How we work

One event, and how Washington, Beijing, Moscow, Brussels, the Gulf, London, Tokyo, Delhi and Brasília each tell it.

What we read

News feeds from outlets in nine blocs, in English, Chinese, Russian, Arabic, French, German, Japanese, Hindi and Portuguese, and the release feeds of governments and institutions. Every feed is listed on the front page under The wires, with its voice (state, public, private, exile or official), its language and a link to the outlet. A feed is admitted only after we have measured it working, its robots.txt allows our crawler, and any machine-readable reservation of text-and-data mining has been read. We read the headline and the standfirst; nothing behind a paywall, and no article text.

What we write

For each event we write a headline, a short account of what happened, and one line per bloc characterising how that bloc's outlets told it — every line ours, with the outlets linked. We never show a publisher's headline or text. Contested claims are attributed (“Xinhua reports…”, “the State Department denies…”); a single-source allegation about a named person is left out, not softened.

State outlets

A bloc's state press is how that bloc tells a story, so it is on the map — labelled STATE on every row, for ever, and never merged with private press into one line. An outlet enters only after a recorded check of its legal position in the UK, the EU and the US. Outlets under sanctions or broadcast bans there — RT, Sputnik, RIA Novosti, Izvestia, Press TV and others — are refused, in text and in video.

Press in exile

Some outlets cover a bloc from outside it — Meduza, The Moscow Times and Novaya Gazeta Europe publish about Russia from Riga and Amsterdam. They do not speak for Moscow, and their framing usually sits closer to London's and Washington's than to the Kremlin's, so counting them as Moscow would put a Western weight on Moscow's side of the scale. We print them in a lane of their own, under the blocs and labelled IN EXILE. They never light a bloc, never count towards its total, never shape its line, and never end its silence: when the Kremlin's own press says nothing, Moscow shows silent, and that silence is part of the story. They do count among the independent confirmations of the facts, because they are separate newsrooms that reported them.

The numbers

The coverage rail counts shared stories: stories that the press of two or more blocs reported in the last 24 hours. For each bloc it shows how many of them its own press covered. A story counts once per bloc, however many items are filed about it. We used to count items, and that measured how our feeds publish rather than how much a bloc says: one Russian newswire, which files every statement and briefing, made Moscow read three times any other bloc. Shared stories still reflect our sources, not the whole of a bloc's press — a bloc we read through more feeds can cover more stories — so the feed count sits beside each row, and the raw item counts are on Sources as a check on the feeds. The languages and voices beneath count the same way: each outlet once for each shared story it covered (and each language it used), so the share of state press is the share of outlets' reports, not of the items a fast wire files. The chart above the rail plots each bloc's shared stories per hour as an average over the six hours up to each point, because a single hour's count swings by chance; hover it for each hour's own count. The right rail shows who is live on television now, the official releases of the day, and any feed we have not heard from.

Coverage spikes

The coverage spikes come from the GDELT Project, which reads news sites worldwide every fifteen minutes and places the events they report by machine. We keep one number per country per hour — how many articles GDELT placed there — and show a country when its last day runs at twice its own rate over the week before. It is a signal that the press is looking somewhere, never a fact about what happened: GDELT's coding is automatic, and nothing from it is ever written into an event. Data: The GDELT Project.

Times

Every printed time is UTC. Ages (“12m”) are counted by your browser from those times, so the page never claims a freshness it does not have.

More

Every feed and its ownership group: Sources. What we collect about you: Data & privacy. What our crawler does and how to stop it: Our crawler.