Summary

  • APNIC’s conference report and current proposal index say prop-170 reached consensus at APNIC 62, but the captured proposal detail page still says “For Discussion” and the public-drafts page says no document is open for editorial comment.
  • APNIC’s own Policy Development Process puts more controls after conference consensus: a final comment period, EC endorsement, then Secretariat implementation. It says implementation usually starts no sooner than three months after endorsement; the prop-170 impact material estimates about six months if consensus is reached.
  • The current APNIC-127 remains the operative public rule observed on 14 September. It retains a /32 minimum, documented larger needs and one-bit doubling for subsequent allocations; it contains no nibble-alignment rule.
  • The correct operational response is neither to dismiss the conference result nor to automate against it. Treat consensus, endorsement, published text, effective implementation and synchronized guidance as separate evidence states.

Three pages, three kinds of state

The smallest useful fact is a disagreement between labels. APNIC’s report for its Mumbai conference lists prop-170, “Nibble-Boundary Alignment for IPv6 Allocations”, under the proposals that reached consensus. The main policy-proposals index, captured on the same day, has also moved prop-170 into a section headed “Reached consensus at APNIC 62”. Yet the proposal’s own detail page still shows its current status as “For Discussion at APNIC 62 OPM”.

It is tempting to ask which page is correct. That is the less useful question. The report is an outcome record. The index is a portfolio view of proposals. The detail page is a status record whose label has not moved with the other two. Each page has a different publication workflow, and the public evidence does not show why they diverged. Calling the detail page stale is a reasonable description of the observed sequence, but calling the lag negligent, contested or deceptive would require evidence that is not available.

A fourth surface adds a different negative signal. APNIC’s drafts-for-public-comment page said there were no documents currently open for editorial comment and contained no prop-170 entry in the captured bytes. That does not prove that a final comment notice was never sent, that no mailing-list step had begun, or that the page would remain empty. It establishes only what the official drafts surface displayed on 14 September.

These distinctions matter because a proposal has at least three identities after a conference. It can have a meeting outcome, a procedural status and an operative effect. A reader who collapses them into a single green light will treat the first published success signal as if it were the final rule. A reader who sees the inconsistent labels and dismisses the whole process will make the opposite mistake. The task is to identify which evidence controls which decision.

Consensus is a checkpoint, not the executable rule

APNIC’s Policy Development Process supplies the sequence. Before a conference, proposals receive at least four weeks of Policy SIG mailing-list discussion. At the Open Policy Meeting, the SIG chairs decide whether the discussion has reached consensus. If the SIG accepts a proposal, the result is reported to the APNIC Member Meeting, where the membership is also asked to reach consensus.

The post-conference path then begins. If consensus exists among the SIG participants and the membership, the proposal goes to the mailing list for a final comment period. If the text changed during the meeting, this stage gives people who were not in the room an opportunity to examine the revised version. If consensus is maintained, the SIG chair asks the APNIC Executive Council to endorse the proposal. Only after EC endorsement does the Secretariat implement it. APNIC says implementation usually occurs a minimum of three months after endorsement.

The conference report’s phrase “reached consensus” is therefore meaningful. It says more than “was discussed” and separates prop-170 from proposals that did not reach consensus or received no consensus call. But the report does not provide a vote count, and APNIC consensus is not a numerical ballot or a claim of unanimity. Nor does the short result line disaggregate every procedural sub-step for a reader. It should not be stretched into evidence of a final-comment notice, an EC decision, an edited policy document or an effective production change.

Prop-170’s impact assessment makes that boundary unusually visible. The Secretariat says APNIC-127 would need changes, internal guidance and assessment processes would need updating, and implementation would take approximately six months if the proposal reached consensus, subject to editorial comment. The conference presentation repeats the implementation surfaces: request-evaluation procedures, registration systems, internal tools and guidance for APNIC and National Internet Registries may need updates.

“Approximately six months” is neither a promised completion date nor a clock that begins automatically with the conference report. The process page gives a usual minimum after endorsement; the impact page gives a conditional estimate for the proposal’s work. The two durations describe different parts of the chain. Combining them into a release date would require dates and decisions the sources do not provide.

The current rulebook shows what has not changed

The clearest operational check is the current policy itself. In the captured APNIC Internet Number Resource Policies, Section 8.1 still says the minimum IPv6 allocation is /32. Larger initial allocations may be justified through comprehensive documentation, and the allocation fulfils the calculated address requirement under the HD-Ratio policy. Section 8.3.4 still says an account holder that meets acceptable utilisation may receive an additional allocation that doubles its existing space, adjacent where possible. Larger needs require documentation based on users, infrastructure, network structure, security segmentation and planned longevity.

The captured policy page contains no occurrence of “nibble”. That negative fact is narrower than it sounds. It does not mean APNIC staff cannot discuss address-plan alignment, that no applicant has ever received an aligned prefix, or that prop-170 has failed. It means the general conditional mechanism proposed by prop-170 was not yet present in the public operative text examined on that date.

The proposed rule is different in a precise way. Prop-170 would allow an LIR to explicitly request nibble-boundary alignment and support the request with a long-term IPv6 address plan. After the underlying requirement is assessed, APNIC would normally delegate the smallest nibble-aligned block capable of satisfying that calculated need. For subsequent allocations, one-bit doubling remains the default; a larger aligned result requires documented and validated need. Section 9 assignments remain outside the proposal.

That proposal has already received detailed scrutiny elsewhere. The size between a non-aligned calculated requirement and an aligned outcome can affect holdings and potentially fees; the interaction with prop-164 raised a drafting question if both proposals advanced. Those are real issues, but they are not the question here. The present question is whether a conference outcome alone changes the rule an applicant or automation should use. The current policy text says it does not.

A status model for operators

An operator does not need to become a constitutional lawyer to avoid premature action. A five-state record is enough.

First, record the conference outcome with its date and source. For prop-170, the report and index supply that evidence. Second, record the final-comment state: notice URL, text version, opening and closing dates, and whether substantive changes appear. Third, record EC endorsement with the resolution or minutes that make it public. Fourth, hash or version the resulting policy text and note its effective date. Fifth, record implementation evidence: updated request guidance, forms or APIs, and any separate NIR procedure that applies to the account.

Each state answers a different question. Conference consensus answers whether the community advanced the proposal. Final comment answers which wording survived broader review. EC endorsement answers whether the governance body approved implementation. The published policy answers what rule is authoritative. Implementation guidance answers how the rule is exercised in an operational workflow.

For a human applicant, the safe action is simple: use the current policy and ask APNIC or the relevant NIR whether a new mechanism is effective before relying on it. For software, the standard should be stricter. Do not infer eligibility or request shape from a conference report, a blog post or a proposal document. Bind automation to a versioned operative policy and a documented service contract. A proposed field such as “request nibble alignment” should not appear in a production request merely because the proposal’s meeting state turned green.

This is not an argument for freezing systems until every web page is identical. Editorial pages will update at different speeds. It is an argument for explicit precedence. Outcome pages can trigger monitoring; they should not silently override the rulebook. A proposal page can explain intent; it should not be mistaken for an effective contract. The rulebook can establish authority; it still may not prove that every form, API and NIR workflow is ready on the same day.

What the mismatch does—and does not—show

The divergent pages are useful because they expose the seams. They show that APNIC’s public policy record is not one synchronized database view. They do not show that the consensus was invalid, that the Secretariat missed a deadline, that an applicant was misled or that any allocation was mishandled. No MyAPNIC workflow, internal registry system, request file or NIR implementation was tested for this article.

They also show why a binary “passed/failed” field is inadequate for public policy monitoring. A proposal may have passed a meeting checkpoint while awaiting a final text. It may be endorsed while awaiting implementation. It may be implemented centrally while an NIR updates a local procedure. Collapsing these conditions into “approved” transfers the burden of interpretation to applicants and developers.

The proportionate remedy is not a dramatic accusation. It is a dated state trail with named owners and source links. APNIC controls its process, policy text and central implementation. The Policy SIG chairs control consensus calls and the final-comment transition. The EC controls endorsement. The Secretariat and NIRs control different operational surfaces. A public status system should let a reader see those hand-offs without pretending they occur simultaneously.

For prop-170, the next decisive evidence is not another summary of hexadecimal address planning. It is a final-comment notice tied to an exact text, followed by an endorsement record, an updated APNIC-127 and an implementation date. Until those signals appear, the most accurate sentence is deliberately two-part: APNIC says prop-170 reached consensus, and the current rulebook still does not say “nibble”.

Sources

  1. APNIC 62 conference report: https://conference.apnic.net/62/report/index.html
  2. APNIC, prop-170 proposal page and Secretariat impact assessment: https://www.apnic.net/community/policy/proposals/prop-170/
  3. APNIC, prop-170-v001 proposal text: https://www.apnic.net/wp-content/uploads/2026/07/prop-170-v001.txt
  4. APNIC Policy Development Process: https://www.apnic.net/community/policy/process/policy-development-process/
  5. APNIC Internet Number Resource Policies (APNIC-127): https://www.apnic.net/community/policy/resources
  6. APNIC drafts for public comment: https://www.apnic.net/community/policy/drafts/
  7. APNIC Blog, “APNIC 62 OPM: Just the facts”: https://blog.apnic.net/2026/08/21/apnic-62-opm-just-the-facts/
  8. APNIC policy proposals index: https://www.apnic.net/community/policy/proposals/
  9. APNIC 62, prop-170 presentation: https://conference.apnic.net/62/assets/presentation-files/381abbc6-92d6-4e05-ab2b-1d0003d6e9a9.pdf