Since 1972, the International Bureau of Weights and Measures has been adding an extra second to the clock whenever Coordinated Universal Time drifts a full second ahead of Earth's rotation. The fix was elegant for an era of paper calendars. It has been a rolling disaster for the era of networked computing.

Leap seconds have crashed Meta, Reddit, and Cloudflare. They have grounded commercial flights. They have scrambled high frequency stock trading. The problem is simple: computers process messages in order, and inserting a second that the system clock does not expect makes "in order" a philosophical question for a few dangerous moments.

Some developers gave up on the official standard and rolled their own workarounds. Google built "smear seconds" that stretch the adjustment across a longer window so nothing hiccups. That is the polite way of saying the world's biggest tech company decided the international standard for time was too unreliable to use.

Starting around 2016, Earth's rotation began speeding up. Not by much. But enough that no leap seconds have been needed since. More importantly, enough that timekeepers are now staring at a scenario they never planned for: the first ever negative leap second.

A positive leap second (adding one) already broke things. A negative leap second (subtracting one) has never been tested against any production infrastructure, anywhere. The potential for catastrophic failures in systems that were never designed to handle a missing second is, to put it gently, non trivial.

At the 2022 General Conference on Weights and Measures, 64 member states agreed to find a way to extend the allowed gap between UTC and Earth's position from one second to something larger by 2035. Now, ahead of the 2026 CGPM meeting, the leading proposal has crystallized: replace the leap second with a leap hour.

A leap hour would not need to be added for hundreds of years. Maybe longer. The proposal under consideration would aim for implementation as early as 2027. This is, in effect, abolishing the leap second by replacing it with a correction so vast that no living human will need to deal with it.

It is worth admiring the engineering elegance here. The solution to a problem caused by a too frequent correction is to make the correction so infrequent that it becomes someone else's problem entirely. Several centuries worth of someone elses.

This is a story about technical debt. The leap second was a perfectly reasonable solution in 1972. Computer networks were not yet the substrate of civilization. Adding a second to a clock was a clerical matter, not an infrastructure event.

But the world changed, and the standard did not. For decades, the fix was to keep patching around the problem: smear seconds, custom implementations, crossing your fingers during leap second events. The standard became a polite fiction that everyone pretended to follow while quietly building their own workarounds.

The leap hour proposal is what happens when the workarounds finally become more expensive than fixing the original design. It took 54 years.

There is a lesson here that extends well beyond timekeeping. Every industry has its leap seconds: standards and processes that made perfect sense when they were created, that everyone knows are causing problems now, and that persist because the cost of changing them feels larger than the cost of another workaround. Until it doesn't.