Summary

  • LACNIC's policy register lists Sergio Kleiman Schwarzstein as author of LAC-2026-3 Version 1, carries his own problem statement on missing ROA coverage, and then closes the file with "Did not reach consensus 04/11/2026" and "Withdrawn 06/11/2026" and no stated reason for either.
  • The same registry records grounds in full elsewhere: its 24 March 2026 board minute ratifies LAC-2025-5 v2 by recorded resolution, names the directors present and notes that one of them left the room because he was a co-author. The bare register is a choice of form, not a limit on the institution's ability to write.
  • What the record decides and what it leaves open are two different things. The instrument was withdrawn. The problem it described — prefixes in the region without cryptographic origin protection — was never adjudicated, and the register's silence is a silence about the ground, not a verdict on the claim.

A file with a name on it

The LACNIC policy system publishes LAC-2026-3, "Require Valid ROA Coverage for Requesting Additional Resources", under a named author: Sergio Kleiman Schwarzstein, Version 1 (policy register, English; original Spanish). That is a small administrative fact with large consequences for how the rest of the record reads, because the page does not present the proposal as an anonymous output of a committee. It presents a person's claim, attached to a person's name.

The claim itself is precise. The proposal would require at least 80% valid ROA coverage for the IPv4 and IPv6 resources an organisation has already been assigned or allocated by LACNIC as a condition of approving a subsequent allocation. In plainer terms: prove you have signed your existing address space before you ask for more.

And the page carries a separate section headed "Rationale (Describe the problem you intend to solve)", which states that issuing ROAs is a recommended practice, that no policy requirement currently promotes high ROA coverage for resources registered in the region, and that some prefixes therefore remain without cryptographic origin protection. That is a problem statement written in the author's own frame: a harm named, a population said to bear it, and a mechanism proposed.

A problem statement is not a finding. The author asserts; the page prints the assertion. But it is the author's assertion, which is what makes the next part of the file worth reading carefully.

The questions came back, and the answers were kept

Jordi Palet opened the discussion on 8 October 2026 with "En principio de acuerdo con la propuesta" — in principle in agreement — and then asked the practical questions: does this apply to incoming transfers, what about IPv6 rectifications, what about direct assignments, and where does ASPA fit (politicas archive, 8 October 2026)?

Three days later the author answered in his own name: "Sí, como está redactado aplica también para transferencias entrantes" — as drafted, it covers incoming transfers too — and added that the wording could be extended to sections 4.4.1.4 and/or 4.4.4.3 for clarity and could incorporate ASPA now that miLACNIC supports ASPA objects (author's reply, 11 October 2026).

That exchange is the ordinary machinery of policy development working as advertised: a question, a dated answer, a named respondent, a drafting consequence.

Douglas Fischer came at the proposal from a different direction on 14 October 2026 and closed his message with "Sendo assim, respeitosamente, sou contra a proposta apresentada" — respectfully, I am against the proposal as presented (politicas archive, 14 October 2026). His reason was jurisdictional rather than technical: a desirable operational practice does not belong in the policy manual. That objection is the interesting one, because it does not dispute the harm. It disputes the instrument.

On 30 October 2026 Kleiman posted his own summary of the comments raised at the LACNIC 46 public policy forum, recording both an interest in promoting the creation of ROAs and the view that promoting ROA creation may not be a matter for the Policy Manual, and ended by asking the list directly: "¿Creen que deberámos continuar con la discusión de la incorporación de incremento de ROAs en el MDP (a través de esta propuesta u otra) o no es necesario?" (author's forum summary, 30 October 2026). The regional and NIC México figures he offered in that message were hedged in the message itself as recollection, and this article does not treat them as verified statistics.

His question to the list is worth reading twice, because it separates the instrument from the objective: this proposal or another, or none at all. An author asking whether the problem should be pursued after his own instrument is doubted is doing something a status field cannot do.

Fulgencio B. Espinal Mckinney of INDOTEL answered the same day, agreeing that the ROA-coverage idea merits continued evaluation, warning that the 80% threshold should not become an automatic obstacle to legitimate requests, asking how coverage would be calculated and whether invalid ROAs could be corrected before a decision, and asking that ASPA be kept separate (politicas archive, 30 October 2026).

Two weeks earlier the moderators had told the community what to expect. Their LACNIC 46 forum report of 28 October 2026, signed by Tomas Lynch and Sergio Rojas, recorded that the proposal's eight-week initial discussion period finished on 21 October 2026 and that, from the end of the forum and for two weeks, the moderators would inform the community whether the proposal had reached the first consensus (forum report, 28 October 2026). So the file names the deciders, names the deadline, and states the question they were going to answer.

Then two dated words

The register's own procedural history reads, in four lines: in discussion 12/08/2026 to 21/10/2026; first consensus assessment 21/10/2026 to 04/11/2026; did not reach consensus 04/11/2026; withdrawn 06/11/2026. The status label is "Withdrawn" on the English page and "Retirada" on the Spanish original, where the timeline reads "No alcanzó consenso".

What the page as read on 19 November 2026 does not carry is equally specific: no reason for the failed first consensus assessment, and no reason for the withdrawal. Not a summary of which objection carried. Not a line saying the threshold was unworkable. Not a line saying the instrument was the wrong home for the problem. Two dates and two status tokens, on a page that elsewhere prints the author, the proposed text and his rationale in full.

The page's own "Last modified" field reads 07/08/2026, which precedes the discussion start date printed on the same page. The register does not explain the ordering, and this article draws no conclusion from it beyond noting that the record's self-description does not reconcile.

One boundary belongs here rather than at the end. The November 2026 messages themselves — the moderators' announcement of the assessment and any message recording the withdrawal — were not obtained for this article. The absence described above is an absence on the register page, and nothing here asserts that the mailing-list archive is silent. The register is where a stranger goes first, and the register is where the gap is.

The same building writes reasons

Nine months before those two lines appeared, LACNIC's board recorded a policy decision in a completely different form.

Its minutes of 24 March 2026 carry, as resolutive item 2.1, the ratification of Policy LAC-2025-5 v2, "Sub-asignación de Recursos IPv4 a Terceros", approved unanimously among the directors present. The same document names who was present — Esteban Lescano, Evandro Varonil, Carmen Denis, Gabriel Adonaylo, Wardner Maia, Rafael (Lito) Ibarra and Edmundo Cázarez, with Executive Director Ernesto Majó — and adds one sentence of ground that has nothing to do with the policy's substance and everything to do with the record's quality: director Cázarez left the session because he stated he was a co-author of that policy (board minutes, 24 March 2026).

The contrast is not rhetorical. One document class in this registry states the decision, the deciders, the vote, the recusal and the reason for the recusal. Another document class, for a proposal that failed, states the date and the status. The register shows no ratification step at all for LAC-2026-3; its printed sequence ends at the withdrawal, whereas the LAC-2025-5 record carries a board ratification date. Both records are first-party. Neither is a draft.

That is what makes the missing paragraph a matter of form rather than capacity. An institution that writes "left the room because he was a co-author" has the vocabulary to write why an eighty-percent threshold did not survive its first assessment. It chose, in this file, not to.

The instrument died; the problem did not

It would be easy, and wrong, to read "withdrawn" as a verdict on the claim that prefixes in the region lack cryptographic origin protection. Nothing in the register adjudicates that. What failed was a mechanism — an 80% coverage gate placed inside the Policy Manual — and Fischer's objection targeted exactly the placement, not the harm. Espinal's objection targeted exactly the calibration, not the harm. The author's own closing question to the list conceded the same distinction: the instrument could change, or end, and the objective could still be worth pursuing.

So the accountable reading of this file is narrower than a scandal and more useful than a shrug. A named author put a problem and a fix on the record. The community answered with a jurisdictional objection and a calibration objection, in dated and attributable messages. The moderators published a deadline for their own verdict. The verdict arrived as a date, and the reasoning did not arrive at all.

For anyone who has to decide whether to spend eight weeks drafting the next proposal, that is the operative fact. LACNIC's policy register will record your name, your text and your rationale. It may not record why your text died — and the registry that can name a recused director's reason has no structural excuse for leaving that blank.

Where this registry's own record is read, the entry point is the directory card: LACNIC.