted in principle.” APNIC’s own status is Abandoned. The 2011 APNIC Policy SIG and Member Meeting actions say it had been superseded by prop-097 and should be abandoned. The first proposal had real regional progress, but it did not become global policy.

The opposite simplification is also weak. Abandonment is not evidence that the authors’ work was irrelevant. The first and second texts exposed conflicts that the third proposal then avoided by changing the scope. The correct relation is a documented sequence of attempts, each with a distinct design, not a retrospective credit transfer from the final policy to the first team.

The APNIC records also show how proposal numbers can conceal this changing mechanism. A reader can see prop-069, prop-086 and prop-097 as consecutive items in a registry of policy proposals. The numbers are identifiers for separate proposals, not revisions to a single codebase or a vote count. The APNIC page calls prop-069 abandoned and prop-097 implemented. The ICANN report classifies the broader efforts as first, second and third. Those status labels provide a useful record of institutional outcomes, but they do not explain the design choices unless the proposal texts are read alongside them.

The biographical conclusion

Paul Wilson’s place in this episode is not that of a solitary policy maker. It is the place of a named author in a cross-RIR team, during a period when APNIC was trying to adapt a regional allocation system to a world without IANA’s former free-pool supply. His work history and long leadership role explain why he was positioned in that environment. The available sources do not show that he alone wrote the proposal, directed the five regional processes, or authored the later rule.

The policy history becomes clearer when the human and institutional levels remain connected but distinct. A team of nine submitted prop-069. Regional communities edited and evaluated local versions. Four-region consensus did not erase the fifth version’s disagreement. A second draft revised the trigger but created a possible first-eligible-region problem. A third separated returns from redistribution and adopted an equal-allocation rule. Five regions approved that design, the ASO transmitted it, and ICANN ratified it in May 2012.

That outcome gives the profile its central point: a global Internet rule did not emerge because one person or one organization spoke for the whole system. It emerged when a common mechanism could be made narrow enough to survive separate regional policy decisions and precise enough to operate between them. Wilson was one of the authors at the beginning of that sequence. The public record gives him that contribution—and no reason to claim more.

Primary records and biographies