Summary

  • ARIN-prop-353, posted 2 August 2026, credits Chris Woodfield as originator and defines out-of-region use as use "in, or advertisement from, locations outside of the ARIN region"; Draft Policy ARIN-2026-5, current text 22 September 2026, drops "advertisement from," refers to the "ARIN service region," and names only its two shepherds.
  • The Advisory Council minutes of 17 September 2026, still a draft pending approval, record that a shepherd "spoke to the author and made an amendment to problem statement" — yet the published problem statement is identical on both pages, and the definition is the text that moved. No dated objection by the originator appears in the September announcement or the October and November mailing-list indexes.

The 2 August 2026 page for ARIN-prop-353 carries a date, a problem statement, and one line of credit: "Proposal Originator: Chris Woodfield." Its definition of out-of-region use is written with a second verb in it. "The sole use of ARIN-allocated resources in, or advertisement from, locations outside of the ARIN region," it reads, and its exception turns on "the use of, advertisement from, locations within the ARIN region in addition to other locations." Both halves measure a location against "the ARIN region" (ARIN-prop-353).

The draft page that replaced it carries a status, "Under Discussion," two names, "Shepherds: Gus Reese, Bill Herrin," a current text dated 22 September 2026, and a shorter definition: "The sole use of ARIN-allocated resources in locations outside of the ARIN service region." The exception was rewritten as well, into "the simultaneous use of a given ARIN resource both within the ARIN service region and in other RIR service regions" (Draft Policy ARIN-2026-5).

Read the two definitions side by side and the movement is specific rather than atmospheric. One verb, advertisement, leaves the operative sentence. One noun changes: the ARIN region becomes the ARIN service region. What survives is the part the two-page comparison cannot compress — an out-of-region use is still something a holder does with ARIN-allocated resources somewhere else. My 19 November 2026 briefing on ARIN-2026-5 set out that comparison and the missing originator line.

This one asks the question that follows from it, and gets an answer the earlier piece did not have to reach: what record did the change leave, and what can the person whose sentence moved do about it?

The minutes record an edit. The pages record a different one.

The Advisory Council met on 17 September 2026. The minutes of that meeting are published, marked as reviewed by the Council and pending approval at its next meeting, and they contain one sentence about the text: "GR stated that he spoke to the author and made an amendment to problem statement and posted the language to the AC Wiki." The same minutes record that the motion to move ARIN-prop-353 to Draft Policy carried unanimously by roll call (AC minutes, 17 September 2026).

Now put that sentence beside the pages it describes. The problem statement printed in the 22 September announcement is the proposer's problem statement. It is also the problem statement on the draft page. The text that reads differently between the proposal and the draft is the definition (22 September 2026 announcement).

That is a mismatch a reader can check without taking anyone's word for anything: the amendment the record describes was made to a portion of the proposal that the published draft reproduces unchanged, while the portion that did change is not mentioned in the minutes at all. Three readings are possible and none of them is established by any page read on 20 November 2026 — the minutes may abbreviate, the wiki language may have been folded into the policy statement rather than the problem statement, or two amendments may have occurred and only one been written down. What is not possible is to reconcile the record with the draft as printed.

The minutes are draft, and they remain unapproved.

What the rules give the author, and when it stops

The ARIN Policy Development Process is not vague about this. Its section 2.2.1 says: "The Proposal Author controls the language of its Policy Proposal until it is accepted by the Advisory Council as a Draft Policy" (Policy Development Process). That single sentence contains both the grant and the expiry. Before acceptance, the author's words are the proposal. After acceptance, they are an input.

What replaces them is section 2.3.3.3: "The Advisory Council may edit the Draft Policy in response to feedback from the Internet Community; in response to feedback received during the staff and legal review; and to better align the Draft Policy with the criteria outlined in section 2.3.1." The Council may also abandon a draft that cannot meet the section 2.3.1 criteria, or where an alternative draft draws more community support. None of that is hidden or irregular. It is the procedure, and the procedure is why a draft's operative language can differ from a proposer's.

What the inspected text does not provide is equally plain. It gives a Proposal Author no post-acceptance right to approve, veto or correct the draft's wording, and no author-withdrawal procedure at the draft stage. The recorded status "Withdrawn by the author" exists, but it sits on the proposals index and against older entries — ARIN-prop-301 from 2021 and ARIN-prop-294 from 2020 among them — while the 2026 proposals displayed carry "Under Discussion" or "New Proposal" (Policy Proposals index). The drafts index lists ARIN-2026-1 through ARIN-2026-5, all as "Draft Policy," and offers no withdrawal status value at all (Draft Policies index). Author withdrawal is a proposal-stage instrument. Once the Council has accepted a draft, the record carries no exit the author controls.

The doors that exist, and the size of the key

Two routes are published, and both are petitions.

Section 2.3.4.1 covers abandonment. If the Council abandons a draft, a community member may ask that a petition be initiated to reverse the decision, within seven days after the announcement date of the Council's action. A successful petition returns the draft to Draft Policy status, triggers a staff and legal review, and bars the Council from revising or abandoning the draft again until it has been presented to the community at a Public Policy Consultation. Every clause in that section assumes the draft has already been abandoned. It does not reach a draft that is alive and worded differently than its author wrote it.

Section 3.1 is broader on its face. "A member of the Internet Community may petition the Board of Trustees to overrule a decision made by the AC that an Internet Community member believes does not properly follow the Policy Development Process as described in this document," it reads, and it adds a restriction that cuts the other way: "ARIN Staff and ARIN Board of Trustees may not initiate or be counted in support of petitions." The author of a proposal is a member of the Internet Community, so Woodfield could open that petition himself.

Sections 3.1.1 to 3.1.4 then set the terms under which it succeeds: initiated by web form with an issue statement, a seven-calendar-day window after the announcement on the public policy mailing list, support given only by registered points of contact of ARIN member organizations, and success at 15 valid support forms from 15 different member organizations.

So the door is real, and the person whose sentence moved may stand at it. He cannot walk through it alone. The instrument that lets an individual complain is the same instrument that requires fifteen organizations to agree before the complaint does anything. That is not a contradiction in the text; it is the text, and it tells you whom the process treats as capable of reversing the Council.

What the record does not contain

Here the answer runs against the framing I set out to test. There is no dated written objection by the originator of ARIN-prop-353 to the rewording of his definition anywhere in the pages this cycle publishes.

The 22 September 2026 announcement that introduced ARIN-2026-5 reproduces the new definition and contains no author statement and no objection attributed to the proposer (announcement). The public policy mailing list's October 2026 index lists 52 messages between 7 and 30 October 2026; no message in that index is authored by Chris Woodfield, and no subject line records an author objecting that published draft text differs from submitted text (October 2026 archive). The November index lists 16 messages between 2 and 18 November 2026, none of them his, and its policy-text traffic concerns the revised ARIN-2026-4 (November 2026 archive). His only message in the whole August-to-November window that these indexes carry is dated Sunday 2 August 2026 at 13:36:14 EDT, and it is a substantive comment on someone else's draft — ARIN-2025-3 — about whether anycast has ever been read as out-of-region use. It ends by suggesting that language making the nuance explicit "would be ripe for a new policy proposal," which is what he then filed (2 August 2026 message).

That absence has a boundary, and the boundary is the point. What these pages show is that in the material ARIN published and I read on 20 November 2026, the originator's disagreement never reached the file. What they cannot show is that no ARIN author has ever objected in writing to an altered draft, or that this one did not write privately to his shepherds or his Council. Correspondence that is not published is not evidence either way, and the minutes' own reference to a conversation with the author shows that such conversations happen.

What a reader can still check

The Number Resource Policy Manual, cited as "Version 2025.1 - 3 March 2026," states that number resource policies in the ARIN region are created under this process (NRPM). That is what makes the traceability question load-bearing rather than aesthetic. If ARIN-2026-5 is adopted, the sentence that governs out-of-region use in the manual will be one that no named person proposed, produced by an edit whose stated ground appears nowhere in the published record.

Two columns would close that gap, and neither requires a new philosophy. One is an originator line on the draft page, so a reader can see whose proposal a draft descends from at every stage of its life. The other is a line in the revision record that says which of the three section 2.3.3.3 grounds an edit rests on, so that a change in operative language is attached to a reason rather than to a silence.

Until then, the checkable comparison stands, and it is a small, sharp inheritance from this draft cycle: on one page a named person's words, on another page the words that replaced them, and in the minutes a description of an edit that does not match the edit a reader can see.