Summary

  • LACNIC publishes Spanish, English and Portuguese versions of the same contact page, each linking weekday office hours to a holiday table.
  • The Spanish table has thirteen 2026 rows. English has ten 2026 rows and three 2025 rows; Portuguese has four 2026 rows and nine 2025 rows.
  • Tourism Week is listed as 2–3 April 2026 in Spanish, 3–4 April 2026 in English and 3–4 April 2025 in Portuguese.
  • The divergence does not prove a wrong closure or deadline. It shows that an external reader cannot recover one authoritative calendar state from the current multilingual record.

The language switch changes the year

The cleanest governance audits often begin with something too small to deserve a committee. Open LACNIC’s contact record in Spanish and the holiday table contains thirteen rows, all dated 2026. Open the English rendering and the table still contains thirteen rows, but three have remained in 2025: the anniversary of Artigas’s birth, Independence Day and Christmas. Open Portuguese and nine rows remain in 2025; only the two Carnival dates, 12 October and 2 November carry 2026.

There is also a date disagreement inside the same named holiday period. Spanish places the two Tourism entries on 2 and 3 April 2026. English places Tourism Week on 3 and 4 April 2026. Portuguese shows 3 and 4 April, but in 2025.

The three public versions were captured together on 29 August. This is not a comparison of an archived page with a current page, or of three unrelated publications. It is a simultaneous reading of one contact record as LACNIC offers it in three languages.

The three versions agree on the institutional setting. They identify the Casa de Internet address in Montevideo. They give Monday-to-Friday hours equivalent to 09:00–18:00. Each connects those hours to closure on holidays. The divergence begins where the calendar should become concrete.

No responsible conclusion can jump from that observation to an operational incident. The pages do not prove that LACNIC opened on a listed holiday, closed on an unlisted one, delayed a request or calculated a deadline incorrectly. They do not state which language has priority. They do not reveal how a staff schedule was actually administered. What they prove is narrower and sufficient: the public record offers three answers to a question that should have one versioned source.

This is an operating surface, not festive decoration

A holiday table on a generic corporate “about” page might be dismissed as low-stakes housekeeping. This one sits below the contact routes through which users are directed toward number-resource requests, IPv4 and IPv6 matters, autonomous system numbers, RPKI, bulk WHOIS or RDAP access, membership records, elections, invoices, technical support, measurement projects, events, training, policy work, grants and security-incident reporting.

That does not mean every channel stops whenever the physical office is closed. An email system may accept messages. A security channel may operate differently from a billing desk. Automated RPKI or registry services may continue. A procedural filing may be deemed received even when no person answers until the next day. The current page does not type those differences, and that is precisely why the table needs a modestly better data model.

Four objects are being compressed into one phrase. The first is a civil holiday: a date recognised under a jurisdiction’s rules. The second is an office closure: a decision about staff presence at a named location. The third is service availability: which phone, email, portal or emergency channel remains active. The fourth is procedural effect: whether the date pauses, extends or otherwise changes a clock in a named rule.

Those objects may coincide, but they are not synonyms. A civil holiday can exist without shutting an automated service. A closed office can keep an incident channel open. A business-day rule can incorporate a calendar different from the one used for normal reception hours. A public holiday label can be translated while its date remains a language-neutral fact.

LACNIC’s other documents make the distinction worth preserving. Its IPv4 exhaustion material uses “working day” in allocation timing. Its election conflict regulation uses business-day intervals. Its Code of Ethics publishes a five-business-day response period. None of the checked text says the contact-page table governs those clocks. That absence prevents an accusation; it also reveals the missing relationship. A reader should not have to guess whether “closed on holidays” is merely reception information or an input to a formal time calculation.

Translation has become version control

The likely publishing pattern is familiar. A page body exists in several languages. Someone edits one body to roll the calendar into a new year. Other bodies are edited partly, later or not at all. Because every language is stored as a complete document, the operational date is duplicated along with the prose.

That is an inference from the public state, not a claim about LACNIC’s internal workflow. The pages do not identify the editor, the sequence of changes or the reason. Perhaps one table was corrected and another was waiting for review. Perhaps dates were copied from different sources. Perhaps a current schedule exists elsewhere. The important point is that the current arrangement lets language determine which state a reader receives.

Translation should govern the name and explanation of a holiday. It should not govern the year, effective period or scope. When those fields live only inside rich text, a linguistic edit and an operational amendment look the same. There is no stable identifier for one holiday across languages, no public version number and no machine-readable way to say that a date supersedes an earlier entry.

The result is not just stale copy. It is a chronology without receipts. A reader can observe that the states disagree but cannot tell which state was approved, when it took effect or whether an earlier state remains valid for a specific procedure. If the Spanish table is updated tomorrow, the disagreement may disappear while the history of the correction disappears with it.

One source record, three human renderings

The repair does not require a new governance body or an elaborate platform. A small locale-neutral calendar object would be enough. Each entry could carry:

  1. a stable holiday identifier, calendar year and civil jurisdiction;
  2. the civil date and the source from which it was adopted;
  3. the relevant time zone and, where useful, start and end instants;
  4. the office, desk or service channels affected;
  5. a procedural-clock field naming any affected rule, or explicitly saying not affected or unspecified;
  6. the approving role, effective date and publication date;
  7. the version superseded and the reason for any correction; and
  8. localized names and explanations rendered from that same record.

Spanish, English and Portuguese pages would then be views, not competing ledgers. A change to the civil date would generate one amendment. A correction could point to the old value without erasing it. The public view could show “last reviewed” and expose a compact fingerprint for the complete annual set. An operator, member or auditor could verify that every locale was rendering the same version even if the words differed.

This separation also prevents overreach. The record need not claim that every holiday suspends every process. It can state that the Montevideo office is closed while a portal remains available, or that procedural effect is unspecified until the relevant rule says otherwise. Precision is not the same as institutional expansion. It limits what the calendar is allowed to mean.

The small receipt protects the narrow institution

Internet number registries administer records whose value depends on continuity and legibility. That does not turn every contact-page defect into a registry crisis. It does mean that the habit of record-keeping matters. A body that expects members to respect exact identifiers, dates and procedural states should make its own small operational states reconstructable.

The strongest response would not be a statement that one language was “only a translation.” Readers use the offered language as the public interface. If a translated page can route an invoice question, an RPKI request or an incident report, its calendar cannot be treated as decorative whenever it diverges.

Nor should the answer be to remove the English and Portuguese tables. Multilingual access is not the cause. Independent operational copies are the cause. The right design keeps all three languages while giving them one factual spine.

The present evidence supports a restrained conclusion. One LACNIC office is currently described by three public calendar states. No harm is proven. No controlling language is identified. The missing item is the receipt that would let any reader move from a localized page to one approved calendar, understand its scope and see what changed. That is a small obligation. It is also the difference between translating an institution and accidentally multiplying it.

Sources