About

Overton Catcher measures how stories spread across newsrooms that are meant to be independent of one another.

Two choices decide everything else. The unit of analysis is a story — a cluster of articles close in meaning — rather than a keyword: one event splinters into a dozen unrelated words, and counting words measures how big the news was, not how it travelled. The unit of independence is the owner rather than the outlet: two papers under one parent company are one newsroom, and an outlet that ran an agency’s copy word for word made no editorial decision at all.

Both are measured inside a market — a set of newsrooms competing for the same audience — because synchrony means nothing outside one. Two papers on different continents covering the same summit are covering a summit; two papers in the same city doing it within the hour are doing something else. Reports currently cover the British national and regional press, the American national press, and a market made up of state-funded international broadcasters. Collection runs wider than publication.

Spreading widely proves nothing on its own, and that is the firmest result here so far. The market of state broadcasters — the closest thing this project has to a labelled example of coordinated promotion — turned out to be indistinguishable from an ordinary national market on every measure of synchrony tried. Real news also reaches everyone within hours. Where a report shows a story carried by many owners at once, the honest reading is usually that it was a big story.

The name promises a second thing: the boundary of what can be said without apology. That work has barely begun. One section of each report counts how often a contested term appears inside quotation marks rather than in a newsroom’s own voice — the difference between reporting that someone said it and saying it yourself. A single week of that is a level, not a movement, and it will take months of archive before the series says anything.

How the figures are produced is described at the end of every report. They are generated once a week without human editing, and the method is still moving: read them as work in progress rather than as findings.