Clocks Are Too Precise (and People Don't Know What to Do About It)

No big deal, but abolishing leap seconds could unmoor us from the sun forever.

Demetrios Matsakis at the U.S. Naval Observatory (Katherine Wells/The Atlantic)

Presenting to fellow scientists on the thorniness of time, Demetrios Matsakis will sometimes show a passage from Chaucer. It’s filled with the strange consonants and unexpected vowels of six-century-old English. Who here, he asks, can read these words like they would their own language?

Only a few hands go up.

That we struggle to parse Chaucer is unremarkable, Matsakis says, because we all understand that language, technology, and the value of money change over time. But Matsakis believes our understanding of time itself must now change over time, too.

Matsakis is chief scientist for Time Services at the U.S. Naval Observatory. With its atomic Master Clock, the Observatory provides the official time of the U.S. government. GPS, the military, and countless businesses—much of modern technological society—depend on the Observatory’s work and its clocks. If you have a smartphone, its time comes from the Observatory, too.

We are talking because, in his role as a senior American horologist, Matsakis believes that the world should abandon “leap seconds,” the global fiction that keeps a year of Coordinated Universal Time the same length as a solar year.

Leap seconds arise from the fact that humanity no longer turns to the sun and moon to make sense of time. Instead, we derive the length of a second from a specific number of oscillations of a cesium atom, then multiply that by certain integers to arrive at longer standards of time. Count 31,557,600 of these SI seconds and you arrive at a Julian year.

But the Earth, Moon, and Sun—massy orbs of rock and fire—do not behave so predictably. Some years, interactions between the three make the Earth orbit a little slower; sometimes, a little faster. In order to keep UTC noon synced to solar noon, the International Earth Rotation and Reference Systems Service adds or subtracts a second to the global UTC every so often. On Monday, it did this most recently, decreeing that June 30, 2015, will last one second longer than the usual 86,400.

But even more than that extra instant in June, leap seconds are having a moment right now. In November, the International Telecommunication Union (ITU) will meet to decide whether to abolish the leap second. The group, an agency of the United Nations, last debated the issue in 2012. It chose to preserve the status quo and re-visit the issue in three years.

The United States is officially against leap seconds. The national position on such matters is determined by the State Department, Matsakis said, but as an advisor on the matter he said he could summarize why the U.S. supported their abolishment: “the real world impossibility of reliably implementing leap seconds.”

“You know [the leap second] exists,” he told me. “Your readers will know it exists for a month, then they’ll forget.”

But most people—including commercial programmers, who write the critical software that controls public and private infrastructure—don’t know about the leap second, Matsakis said, and that means their code doesn’t account for it. So when a new leap second rolls around, things break. Reddit, LinkedIn, and Yelp all suffered issues related to the last leap second in 2012. And, more seriously, computer booking systems used by Qantas Airlines all struggled, delaying flights by hours.

In some cases, it is impossible to update systems before the next leap second arrives. Matsakis spoke of a Switzerland power company whose backup systems only turn on when needed—otherwise, they sit disconnected from the network. When they were activated in a test after the last leap second, they crashed.

Had they been needed at the time, Matsakis said, parts of the country would have suffered a blackout. As a Naval Observatory spokesman told the New York Times in 2012, getting rid of the leap second “removes one potential source of catastrophic failure for the world’s computer networks.”

But wouldn’t abandoning the sometimes-second eventually divorce the Earth from the sun? This is one of the reasons why Canada, China, and Britain all cling to the leap second, or, at least, did at the last global meeting.

Matsakis thinks such a link has already been lost.

“​We don’t have a connection,” he told me. “We lose the connection twice a year when we go on daylight saving time.”

By the end of this century, the accumulated gap between UTC and what-should-be-the-solar-time would only be about one minute.

“What will happen is that in about 1,000 years, instead of the sun being overhead at noon, it will be overhead at 1 p.m. But by then society will have shifted.”

Hence the Chaucer. “Almost no one can read Chaucer,” Matsakis says, but we don’t freak out about it. Instead, we understand language to be one of those systems that has shifted imperceptibly over the centuries. In 600 years, when scholars translate texts from before the 21st century, they will just know that—in addition to translating or annotating monetary values so they make sense for contemporaneous readers—“noon” needs to become “1 p.m.”

Such a slow shift over time would be worth it, Matsakis said, for all the network failures it would prevent.

“You would [think it was worth it] too if you were one of the people stranded in Australia when Qantas Airlines went down,” he told me.

He recalled how American railroad companies only invented timezones after crashes forced them to. (Between 1831 and 1853, there were 97 railroad crashes—often because two trains, scheduled at close intervals on the same length of crowded track, disagreed about the time.) And perhaps it would be the same for the leap second.

“People understand crashes,” Matsakis said. “They understand people dying. If there is a plane crash, leap seconds will go away, and maybe that’s what it will take.”

The Atlantic last talked to Matsakis in February, when he described his job and the functioning of the American atomic master clocks: