November 10, 2005 10:26 AM   Subscribe

In one corner, precise astronomers who just want to keep things as they are. In the other, revisionist telecommunications officers. Fight!
posted by goodnewsfortheinsane (25 comments total)


Tsk. Americans...
posted by Artw at 10:35 AM on November 10, 2005

"If you get rid of the leap second you are destroying the notion of what time is," Dr Lippincott said.

Dear god! Somebody stop us! We'll be after the notion of space next!
posted by freebird at 10:41 AM on November 10, 2005

"If your navigation system causes two planes to crash because of a one-second error, you have worse problems than leap seconds," said Steve Allen, a University of California astronomer who maintains a Web site about leap seconds.

He's got a good point there. And then when you consider that the U.S. is changing around the start and end for Daylight Saving Time (in 2007, I think) and that you're going to end up off by an hour in legacy code, I think the best thing to do is to make sure that code can handle the leap second (which it should, since it's "official time").
posted by Godbert at 10:42 AM on November 10, 2005

As some additional background, the method of determining leap seconds is as precise as it can be, but it's inherently a rather imprecise process.

The Earth isn't a ball of solid rock, spinning serenely in a vacuum... it's a semi-solid ball with a lot of sloshing about. The Moon is a constant source of much of that interference. Any given 'sunrise' will vary somewhat from the official sunrise time, simply because every individual rotation the Earth makes is unique.

Now, these are very small changes, but they accumulate over time, and every once in awhile, the Powers That Be decide that enough slop has accumulated to add a leap second. Sometimes the Earth doesn't appear to slow down much for a long time, and a long while elapses between leap seconds. Sometimes we add them more quickly.

Time, from the viewpoint of an astronomer, is a very analog thing. It's defined as 'the precise moment Star X can be seen to rise over Feature Y on Date Z'. And they get slightly different results every time they measure it.

Fundamentally, this is a fight for control between time derived from astronomy (which varies) and atomic time (which varies not a bit.)

I think, overall, it would be most useful to humanity to go to atomic time. Time used to mean 'when the sun rises', but that's just not true anymore. But this transition, if we decide we want one, should be discussed very publicly, and done very slowly. As the astronomers are pointing out, it may require expensive changes to their hardware.

There's no rush, after all. We have all the time in the world. :)
posted by Malor at 10:51 AM on November 10, 2005

A clarification from one of the linked articles:
The current down-link data format for the GPS satellites stores the difference between GPS system time and UTC using a signed 8-bit quantity. This means that the maximum difference ( GPS - UTC ) that can be stored is 127 seconds. If leap seconds continue to be inserted into UTC, then the current fleet of GPS satellites and all receivers of GPS signals will become obsolete.
posted by nobody at 10:55 AM on November 10, 2005

better solution: slow down earth's rotation with equator-mounted rockets so that one new day = 1.426758 old days. There will happen to be exactly 256 of them in a year. Each new day has 32 new hours (about 35 old hours). Each new hour has 64 new minutes (about 71 old minutes). Each new minute has 64 new seconds (about 60 old seconds). Each new second is about 0.941 old seconds.

Instead of 365.2514159265, 24, 60, 60, we how have 256, 32, 64, 64... nice round numbers.

Earth's rotation is adjusted as needed to keep the numbers in line.

To have shorter days, but still 256 of them in a year, requires Earth orbit adjust, which is probably infeasible at this time.
posted by kurumi at 10:57 AM on November 10, 2005

good reading
posted by poppo at 10:58 AM on November 10, 2005

To have shorter days, but still 256 of them in a year, requires Earth orbit adjust, which is probably infeasible at this time.

Just build directionality into those equator mounted rockets
posted by poppo at 11:12 AM on November 10, 2005

I find it interesting that apparently so few clocks implement the Leap Second. I had a project to basically make one of those WWVB controlled clocks, and I went and included the leap second. It wasn't very hard.
posted by TheOnlyCoolTim at 11:22 AM on November 10, 2005

If you have an instrument that can't handle UTC, with the associated leap seconds, you use TAI, which doesn't have them.

(Short version: TAI is time defined by atomic clocks. UT, Universal Time, is time based on the rotation of the earth. UT seconds aren't nearly as consistent as TAI seconds, so the scales shift. UTC fixes this by using TAI second, but when the difference between UT and UTC exceeds .9 seconds, it jumps a second to narrow the difference.)

This is a non-problem. UTC is used for civil time, because the general population likes the mornings to be in the morning, thus, leap seconds and leap years. TAI is used for things that have to have one second follow another, and can't have skips or repeats.

If you are too stupid to use the right time scale for you application (and there are far more than UTC and TAI), then you are too stupid to make judgements about the usefulness of leap seconds. If you are smart enough, you don't have a problem with UTC skipping, because you either work around it with a lookup table, or you avoid the whole problem completely with another time scale.

(More. UT is actually four time scales. The directly observed scale is UT0. The problem -- the earth isn't spherical, it is oblate, so we correct for that, and get UT1. Problem: UT1 isn't very smooth, because the earth's rotation is affected by tidal forces from the moon, the sun, and other planets. Thus, UT2, a smoothed version of UT1. Finally, use that to correct TAI, and you get UTC.)
posted by eriko at 11:27 AM on November 10, 2005

It makes sense (to me, anyway) to standartize satellites and navigation systems to atomic time, and everything else (e.g., when you're supposed to show up for work) to astronomical time. Instead of updating the former with leap seconds, we could just keep track of the difference between the two.

This policy will keep life on Earth simple, but has better forward-compatability with the day when there's a permanent human presence off the planet.
posted by Eamon at 11:28 AM on November 10, 2005

Damn live preview. Looks like we're already doing that. Thanks eriko and TheOnlyCoolTim.
posted by Eamon at 11:29 AM on November 10, 2005

"Time has basically always really meant what you measure when you put a stick in the ground and look at its shadow,"

Time is just a magazine. geez...!
posted by mikhail at 11:41 AM on November 10, 2005

This is why the Unix epoch is such a good idea: it doesn't matter what time it is "now" because you just measure in terms of how long it's been since "then".
posted by NinjaPirate at 12:30 PM on November 10, 2005

I have always wished my birthday was the epoch
posted by poppo at 12:31 PM on November 10, 2005

And this, ladies and gentlemen, is why you should always use libraries when implementing date and time.
posted by Deathalicious at 2:22 PM on November 10, 2005

Interesting. I always thought horology was something completely different.
posted by kyleg at 2:40 PM on November 10, 2005

What? The current U.S. administration taking a position that infuriates scientists, and doing it in secret? I'm shocked! Shocked I say!
posted by theonetruebix at 4:03 PM on November 10, 2005

I'm being obtuse here, but there are about 30 million seconds in a year. So if we've add one second, say every two years...

Run it backwards and it seems the earth could only be 60 million years old.

OK, maybe it's exponential, which is what you'd expect. That seems to make it even younger.

What gives?
posted by grahamwell at 4:38 PM on November 10, 2005

Thankyou, grahamwell, for destroying my mind.
posted by Jimbob at 5:58 PM on November 10, 2005

I say we get TimeCube Guy to adjudicate.
posted by pompomtom at 8:03 PM on November 10, 2005


Time cannot run backwards. There, that's settled.
posted by goodnewsfortheinsane at 3:13 AM on November 11, 2005

I'm being obtuse here, but there are about 30 million seconds in a year. So if we've add one second, say every two years...

Run it backwards and it seems the earth could only be 60 million years old.

Leap seconds don't add a second to the length of every day. They each add a second to the length of one day, so that subsequent days "start" at the right time.
posted by Eamon at 7:34 AM on November 11, 2005

AND .. since there are only 365 days in the year and we add an extra day every four years, History MUST have started about 1,200 years ago.

Think I'll have some fun with this one but yes, thanks Eamon, I had to screw my brain up tight but I do see.
posted by grahamwell at 12:18 PM on November 11, 2005

« Older Crazy Swiss   |   THINK! Newer »

This thread has been archived and is closed to new comments