The Last Leap Second History
The second was the sky.
Then the second was an atom.
For most of history, a second was simply a slice of the day. Then we built clocks better than the planet, and spent fifty years patching the gap with single seconds. Here is how we got there — and what those seconds cost.
A short history of the second
Loading the timeline…
The cost of one second
When a second broke the internet
Every leap second was a tiny, announced step in time — and a surprisingly reliable way to break computers that assumed time only ever moves forward smoothly.
Loading the incidents…
Leap smearing
Rather than let a clock jump or repeat a second, Google and later Amazon spread the extra second across a window of hours so every tick is a fraction longer — a "leap smear". It works beautifully, and it means two servers that both claim to be exactly on time can disagree by nearly a second. That proliferation of subtle timescales is one more reason the world chose to end the practice altogether.
“Every leap second is a major source of pain for people who manage hardware infrastructures.” Engineers at Meta, quoted by Al Jazeera · 2022
Why ±0.9 seconds?
The IERS aimed to keep |UT1 − UTC| under 0.9 s. Once it approached 0.6 s, a leap second was announced six months ahead and inserted at the end of June or December, briefly making a day 86,401 seconds long. That 0.9 s is the tight little band this whole site's game asks you to defend.