The divergence score measures the framing gap between the world’s newsrooms on the same event. The higher the score, the more the narrative fractures along borders.
Across sixteen stories and forty countries, the median gap between framings falls back to fourteen points, against twenty-five in each of the two preceding weeks, and the maximum drops from fifty-eight to twenty-nine. It is the sharpest fall in a month, and it returns the week to W30 levels. A quick reading would see a world coming to agreement. The three points that follow show that would be hasty.
Three stories carry an explicit numerical disagreement and two of them sit among the lowest scores of the week. On TikTok, the settlement figure ranges from four hundred to five hundred and fifty-nine million dollars depending on which country reports it, with no newsroom flagging the gap. In Colombia, the presidency announces three hundred and fourteen deaths while the forensic institute identifies three hundred and nineteen, and there some newsrooms do note it — the story rises to eighteen. Numerical disagreement exists independently of what our scale says about it.
Weak signals and narrative shifts are the Prism’s deep analysis: what newsrooms have not yet named. Subscribers only. The blind spots stay free.
No Zambian outlet appears in the monitored press corpus. The election is told by its neighbours and by two European capitals, never from Lusaka. It is a collection limit, and it falls on the most divergent African story of the week.
As the forty-trillion-dollar mark is crossed, the American press treats the threshold as a domestic budget figure and does not mention China and Japan among the main holders. It is Dhaka, Abuja, Islamabad and Paris that read the figure through its external consequences.
No official Japanese reaction to a decision affecting the region's security architecture, nor any examination of its consequences for Japan itself. The country most directly concerned after Korea is absent from the debate.
Dated, falsifiable predictions, judged publicly. The statement and deadline are free — the reasoning behind them is for subscribers. We settle them later, won and lost alike.
The statement is free. Reading the world, less so. Access the weak signals, the narrative shifts and the full reasoning behind every prediction.
Complete weak signals, narrative shifts with evidence, and the reasoning behind every prediction.
$4.99/month$39.99 /year · save 33%
Show your friends how the world sees the same news differently.
Get the Weekly every Saturday morning. Free.
The counter-example of the week, and it is indispensable: the instrument is not faulty, it has a blind spot. On the Kenya helicopter crash it detects a selection divergence perfectly — Quito centres its account on Ecuador's intelligence chief, Nairobi on the safety of its air tourism, Brussels and Paris on other victims again. On the Zambian re-election it cleanly separates the coverages that lead on restructured debt and falling inflation from those that lead on the opponents arrested. When disagreement lies in the account, the measure finds it. When it lies in the number, it misses it.
Eight of the week's sixteen stories have Washington as their actor, and several follow the same movement: an American announcement, then a correction from the party concerned. South Korea's defence ministry states that manoeuvres are continuing without notable change after the reduction order. The Canadian tariff truce is presented across the Atlantic as the fruit of an agreement in principle, and in London as a conditional gesture with no text. Yet the mean gap of those eight stories — thirteen point three — is the same as that of the other eight, thirteen point nine: the identity of the actor predicts nothing.