Summary

  • LACNIC published LAC-2026-6 for discussion on 25 September 2026. Version 1 repeatedly replaces generic references to a policy proposal with references to a version of that proposal. It has not reached consensus or become policy.
  • The author says the revisions clarify how the PDP already operates. He also says the exact technical reason a discussion clock cannot restart without a new system version has not been confirmed by staff. The proposal's explanation and a staff finding are different kinds of evidence.
  • The practical test is whether the text assessed at each stage can be traced back to the version that was published and debated, with an explicit account when minor or editorial changes follow a different path. A stable proposal code alone does not answer that question.

One code, more than one text

The public page for LAC-2026-6 names one proposal and one present version. That is enough for an initial discussion. It is not enough to describe the journey of a proposal after revisions. The identifier can stay in place while the text participants saw at the beginning differs from the text placed before a forum, tested for consensus, sent to a last call or considered by the Board. The identity of the proposal and the identity of the text are related, but they are not interchangeable.

The draft acknowledges the distinction by changing the nouns throughout selected PDP provisions. A version would be received and published on the Policy List. A version would meet the two-week availability rule before the Public Policy Forum. Moderators would assess discussion and suggest consensus on a version. A version would be presented and debated at the forum. The Board would accept or reject a version. The sequence matters because each verb may carry forward an authority that the earlier verb did not confer. Publishing a text makes it available; it does not mean that objections were addressed. Presenting it does not establish consensus. Consensus does not make it Board-ratified or implemented.

The strongest reading of the proposal is a modest one. Its author, Jordi Palet Martinez, says the edits make the written PDP match practices already used, not redesign the PDP. That can be a valuable repair. If the system already distinguishes versions but the published rules sometimes say only “proposal,” careful language can reduce avoidable disputes. One need not infer a hidden power grab from a better noun.

Still, a clarification is consequential precisely when it determines what the next reader believes happened. For an operator affected by a resulting number-resource rule, the question is not whether an internal system displays a version number. It is whether an objection to one text can be treated as disposed of when a different text later receives the decision. A family code cannot answer that on its own.

A clock belongs to a text, not to a label

LACNIC's current PDP places at least eight weeks of discussion before consensus assessment and a four-week last call after it. It also requires forum presentation in the ordinary path. These are opportunities to examine a particular text, not abstract waiting periods attached to a code. The new draft's version language makes that relationship visible across the Policy List, forum and moderators' duties.

The awkward case is a revision during the journey. If discussion ends without consensus, the proposed wording says the authors can continue with a new version or withdraw. If debate prompts changes before the decision, it distinguishes restarting discussion for a revised version from minor changes that may be introduced during Last Call. Existing PDP history, including LAC-2020-6, separately treats editorial changes during Last Call as a reason to publish a new version and restart that stage. Those paths should not be flattened into “the proposal stayed open.” The relevant public facts are which words changed, which stage was repeated and why an exception did not require a fresh forum presentation or eight-week debate.

The author adds a practical explanation: the policy system apparently cannot restart the discussion period without a new version. He expressly says the precise reason has not been confirmed by staff. A careful reader should therefore distinguish the observed proposal to require a new version in that situation from a verified technical diagnosis of the platform. Even if the platform constraint is confirmed, it would explain an administrative implementation choice; it would not, by itself, establish that every community objection to the new words was heard.

The draft also moves the code-assignment moment. Its current-text comparison says a proposal is reviewed and then assigned an identifier; the proposed text says the system assigns the identifier automatically before review and publication. That is a small but useful separation. A system ID proves that an object entered a workflow. It does not prove that the text was publicly available, qualified for a forum agenda, reached consensus or survived Board review. An automated timestamp can help reconstruct the path, but only if it is not mistaken for an approval.

What the record would have to show

Consider a proposal first discussed as version 1 and later revised as version 2. The question is not whether all revisions are forbidden; a living process must permit corrections. The question is what version 2 inherits. If it changes a substantive requirement, the affected parties need to know whether a new discussion and presentation were required. If it is minor enough for Last Call, the record should identify that judgment and the text difference. If a last-call editorial repair requires a new version and a restarted last call, the record should show both.

None of those outcomes is inferred to have happened in the LAC-2026-6 case; they are the cases against which its proposed language should be tested.

A compact public stage map could link the proposal-family code to each exact version, its publication date, the forum at which it was presented, the objections and their disposition, the change classification, the consensus call, the last-call interval and the Board's eventual decision. It could record a superseded version without pretending it vanished. Such a map would not make attendance into authorization, or require private operator submissions to be published. It would show whether the text that acquired authority passed through the advertised tests. This is an editorial test, not a feature LACNIC has announced.

LAC-2026-6 remains under discussion. Its author may be right that it only states existing practice more accurately. The useful test of that claim is not an argument over the word “clarification.” It is a reconstructible chain from the version people could contest to the version a decision would bind. If that chain is visible, the new nouns have done real work. If it is not, the stable proposal code will continue to hide the most important change in the room: the text itself.

Sources