International Decision to Avoid "Negative Leap Second" as Atomic Clocks and Earth's Discrepancy Reach Limit

International Decision to Avoid "Negative Leap Second" as Atomic Clocks and Earth's Discrepancy Reach Limit

Will the Disappearing "One Second" Stop the World? The Truth Behind Negative Leap Seconds and Infrastructure Crisis

When the clock shows 23:59:58, the next moment it jumps directly to 00:00:00 of the next day, skipping 23:59:59 altogether. This "non-existent one second" is being discussed as a possibility in the coming years.

This is called a "negative leap second." Until now, the world has added a second as needed to adjust the discrepancy between the Earth's rotation and atomic clocks. However, in recent years, the Earth's rotation has temporarily sped up, and for the first time, it may be necessary to subtract a second. While this is negligible to human perception, for computers, time is a common language that supports data sequencing, transaction completion, communication synchronization, power plant control, and satellite positioning signals. If a character suddenly disappears from that language, it could trigger a chain of unexpected behaviors.

However, interpreting this as "a major disruption is imminent" is not accurate. The implementation date for the negative leap second has not been set, and it has been confirmed by the International Earth Rotation and Reference Systems Service that it will not be inserted at the end of December 2026. Rather, international time standard stakeholders are trying to change the system itself before executing an unknown adjustment. The focus is not on the panic of a second that may or may not occur, but on whether the "design problem of time" that the digital society has faced for years can finally be resolved.


Atomic Clock Time and the Rotating Earth's Time

The basis of the time we use daily is Coordinated Universal Time (UTC). UTC progresses at the same rate as International Atomic Time (TAI), which is based on highly accurate atomic clocks. On the other hand, the Earth's rotation, which creates day and night, is not a perfect clock. The length of a day slightly fluctuates due to the moon's tidal effects, ocean and atmospheric movements, fluid motions within the Earth, and mass transfer from ice sheets to the sea.

To prevent the difference between the Earth's rotational time UT1 and UTC from becoming too large, a system was adopted to introduce one-second steps into UTC. This is the leap second. Since the system started in 1972, 27 seconds have been added. The most recent was on December 31, 2016, and all adjustments so far have been "positive leap seconds" where a second is added.

In a positive leap second, the day is greeted by inserting the normally non-existent 23:59:60. In a negative leap second, the last second is skipped. While this may seem like a simple operation when looking at the clock hands, modern information systems have implicit assumptions embedded, such as "a minute always has 60 seconds," "time increases at a constant interval," and "every specified time exists." The negative leap second breaks the assumption that "every time occurs once."


Why Has the Earth Suddenly Sped Up?

Over long time scales, the Earth's rotation gradually slows down mainly due to the moon's tidal effects. However, changes over short to several decades are not unidirectional. When the liquid outer core transfers angular momentum to the mantle, the rotation speed of the solid part, including the Earth's surface, also changes. Winds, ocean currents, and seasonal water movements subtly alter the length of a day.

A study published in the scientific journal Nature in 2024 analyzed that changes within the Earth are speeding up the rotation of the solid part, while mass transfer due to rapid melting of ice in Greenland and Antarctica is mitigating that acceleration. As water that was near the poles spreads toward the equator, mass moves away from the rotation axis. This is similar to the effect of a figure skater slowing down when spreading their arms.

The study indicated that if the current UTC system continues, negative adjustments will be needed by 2029. Without ice sheet melting due to global warming, the necessary timing might have been about three years earlier. However, this is not a definitive forecast that can be written into a calendar. There is irregularity in changes within the Earth and the atmosphere and oceans, making it difficult to accurately predict future rotation speeds on an annual basis. The number 2029 should be read as a scenario indicating the urgency of system change, not as a prediction that "one second will definitely disappear on that day."


Why a Single Second Can Shake Infrastructure

The danger is not that people will be unable to do anything for one second. It's that multiple devices may interpret time differently, leading to discrepancies in event order and elapsed time.

In communication networks, if the timing of logs, certificates, authentication tokens, and packet processing is out of order, normal processing may be judged as fraudulent, or the causal relationships in fault analysis may collapse. In power grids, phase measurements and protective devices at distant points rely on high-precision synchronization. Satellite positioning depends on time standards because it calculates positions based on the time difference it takes for radio waves to arrive. In financial markets and settlements, the order of orders concentrated at the same time, fair records, and deadline judgments become issues. In clouds and databases, distributed servers may be unable to determine "which update came first."

Of course, robust systems use monotonic clocks that do not go backward to measure processing time, taking into account time synchronization fluctuations and network delays. However, real infrastructure is a mix of new and old devices, different operating systems, proprietary specifications, and embedded devices that cannot be updated. Not everything is implemented according to the textbook. Furthermore, even if a device is correct on its own, if one operator skips a second instantly while another speeds up the clock gradually over several hours, discrepancies will arise at connection points.

This method of "absorbing gradually" is called leap smear. While it can avoid sudden steps, if each company has different adjustment times and curves, clocks pointing to the same moment may not match. The problem is not that there is no correct workaround, but that there is no globally common implementation method.


What Past Leap Seconds Have Broken

Concerns are not just theoretical. During the positive leap second in 2012, a bug caused CPU usage to spike on Linux servers, affecting several services like Reddit. In 2017, Cloudflare's DNS software experienced failures because the assumption that elapsed time could not be negative was broken. According to the company's post-incident report, the most affected devices were fixed in about 90 minutes, and the global response deployment was completed hours later.

The important point is not that "the world broke in the past." Even if the impact was limited, the fact that a common abnormal input reaches many systems at the same moment is troublesome. Normal bugs often surface gradually, but leap seconds are a global simultaneous event. Moreover, there is no precedent for negative leap seconds in actual operation, and it is impossible to fully grasp how much code and time distribution equipment remain that only assumes positive leap seconds.

On the other hand, it cannot be simply stated that "negative leap seconds are more dangerous than positive ones." In terms of UTC display, negative adjustments skip a second and move forward, which can be easier to handle than repeating the same time in positive adjustments. Different types of bugs, such as scheduled processes firing one second early, jobs set for non-existent times being skipped, or elapsed time being calculated with an extra second, are possible. Which is more serious depends on the OS, time distribution method, and application design.


Diverging Views on "Fear" and "Calm Reasoning" on Social Media

In the 2026 discussions on Hacker News, which attracts engineers, posts from those who have experienced past leap second failures stand out. One participant recalled the experience of a server's CPU suddenly saturating without any changes or access surges in 2012, describing it as "the worst bug in 20 years of career." Another participant learned the cause for the first time by seeing the same second recorded twice in logs and has been checking leap second announcements every six months since then. Reactions based on field experience convey that even minimal time corrections can lead to significant confusion and recovery work until the cause is identified.

 

On the other hand, there are also opinions that negative leap seconds should not be feared excessively. Comments include "time jumps forward, so it's simpler than a 'time loop' caused by repetition," "even if a process doesn't run at the specified second, many systems regularly handle similar omissions due to load or delay," and "elapsed time should be measured with a monotonic timer rather than subtracting UTC." While schedulers and real-time control require attention, if designed properly, the impact can be localized, representing a calm stance.

In Reddit's programmer community, there are voices saying, "Important systems should test negative leap seconds," and "using UTC as an all-purpose continuous time for internal computer processing is itself problematic." Conversely, there are counterarguments that it is wrong to blame only leap seconds, given existing options like atomic time and monotonic clocks. This reflects skepticism about trying to satisfy both public time close to astronomical time and uninterrupted time required by computers with the same mechanism.

These posts are not a representative survey of all users. They are open discussions voluntarily written by engineers with a high interest in the topic. Even so, the split in reactions itself indicates the essence of the problem. A simple one second in one implementation can become a major issue in another. Changing common standards is not just about "writing the correct code," but a coordination problem of transitioning diverse systems worldwide simultaneously.


Will World Time Change Significantly in October 2026?

In 2022, the General Conference on Weights and Measures decided to allow a larger difference between UTC and UT1 and to end reliance on leap seconds by 2035. Furthermore, at the 28th General Conference to be held from October 13-15, 2026, a resolution to significantly advance the transition will be deliberated.

The published proposal suggests making UTC a continuous time scale from May 20, 2027, and expanding the upper limit of the difference with UT1 to 3600 seconds, or one hour. If adopted, there will be no need for one-second corrections over a long period, even if the Earth's rotation speeds up or slows down slightly. According to experts, there is at least a few centuries, or potentially a millennium, of leeway before reaching a one-hour difference.

The expression "stop leap seconds and suddenly add leap hours someday" is easy to understand, but it does not mean that adding one hour at once in the near future has been decided. The 3600 seconds is a new upper limit set for the discrepancy between UTC and Earth rotation time, and the specific adjustment method for the future will be decided by the society of that era. The aim is to eliminate the burden of irregular one-second processing every few decades and bring UTC closer to continuous time suitable for computers.

However, as of September 2026, this proposal is a "draft resolution" and not yet a final decision. The results of the vote and transition rules need to be confirmed. Additionally, in fields like astronomy, geodesy, and some satellite and radio equipment that require UT1, transitioning to a system that acquires the difference from UTC as separate data and reflects it in calculations is essential. Transition measures for countries and operators where updating old equipment takes time will also be a challenge.


What the "End of One Second" Questions

The debate over the negative leap second is not about the Earth spinning at an abnormal speed and the world collapsing. The changes in the Earth's day are on the millisecond scale and are not something humans can feel. The issue is that the time created to match nature's irregular movements and the time needed by the digital society for uninterrupted numbers are colliding within the same UTC framework.

Even if the system change is adopted, it does not mean that preparation becomes unnecessary. Organizations in time distribution services, OS, cloud, communications, finance, power, and satellite operations need to share specifications by the transition date and ensure that different time scales do not coexist. It will be an opportunity to review basic designs, such as using monotonic clocks for elapsed time, allowing wall clock jumps, not determining the entire order of events by timestamps alone, and documenting synchronization methods with external services.

Humanity has long determined time by observing the movement of the sun. Now that the precision of atomic clocks and network synchronization have become the foundation of society, world time is trying to take a slight distance from the Earth's rotation for the first time. The one second that might have disappeared may actually change the history of the system without disappearing. Its conclusion will move significantly at the international conference in October 2026.


Source URL

Related Articles