Why Abolishing Leap Seconds Is a Quiet Infrastructure Bill
A vote in Versailles could end 50 years of leap seconds — and the decision ripples through finance, satellite networks, and every system that ticks. Here is who wins, who pays, and what happens next.
The Versailles Meeting That Nobody Will Notice
Delegates from the International Bureau of Weights and Measures are meeting in Versailles from October 12 to 15 to consider a proposal that sounds almost too simple to matter. Abolish the leap second.
It has been 50 years since the first leap second was added in 1972. Twenty-seven have been inserted since, the most recent landing at the end of 2016. They exist to bridge a gap between the smooth tick of atomic clocks and the fickle rotation of Earth. Remove them, and UTC stops chasing the planet. A leap hour — or something like it — could replace the practice when the drift finally accumulates enough to demand attention.
The vote will pass. The practical pressure is overwhelming. But the aftermath matters more than most realise. This is not a technical footnote. It is a structural adjustment to the time layer beneath the global economy.
How Time Became a Problem
Atomic clocks based on caesium atoms define UTC. They do not wobble. Earth does. Tides, ocean currents, atmospheric shifts, melting glaciers, volcanic activity, and the gravitational tug of the Moon all nudge the planet’s spin rate. Since roughly 2020, Earth has been rotating faster than its recent average, which complicates the math even further.
Scientists track rotation using radio telescopes pointed at distant black holes as fixed reference points. When the drift between atomic time and solar time grows large enough, a leap second is inserted. It keeps clocks honest. It also breaks software.
Most computer systems assume time moves forward and that every minute contains exactly 60 seconds. A leap second disrupts both expectations. In 2012, it took down parts of LinkedIn and Reddit. In 2017, Cloudflare suffered a major outage after a leap second triggered a bug in its NTP stack. These were not isolated incidents. They were symptoms of a world built on a timing standard that occasionally stutters.
The Negative Leap Second Scare
Earth’s acceleration has added urgency. If the planet continues spinning faster, the difference between UTC and solar time shrinks. That opens the door to a negative leap second — a second removed rather than added.
No modern system has been stress-tested for that scenario. Engineers have spent decades patching around positive leap seconds. The negative variant is uncharted. Removing a second means timestamps collapse. Scheduling logic, database transactions, and logging systems all depend on monotonic progression. A regression at the scale of one second can cascade through any architecture that does not account for it.
The 2022 decision by the BIPM’s governing body to target abolition by 2035 was a response to this kind of risk. The latest discussions are shaped by the same fear: the more complicated the solution, the more likely it is to fail somewhere.
The Leap Hour Proposal
The leading alternative is straightforward in concept and vast in implication. Let UTC drift. Allow atomic time and solar time to diverge until the gap reaches one hour, then adjust with a leap hour.
At the current rate of Earth’s rotational deceleration, that would take well over a millennium. The proposal is essentially a long-term deferral. It removes the biannual or annual headache and replaces it with something so distant that no engineer alive will witness it.
Supporters argue that precise timekeeping can continue independently of solar alignment. Scientists still measure rotation. Astronomers still need it. The change only affects how the public standard is labelled, not how the planet is observed. Critics raise a philosophical objection: detaching official time from day and night severs a cultural tether. But the argument against leap seconds is pragmatic, not poetic, and pragmatism is winning.
The Hidden Bill: Who Inherits the Change
Abolishing leap seconds is effectively a quiet infrastructure bill. It shifts costs, reallocates risk, and rewrites assumptions across sectors that have never had to think about time as a system variable.
Financial markets run on timestamps measured in nanoseconds. Traders, clearing houses, and exchanges already fight over who wins when a trade lands in tick one versus tick two. Leap seconds are fringe events in that war, but they are also unpredictable ones. A system that crashes during a leap second can generate mismatched records, failed reconciliations, and regulatory exposure. Eliminating the event removes a source of operational risk. That benefit accrues to every institution that holds a position at midnight.
Satellite networks depend on precise timing for navigation, coordination, and ground-station handoffs. GPS, Galileo, and similar systems carry their own internal time scales and correct for relativistic effects continuously. Leap seconds are injected at the application layer, not the signal layer. Removing them simplifies the integration work that ground operators perform, reduces the chance of a misaligned injection, and tightens the margin for error in systems that cannot tolerate surprise.
Cloud providers and CDN operators absorb the blast radius of every leap-second mishap. Cloudflare and Reddit proved that much publicly. The cost of a crash is not just downtime. It is lost traffic, eroded trust, and engineering hours diverted from product work to emergency patching. Abolition converts those sporadic crises into routine maintenance.
Data centres and database administrators face the same pattern. Logs, audits, and replication chains all rely on monotonic time. A negative leap second could cause timestamps to overlap or regress, triggering alert fatigue, corrupting sequential identifiers, and forcing emergency rollbacks. The leap-hour framework removes that class of failure entirely.
Scientific instrumentation is the sector that loses the most in abstract terms. Paleoclimatologists, geophysicists, and astronomers who correlate measurements with solar time will need new conventions for labelling data. The measurements themselves do not change. The labels do. Over decades, that distinction matters less, but during the transition it creates friction.
What Happens Next
If the vote passes, the clock starts on a phased transition. The BIPM’s earlier target of 2035 remains the working deadline. Systems that currently apply leap seconds manually or through distributed update mechanisms will need to adjust their code paths. Engineers will write new parsers, update libraries, and retire legacy handling logic. It is unglamorous work. It is also essential work, and it will happen whether or not this vote succeeds.
The more interesting question is what the decision signals. Timekeeping has always been a negotiation between physical reality and human convention. The leap second was a compromise that kept both sides honest. Abolition declares that the compromise is no longer worth the cost. That shift will reverberate through every industry that builds on UTC — which is to say, every industry that builds on modern infrastructure.
Earth will keep rotating. The Moon will keep pulling. Atomic clocks will keep ticking. The only thing that changes is what we call the second.