We measure how the world's news covered people being harmed — and the help and hope within it.
235 of its 513 stories (46%) were about people being harmed, usually 36%. Dhaka accounted for 127 of Bangladesh's 235 stories about harm, carried by 86 outlets. 77 of Dhaka's 127 stories were about protests or strikes. Its coverage was about twice as angry as the world's.
494 of its 1,453 stories (34%) were about people being harmed, usually 26%. Bondi Beach accounted for 295 of Australia's 494 stories about harm, carried by 476 outlets. Flu was named in 10 of its stories about harm, about three times usual. Its coverage was about one and a half times as fearful as the world's. Anger in its coverage rose to about one and a half times its own usual.
29 of its 155 stories (19%) were about people being harmed, usually 9%. Reports of people killed ran at about three times the usual rate (15 stories).
14 of its 47 stories (30%) were about people being harmed, usually 16%. Earthquake was named in 11 of its stories about harm, about three times usual.
235 of its 643 stories (37%) were about people being harmed, usually 21%. Moscow accounted for 165 of Russia's 235 stories about harm, carried by 308 outlets. Its coverage was about three times as grief-stricken as the world's.
137 of its 366 stories (37%) were about people being harmed, usually 26%. Reports of people injured ran at about 5 times the usual rate (46 stories). Its coverage was about one and a half times as grief-stricken as the world's.
84 of its 290 stories (29%) were about people being harmed, usually 23%. Reports of abductions ran at about 8 times the usual rate (8 stories). Its coverage was about one and a half times as angry as the world's.
34 of its 62 stories (55%) were about people being harmed, usually 43%. Phnom Penh accounted for 22 of Cambodia's 34 stories about harm, carried by 42 outlets.
16 of its 59 stories (27%) were about people being harmed, usually 17%. Its coverage was about one and a half times as grief-stricken as the world's.
Reports of people injured ran at about three times the usual rate (17 stories). Its coverage was about one and a half times as angry as the world's.
Also named far more than usual: accidents in Japan (22 of its stories, more than ten times its usual); cyclone in US (21 of its stories, about 4 times its usual).
Stories about people being harmed, in every 100 stories, the 30 days to 22 Dec. The big dot is this day; the shaded band is a normal day (the 30 days before it). Click a day to open it. Based on 12,286 stories about harm among 66,789 counted on 22 Dec.
The news turned to Bangladesh for the fourth day running. 18 in every 100 of the day's 66,789 news stories were about people being harmed, as usual. In Papua New Guinea's news, mentions of the earthquake ran about three times their usual. Bangladesh stood out: 46% of its news was about people being harmed, usually 36%. On 22 December 2024, a year earlier, the share was 18 in every 100, the same as today.
Written by code from the day's numbers.
Countries with at least 50 stories that day.
Stories, not people.
The day's biggest events: attacks, accidents and disasters, people displaced, protests, outbreaks. Bigger dot: more stories. Ranked by the share of the day's outlets that reported it and how clearly it was about one thing, the same way for every place and year; the news data does not say who attacked whom, so both sides of a war appear where the news put them. Click a dot for its stories.
Of the stories about people being harmed, what part of life each was about. The mark is usual.
Stories that report it, not people: we count what reporters wrote, not what happened.
The words in the day's stories were about as usual. Filled dot: today. Ring: usual.
From the words reporters used across the whole day, never one story. Fear is the weakest of the four measures. How?
Unlock a year of trends, free. Follow up to 5 countries, with a bell when their news changes, and 👍 vote for Pro — ten years of the archive, any period, daily alerts.
Create a free accountSign in · a username, an email and a password.