The Global Policy That Never Reached the Board

Summary

• A global number policy has a narrow subject: the rules by which the IANA functions operator issues Internet number resources to the RIRs. Calling a proposal global does not bypass regional policy development. The same common policy must survive the autonomous procedures of all five RIR communities before the NRO and ASO can transmit it to ICANN's Board.
• The ASO's consolidated archive identifies two major post-2004 proposals that never reached Board ratification. The 2009 recovered-IPv4 proposal failed because the regions did not adopt the same substantive text. Its 2010 successor was adopted in ARIN but abandoned, withdrawn, or allowed to expire elsewhere. These were not Board rejections; the Board never received a final common proposal.
• A narrower 2011 successor succeeded after separating the rules for returning address space from the rules for distributing whatever reached the recovered pool and replacing needs-based allocation with deterministic equal distribution. It reached the Board in March 2012 and was ratified in May. The contrast shows that scope control and textual convergence, not only technical merit, determine survival.
• The present record is strong enough to prove structural veto points but not to calculate a complete attrition rate. Local proposals that are never recognized as global, never propagated to all regions, or never assigned a consolidated global identity may remain visible only in regional archives. Accountability therefore requires a public proposal census, version history, terminal reason, and handoff record at every stage.

The Board is the visible gate, not always the decisive one

The formal global policy route ends at ICANN's Board. That makes the Board an obvious target for scrutiny. It can accept a proposal by simple majority, reject it by a two-thirds vote, request changes, or take no action and allow ratification under the prescribed rule. The Board's resolutions are public, and a rejection must identify concerns. These features give the final stage a constitutional prominence that earlier discussion does not share.

Yet the Board can act only on a proposal that reaches it. Under the ASO global policy development process, a proposal first enters one regional forum or the ASO Address Council. The Address Council determines whether the subject qualifies as global. It must then move through all five RIR policy fora. Regional outcomes may differ, staff and proposers must identify common elements, and a common text must be ratified by each RIR through its own method. The NRO Executive Council then advises the Address Council that all regions have adopted it. The Council reviews whether each process was followed and whether significant viewpoints were considered. Only a favorable Council action sends the text to ICANN.

Every stage before that transmission can end the effort. A proposal can be judged regional rather than global. It can fail to enter one policy forum. It can lack an active author in several regions. One community can reject it or simply fail to find consensus. An RIR board can decline a ratification required under its local process. Two regions can adopt propositions with the same title but materially different obligations. Reconciliation can fail. The NRO EC can lack the basis to certify universal adoption. The Address Council can identify procedural defects and return the proposal.

None of those outcomes is an ICANN Board veto. Some are community judgments; some are institutional quality controls; some are failures of coordination or time. Their common effect is to keep a proposal off the Board's agenda.

The difference matters for accountability. Blaming ICANN for a policy that never arrived confuses the location of formal authority with the location of effective power. Conversely, describing early death as merely "no consensus" can hide which region, text, threshold, or institutional decision made common adoption impossible. The record should allow entities to identify the exact gate.

Global policy is deliberately narrow

The ASO's policy explanation defines global policies as rules governing how the operator of the IANA number functions issues Internet number resources to the RIRs. The RIRs then distribute those resources under regional policies developed by their own communities. This boundary separates a genuinely global allocation instruction from a regional rule about assignments, transfers, documentation, or member eligibility.

The distinction is a jurisdictional safeguard. Without it, labeling a proposal "global" could pull ordinary regional decisions into ICANN's Board. The five RIR systems would become preliminary committees for a central policy authority. The narrow definition keeps the Board's role tied to the IANA-to-RIR interface where one coordinated rule is operationally necessary.

It is also the first veto point. When a proposal is submitted to the Address Council, the Council examines whether it qualifies. If not, the Chair informs the author and the Council takes no further action. When a proposal begins in a regional forum, a regional member of the Council notifies the Chair, and the Council makes the same classification. A proposal can therefore be technically consequential across several regions yet fail the global-policy definition because it regulates what RIRs do after receiving resources rather than what IANA does in issuing them.

That gate is defensible, but its archive is less visible than the later Board record. The consolidated global proposal page lists proposals recognized and tracked as global, including accepted, withdrawn, and abandoned items. It does not purport to be a census of every suggestion that someone may have described as global before classification. A rejected classification can be recorded in Council minutes or a regional thread without appearing as a named global proposal.

This is the first reason to be cautious with counts. The visible archive measures proposals that acquired enough institutional identity to be tracked. It does not establish the denominator of all attempted ideas. Absence from the Board record may indicate regional failure; absence from the ASO proposal table may indicate an even earlier failure of classification, propagation, or documentation.

The structural inference is still strong: agenda control begins before the five regional consensus tests. The numerical inference is bounded: the public consolidated table cannot alone tell us how many ideas vanished at that first gate.

Five autonomous policy systems create a unanimity rule

Once a proposal qualifies as global, it does not enter one worldwide ballot. It enters five separate policy systems. AFRINIC, APNIC, ARIN, LACNIC, and the RIPE community each use their own discussion forum, meeting practices, consensus assessment, review stages, and organizational ratification. The ASO's regional policy page emphasizes that anyone may submit proposals and that membership is not required, but it also directs entities to five distinct processes.

The global rule is stronger than ordinary majority approval. Each region must ratify the same common text. A region with fewer entities does not possess a fractional vote. One unresolved objection in one process can prevent global completion even when four regions are prepared to proceed. This is a form of institutional unanimity among regions, layered over consensus within each region.

The rule protects regional autonomy. A global majority cannot impose an IANA allocation mechanism on a region whose community identified an operational or distributive problem. Each RIR can apply its own safeguards before accepting an obligation that will affect all five. The requirement also tests whether the proposal is truly global: if its benefits depend on conditions present in only some regions, universal adoption should be difficult.

The same protection creates a veto. It need not be exercised through a dramatic "no" vote. Failure to reach consensus is enough. So is withdrawal by authors when opposition makes success implausible. Expiration under a regional timetable can have the same global effect. A material local amendment can produce regional acceptance but defeat the identity requirement. The global mechanism treats these distinct terminal states alike at the handoff point: there is no common text to send.

This is not necessarily a flaw. Consensus systems are designed to prevent premature coercion. But accountability requires naming the effect. A regional community is not merely discussing a local version; it is exercising one of five indispensable approvals. Entities should know when a local amendment changes the worldwide proposition, when delay will terminate the global effort, and when an apparently regional decision has closed the Board route.

The formal equality of regions also shifts agenda power toward the least persuaded region. Authors must solve the strongest objection, not merely assemble the largest coalition. That can improve policy. It can also encourage a lowest-common-denominator text stripped of obligations that are important but controversial.

The 2009 recovered-space proposal failed through textual divergence

The first decisive case in the ASO's post-2004 archive is GPP-IPv4-2009, the Global Policy Proposal for the Allocation of IPv4 Blocks to Regional Internet Registries. The proposal addressed a foreseeable post-exhaustion problem. RIRs or other holders might return IPv4 space after IANA's ordinary free pool was depleted. IANA needed authority to receive recovered blocks and a rule for reallocating them.

The proposal attempted to build a recovered pool and distribute from it according to need. Its most difficult question was upstream of allocation: whether RIRs had to return recovered address space to IANA. The initial text made return mandatory for qualifying recovered blocks. That obligation reached into choices about space under regional administration and exposed different views about legacy resources, recovery, incentives, and regional control.

Regional records show why a shared title was limited public evidence. The ARIN archive for Draft Policy 2009-3 records objections to mandatory return and a revision under which an RIR could designate recovered space for return according to its own policies and strategies. The revised language made return optional at the regional level. The archive expressly recognized that this was likely a substantive change requiring reconciliation with versions passed elsewhere.

The ASO consolidated table states the outcome plainly: the same version was not adopted by all five RIR communities. Its regional notes say ARIN's adopted version did not include mandatory return, while AFRINIC, LACNIC, and RIPE recorded approvals of their versions. The RIPE 2009-01 archive labels the proposal accepted locally but adds that differences among global texts meant it would never be globally implemented. APNIC's prop-069 archive records initial regional endorsement and later abandonment after a successor superseded the effort.

This was not a case in which four regions voted yes and a fifth simply voted no on an identical proposition. The more revealing event was successful local modification. ARIN addressed a regional objection by changing the obligation. That change improved acceptability in one forum while destroying the common text necessary for global completion. Regional policy development worked on its own terms and global policy development failed on its.

The proposal never became a final common text for the ASO AC to transmit. The ICANN Board did not decide whether mandatory return was fair, whether optional return would supply enough space, or whether needs-based reallocation was administrable. Those questions were resolved, deferred, or displaced below the Board layer.

Local acceptance can still equal global failure

The 2009 history exposes a status problem. AFRINIC's AFPUB-2009-v4-002 page describes its regional policy as implemented. RIPE describes 2009-01 as accepted and publishes the resulting document. ARIN records an adopted regional policy. APNIC records that prop-069 reached consensus and endorsement before it was eventually abandoned as superseded. A reader opening one archive could reasonably conclude that the policy succeeded.

At the global layer, it did not. A regional system can mark a text accepted because its own consensus and ratification stages were completed. That status does not certify identical adoption elsewhere. If the text was intended to operate only after ICANN Board ratification, regional publication may have no global operational effect. "Implemented" can therefore describe incorporation into a regional manual or completion of a local decision while the condition for worldwide operation remains unsatisfied.

This semantic mismatch obscures attrition. Consolidated reporting calls GPP-IPv4-2009 abandoned. Regional archives preserve accepted or implemented descendants. Both descriptions can be accurate, but only if the reader knows which unit is being measured: the local proposal, the local policy text, or the coordinated global proposition.

A robust record should assign separate fields. Regional status should show introduction, consensus, final call, board action, and local publication. Global status should show classification, propagation, common-text version, universal ratification, NRO transmittal, ASO review, Board submission, and Board outcome. A dependency field should state whether local implementation is conditional on global ratification. Supersession should link the old proposal to the replacement without rewriting history.

The distinction also matters politically. It is tempting for a regional institution to say it approved the idea and another region blocked it. That may be true, but textual divergence can make the claim too simple. If one region approved an optional-return rule and another approved mandatory return, they did not approve the same policy. The failure belongs to reconciliation as much as to any individual veto.

Global policy requires more than overlapping intent. It requires an executable common obligation.

The 2010 successor died through distributed non-consensus

The next case, GPP-IPv4-2010, tried again. It retained a reclamation pool and needs-based allocation while responding to issues raised by the 2009 proposal. Its regional identifiers included APNIC prop-086, ARIN 2010-10, LACNIC 2010-04, RIPE 2010-05, and an AFRINIC proposal. The coordinated title suggested a renewed worldwide effort; the outcomes show how quickly that identity fragmented.

ARIN adopted its version. The ARIN Advisory Council record documents a recommendation that the Board of Trustees adopt Draft Policy 2010-10 after discussion and last call. In one region, the proposal cleared the relevant gates.

APNIC did not. Its prop-086 archive records presentation at APNIC 30, return to the mailing list, revised versions, and failure to reach consensus at APNIC 31 on 24 February 2011. The proposal was abandoned. LACNIC's policy proposal archive records no consensus at the October 2010 forum and withdrawal by the authors in May 2011. The RIPE 2010-05 page records withdrawal on 22 March 2011. AFRINIC's proposal page records expiration.

No single terminal verb captures this death. "Abandoned" describes APNIC and the global record. "Withdrawn" describes LACNIC and RIPE. "Expired" describes AFRINIC. "Adopted" describes ARIN. The global proposition failed because universal regional ratification became impossible, not because one central authority issued a rejection.

The variety matters. No consensus is a community outcome. Withdrawal is an authorial response to the state of discussion. Expiration can reflect a time rule and lack of sufficient progress. Each indicates a different accountability question. Were objections substantively unresolved? Did authors lack capacity to continue in five fora? Did an emerging successor make further work irrational? Did procedural clocks turn inactivity into termination? A consolidated terminal status should preserve those distinctions.

By the time the global record declared the proposal unable to advance, the Board had no complete instrument to evaluate. The strongest veto was cumulative: one regional success could not offset four different forms of failure.

Proposal authors carry an unusually heavy coordination burden

The global procedure expects authors to participate in every regional forum and help communities understand deliberations elsewhere. If an author cannot travel, the relevant RIR is expected to arrange presentation. The Policy Proposal Facilitator Team monitors progress, while RIR staff help identify and reconcile common language. These supports matter, but the intellectual and political burden remains high.

An author must explain the same mechanism to communities with different current policies, address inventories, transfer rules, institutional histories, meeting schedules, and consensus cultures. A revision welcomed in one region may reopen discussion in four others. A phrase treated as editorial in one forum may be substantive in another. Time zones, language, travel, volunteer capacity, and the sequence of meetings shape whether the same text receives comparable attention.

This burden creates a quiet veto point: exhaustion of sponsorship. A technically sound idea can disappear because no one can sustain coordinated advocacy through five cycles. Withdrawal may reflect substantive defeat, but it can also reflect a rational decision that the next revision would require restarting too much work. The archive rarely quantifies how many author hours, meetings, or version reconciliations preceded the terminal status.

Institutional entities have an advantage. RIR staff and experienced policy figures can follow every forum, understand procedural differences, and draft language with global consequences in mind. A network operator or researcher with one regional base may be formally entitled to propose but practically unable to shepherd the proposal worldwide. "Anyone may propose" describes access to the entrance, not equality of capacity through the route.

That asymmetry can narrow the agenda to subjects backed by established organizations or unusually persistent volunteers. It can also make early staff coordination decisive. If the PPFT and RIR secretariats identify divergence quickly, authors can decide whether to reconcile, split scope, or withdraw. If divergence is discovered after several regions have completed local action, the cost of common text becomes much higher.

A global system need not finance every proposal. It should disclose the coordination burden and offer a durable common workspace: authoritative text, regional issue matrix, version comparison, meeting calendar, named facilitators, and explicit statement of which changes would require reconsideration elsewhere. Transparency can reduce accidental attrition even when substantive vetoes remain.

The 2011 proposal succeeded by narrowing the bargain

The third proposal in the sequence, GPP-IPv4-2011, demonstrates that failure did not mean the recovered-space problem was imaginary. The successful text changed the structure of the bargain. APNIC's prop-097 page states that the proposal described how IANA would allocate IPv4 resources after exhaustion while placing the process by which space entered the recovered pool outside scope.

That exclusion was consequential. The 2009 fight over mandatory return had reached into regional control of recovered space. The 2011 proposal did not require an RIR to recover or return particular blocks. It governed the disposition of whatever address space arrived at IANA by any means. The proposal also used deterministic equal distribution among the five RIRs on a regular schedule rather than a needs test whose application could reproduce disputes over eligibility and regional advantage.

The narrower text moved quickly. APNIC records introduction in January 2011, consensus in February, final call, executive endorsement in May, ASO AC submission to the ICANN Board on 13 March 2012, public comment, and Board ratification on 6 May. ICANN's ratification notice summarizes the mechanism as return to an IANA-managed pool followed by equal, deterministic allocation to all five RIRs.

The success does not prove that every objection to the earlier texts was resolved on its merits. Some were moved outside the global policy. Regional communities retained control over whether and how space would be recovered or returned. Equal allocation avoided the institutional need to compare regional need, but it may not have mirrored actual demand. The winning bargain traded ambition for convergence.

That is often how multi-level governance advances. A proposal survives not because the central authority overcomes regional vetoes, but because the regulated issue is decomposed until all regions can accept the shared portion. The remaining issues stay regional, contractual, operational, or available for later work.

The lesson is not that narrow policy is always superior. It is that scope is a constitutional variable. A global proposal that combines an indispensable IANA rule with contested regional obligations gives every region a reason to block the entire package. Separating them can preserve the genuinely global rule without pretending that the disputed regional questions were settled.

Textual identity is a substantive safeguard, not clerical neatness

The requirement for common text can look formalistic. If five regions agree on the general idea, why should different wording prevent the Board from reconciling it? The answer lies in authority. The ICANN Board is supposed to ratify a policy developed by the number communities, not negotiate a new compromise that no region approved.

Differences in words can move resources and obligations. "Shall return" and "may designate for return" express different control over recovered address space. "Based on need" and "in equal parts" allocate scarcity differently. Definitions of exhaustion, eligible inventory, reserved space, minimum block size, activation, reporting, and transfer rights can change who receives resources and when. A Board-created synthesis would make ICANN the policy author at the final stage.

Common text therefore protects bottom-up legitimacy. Every region can know the exact instruction that will bind the IANA function. If Board review raises a substantive change, the procedure sends that change back through the RIRs rather than allowing silent amendment. The cost is that regional adaptation can become a veto even when it addresses a genuine local concern.

The right response is not to weaken identity but to manage versions better. Each global proposal should have an authoritative text with a public cryptographic hash, a version number, and a matrix linking every regional identifier to that version. When a region proposes an amendment, the record should classify it as editorial, interpretive, or substantive and explain who made that classification. Substantive change should automatically notify all regions and identify which completed stages must be repeated.

The 2009 archive leaves readers reconstructing divergence from several pages. A modern record could show it directly: common version, ARIN modification, affected clause, views from other regions, reconciliation attempt, and terminal decision. That would make the veto visible without treating amendment as misconduct.

Textual identity is where distributed consent becomes one executable policy. It deserves better evidence than matching titles.

The ASO AC is a process auditor with its own gate

Even after all five RIRs ratify a common text, the proposal does not automatically go to the Board. The NRO EC transmits it to the Address Council, which has a defined review period. The Council examines whether each regional process was followed and whether significant viewpoints were adequately considered. It may send the proposal to ICANN, report concerns and request further regional review, or seek more time.

This gate is narrower than a sixth merits vote, at least in design. The Council should not replace five regional judgments with its policy preference. Its function is to validate that the distributed process produced the claimed common agreement. That includes checking the procedural history and reconciling language.

Process review is nevertheless powerful. A finding that one region did not follow its PDP can return an otherwise complete proposal. A concern that significant views were not adequately considered can require more deliberation. A request for time delays Board access. Council members who monitored the proposal through the PPFT possess informational influence over how these questions are framed.

The safeguard is justified because universal ratification can be asserted too easily when regional statuses use different terms. Someone must verify that "consensus," "accepted," "endorsed," and "ratified" satisfy the required local stages and all attach to the same text. The Council is the institution positioned to do that.

Accountability requires its report to be public and specific. It should list the version reviewed, each regional decision and date, the evidence that local procedures were followed, material objections, reconciliation history, and the Council vote. If the proposal is returned, the report should identify the defect and the cure required. Otherwise, procedural review can become an opaque policy veto exercised after communities believe their work is complete.

The abandoned 2009 and 2010 proposals did not reach this final favorable handoff because common regional ratification was absent. Their history still shows why the audit gate matters: local statuses alone did not establish a final global proposition.

Silence and time can exercise power without owning it

Multi-level consensus often imagines an objection stated, debated, and either resolved or sustained. In practice, a proposal can fail because discussion fades. Authors stop revising. A meeting records no consensus and returns the item to a list that produces little new evidence. A regional deadline expires. Another proposal overtakes the same subject. No institution declares that the underlying problem lacks merit, yet the route closes.

This is a form of power without a clear owner. A entity who raises a decisive objection may not be responsible for proposing replacement text. Chairs may be reluctant to declare consensus without visible support. Authors may conclude that continued effort is wasteful. Staff may accurately apply expiration rules. Each actor performs a defensible role; collectively, the proposal disappears.

The 2010 case contains several such mechanisms. APNIC made an express no-consensus finding and abandoned the proposal. RIPE and LACNIC record withdrawal. AFRINIC records expiration. Those endings should not be collapsed into a story of one hostile region. They show that a global proposal must maintain positive momentum everywhere. Four absences of completion outweigh one adoption.

A terminal report should therefore answer more than "approved or rejected." It should identify the last active version, unresolved issues, participation level, author decision, applicable clock, superseding proposal, and whether any region requested further work. It should state who can revive the idea and whether revival requires a new identifier.

Such a report would not force consensus. It would prevent institutional silence from erasing policy learning. Future authors could see which objection blocked progress and which drafting changes had already failed. Communities could distinguish lack of support from lack of capacity. The Board and wider ICANN community could understand why an issue never reached them without intruding into regional authority.

In a distributed system, non-action is often the effective veto. It should leave an auditable trace.

The archive proves cases, not a complete attrition rate

The ASO global proposal page is unusually valuable because it places accepted, abandoned, and withdrawn proposals together. It shows GPP-IPv4-2009, GPP-IPv4-2010, and the successful GPP-IPv4-2011 sequence with regional identifiers and outcomes. The current-policy page separately records ratified global policies for IPv6, ASN blocks, and post-exhaustion IPv4 allocation.

That evidence supports a bounded conclusion: at least two formally tracked global IPv4 proposals in this sequence failed before Board ratification, and their regional records identify textual divergence and distributed non-consensus as the causes. It does not support a claim that only two global ideas have ever died before the Board, or that a fixed percentage of all proposals is rejected regionally.

The missing denominator has several sources. An idea may be proposed in one RIR as a regional policy even though its effects would benefit from coordination. It may be called global by an author but fail the ASO definition. It may never be introduced in all five regions. It may be split, renamed, or superseded before the ASO assigns a consolidated identifier. Older archives may use different status vocabulary or preserve evidence in meeting minutes and mailing lists rather than one index.

Survivorship bias also shapes attention. Ratified policies receive durable ICANN pages, implementation records, and current-policy links. Failed proposals remain dispersed across archives whose URLs and formats change. The more successful the eventual replacement, the easier it is to treat earlier failures as drafting prehistory rather than governance evidence.

A proposal census should begin at classification, not Board submission. Every purported global proposal presented to an RIR or the Council should receive a durable tracking record, even if the Council concludes it is not global. The record should link regional instances, authoritative versions, decisions, reasons, supersession, and final disposition. It should distinguish an idea family from each formal proposal so repeated attempts can be studied without double-counting.

Until such a census exists, qualitative institutional analysis is stronger than attrition statistics. The official cases reveal the veto structure clearly. They do not justify a false precision about frequency.

Early vetoes can improve policy and still require accountability

The term "veto point" can sound accusatory. In the global numbers system, many early gates serve legitimate purposes. The definition test protects the boundary between regional and global authority. Five-region ratification protects local communities from central imposition. Common text ensures that the Board ratifies what the regions actually accepted. The ASO review protects procedural integrity. Withdrawal allows authors to stop an unproductive attempt.

The 2011 success suggests that the gates generated policy learning. The contentious obligation to return recovered space was removed from the global allocation rule. Deterministic equal distribution replaced a disputed needs model. A narrower proposition obtained universal support and became operational policy. The earlier failures were not wasted if they exposed the terms no common agreement could carry.

But a beneficial veto remains an exercise of power. Entities should know who made the decisive judgment, under which rule, on what text, and with what effect. A region that rejects mandatory return should own and explain that position. An author who withdraws after repeated no-consensus findings should leave a reason. A Council that classifies a proposal as regional should publish the boundary analysis. A process auditor that returns a proposal should identify the defect.

Accountability is especially important because the Board stage has formal reason requirements. A Board rejection requires a supermajority and specific concerns, and the procedure contemplates further work and possible mediation after repeated rejection. Early failure can be less demanding: no common text means no Board record, no central vote, and no Board explanation. The policy can die where reasons are distributed across five archives.

The appropriate reform is not to let the Board rescue proposals that lack regional consent. It is to require early gates to produce evidence comparable in clarity to the final gate. Distributed authority should produce distributed reasons, then a consolidated terminal account.

A majority fallback would change the constitution, not fix the record

One apparent solution is to allow four of five regions to send a proposal to the Board. That would reduce the veto power of one region and make Board review more common. It would also transform the global policy bargain. The dissenting region would be bound at the IANA-to-RIR interface by a policy its community did not ratify, and the Board would become the arbiter of whether the majority should prevail.

Such a rule might be defensible for a narrowly defined emergency, but it is not a transparency repair. It changes substantive authority. The present system treats regional unanimity and common text as the source of the Board's mandate to ratify. Removing either would require a new legitimacy account, safeguards for dissent, and a decision about whether an RIR can opt out of an allocation rule that operationally depends on all RIRs.

The better initial response is to distinguish deliberate dissent from accidental attrition. A region that rejects a common text after full consideration is exercising the constitutional veto the system grants it. A proposal that expires because version notices were unclear or authors could not manage five calendars reflects avoidable coordination failure. Better tracking, facilitation, and terminal reports can reduce the second without weakening the first.

Time limits may still help. Each region could publish an indicative schedule after accepting a global proposal for discussion, and the PPFT could issue a common progress report when a proposal stalls. Extensions should be reasoned. Inactivity should trigger a status conference rather than silent disappearance. None of these measures compels approval.

Issue decomposition is another non-coercive tool. If four regions support a package and one rejects a particular obligation, the facilitators can ask whether the indispensable global rule can be separated from the disputed regional rule. The 2011 proposal shows that narrowing can convert a vetoed package into a common core.

Consensus should be difficult to achieve because global number policy is consequential. It should not be difficult to observe.

The final Board rules make the earlier opacity more conspicuous

The contrast with the ICANN stage is instructive. Once a proposal is formally submitted, the Board operates against a defined clock. It may accept by simple majority, reject by a two-thirds vote, request changes by simple majority, or take no action, in which case the text is deemed ratified. A rejection must be accompanied by specific concerns. A request for substantive change returns to the distributed policy system; it does not authorize the Board to revise the bargain unilaterally.

These rules do more than constrain the Board. They allocate the burden of action. Acceptance is easier than rejection. Silence favors the community-developed text rather than institutional delay. A proposed change must face renewed regional consent. Repeated rejection can lead toward mediation. The final gate is powerful, but its power is disciplined by outcome categories, thresholds, reasons, and time.

Before submission, the burden often runs the other way. A proposal needs affirmative completion in every region. Silence does not ratify it. Expiration does not require a global explanation. Withdrawal can end one regional instance without a consolidated account of consequences. A locally accepted but divergent text does not move forward. The earlier system is therefore more demanding of proponents and less demanding of a single terminal decision maker, because there may be no single decision maker.

That asymmetry may be inevitable in a bottom-up federation. Five communities cannot be forced to deliberate on one synchronized clock merely to imitate a corporate board. But the quality of the record need not be asymmetric. Each regional terminal act can identify reasons; the PPFT can consolidate them; the ASO can state why the Board submission condition was not met.

The Board stage demonstrates that procedural clarity does not require centralizing policy authorship. The same principle can govern non-arrival: explicit state, fixed responsibility for notices, preserved text, and a reasoned account of what happens next.

A public ledger should follow the proposal through every handoff

The global policy system needs one authoritative record that begins when an idea claims global scope and ends only with ratification, rejection, withdrawal, abandonment, expiration, reclassification, or supersession. The record should not replace regional archives; it should connect them.

For each proposal, the ledger should state the problem, the claimed global-policy basis, the Council's classification, the authoritative text and hash, authors, facilitators, and all regional identifiers. It should show the current stage in each RIR, the next event, the latest decision, and the version under consideration. A change log should identify substantive divergence immediately.

At a terminal stage, the ledger should identify the decisive condition. "Not adopted by all five" is accurate but incomplete. The record should say, for example, that one region adopted materially different return language, another later abandoned a superseded version, and no reconcilable common text remained. For the 2010 family, it should preserve adopted, no-consensus, withdrawn, and expired outcomes separately.

The ledger should also distinguish proposal family from proposal instance. GPP-IPv4-2009, GPP-IPv4-2010, and GPP-IPv4-2011 addressed related recovered-space problems but offered different bargains. Linking them would show policy learning without implying that the successful 2011 text was simply another version of the failed 2009 obligation.

Finally, the record should capture handoffs. NRO EC transmittal, ASO AC procedural report, submission to ICANN, Board acknowledgment, public comment, questions, decision, and implementation should each have dates and source documents. If a handoff never occurs, the ledger should say why.

This evidence would change institutional debate. Entities could test claims that "the Board blocked" a policy, identify regions where engagement is needed, compare time spent at each gate, and study whether certain subjects repeatedly fail through scope or text. It would also protect regional autonomy by documenting rather than centralizing decisions.

The infrastructure of accountability is often less dramatic than a new voting rule. In this case, a durable, version-aware public record would expose the real constitution already operating.

The real test is whether the system can explain non-arrival

For a policy that reaches ICANN, the final chronology is relatively easy to reconstruct. A common text is transmitted, the Board receives it, the public can see a ratification or rejection, and implementation follows. For a policy that never arrives, evidence fragments. One regional archive says accepted. Another says abandoned. A third says withdrawn. The consolidated page says the proposal could not advance. The Board agenda is silent because there was nothing final to place on it.

The governance test should therefore begin with non-arrival. Can the ASO identify every proposal that claimed global scope? Can it show the classification decision? Can a reader see the exact text each region considered? Can the record distinguish lack of consensus from withdrawal and expiration? Can it identify whether a local amendment was substantive? Can it explain why reconciliation failed? Can it link a successor and state which disputed obligations were removed or changed?

The 2009 and 2010 recovered-IPv4 proposals answer enough of those questions to reveal the architecture. The first failed because the regions did not approve the same return obligation. The second failed because one adoption could not overcome no consensus, withdrawals, and expiration elsewhere. The 2011 successor reached the Board after narrowing the global obligation and adopting a deterministic allocation rule.

Those cases locate power more accurately than a Board-centered account. The ICANN Board holds the final legal gate, but regional communities, authors, RIR boards, facilitators, the NRO EC, and the ASO AC determine whether a complete proposition exists for the Board to judge. Their vetoes may be principled, procedural, textual, or passive. All are structurally real.

A mature global policy system should not promise that every proposal reaches the Board. Most ideas should be tested and some should fail. It should promise that failure leaves reasons proportionate to the power exercised. When a proposal vanishes, the public should not have to infer the veto from five incompatible status pages.

The Board's absence is itself an outcome. The institutions below it should be able to explain who produced that outcome, on which text, and why.