"Deadly shark attacks doubled." Did the oceans get twice as dangerous?
The headline was true: worldwide deaths from unprovoked shark bites went from 5 to 10. But when the starting number is tiny, a few more or a few less makes an enormous percentage, up or down.
Deaths from unprovoked shark bites, worldwide
Four years, three dramatic headlines
Deaths from unprovoked shark bites, worldwide, with the change from the year before
A big percentage built on a small number is mostly noise. Going from 5 deaths to 10 is a 100% jump, but it's 5 more deaths in a whole year, across the entire world, and the very next year the number fell by more than half. Every year here sits close to the usual 8. Nothing about the oceans doubled or collapsed; a few unlucky or lucky events just moved a tiny number around.
The same thing happens with rare events anywhere: a small town's murders, a school's number of perfect scores, plane crashes in a single year. When a percentage sounds dramatic, ask for the raw numbers behind it, and whether they've swung like this before.
What "doubled" or "cut in half" headline might have been a change of just a handful?
Bites vs. deaths: bigger numbers swing less
| Year | Unprovoked bites | Change | Deaths | Change |
|---|---|---|---|---|
| 2022 | 57 | – | 5 | – |
| 2023 | 69 | +21% | 10 | +100% |
| 2024 | 47 | −32% | 4 | −60% |
| 2025 | 65 | +38% | 9 | +125% |
Bites, which run in the dozens, swing much less in percentage terms than deaths, which run in single digits. The shark researchers who track these numbers described 2023's bites as "within the range of the normal number." The more events behind a number, the less its percentage bounces around from luck alone. How much does an 18% higher risk really mean? looks at a related trap.
Related exhibits
Sources: Florida Museum of Natural History, International Shark Attack File yearly worldwide summaries, 2022–2025 (unprovoked bites and fatalities; 2020–2024 five-year averages); CBS News, "Deadly shark attacks doubled in 2023," February 2024.