Daybrief is a daily world briefing built from two open datasets: what the world's press covered (GDELT) and what people looked up (Wikipedia pageviews). GDELT publishes each day's complete event file the following morning (07:00 UTC), so the briefing always shows the most recent finished day. The site is rebuilt twice a day.
How stories are ranked. A story's rank is the number of distinct news outlets GDELT recorded covering it that day — not clicks, not an editor, not a personalised feed. Headlines are shown as published, linked to the original publisher. We never show article text and never rewrite headlines.
Filters we apply.
Coverage anomalies. For each country we compare yesterday's story volume with the median for the same weekday over the previous four weeks — press volume drops sharply at weekends, so comparing a Sunday with a Wednesday would mislead. A spike is only shown with the story driving it. If that story actually belongs to another country and was merely syndicated with this country's tag, the spike is treated as an artifact and hidden. (We learned this the hard way: one widely syndicated Ukraine story once made Norway look 30× louder than usual.)
What the world suddenly looked up. Wikipedia articles that entered a language edition's most-read list after at least a week off it (NEW), or that reached 1.5× their previous peak (rising). Evergreen pages that are always popular are left out. Short descriptions come from Wikidata (CC0).
Known limits. GDELT's country tags and outlet counts are machine-generated and sometimes wrong. Duplicate detection misses synonyms. English sources dominate. If you spot something misleading, please tell us via the Contact page.
Weekly recaps are generated automatically from the same data.