Earth Is Changing Speed: Why Milliseconds Matter for Satellites and the Internet
On Sunday, July 26, 2026, astronomers and metrologists around the world quietly marked what mid month predictions from the International Earth Rotation and Reference Systems Service, known as IERS, suggested would be the shortest day of the year. Earth was expected to complete a full rotation about 0.65 milliseconds faster than the standard 86,400 second day. To a person going about an ordinary Sunday, that difference is unmeasurable and irrelevant. To the network of atomic clocks, satellites and synchronized computer systems that quietly run global communication, navigation and finance, it is part of a pattern that has kept scientists watching the sky and the calendar with unusual intensity for the better part of six years.
This is not a new phenomenon. Earth's rotation has never been perfectly steady. What has changed is the direction and the stakes. For most of human history, the planet's spin has been slowing down, gradually, over centuries, due to the tidal drag of the moon. But since around 2020, Earth has instead been speeding up, setting one record after another for the shortest days ever measured by atomic clocks, and in doing so it has forced a question nobody in the timekeeping world ever expected to face this soon. What happens to the world's clocks the first time humanity has to subtract a second from them, rather than add one.
The Shortest Days in Recorded History
The story begins in 2020, when Earth quietly recorded the twenty eight shortest days of the entire atomic clock era, a run of speed that took most geophysicists by surprise. New records followed in 2022, and then again through 2024 and 2025, culminating in what is currently the fastest rotation ever measured, on July 5, 2024, when the planet completed its spin about 1.66 milliseconds ahead of schedule. A year later, on July 10, 2025, Earth turned in 1.37 milliseconds under the mark, another entry in an increasingly strange ledger.
Then, just as suddenly, the trend appears to be easing. The July 2026 predictions from IERS point to a shortest day of only about 0.65 milliseconds under standard length, roughly half of what was recorded the previous July. Analysts at TimeAndDate.com and reporting from Forbes both describe this as a sign that Earth's recent burst of speed may have peaked, with days gradually lengthening again through the remainder of 2026. Nobody is certain this trend will hold. Earth's rotation is influenced by a tangle of forces operating on wildly different timescales, and a slowdown this year does not rule out another burst of acceleration next year or the year after.
Earth's Rotation in Numbers
- July 5, 2024: fastest rotation ever recorded by atomic clocks, about 1.66 milliseconds under 24 hours
- July 10, 2025: second fastest day on record, about 1.37 milliseconds under 24 hours
- July 26, 2026: predicted shortest day of the year, about 0.65 milliseconds under 24 hours
- Twenty seven leap seconds added to UTC since 1972, none since 2016
- GPS time currently runs a fixed 18 seconds ahead of UTC and does not use leap seconds at all
Why Is Earth Speeding Up
Several forces act on Earth's spin at once, and scientists are still untangling how much each one contributes. Over the short term, the moon's monthly orbit plays a direct role. Earth spins slightly slower when the moon sits above the equator and slightly faster when the moon moves north or south of it, a rhythm physicist Judah Levine of the National Institute of Standards and Technology has described as producing short bursts of correlated speed that fade out over longer intervals.
The deeper explanation, according to geophysicist Duncan Agnew of the Scripps Institution of Oceanography, lies far below the surface. Earth's liquid outer core appears to be decelerating relative to the solid mantle above it, and by the same principle that makes a spinning ice skater speed up when drawing in their arms, a slower core allows the solid outer layers of the planet to spin faster. Agnew's research, published in Nature in 2024, calculated that this core driven acceleration has actually been building for roughly fifty years, but its effect on the surface was masked for decades by an opposing force, the redistribution of mass caused by melting polar ice.
From Leap Seconds to a Possible Leap Hour
Civil time, the Coordinated Universal Time or UTC that phones, computers and broadcast networks rely on, is based on atomic clocks that never waver. Earth's actual rotation, called UT1 by astronomers, wanders slightly. To keep the two aligned within one second of each other, the system of leap seconds was introduced in 1972 under standards set with the National Institute of Standards and Technology and the International Bureau of Weights and Measures, or BIPM. Whenever UTC and UT1 threaten to drift more than a second apart, IERS in Paris issues a bulletin instructing the world to add, or in theory subtract, a single second at the end of June or December.
In practice, only additions have ever been needed. Twenty seven leap seconds have been inserted since 1972, nine of them in the 1970s alone, but the pace slowed dramatically as Earth's rotation behaved less predictably, and no leap second has been added since 2016. On July 6, 2026, IERS issued Bulletin C number 72, confirming that no leap second would be introduced at the end of December 2026 either, a detail that circulated quickly among engineers and made its way onto technology forums such as Hacker News within hours.
The bigger decision now sits with the General Conference on Weights and Measures, or CGPM, the body that governs global measurement standards and meets roughly every four years. In 2022, its member states adopted Resolution 4, agreeing in principle to widen the allowable gap between UTC and UT1 to something larger than one second, with a deadline for implementation no later than 2035. According to reporting from Scientific American, the CGPM is expected to vote in October 2026 on a proposal that would go further still, replacing the leap second entirely with a far rarer leap hour, a change so infrequent that Patrizia Tavella, director of the Time Department at BIPM, and other experts believe it might not be needed for centuries. Supporters, including delegations from the United States, Canada and France, argue this would finally remove the operational risk that leap seconds pose to modern computing. Russia has pushed back, citing the technical burden the change would place on its GLONASS satellite navigation system, which unlike GPS still incorporates leap seconds directly into its signal.
Why Milliseconds Matter for GPS and Satellites
It is tempting to dismiss a millisecond, or even a full second, as too small to matter. Satellite navigation systems suggest otherwise. GPS satellites determine a receiver's position by measuring the time it takes radio signals to travel from multiple satellites to the ground, and because those signals move at the speed of light, an error of just a few billionths of a second can translate into a positioning error of several meters. GPS deliberately sidesteps the leap second problem altogether. Rather than following UTC directly, GPS time runs on its own continuous atomic scale and simply broadcasts a fixed offset, currently eighteen seconds, that receivers use to convert GPS time into civil UTC. That elegant workaround is precisely why a mishandled leap second, positive or negative, worries engineers more than it worries astronomers. A GPS synchronized system and a UTC synchronized system that are supposed to agree on the time could suddenly disagree by a full second if one side implements a change the other has not yet accounted for.
Telecommunication networks, power grids, and high frequency financial trading systems share the same vulnerability. Many of these systems depend on the Network Time Protocol to distribute UTC across the internet, and all of them assume time moves forward in a single, unbroken direction. When the last leap second was added in 2016, engineers at major technology companies still had to patch systems that briefly stumbled over the extra second, and the 2012 leap second event caused documented disruptions across Linux servers and other platforms worldwide. Engineers at Meta have publicly warned that a negative leap second, subtracting a second that has never been subtracted before, carries a materially higher risk, precisely because no software in existence has ever been tested against a clock that briefly runs backward.
The International Response
The response has unfolded across several overlapping institutions, each with a different mandate and a different level of urgency. IERS in Paris tracks Earth's rotation in real time and issues the bulletins that trigger, or in 2026, repeatedly decline to trigger, a leap second. BIPM, based just outside Paris, is the custodian of the global time standard itself and has been pushing hardest for a swift resolution, with Tavella describing the situation as sufficiently urgent to justify moving to a leap hour as soon as possible. The International Telecommunication Union, or ITU, ceded formal authority over leap second policy to the CGPM in 2015 but retains control over how UTC signals are actually disseminated across broadcast and telecom infrastructure worldwide, giving it an informal veto that former BIPM Time Department director Felicitas Arias has said still makes some officials nervous.
National positions have diverged along practical, rather than ideological, lines. The United States, Canada and France have generally supported retiring the leap second on an accelerated timeline, reflecting the exposure of their extensive satellite, cloud computing and financial infrastructure to timing errors. Russia has been the most visible holdout, not out of skepticism about the science but because GLONASS, its answer to GPS, was built around a design that does incorporate leap seconds, meaning a rule change could force Russia to modify satellites and ground stations at real cost. That disagreement is part of why the 2022 resolution set 2035 as a ceiling rather than a fixed date, and why the vote scheduled for October 2026 in Paris will be closely watched well beyond the small community of professional timekeepers who usually pay attention to these meetings.
What Happens When the Clocks Diverge
For most people, none of this will ever be noticed directly. Felicitas Arias has pointed out that the seasonal one hour shift between standard time and daylight saving time already exceeds any leap second adjustment by a wide margin, and the public absorbs that twice yearly change without incident. The real exposure sits deeper in the technology stack, in the scheduling libraries, database timestamps, satellite ephemeris calculations and cryptographic systems that quietly assume UTC never runs backward and never contains a sixty first second in any minute.
Precision timing errors have already had visible, if minor, consequences. Reporting from National Geographic notes that small shifts in Earth's rotation altered the calculated path of a recent solar eclipse closely enough to exclude certain cities in Texas from the band of totality that earlier, less precise calculations had predicted. If a fraction of a millisecond can move an eclipse path measurably, the same imprecision, multiplied across millions of GPS guided vehicles, aircraft, ships and financial transactions, becomes a genuine engineering concern rather than an academic curiosity.
Looking Ahead
Earth's rotation will keep changing regardless of what any conference decides. The moon will keep tugging at the tides, the core will keep exchanging angular momentum with the mantle, and the ice sheets will keep melting at a pace shaped by decisions made far from any physics laboratory. What the October 2026 vote in Paris can control is not the planet's behavior but humanity's tolerance for its unpredictability, whether the world continues patching its clocks one second at a time or accepts a looser, more forgiving relationship between atomic time and astronomical time that will not need revisiting for a very long while.
There is something quietly humbling in the fact that the most advanced timekeeping infrastructure ever built, the atomic clocks accurate to one second in tens of millions of years, still has to negotiate with a spinning rock that refuses to keep a perfectly steady beat. The milliseconds at stake are too small for any person to feel. The systems built on top of them, from a phone's blue dot on a map to the split second execution of a global trade, were never designed to feel anything at all. They were designed to trust the clock completely. The question the world's timekeepers are racing to answer before 2035, and perhaps sooner, is what happens the day that trust is tested for the very first time.
0 Comments