← All posts

Who presses the button at 04:00:00

When ten outlets run one story in the same second, it looks like the world confirmed it independently. We counted it to the second: almost always there is a single press service or news agency behind it.

On the morning of 3 September 2026, ten outlets from six countries ran the same story — a UN report that overshooting the 1.5 °C target had become unavoidable. Here is exactly when:

In a minute and a half, ten newsrooms reacted in four languages. The first — at exactly 04:00:00.

A picture like this looks like proof of collusion, and that is exactly why we decided to study it closely. The answer surprised us: it turned out to be unexpected and predictable at once.

01Beat the 60 seconds

We have each record's publication time at the source — not when we saw it, but what the source itself declared. For every story we know when each outlet joined it.

A simple count gives you nothing here, and that is the first thing to know. Within a single story the typical gap between outlets arriving is 264 minutes. Pairs that fall within one minute — 0.5%; within five minutes — 1.8%; within an hour — 19.4%. So "several outlets ran it close together" is not, by itself, an event: at that spread, rare coincidences are bound to happen.

So we measured not "how many outlets ran it close together" but how improbable that is given the story's own pace. Six outlets within two minutes is unusual for a story that drags on for a day, and normal for a story that fit entirely into ten minutes. The same picture means different things. It all depends on how fast the story was moving in the first place.

We called a burst anything that, by chance alone, would occur less than once in a thousand times. The exact formula (a binomial tail on the story's pace) is in the "Method".

02The week's bursts

Over seven days, among 3,251 stories with five or more outlets:

rarer than 1 in 20158 · 4.9%

rarer than 1 in 100108 · 3.3%

rarer than 1 in 100060 · 1.8%

rarer than 1 in 10,00033 · 1.0%

rarer than 1 in a million14 · 0.4%

Among stories that reached the feed the share is higher — 3.4%. Among large ones, ten outlets and up — 3.7%. The bigger the story, the more often it contains a moment when everyone runs at once. The densest burst of the week: 10 outlets out of 42 within two minutes.

03Diagnosis — an embargo

We looked at which second the dense window starts on. It turned out that 19 bursts out of 60 (31.7%) start within five seconds of a round hour.

A round mark is a mechanical signature. No newsroom independently decides to publish a piece at exactly 04:00:00. That is what it looks like when a publication ban, agreed in advance with the source, is lifted.

Game reviews, 15:00:00. Five outlets — Eurogamer, Gamekult, IGN, Numerama, PC Gamer — published to the second. One second later Golem, three later VGC, six later Ouest-France, nine later Le Monde. Nine outlets, four countries, nine seconds.

UN report on El Niño, 06:00:00. El País, Libération, RTÉ and Wired — right on zero; G1 after 45 seconds, Daily Mail after 49, BBC after a minute fifty-four.

This is not a conspiracy. It is a press service that sent the material out ahead of time on a "do not publish before" basis. The practice is legal, well known and useful: it gives newsrooms time to read the document and adapt it to their own format, rather than paraphrasing a press release on the fly.

But for a reader looking at the feed, the effect is the same as collusion: the whole world says the same thing at once. In that moment it looks like independent confirmation. It is not. The ten outlets that ran at 04:00:00 read the same document and the same accompanying note.

04At Messi's speed

The bursts that did not land on a round mark are nearly all explained more simply: a news agency's wire fired, and everyone subscribed to it ran within a minute.

Messi leaving the Argentina squad: nine outlets across eight countries in 1 minute 45 seconds (15:24:53 → 15:26:38). The death of former French PM Balladur: six outlets in 1 minute 41 seconds, five of them French. Starmer leaving parliament: Belfast Telegraph and Evening Standard — in the very same second, 11:27:08, then Daily Mail, RTÉ, Financial Times, Times of India.

05Dozens of outlets — one voice

Here is where it gets most awkward for intuition. Within each burst we grouped outlets by how similar their texts are. If two pieces are almost word for word the same, that is one wire that spread across storefronts, not two independent witnesses.

In some cases it is a complete collapse. The Messi story, which gathered more than a hundred outlets, turned out to be one group: a hundred storefronts, but one text. The story "Justice Ministry adds a comedian to the foreign-agent register", where eight Russian outlets ran within two minutes, consists of 30 sources and collapses into two groups. A road accident in Stavropol Krai: 24 outlets — one group.

06The odd one out

We went looking for a third class — coordination explained by neither an embargo nor an agency wire. We did not find it. It all turned out more prosaic.

After filtering out round marks and requiring the burst to happen inside a single country or language audience, 23 stories a week are left — 3.3 a day. But even they fall apart if you count independent voices rather than outlets. Collapse the near-identical texts into one voice and a quarter of the bursts survive, and six of those are product announcements and lifted embargoes (Range Rover, Huawei, Lenovo, Nvidia, Android, a game review). The rest are genuine breaking news.

We are honest about the method's own fragility, too: the share of surviving bursts depends heavily on where you draw the line for "texts too similar".

0.92 · "similar"16 / 64

0.9532 / 64

0.9750 / 64

0.99 · "almost verbatim"61 / 64

A quantity that changes fourfold with the third decimal of a constant does not describe a stable phenomenon, and we chose not to make a reader-facing metric out of it.

07So was there collusion?

Here are three takeaways from our small study that may add a little understanding of the information field and how media work in general.

Simultaneity is not proof. It happens with embargoes, with a central-bank decision at 13:30, and with a goal in the ninetieth minute. The observation "they ran at once" describes behaviour, not motive, and confusing the two is the most common mistake in conversations about the media.

The number of outlets is not a measure of confirmation. A hundred and six storefronts can be one voice. The only way to tell is to compare the texts, not count the logos.

Whoever looks for a conspiracy will find it in a press service's schedule. Exactly 04:00:00 looks far more suspicious than 04:00:33 — and means exactly the opposite: an open, documented practice, visible precisely because it is tidy.

That is why at digesta news "how many outlets" is not a trust figure: we compare texts rather than count logos, and we show where a hundred storefronts are one voice.

---