Our Methodology

Each news desk is placed on a pro-govt ↔ neutral ↔ critical spectrum from its editorial framing history, not from a single article. A desk is pro-govt when its framing consistently centres the government's narrative and downplays accountability; critical when it consistently foregrounds accountability, opposition, and affected voices; neutral when its coverage is descriptive wire-style reporting with minimal editorial framing.

The label describes framing tendency, not factual reliability — a critical desk and a pro-govt desk can both report accurately. We include desks across the whole spectrum (from OpIndia and Swarajya on the right to The Wire and Scroll on the critical flank) precisely so no single worldview dominates the picture.

A fact-check desk is included only if it is an independent, IFCN-aligned or equivalently transparent checker that publishes a clear rating, a public methodology, and a corrections policy. Our network spans English and regional-language desks — Alt News, BOOM (and BOOM Hindi), Vishvas, Factly, Fact Crescendo, The Quint WebQoof, India Today Fact Check, NewsMeter, The Logical Indian, Newschecker, AFP/PTI Fact Check, and DFRLab.

When a claim can't reach a dedicated checker API, we widen recall with a Google News search across these same desks, so regional and Hindi claims are caught too.

Claims are matched to fact-checks deterministically first — on entity, numeric, and lexical agreement — so verification keeps working even when AI quota runs out. A language model is used only to break genuine ties, never to gate a clear match. The consensus verdict reflects the balance of independent checker ratings: a claim most checkers rate false is labelled false, and conflicting ratings are shown as disputed rather than forced into a single call.

The radar ranks claims by heat = falsity × velocity: how false checkers rate it, multiplied by how fast it is spreading across desks. Because we don't have WhatsApp or X firehoses, virality is inferred from cross-desk mention velocity — a good-enough, zero-cost signal. Every blip links to the exact desks amplifying it, so you can see who is pushing a false claim.