The Leap Second Dies — And No One Noticed in English Media
In a decision made with barely a ripple in English-language coverage, the CGPM voted to abolish the leap second. What changed, who wins, and what breaks when the clock shifts.
The vote happened in Versailles. No press release followed.
On November 15, at the 28th meeting of the General Conference on Weights and Measures (CGPM) in Versailles, France, delegates voted to abolish the leap second. The practice dates back to 1972. It ends now, with implementation starting May 20, 2027 — eight years earlier than originally planned.
There was no splash in English-language media. No front-page headline. But this is one of the most consequential technical decisions made in decades, and almost nobody outside Korea, Japan, and a handful of metrology circles wrote about it with any detail.
Here is what happened, and why it matters far beyond timekeeping.
Why the leap second existed
The leap second exists because two clocks disagree. World Standard Time (UTC), the backbone of global computing, is derived from cesium atomic oscillations — unforgivingly stable. Solar time, which governs sunrise and sunset, is governed by Earth’s rotation — which varies.
Since 1972, the International Earth Rotation and Reference Systems Service (IERS) has inserted or subtracted a second whenever the gap between UTC and Earth-rotation time (UT1) approached 0.9 seconds. There have been 27 positive leap seconds — seconds added, never removed.
Each one arrived on June 30 or December 31, at 23:59:59, followed by a second that should not exist: 23:59:60.
That extra second has repeatedly broken things.
What broke before
In July 2012, a leap second insertion caused the reservation system for Qantas Airlines to fail, grounding over 400 flights and stranding passengers in airports across Australia and New Zealand. Reddit, Slack, and Mozilla reported disruptions. Firms that had invested in mitigation still found edge cases in their code.
The problem is not theoretical. It is structural. Many systems assume a minute contains exactly 60 seconds. When it does not, log timestamps collide, transaction ordering reverses, and some processes hang waiting for a clock that refuses to advance.
A negative leap second — removing one second rather than adding it — was technically cleaner on paper but operationally untested. For the first time in 54 years, Earth’s rotation has been accelerating since 2020, making a subtraction likely. On July 5, 2024, the shortest day on record was logged: 0.00166 seconds shorter than the 24-hour standard.
The fear was real. No engineer had built, tested, or deployed code for a second deletion. The risk of cascade failure across financial exchanges, satellite navigation, and telecommunications networks was significant.
What the CGPM actually decided
The resolution did two things simultaneously. It eliminated the leap second entirely and raised the allowable drift between UTC and UT1 to 1 hour — up from the current 0.9-second threshold.
One hour of drift sounds dramatic. It will not arrive quickly. At current rates of Earth-rotation change, it would take centuries for the gap to accumulate to that magnitude. The decision is precautionary, not emergency-driven.
The implementation date — May 20, 2027 — was accelerated from the previously discussed 2035 horizon precisely because of that acceleration in Earth’s spin. Waiting for a negative leap second under the old rules carried unacceptable risk.
Park Yeon-gyu, deputy director of the Korea Research Institute of Standards and Science (KRISS), was quoted in the Chosunbiz report acknowledging the uncertainty: “Predicting what will happen when we try to subtract a second is impossible. We simply did not have the nerve to proceed.”
The cost calculus
The CGPM’s working group estimated that preparing global IT systems for a negative leap second could rival the cost of the Year 2000 (Y2K) remediation — widely cited at over $300 billion worldwide. That figure included banking, aviation, utilities, and government infrastructure. The leap second preparation would have required similar coordination across every sector that relies on timestamped transactions.
The decision to abolish the leap second is, at bottom, an economic one. It is cheaper to let UTC drift from solar time by up to an hour over centuries than to force every server on Earth to handle a second that cannot be predicted with certainty.
Who wins and who loses
Financial markets are the clearest winners. High-frequency trading firms, clearinghouses, and exchanges rely on microsecond-precision ordering. The leap second introduces an anomalous interval that can desynchronize servers, reorder trades, and create audit failures. Its removal eliminates that risk permanently.
Satellite navigation systems — GPS, Galileo, BeiDou, GLONASS — also benefit. These systems encode leap second adjustments in their broadcast signals. Engineers have spent decades building fallback logic. Removing the adjustment simplifies their architecture.
Telecommunications and cloud providers gain as well. NTP (Network Time Protocol) servers, data-center synchronization, and CDN origin validation all become marginally more stable.
The losers are astronomers, maritime navigators using celestial references, and anyone whose work depends on the direct alignment of UTC with solar time. For those groups, the gap between clock time and sun time will gradually widen. It will remain negligible for generations, but the conceptual tether is cut.
What happens next
The transition is straightforward in principle and complex in practice. Starting May 20, 2027, no further leap seconds will be inserted or removed. UTC will continue to run on atomic time alone. UT1 will drift freely until the 1-hour threshold is reached, at which point a new adjustment mechanism — still undefined — will intervene.
Organizations with critical infrastructure should treat this as a deadline, not a deadline to ignore. Systems that were patched for the 2012 and 2016 leap seconds remain vulnerable to edge cases around the new regime, particularly in embedded firmware and legacy operating systems that assume a fixed 86,400-second day.
The decision itself — made in Versailles, reported primarily in Korean media — reveals something about how technical governance flows in the digital age. The most impactful decisions are often silent. They happen in rooms where metrologists and standards bodies deliberate, far from the headline cycle, yet they rewrite the operating assumptions of every networked system on Earth.
The leap second served its purpose for half a century. It kept atomic time and solar time within a reasonable margin while computing infrastructure was crude and few systems depended on nanosecond precision. That era is over. The next one runs on a clock that no longer answers to the Earth.