Summary

  • Alejandro Acosta has defined LACNIC's IPv6 figures himself, one counted noun at a time: end-user penetration in a June 2024 post, autonomous systems above a 1% APNIC adoption threshold in a June 2021 post, and member organisations holding resources versus configuring them in an October 2020 post.
  • The statistics page that displays the network percentage prints no threshold and does not name the post that supplies one, and its "Last updated" line is a weekly regeneration mark rather than a measurement date. Across the five documents reviewed here, no single page prints two of the three counted nouns.

The sentence that says what the percentage counts

In June 2024, Alejandro Acosta — LACNIC's R+D coordinator — did what most people who publish a regional adoption number never do. He wrote down what the number counted. "The numbers you will see in the following images represent IPv6 penetration rates among end users," he wrote in the ten-year review post, and then, because a rate over people is only as clear as its example: "30% means that out of every 100 people, 30 already have IPv6."

That is a definition a stranger can carry in his head. It also fixes the denominator at people, not networks, not allocations, not traffic. Everything else the post shows — country rankings, the regional level, Uruguay above half — rides on that one sentence. The figures themselves sit in images that the post's text does not display, so no country percentage is used here; the sentence is the thing worth having.

The rule that lives in another post

Three years earlier, in June 2021, he had done the same job for a harder noun, in a post announcing a new statistics page. "A efectos de las estadísticas mostradas se considera un ASN con IPv6 aquellos que reporten más de un 1 por ciento (>1%) en su adopción (valor tomado por APNIC)," reads the launch post — my translation: for the statistics shown, an autonomous system counts as having IPv6 when it reports more than 1% adoption, a value taken from APNIC.

Read that twice. An autonomous system that holds IPv6 addresses, announces a prefix and carries a little traffic still counts as not having IPv6 if the measured adoption sits at or below one percent. The threshold is what turns a continuum into a countable noun, and it has teeth. The parenthetical is his own sourcing note, placed there by the same hand that set the bar.

The page that shows the number and not the rule

The page this rule governs says far less. "Este set de datos muestra de una manera muy sencilla de visualizar el porcentaje de ASs que han desplegado IPv6 vs ASNs que no han desplegado IPv6," reads the ASN dashboard — the dataset shows, very simply, the percentage of ASs that have deployed IPv6 against those that have not.

In the text that page served on 21 November 2026 it prints no threshold, no mention of one percent, and no pointer to the 2021 post that supplies the rule. What it does name is its upstream: a "Stats based on" line crediting APNIC's IPv6 statistics site, plus a link to the underlying open data. So a reader who arrives at the figure can learn which laboratory's measurements feed it, and cannot learn from that page what "deployed" requires. The word doing the work is undefined on the page that uses it.

A date that dates the page, not the measurement

The same page carried "Last updated: 14/Nov/2026 15:59:48" on 21 November 2026. An earlier reading of that URL printed "07/Nov/2026 15:59:47" — seven days before, at the same clock minute. The 2021 launch post states the refresh schedule in his own words: "todos los días sábados a las 3:59 pm UTC -3", every Saturday at 3:59 pm UTC-3.

Mine, and only this: those two stamps sit seven days apart at 15:59:4x, and 7 and 14 November 2026 both fall on a Saturday. A weekly regeneration at a fixed clock time, announced in the post that launched the statistic, is what that line records. It says when the page rebuilt itself. It does not say which period the percentage describes, and the page prints nothing that does.

The year the definitions did not travel

In June 2026, Acosta opened the IPv6 panel at LACNIC 45 with a question the panel write-up reports directly: "Why do we still need to promote IPv6 in 2026?" The same text discusses more than 50% of connections to Google services arriving over IPv6, and scopes that figure — connections to one company's services, not the region's traffic and not its networks.

In the text that post served on 21 November 2026, neither definition appears. Not the end-user sentence. Not the one-percent rule. The newest document in this set, the one carrying a 2026 date and his name at the front, is the one that defines nothing at all.

The third noun: organisations

A third counted thing lives in an older post. "Currently, 96% of LACNIC member organizations have received IPv6 resources, yet only 47% have configured the protocol in their networks," reads the 2020 piece about a LACNIC 34 tutorial Acosta led with Uesley Correa and José Cotúa. Note the shape: a member organisation counts as having IPv6 once it holds the resources — a state that the very next clause distinguishes from having switched the protocol on. Two percentages, one population, two states.

That sentence belongs to the publication rather than to a bylined definition, and it is not a people rate and not a network rate. It is also the pair that later texts keep carrying forward, which is the whole argument in miniature: a number can outlive the clauses sitting next to it.

What the reader has to assemble

Count the nouns across these five documents. End users, on one page. Autonomous systems above one percent, with the rule on a second page and the percentage on a third. Member organisations holding resources against configuring them, on a fourth. The 2026 panel text adds a connection-level figure and no definitions.

Mine, and no more than this: three distinct counted nouns, defined in three separate documents, none of which prints the current value of another's series. No reviewed page holds two of them together. A reader who wants to know how the region is doing on IPv6 must decide, unaided, which pulse to take, and must carry each qualifier from a page he is not standing on.

This is not a story about a wrong number. Acosta's own sentences are careful, sourced and, in the 2021 case, unusually explicit about where the measurement comes from. The finding is structural: the care is distributed, and no single surface reassembles it. The most quotable form of the region's IPv6 story is the one with the least definition attached to it — a percentage on a page whose only stated timestamp is a machine rebuilding itself on a Saturday afternoon.

What these documents do not settle

No rendered browser copy of these five pages was retained; each quotation is the wording the page served on 21 November 2026. The 2024 post's country figures sit in images its text does not display, so no country percentage is cited. Whether the dashboard once printed the one-percent rule and later lost it cannot be established from this record, and I assert no such history. Whether another LACNIC author would produce the same split between a defining post and a displaying page cannot be tested from five documents. And the 50% figure is a connection measure for one company's services, not a regional adoption rate.

Acosta wrote the definitions down, twice, in his own sentences. The pages never carried them together. Until one does, this region's IPv6 rate is a number you assemble yourself — and the assembly is the argument.