Summary

  • ICANN fixes the comment window precisely — all dates and times UTC, open proceedings closing at 23:59 UTC, a 40-day floor unless the Bylaws or a community's procedures say otherwise — and reports the count loosely, in three units across three summary reports.
  • The units are 181 "comments" attributed to six community groups and two individuals, 18 "comments" decomposed from four submissions, and 23 "Public Comments" that fold two late filings into the headline; the grouping authority also moves between the requester's Guided Submission Form and ICANN's own themes.
  • The method is published, but as section prose of varying depth rather than a header field, and a 2018 ICANN summary report carried a named preparer line the 2026 template does not.

Start with the part of ICANN's Public Comment machinery that cannot be misread. "All dates and times are based on UTC, and Open Proceedings close at 23:59 UTC," the Public Comment landing page states, and adds the floor: a proceeding runs a minimum of 40 days unless the Bylaws or community operational procedures provide otherwise. Proceedings are opened by the ICANN Board, a Supporting Organization, an Advisory Committee or ICANN org. Whatever else is loose here, the window is not. A reader in any time zone can compute the exact minute their text stops counting.

The Draft PTI FY27 Operating Plan and Budget proceeding shows how much of the record is genuinely checkable. Its timeline prints Open for Submissions 16 December 2025, Closed for Submissions 12 February 2026 and Report Due 2 April 2026, all in UTC. Its submissions list names each filer beside a "Submitted on" date. The report is downloadable at 189.02 KB. And the page carries a 21 April 2026 update recording that the PTI Board adopted the FY27 plan on 17 April 2026. Window, filings, report, decision, on one page: a closed loop.

Then comes the count.

The summary report for the ICANN FY27–31 Operating and Financial Plan and the ICANN/IANA FY27 plans and budgets, published 2 April 2026, states: "ICANN received 181 comments from six community groups and two individuals on the Draft Plans. Six submissions included responses to the questions asked in the Guided Submission Form." Read that twice. The number is 181. The population named beside it is eight commenters — six groups and two individuals — and six form-answering submissions. The report never prints a total submission count, and converting eight commenters into eight filings would be an invention, so I do not.

The companion PTI summary report, published the same day, counts something else: "There were four submissions to the Public Comment proceeding on the Draft PTI FY27 OP&B. The submissions constitute a total of 18 comments. To gain a better understanding of the comments, ICANN grouped them into three main themes." Nine, six and three; the arithmetic closes. Four filings became 18 comments and every comment has a home. This is the counterweight, and it belongs before any criticism: one of the two reports published that morning reconciles end to end.

The Review of Reviews Draft Report summary report, published 25 August 2026, prints a third construction: "In total, 23 Public Comments (21 comments in this proceeding and two late submissions from the Noncommercial Stakeholder Group (NCSG) and Public Internet Registry (PIR)) were submitted by stakeholders from across the community." The parenthetical is the honest part — the rule is published, and a reader can subtract. But the umbrella noun is new. "Public Comments" now covers filings that arrived after the window shut at 23:59 UTC, so the headline says 23 while the proceeding counted 21. Both figures are true. Neither is the same instrument as 181.

Three reports, three units: comments attributed to a named commenter population; comments decomposed from a filing count; Public Comments that absorb late arrivals. No single report is wrong. Together they mean a reader cannot compare proceedings, because the noun "comment" does different work in each.

The grouping authority moves too. The FY27–31 report lists eight categories and states that "These categories follow the structure provided in the Guided Submission Form utilized for the Public Comment Proceeding." The buckets are the requester's form. The PTI report takes the other road — "To gain a better understanding of the comments, ICANN grouped them into three main themes" — buckets authored by staff. Same publisher, two reports published the same day, two different authors of the placement.

A commenter who answers the form is counted in the requester's rooms; a commenter who ignores it is counted in staff's.

Is the method published at all? Yes — in section prose whose depth varies. The Review of Reviews report opens its summary section by describing what the summary does: organized by theme and related sub-topics, attempting "to highlight any areas of broad support as well as principal concerns raised and points of divergence among commenters," and stating that the summary "does not replace or supersede the comments themselves." The Name Collision summary report goes further one way and less another: its summary section "summarizes the submissions to your open proceeding but does not address every specific position," while a separate Analysis of Submissions section counts 25 distinct comments from four commenters and says which comments fell outside the scope of the procedures under consultation. Where a proceeding asks a yes-or-no question, the third AGB proceeding report prints the tally and appends the rule: "The yes or no response is informational only and is not a vote on the topics or language."

So the finding is narrow, and it is not the one a critic might prefer. ICANN publishes a method. It publishes it in prose, at different depths, in different sections, of different reports — never as a fixed header field. A reader who needs the unit, the grouping rule and the late-filing treatment must find the right section of the right report and hope it is there.

One more field used to exist. The 2018 summary report on the O.COM single-character label prints a publication date and a "Prepared By" line naming Karla Hakansson. The 2026 template prints "Category: Operations," "Requester: ICANN org" and an organisational contact address, and no individual preparer line in the material I read. That 2018 document is a different lineage — the Registry Service Evaluation Process — so this is not a claim that ICANN deliberately stopped naming a preparer. It is a claim about what two documents print.

Who summarises is less mysterious than it sounds, because ICANN says it plainly: "ICANN org summarizes the submissions and identifies common themes in a summary report. The group that opened the proceeding, reviews the submissions and summary report, addresses the input, and proposes next steps" (About Public Comment), with at least 14 days for org to review submissions before the summary stage. In the FY27 proceeding the requester printed on the report is ICANN org — the body that drafted the plans summarised the comments on its own draft. That is the structure, not a scandal. It is exactly why the missing header field matters: when one organisation writes the proposal and the summary, the summary's method is the only independent check a reader has.

And the checking end exists, at the level of one filing. The GAC's submission to the Review of Reviews proceeding prints Name, Date, Affiliation and an Email field, an Attachment block naming "GAC Comment on Review of Reviews Draft Report.pdf" at 163.86 KB, plus Summary of Attachment and Summary of Submission. A stranger can verify that a commenter filed and what they filed. What a stranger cannot do is reconstruct how 23 Public Comments became themes, or how 181 comments became eight categories, without prose that may or may not be in the document.

The fix is a header, not a new duty — five fields, each of which some ICANN report already prints somewhere. Define the unit of the headline number instead of letting the noun imply it. Print the submission count and the distinct-commenter count beside it. State the late-submission rule the way the Review of Reviews parenthetical already states it. State the grouping authority — form-derived or staff-derived, and the rule. Carry a named preparer or sign-off line, as the 2018 report carried.

A sixth line would make the best existing practice universal: the count of distinct comments analysed and a statement of what fell outside scope, which the Name Collision report already does.

Bound what was read. The summary reports were read through the text served for their exact published addresses rather than by opening each PDF, so a labelled methodology section further inside any one of them cannot be ruled out; the finding concerns the sections described. Only one individual submission page was inspected, so nothing here asserts field presence across submission types. The FY27–31 report prints no total submission count, which is why none appears above.

What this adds to what BTW has already published: my EBERO briefing read ICANN's emergency backstop — hours per week at the entrance, cancelled operations in the middle, and no clock for telling the registrant. Sofia Ren's Reconsideration Request 26-14 read a docket with a filing and no posted disposition. Nia Okafor's material sits in the Empowered Community's rejection-action arithmetic. An open search of existing coverage for ICANN public comment summary reports returned no titles. None of that touched the counting of the comment corpus — which is where the accountability gap actually sits, not in whether comments were read, but in whether anyone outside can tell what was counted. Directory entry