Say Goodbye to the Leap Second: Why Tech Is Finally Fixing Broken Time

Every once in a while, humanity discovers that the systems it built to organize reality do not quite align with reality itself. For more than fifty years, our global timekeeping infrastructure has relied on a strange, awkward compromise known as the leap second. It was designed to keep atomic clocks tied to the irregular rotation of our planet. However, in our interconnected digital world, this tiny sliver of time has caused immense frustration, crashed major websites, and kept network engineers awake at night.

Now, the verdict is finally in. Scientists and government representatives from around the world, gathered at the General Conference on Weights and Measures (CGPM) outside Paris, have voted to scrap the leap second. By 2035, this irregular adjustment will be abandoned or paused for at least a century. This decision marks a historic turning point in how humanity measures the passage of a day. It separates the cold, perfect ticking of atomic clocks from the wobbling spin of planet Earth.

Here at TechRook, we follow the intersection of pure science and practical technology. In this in-depth guide, we will break down what the leap second is, why software engineers and cloud giants fought so hard to kill it, and what our digital future looks like without it.

What Is the Leap Second and Why Was It Created?

To understand why the leap second is on its way out, we first need to understand why it was introduced in the first place. For thousands of years, time was measured by the sky. A day was simply the time it took the Earth to rotate once on its axis, bringing the sun from its highest point at noon back to its highest point the next day. We divided that day into 24 hours, each hour into 60 minutes, and each minute into 60 seconds.

This system, known historically as solar time and refined in modern terms as Universal Time (UT1), worked well for an agrarian world. However, the middle of the twentieth century brought a revolution in physics: the invention of atomic clocks. In 1967, scientists officially redefined the second. Instead of basing it on a fraction of an astronomical day, they defined it based on the steady quantum behavior of cesium-133 atoms. Under this definition, one second is exactly 9,192,631,770 cycles of radiation produced by the transition between two energy levels of a cesium atom. This standard gave birth to International Atomic Time (TAI).

Atomic time is ridiculously precise. An advanced cesium or strontium clock will not lose or gain a single second over millions of years. But this extreme precision exposed an inconvenient truth: planet Earth is a terrible clock.

Our planet does not spin at a perfectly constant speed. Its rotation is constantly influenced by several complex natural forces:

• Tidal Friction: The gravitational pull of the moon and sun creates ocean tides, which act like a gentle brake on the planet, gradually lengthening our days by a couple of milliseconds each century.
• Core Dynamics: Sloshing movements of molten iron deep inside the Earth’s core transfer angular momentum back and forth between the core and the mantle.
• Atmospheric Winds: Seasonal jet streams and major weather patterns push against mountain ranges, subtly shifting the planet’s rotational velocity.
• Glacial Rebound: The melting of massive ice sheets from the last ice age causes continental landmasses to slowly rise, altering the planet’s shape and distribution of mass.

Because the Earth slows down and speeds up in unpredictable ways, astronomical time (UT1) and atomic time (TAI) inevitably drift apart. By 1972, metrologists created Coordinated Universal Time (UTC) to bridge the gap. They decreed that UTC would tick at the exact, steady rate of atomic clocks, but whenever the difference between UTC and astronomical time approached 0.9 seconds, an extra second would be injected at the end of June 30 or December 31. That extra tick was the leap second.

The Problem with 23:59:60

On paper, adding a leap second seemed like an elegant solution. On a digital clock face, the final minute of the designated day would display an unusual pattern. Instead of rolling over from 23:59:59 directly to 00:00:00, it would count an extra value: 23:59:60.

Human beings barely notice an extra second. If you are watching television, cooking dinner, or sleeping, the leap second passes without leaving a trace. For computer systems, however, that single anomalous digit is a waking nightmare.

Computers do not think like humans. Operating systems and computer networks are built on a strict, monotonic assumption: time always moves forward at a predictable, regular rate, and each minute contains exactly sixty seconds numbered from 0 to 59. When an operating system receives a time signal containing 23:59:60, software routines often have no idea how to interpret it. The computer may see time standing still, it may see time rolling backward by one second, or it may crash outright because the programmers never wrote code to handle a 61st second.

Modern computing is also built heavily on POSIX time, which is the standard across Unix, Linux, macOS, and countless embedded devices. POSIX defines time as the total number of seconds elapsed since January 1, 1970 (the Unix Epoch). By definition, the POSIX standard does not account for leap seconds. When a leap second occurs, POSIX systems repeat the second 23:59:59 twice. Repeating a timestamp can completely break distributed databases, which rely on unique, forward-moving timestamps to order financial transactions, confirm user logins, and verify database writes.

A History of High-Profile Leap Second Disasters

Between 1972 and 2016, international timekeepers added 27 leap seconds to UTC. In the 1970s and 1980s, when computers were isolated mainframes, this practice rarely caused global chaos. But as the internet expanded and global services became tightly coupled across distributed cloud infrastructure, leap seconds began breaking mission-critical systems.

Here are just a few memorable incidents caused by the leap second over the last decade and a half:

1. The Great Leap Crash of 2012

On June 30, 2012, a leap second was added to UTC. Within minutes, multiple prominent websites and services tumbled offline. Reddit suffered a major outage when its Linux servers experienced sudden CPU lockups caused by the kernel timer mishandling the repeated second. Mozilla, Gawker, Yelp, LinkedIn, and Foursquare reported significant disruptions. In Australia, the Qantas airline booking system crashed completely, forcing staff to check in passengers by hand and delaying more than 400 flights worldwide.

2. The Cloudflare Glitch of 2017

On New Year’s Day in 2017, another leap second went live. Cloudflare, one of the internet’s largest content delivery networks and security providers, experienced an internal software bug. The DNS servers they operated compared two different clock sources: one that accounted for the leap second and one that did not. Because the calculation resulted in a negative number, the software crashed, dropping roughly 0.2 percent of all DNS queries routed through their edge network and affecting roughly 1 percent of their overall web traffic.

3. Financial Market Havoc

High-frequency trading algorithms execute millions of transactions per second. In the financial sector, a difference of a few microseconds can mean millions of dollars won or lost. Because a leap second typically arrives at midnight UTC (which is 8:00 PM in New York and 9:00 AM in Tokyo), Asian markets and after-hours currency trading desks are often actively trading when the leap second hits. To prevent glitches, several global stock exchanges have had to suspend operations or shut down their data feeds during leap second intervals.

The Patchwork Era: How Tech Giants Tried to Solve It

Because international timekeepers showed no urgency in removing the leap second during the 2010s, Silicon Valley engineers decided to take matters into their own hands. The result, unfortunately, was a chaotic patchwork of custom solutions that introduced new risks.

The most famous workaround was created by Google and is known as “leap smearing.” Instead of adding a sudden 61st second at midnight, Google’s Network Time Protocol (NTP) servers spread the extra second across a 24-hour window. Over those 24 hours, Google’s servers run their clocks just a tiny fraction of a percent slower than normal. By the time the window closes, the servers have accumulated the extra second without ever having to freeze time, repeat a second, or show the dreaded 23:59:60 timestamp.

While leap smearing solved the problem for internal Google servers, it created a massive problem for the wider internet. Different companies smeared time in different ways:

• Google smeared the second across 24 hours (12 hours before and 12 hours after midnight).
• Amazon Web Services (AWS) implemented a smear window spanning 24 hours, running from noon before the leap second to noon after the leap second.
• Meta advocated for a 17-hour smear window.
• Other systems did not smear at all, adhering strictly to the official UTC standard by stopping the clock for one second at midnight.

When you have thousands of microservices communicating across different cloud providers, having one server running Google’s smear and another running Amazon’s smear means the two machines can disagree on the current time by hundreds of milliseconds. In a world of distributed databases, that disagreement is an invitation to data corruption, lost transactions, and security token failures.

Comparison of Global Time Systems

The table below illustrates how different modern systems, communication standards, and satellite constellations deal with timekeeping and leap seconds.

System / Standard	Time Base	Applies Leap Seconds?	How It Handles Time Drift
UTC (Standard)	Atomic (TAI derived)	Yes	Inserts a 61st second (23:59:60) as needed
GPS (Global Positioning System)	Atomic	No	Runs continuously without interruption since 1980
Galileo (Europe)	Atomic	No	Runs continuously; transmits an offset flag to receivers
GLONASS (Russia)	Atomic + Astronomical	Yes	Adjusts its satellites directly to follow UTC leap seconds
POSIX / Unix Time	Epoch Counter	No (in design)	Repeats second 23:59:59 to simulate the delay
Leap Smearing (Google/AWS)	Modified NTP	No (step avoided)	Slows down system clock rate over a 24-hour period

The Push to Kill the Leap Second

Over the past five years, the push to abandon the leap second grew from quiet complaints into an active campaign led by major technology firms. Engineers from Meta, Google, Microsoft, and Amazon published white papers, wrote open letters, and attended standards meetings to demand an end to the practice.

Meta engineers Oleg Obleukhov and Ahmad Byagowi published a widely read manifesto on the company’s engineering blog titled “It’s time to leave the leap second.” They argued that the leap second was an obsolete artifact of the pre-internet era. They pointed out that humanity already manages different time zones, daylight saving time shifts, and regional variations without needing micro-adjustments at the fundamental protocol level.

The campaign gained momentum because of a surprising development: the Earth started speeding up. Throughout the 1970s and 1980s, the Earth’s rotation was consistently slow, requiring frequent positive leap seconds. But in 2020, scientists noticed that the Earth was spinning faster than it had in fifty years. Several days in 2020 and 2022 broke the record for the shortest day ever recorded since atomic clocks came into use.

This acceleration meant that the world faced the terrifying prospect of a negative leap second. In a negative leap second, the clock would have to jump directly from 23:59:58 to 00:00:00, skipping second 59 entirely. While software had occasionally been tested for repeated seconds, almost no commercial software had ever been tested for a deleted second. Experts warned that a negative leap second could trigger widespread network collapses across aviation, financial systems, telecommunications, and automated industrial plants.

The Versailles Decision: What Happened at the CGPM?

Faced with mounting pressure from engineers, telecom networks, and navigation specialists, the International Bureau of Weights and Measures (BIPM) took action. In November 2022, at the 27th meeting of the General Conference on Weights and Measures held in Versailles, France, representatives from around the globe voted on Resolution 4.

Resolution 4 proposed a simple, sensible solution: lift the strict requirement that UTC must remain within 0.9 seconds of UT1. Instead of adjusting time every few years, metrologists agreed to allow the difference between atomic time and the Earth’s rotation to increase past one second, postponing any correction until at least 2035.

The resolution passed with overwhelming support from the United States, Canada, the European Union, and several Asian nations. There was one notable objector: Russia. Russia’s satellite navigation system, GLONASS, was specifically built to synchronize directly with UTC leap seconds, unlike GPS and Europe’s Galileo, which operate on continuous atomic time without leap seconds. Adjusting GLONASS to a leap-free standard will require costly software overhauls and hardware upgrades to ground control stations and satellites. Consequently, Russia lobbied to delay the change until 2040. In the end, the conference compromised on a target date of 2035, giving all nations and legacy systems more than a decade to prepare.

What Happens to Human Time Now?

When people first hear that scientists are abandoning the leap second, they often ask a very natural question: will noon eventually drift into midnight? If we stop adjusting our clocks to match the Earth’s spin, will our calendars fall out of sync with the sun?

The short answer is: not in our lifetimes, and not for many generations to come.

The difference between atomic time and the rotation of the Earth accumulates at a snail’s pace. Without leap seconds, the total drift between atomic time and solar time will be roughly one minute over the course of a century. It would take between 2,000 and 4,000 years for the difference to accumulate to a full hour.

By the time a noticeable drift of even a few minutes develops, humanity will have plenty of options. Metrologists have already proposed that instead of injecting small, unpredictable one-second adjustments every few years, we could simply allow the discrepancy to accumulate over several centuries and then make a single scheduled adjustment. We already make artificial one-hour shifts twice a year in many parts of the world for daylight saving time. Correcting for a few minutes once every century or two will be vastly easier to schedule, automate, and communicate than an unpredictable leap second announced just six months in advance.

The Philosophical Shift: Clock vs. Cosmos

The decision to retire the leap second is more than just a win for software engineers; it represents a profound philosophical change in how humanity relates to the universe.

For all of recorded human history, time was tied to the cosmos. Our days were bounded by the rising and setting of the sun, our months were shaped by the phases of the moon, and our years followed the journey of our planet around its home star. Time was an astronomical reality that we observed, recorded, and adapted to.

By untethering UTC from UT1, we are declaring that in the modern era, precision and consistency matter more than astronomical fidelity. Our society no longer runs on sundials; it runs on high-speed fiber-optic links, packet-switched communication, distributed ledger transactions, and orbital satellite positioning. In that world, an unstable, variable measurement of time is not an appreciation of nature. It is a technical hazard.

What Systems Engineers Need to Do Between Now and 2035

While the elimination of the leap second will make life much simpler in the long run, the transitional decade ahead still requires careful attention from network administrators, database architects, and software developers.

1. Audit Legacy Systems: Older enterprise servers, telecommunications hardware, and legacy embedded systems may have leap second handling hardcoded into their system kernels or firmware. Organizations must identify and modernize these systems before 2035.
2. Standardize NTP Configurations: If your organization operates a hybrid cloud setup, ensure that all services use a consistent Network Time Protocol strategy. Avoid mixing standard step-based NTP with provider-specific leap smearing algorithms.
3. Watch for Temporary Volatility: The agreement sets 2035 as the outer deadline. Between now and then, international timekeepers at the International Earth Rotation and Reference Systems Service (IERS) could still theoretically declare an emergency leap second, positive or negative, if planetary spin deviates drastically before the new rules take effect. Monitoring IERS bulletins remains essential.
4. Update Time Libraries: Developers should rely on modern, active datetime libraries in languages like Python, Go, Rust, and Java. These modern libraries are continuously patched to align with BIPM recommendations and handle timezone database updates seamlessly.

Conclusion: Welcoming an Era of Uninterrupted Time

The leap second was born out of a noble desire to harmonize the mechanics of our solar system with the quantum certainty of the atomic world. But as our civilization built an intricate digital architecture spanning the globe, the cost of that harmony became too high to justify.

The decision to phase out the leap second by 2035 is a triumph of common sense over outdated convention. It eliminates a needless source of bugs, protects vital online infrastructure, and ensures that the digital foundation of our global economy ticks forward smoothly, reliably, and without pause.

As we wave goodbye to the leap second, we step into a future where time is no longer interrupted. The Earth may wobble on its axis, ocean currents may ebb and flow, and deep magma may churn beneath our feet, but our clocks will finally remain unbroken.