Method

Raven Wire groups every headline from 60+ newsrooms into stories. Each story is scored out of 100. The verdict says how well it is confirmed, not whether it matters.

How a story gets its score

Independent confirmationup to 45Separate newsrooms. Wire copy re-run by ten outlets counts once. Outlets run by the same state count once.
Source track recordup to 25Best outlet's tier and editorial independence.
Wire service15An established newsroom runs AP, Reuters or AFP copy.
Geographic spreadup to 15Outlets from different countries.
Hedged languageminus up to 20“Reportedly”, “unverified”, “could not be confirmed”.
State media onlycapped at 25RT, TASS, CGTN, Xinhua, Global Times: a record of what a government says, never confirmation.
Rated false by a fact-checkercapped at 20A published fact-check (ClaimReview) rates a matching claim false or misleading.

Measured signals

IODA internet outages (Georgia Tech), USGS earthquakes, GDACS disasters, NWS severe weather, cloud status pages (AWS, Google Cloud, Cloudflare, GitHub and others), CISA actively exploited vulnerabilities. A story links to a signal only when place, time and topic line up.

The analyst

Claude writes a read on each top story (solid → doubtful) plus the “Connecting the dots” brief. It is given only the reports in this feed. It never sets a score or raises an alert level; a “doubtful” read marks the story disputed and caps its level at L2. More on AI use.