Every 30 minutes we cluster the news across 48 outlets and score how much they actually disagree. This is what's happening across 9,278 tracked events.
The single most divergent event from the last week, shown through the outlets that frame it most opposingly. The feed beside it is the latest framings as they land.
Average divergence per category across every tracked event. War and politics tend to top the list; health and economy usually sit lowest.
Divergence and polarization rarely sit still. These two charts show how contested coverage has been, and whether outlets are framing stories more with their own side than across it.
Average divergence score per week, overlaid with the share of articles carrying negative sentiment.
Higher = outlets agree more within their political side than across it. Negative = cross-aisle outlets actually frame stories more similarly.
Bar length is event count; color is how contested that category tends to be. Politics dominates the volume; war and politics dominate the contention.
Outlets do not just pick different stories - they pick different tones. The heatmap breaks down what share of each outlet's coverage lands in each framing bucket, ordered left to right by political alignment.
Outlets ordered left to right by political alignment. Darker cell = higher share of articles with that framing.
For each outlet, how often its framing matches same-side outlets versus opposite-side outlets when they all cover the same event. The bigger the gap, the more that outlet clusters with its side.
Two views of the same thing: outlets that regularly diverge from the consensus, and outlet pairs that quietly agree despite sitting on opposite sides.
How often each outlet's framing diverges from the consensus across events they cover. Requires 10 or more events.
Outlet pairs that frame the same stories similarly despite sitting on opposite sides of the political spectrum.
Where outlets told you wildly different stories about the same thing. Higher score = more contested framing.
Low-divergence stories with broad outlet coverage are useful calibration: sometimes the important signal is that the usual framing fight did not happen.
Stories from the last 7 days covered only by left-leaning or only by right-leaning outlets. The gaps say as much as the overlap. 30-day blind spot index →
The full feed is free, open, and updated every 30 minutes.