Time Committee Procrastinates

The International Telecommunication Union, an arm of the United Nations, recently announced with great fanfare that it intended to settle the "problem" of leap seconds once and for all. At their meeting on January 19th, the committee decided to kick the decision down the road for another three years. Procrastination — what could be more appropriate for an official international committee on time?

Truth be told, this choice was as good as any other. Leap seconds have been added periodically to keep the official time in sync with Earth's uneven, unpredictable rotation, most recently at the beginning of 2009.

Old and modern timekeeping
Precision time, then and now: an antique chronometer and modern atomic clock at the U.S. Naval Observatory in Washington, D.C.
U.S. Naval Observatory
These leap seconds upset techies, because they mean that the internal clocks inside devices can't run autonomously. Instead, somebody has to go in there every few years and set the clock back a second. So people who run computer networks would be much happier if leap seconds were abolished.

However, abolishing leap seconds would just kick the problem down the road for a millennium or two, by which time local noon (the Sun being at its highest) would occur several hours late. People seem (strangely enough) to be willing to put up with daylight saving time, but I doubt anybody would be happy having the Sun highest at 11 p.m.!

What's my opinion? It's a thorny problem without any good long-term solution. But I object on principle to bending people's behavior to suit the convenience of computer techies. Machines were meant to serve us, not the other way around!

For more information, see our online article on time.

17 thoughts on “Time Committee Procrastinates

  1. Sandy Brown

    "…people who run computer networks…" Funny, but my computer is connected to the Internet, and its time is automatically updated via the NIST tie service. Since they run the "atomic clock," where’s the problem???

  2. Brooks Davis

    I have no idea where the canard that network devices are the problem came from, but that mostly isn’t the case. The Network Time Protocol (NTP) can deal with leap seconds though not all time sources keep their list of leap seconds up to date. The more significant issues lie elsewhere.

    One issue is that you can not build a practical non-networked device (including a device with GPS clock) that will definitely display the correct UTC wall clock time at +31536000 seconds. The problem is that there are up to 4 opportunities to change that value during a given year as determined by the International Earth Rotation and Reference Systems Service.

    Another issue is that it is often useful to have a continuous time scale for precision logging and it is further useful if that time scale aligns with wall clock time. With leap seconds, neither is true.

    As the worry about precession of wall time relative to solar time in the indefinite future; we already have a solution: timezones. As solar noon drifts too far outside the timezone we can make a periodic change. At the current rate of change in the speed of earth’s rotation the general public will not even notice for several centuries unless they try to sync their cellphone and high precision sundial. By making a large jump and planning for it well in advance we will be able to keep solar time acceptably aligned with wall clock time while allowing reasonably long lived devices to operate without needless updates.

  3. jeff wahlgren

    "Machines were meant to serve us, not the other way around!"

    And yet most of my day is spent on ‘work-arounds’ for the special cases my machines cannot accomodate.

  4. Frank Reed

    I’m a Time Lord. No really. It was announced on the radio so it must be true. And so is Rob Seaman (see above post). You can listen to many minutes of Rob and a couple of minutes of me here: http://www.startalkradio.net/?p=323.

    As a Time Lord (Ha!), I can assure you that this isn’t just a matter of the time of Nature versus the time of Machines. Atomic Time and GPS Time and the astronomers’ Ephemeris Time (in its many acronymed names) are realizations of a time that is more absolute, more fundamentally-founded in the laws of physics than the UTC that we use today. They are "Physics Time" as opposed to "Earth Rotation Time". The latter is the principle underlying UT, and is based on one very large but slightly wobbly artifact: the Earth itself.

  5. kdconodKC

    As far as allowing time to drift so that eventually we have to shift time zones doesn’t sound like a good idea. People are often ‘time-zone centric’…the major population centers tend to be near the center of the time zone and so people forget about the folks living at the edges. For them they are already putting up with unpleasant disconnects with solar time. If you don’t make small adjustments, then people will have to put up with increasingly unpleasant conditions for centuries.

  6. Frank Reed

    KC, you wrote: "the major population centers tend to be near the center of the time zone and so people forget about the folks living at the edges" …. But it’s not true at all. Time zones have been drifting east since they were first created. And since "Daylight Saving Time" in many countries, especially the US, is not in effect for most of the year, we are FAR from solar time and ‘major population centers’ are no longer even close to the appropriate mean time. Just consider the eastern US. For nearly two-thirds of the year, cities from Boston at the eastern end all the way over to Detroit and Toledo and Indianapolis use a mean time which corresponds to 60 degrees west longitude as far as the actual longitude of the Sun is concerned. Get out a globe (virtual or otherwise) and consider what this means: when a clock in Detroit says 12:00:00 noon, the Sun is over a meridian far out in the Atlantic well east of Bermuda, east of most of Nova Scotia, and even east of Martinique in the Lesser Antilles. We do not keep time by the Sun.

  7. Frank Reed

    Further on shifting time zones. It’s an easy solution well within the normal rules of "mean time" law as it already exists. Consider Michigan. That state has legislated itself onto Eastern (US) Time since the middle of the 20th century. By geography and longitude rules, they ought to be in Central Time but convenience and culture have persuaded them to live on Eastern Time. They can switch "back" to Central Time if they wish at any time. In 200 years, if the slow drift implied by dropping leap seconds were to become impractical, they could easily legislate themselves back to Central Time. It’s almost easier to imagine that law change than it is to imagine Michigan as an autonomous entity in 200 years. And yet there’s still a catch. Time zone shifting cannot stop the movement of the date. Without global agreement, the beginning of the calendar day would slowly drift east. Eventually there would even be confusion over the day of the week. Is it Sunday yet? Would there be any fraction of the population in 500 years who would be concerned over whether the calendar date is really Saturday or Sunday??

  8. Rob Seaman

    Hi Frank. A few points. The ITU proposal makes no mention of timezones at all. Whatever the viability of the notion of encouraging them to drift relative to UTC (we disagree on that apparently), it has not been studied in any coherent way.

    The fundamental issue is that TAI is a metronome and Universal Time is time-of-day, that is, a measure of Earth orientation angle. The question being debated is whether all our clocks should become metronomes. My position is that both kinds of timepiece are necessary.

    Note, additionally, that metronomes already exist and that TAI (and GPS without leap seconds) are *already* available. This proposal isn’t about providing access to atomic time, it is about denying access to Earth orientation time. Telescopes and other software and systems used by both professional and amateur astronomers are dependent on Earth orientation. Implementing new infrastructure for this would be very expensive, and would provide nothing we don’t already have. The expense would be carried by end users as well as communities such as astronomy and aerospace.

    The organizers of "Decoupling Civil Timekeeping from Earth Rotation" made an attempt to reach out to commercial astronomy vendors as we did to other stakeholders such as the very enthusiastic sundial and solar eclipse communities. We were delighted with your participation representing navigational issues! There are many more such niches whose denizens remain unaware of this issue, and this appears to continue to include many vendors of commercial astronomical equipment. Three years is not a long time to build awareness.

  9. Frank Reed

    Hi Rob. You mentioned that the ITU proposal does not mention time zones. And it shouldn’t. On land and in territorial waters, time zone definition and selection is under the control of national governments with some devolution to regional governments (states, e.g.). At sea, time zone standards are suggested, but the captain decides based on convenience. I do, of course, agree with you that there are major costs that would accompany dropping leap seconds from UTC (or UT whatever it might be named… UTx). I don’t think there’s much point going into this further here since S&T’s message board system rather effectively shuts down any discussion at about ten messages. But since it is S&T, I would add that there’s a group that might face some interesting issues that hasn’t been discussed: backyard astronomers with computer-driven telescopes. Most of these systems have simple methods for software updates, but, who knows, there may be some that cannot handle a value for DUT that is greater than one second. That would send all the owners of any such telescopes back to the 1980s and they would actually have to find astronomical objects manually. Oh the horror… the horror… :) I do not agree with one statement that you made: that dropping leap seconds would somehow "deny access to Earth Orientation Time". The latter is observable no matter what UTx becomes. Ensuring funding for the observations is a separate matter.

  10. Rob Seaman

    Frank, just to wrap this up, as you say we don’t know how many applications will be able to handle DUT1 > 0.9s. The bigger issue is that few systems ever implemented DUT1 support in the first place because UTC provided a close approximation to Universal Time. If UTC no longer provides UT, these systems will undoubtedly break. The timezone gimmick is, of course, completely beside the point for either of these cases.

  11. Anthony Barreiro

    The human experience of time arises inextricably from the rising and setting of the Sun and the progress of the seasons. Our ancestors invented written calendars and mechanical clocks to keep track of these rhythms, even when it was cloudy or they were indoors. Over time the clocks became more precise than the planet whose rotation they were designed to emulate, and they were put to more and more sophisticated uses in which a thousandth of a second (1/86,400,000 of a day) can make a significant difference. I understand that leap seconds are a pain for chronometrists, but the rest of the human race needs our clocks to remain synchronized with the Sun. Would it be easier for the technicians if instead of using leap seconds we used leap minutes? A minute one way or the other would not be a noticeable abberation from solar time, and it would only need to be used once or twice per century, rather than several times a decade. (As for daylight "saving" time, let’s just get rid of it.)

  12. Warren Odom

    For a very authoritative description of the problem from a computer point of view, read this (not too lengthy) article from the Association of Computing Machinery.


    Much of the concern revolves around equipment that depends on sub-second time synchronization with other equipment, so just periodically adjusting to the atomic clock isn’t sufficient, unless all computers agree on exactly how to adjust (and they don’t).

    Her are some interesting quotes:

    Air Traffic Control systems perform anti-collision tests many times a second because a plane moves 300 meters in a second. A one-second hiccup in input data from the radar is not trivial in a tightly packed airspace around a major airport.

    On December 8, 2010, a 70-msec power glitch hit a Toshiba flash chip manufacturing facility, and 20 percent of the products scheduled to ship in January and February 2011 had to be scrapped.

    I’m told from usually reliable sources that the entire U.S. nuclear deterrent is in "a special mode" for one hour on either side of a leap second and that the cost runs into "two-digit million dollars."

    We have tried to suggest a compromise on leap seconds that would vastly reduce the costs and risks involved: schedule the darn things 20 years in advance instead of only six months in advance. If we know when leap seconds are to occur 20 years in advance, we can code them into tables in our operating systems.

  13. Frank Reed

    Warren Odom, you wrote: " We have tried to suggest a compromise on leap seconds that would vastly reduce the costs and risks involved: schedule the darn things 20 years in advance instead of only six months in advance. If we know when leap seconds are to occur 20 years in advance, we can code them into tables in our operating systems." Sorry, but this is IMPOSSIBLE. And in fact, it’s the whole point of leap seconds. Leap seconds are inserted because the Earth’s orientation is observed to disagree with clock time. While the number of leap seconds expected in some interval of years can be roughly estimated, the difference between a reasonable guess and actual observation will exceed one second within a fairly short period of time. And as soon as it does, all of the problems associated with dropping leap seconds entirely immediately come into play. The only alternatives that make much sense are: 1) keep UTC as it is, inserting leap seconds as observations require so that the difference between UTC and UT1 (traditional Earth Orientation Time) remains below 1 second, or 2) stop the leap second practice entirely. Both options have their merits.

  14. Frank Reed

    Anthony Barreiro, you wrote:
    "the rest of the human race needs our clocks to remain synchronized with the Sun." … But when was the last time you used a sundial (and did NOT add any adjustments)? It’s been over 200 years since most people in the West kept time, literally, by the Sun. It’s been over a century since most people kept even Local Mean Time (of course, there was outrage in some quarters when time zones were introduced). And with the addition of Daylight Time, now in effect for nearly two-thirds of the year in the US, many places set their clocks to mean time meridians that are over a thousand miles away from their geographic locations. And it’s not just the deserts of western China that we’re talking about here. For MOST of the year, Indianapolis, IN (to pick a random place not in the news today) sets its clocks based on 60 degrees West longitude –a meridian far out in the Atlantic technically correct in terms of mean solar time for places like Martinique and Barbados and the eastern tip of Nova Scotia. From March through October in Indianapolis, the Sun doesn’t reach the meridian until about 2:00 in the afternoon. Similarly Vigo, Spain for most of the year sets its clocks such that the Sun doesn’t reach the meridian until about 2:30 in the afternoon. Dropping leap seconds will only add about a minute to this in 75 years.