Summary

  • APNIC announced seven proposals for discussion at its 10 September 2026 Open Policy Meeting. Its conference report assigns all seven to one of three groups.
  • Prop-169 and prop-170 reached consensus. Prop-164, prop-168 and prop-171 did not reach consensus and will return to the mailing list. No consensus call was held for prop-165 or prop-172.
  • “No call” is not a failed call. A compact proposal-state receipt should record the version considered, why no call occurred, the next action and when the authoritative status page was updated.

Seven proposals, three verbs

Before the Open Policy Meeting, APNIC’s welcome note presented a clean agenda. Seven numbered proposals were up for discussion on 10 September: prop-164, prop-165, prop-168, prop-169, prop-170, prop-171 and prop-172.

The conference report reconciles that list without losing an item. Two proposals, prop-169 and prop-170, “reached consensus”. Three—prop-164, prop-168 and prop-171—“did not reach consensus” and will be sent back to the Policy SIG mailing list for further discussion. For the remaining two, prop-165 and prop-172, the report uses a separate sentence: “There was no consensus call held”.

That third sentence is the story. It is not an empty cell in a results table. It describes the absence of the decision event that would have produced either of the first two outcomes.

This is worth crediting. A weaker summary might have grouped every proposal that did not advance with the proposals that failed to reach consensus. APNIC did not do that. The report keeps an important procedural distinction intact. It therefore gives a careful reader enough evidence not to call prop-165 or prop-172 rejected.

But it does not complete the state transition. It does not say why the calls were not held, which version or discussion condition controlled that choice, what happens next, or when the proposal pages will cease to describe the items as merely “For Discussion at APNIC62 OPM”.

A call is an event, not a mood

APNIC’s own explanation of rough consensus helps define the boundary. Participants with objections are asked to explain them. The community works through those objections or considers whether a proposal can be changed. Consensus is reached when objections have been resolved or given due consideration and the community believes the benefits outweigh the disadvantages. Unanimity is not required.

That process makes a consensus call a procedural event. The chairs ask the community to express a state against a particular proposal text at a particular point in the meeting. A positive result and a negative result differ, but both require the event to occur. If no call is held, there is no result to reverse-engineer from applause, microphone traffic, chat messages or the apparent temperature of the room.

The difference is not semantic fussiness. “Did not reach consensus” can trigger a named next step. In this report the three proposals in that group are explicitly sent back to the mailing list. “No consensus call” carries no equivalent next-step clause. A database that permits only passed and failed will therefore misstate two of the seven agenda items. A news brief that calls them defeated will do the same in prose.

Nor can an observer safely use the proposals’ histories to fill the gap. Prop-165 had previously failed to reach consensus at APNIC 60, was withdrawn at APNIC 61 and returned in a later version for APNIC 62. Those earlier states show that one proposal can move through different procedural positions at different meetings. They do not explain the missing call on 10 September. Prop-172 was newer, with versions and an impact assessment published before the meeting. Its shorter history also supplies no reason.

The only defensible common statement is the one the report makes: no consensus call was held for either proposal.

The strongest case for a short report

A conference report is not a transcript. Its job is to tell a broad audience what happened across plenaries, Special Interest Groups, elections, technical sessions and community events. Requiring it to reproduce every intervention would make the summary less useful. People in the Policy SIG room may also have heard an explanation that the compressed page does not repeat.

This defence matters because the omission of a reason is not evidence of manipulation or procedural failure. Chairs may have had sound grounds to stop before a call. A text may have changed, time may have run out, an author may have requested another step, or discussion may have exposed work that must occur first. These are examples of possible reason categories, not claims about prop-165 or prop-172.

The public-record problem is narrower. The website is the durable interface for a participant who was not in the room, an operator returning months later, or a system trying to track proposal status. For those readers, a result sentence needs a join to an authoritative next state. Otherwise knowledge remains dependent on attendance, memory or a later mailing-list search.

At the evidence freeze, the individual pages for prop-165 and prop-172 still said “For Discussion at APNIC62 OPM”. That wording was accurate before the meeting. Afterwards it no longer answers the important question, because discussion occurred and the report has recorded no call. The stale-looking label does not prove a later status. It simply shows that the meeting report and the proposal register have not yet been joined in a way a remote reader can follow.

Preserve the no-call state

The repair need not turn the conference summary into minutes. APNIC could publish one compact proposal-state receipt for every agenda item.

The receipt would identify the proposal and exact version considered, the session and time, and whether a consensus call was held. If a call occurred, it would record the result and the next step. If none occurred, it would preserve a bounded reason category, link or quote the chairs’ statement, name the next forum or action, and timestamp the update to the authoritative proposal page.

Reason categories should describe procedure, not manufacture sentiment: for example, text revision required, author action pending, agenda time exhausted, prerequisite evidence outstanding, or decision moved to a named later forum. APNIC should use only the category supported by the contemporaneous record. Where no reason was preserved, the receipt should say that rather than retrofitting certainty.

The Boolean field is equally important: call held — yes/no. It prevents no-call from becoming a null that software later interprets as failure. The result field should be conditional on that answer. A blank result with a positive no-call state is more accurate than a completed red or green box.

This control is Theo March’s proposal, not a feature APNIC has announced. It does not second-guess the chairs or demand a vote. It makes the object of decision and the absence of decision equally auditable.

Sources