Slashdot Mirror
The Future of Time: UTC and the Leap Second
rlseaman writes "UTC ("Coordinated Universal Time") is very close to being redefined to no longer track Earth rotation. Clocks everywhere — on your wall, wrist, phone or computer — would stop keeping Solar time. 'American Scientist' says: 'Before atomic timekeeping, clocks were set to the skies. But starting in 1972, radio signals began broadcasting atomic seconds and leap seconds have occasionally been added to that stream of atomic seconds to keep the signals synchronized with the actual rotation of Earth. Such adjustments were considered necessary because Earth's rotation is less regular than atomic timekeeping. In January 2012, a United Nations-affiliated organization could permanently break this link by redefining Coordinated Universal Time.'"
Taco, all you did is quote the article summary. I can spin up an RSS reader to do that.
I want to delete my account but Slashdot doesn't allow it.
Here's the full PDF version.
We can't have this! Think of the children! They'll learn that Noon is when the Sun is high in the sky only to see it pitch black!
By then, children will no longer venture outside. Thanks to Facebook. And Slashdot.
Have gnu, will travel.
In 2012, a new definition of time that is only relative to the Earth's reference frame falls short.
My God, it's Full of Source!
OUTSIDE_IP=$(dig +short my.ip @outsideip.net)
The Download: PDF Only link on the right hand side of the first linked page (Cornell) gives you a PDF of the same article.
Oh no... it's the future.
10 is an arbitrary base and works poorly for time. Earth time works very well in Base-12, which is what we have now.
If you want to think of the children, the Egyptians and Babylonians taught their children to count on the knuckles of their four non-thumb fingers. The 10-finger-number system is just an unfortunate thing that came along with arabic numerals, which are very useful.
I saw we adopt something from the UNICODE set to mean eleven and twelve. Say, they're not 1-teen and 2-teen, are they?
My God, it's Full of Source!
OUTSIDE_IP=$(dig +short my.ip @outsideip.net)
... could we also get rid of Daylight Savings Time?
We say 365 days.
We observe 365.25.
The tropical year (equinoxes+solstices) is closer to 365 solar days, 5 hours 49 minutes 19 seconds
The sidereal year is 1.0000385 tropical years (365.256363004 ) (20m24.5128s longer than tropical year)
So may times...
Slashdot's rate-of-post filter: Preventing you from posting too many great ideas at once.
This has been *progressing*?
This is possibly the stupidest idea in history. Their stated goal: taking complexity out of time handling code -- *cannot happen*: it will *still* have to account for all the years we did this.
And it will break *lots* of stuff.
Seriously this has been an issue for along time - GPS time does not include leap seconds and I am tired of having to write software that let's user adjust for the variable amount of leap seconds - nobody really cares if the earths rotation is synchronized with " UTC"
Those who doesn't care about synchronisation already have the option to use TAI. They should use that instead of redefining UTC.
Yes, think of the humans. The ones wasting millions of dollars a year keeping databases up to date with the leap seconds which, unlike leap years, are not algorithmically predictable, and ensuring that all critical time-dependent software is properly tested and verified.
Leap seconds happen at most twice a year, and typically once every few years. If they happen at their maximum permitted frequency, then in 1000 years the difference between midday and noon will be 2000 seconds, or just over half an hour. In other words, for me the drift due to not living exactly on a time zone boundary line will still be more than the difference due to ignoring leap seconds.
Note that this is a wildly pessimistic prediction. It's more likely that we will drift about 5 minutes over the next thousand years. In 1,000 years, if the position of the sun over Greenwich at 12:00 is still of importance to a significant fraction of the human race, then I'll be very surprised.
I am TheRaven on Soylent News
will simply ignore UTC and continue to show the time based on the earth's rotation. All this means is that UTC loses any small shred of relevance that it once may have had to the common man. So go ahead, redefine UTC and we can all just go back to using GMT for our reports, syslog messages, traps etc. that all have far more to do with the time as experienced by users than the time as experienced by a cesium atom. I expect that soon after the decision is made someone will start gmt.pool.ntp.org and utc.pool.ntp.org and we'll have a choice.
Nullius in verba
?
Leap seconds is already included in the GPS stream. So, with the GPS stream you get GPS time, and UTC time. Most GPS receivers output GPS time until they've received the NAV or other message that has the leapseconds count in it, and then correctly updates to UTC time. It seems like a simple, solved problem. Am I missing something with respect to your leap year stuff?
If UTC would be redefined to no longer be adjusted to Earth's rotation, then what would be the point of having UTC at all? We already have a time scale that counts seconds without adjusting to Earth's rotation: TAI
Please correct me if I got my facts wrong.
It's called TAI. Redefining UTC is just plain stupid - it was created to track sol. If someone doesn't like the fact that doing so requires occasional adjustments, then they chose to use the wrong time scale. Those who use UTC as intended shouldn't have to live with the problems which will result if it is unlocked from solar time, just to keep those who made a poor choice happy.
(GMT hasn't been in use for a long time, although most people use the term interchangeably with UTC).
"National Security is the chief cause of national insecurity." - Celine's First Law
" I am tired of having to write software that let's user adjust for the variable amount of leap seconds"
You're doing it wrong.
"National Security is the chief cause of national insecurity." - Celine's First Law
The situation can be explained in three pictures.
Using already-deployed code, here is one way to solve the problems.
Leap seconds happen at most twice a year, and typically once every few years. If they happen at their maximum permitted frequency, then in 1000 years the difference between midday and noon will be 2000 seconds, or just over half an hour. In other words, for me the drift due to not living exactly on a time zone boundary line will still be more than the difference due to ignoring leap seconds.
It's impossible to predict ahead of time how many leap seconds are necessary. Just because it's been historically ±1 second doesn't mean it will always stay like that in the future. As T increases quadratically, we will eventually see leap minutes and leap hours.
That's not even taking into account of the fact that doomsday style meteor impacts could necessitate adding or removing a whole day every year. That pretty much breaks half of the time-related code that I've written.
Note that this is a wildly pessimistic prediction. It's more likely that we will drift about 5 minutes over the next thousand years. In 1,000 years, if the position of the sun over Greenwich at 12:00 is still of importance to a significant fraction of the human race, then I'll be very surprised.
It's not pessimistic at all. T between the 1000AD and 2000AD was 1600 seconds.
In fact, I'd say that's a very optimistic prediction, since T is monotonically increasing, T for the next 1000 years will be much larger than T for the previous 1000 years.
The old system will end and all new one will start up at the same time with big changes.
The ease of conversion is more when converting compound units. A newton is a kg m/second^2. A Joule is a Newton applied over a meter = kg m^2/second^2. A Watt is a Joule used every second = kg m^2/second^3.
American units have the pound, which is a slug foot /second^2. Except no one uses the slug, so it has to be pound-force = pound mass * 32.2 ft/second ^2. Energy? Sometimes they use foot-pound which is easy enough, but other areas we use kilowatt-hour, or BTU (defined from the heat capacity of water). Power is defined in either kilowatts or horsepower (defined via watts). Air conditioners are rated by BTU (actually BTU/hour, but that's not the way they're advertised).
The metric system is originally based upon an earth derived measurement called the meter. 1m ~ 1/10,000 the distance from the equator to the poles, 1g = the weight of 1cc of pure water at STP, etc. That all the measurements are base 10 is not what makes it "metric", it's that they're derived from the (originally) earth centric meter. Our time system is also derived from earth centric measurements, called day and year.
Base 10 time would be a huge adjustment for society. While using a 100,000 "MetSec" day and a 100 "MetSec" "MetMin" would produce units fairly close to the existing second and minute measurements, a "MetHour" would be much longer or much shorter than an hour, either 14.4 minutes (100 MetHours/day), or 2.4 hours (10 MetHours/day). And that doesn't do anything to address the leap second issue, nor does it alter the ~ 365.25 MSD year.
The bottom line is that as long as we maintain the concept of a day and year and all the associated stuff (seasons, equinoxes, solstices, etc.), all of which are critical to agriculture and survival, there has been no system of time keeping proposed that is significantly better than what we have. The universe is not going to arbitrarily adapt it's cycles to make it easy for our minds and computers to keep track of time.
And that's without considering relativistic time dilation. While most people never have to worry about time dilation effects, the atomic clocks that create UT1 and GPS satellites have to compensate for relativistic differences caused by differences in local gravity and speed, both of which are affected by altitude.
Perhaps Douglas Adams said it best, "Time is an illusion, lunchtime doubly so."
make imaginary.friends COUNT=100 VISIBLE=false
Oh groan, best use of that tag on /. so far.
MilkMiruku
Exactly. 2,3,4, and 6. I propose that we go even further and use also use 60 so that 5, 10, 12, 15, 20, and 30 are also factors. In fact, my entirely novel suggestion is that we use base-12 and base-60 for time-related purposes. Who's with me?!
Um ... don't you ever use the phases "half past ten" or "quarter to six"?
How can we continue to believe in a just universe and freedom to eat crackers if we have no ale?
You don't need fractions though? That's the point of metric?
A day is 100 centidays long.
Go to work at 37.5cd (375millidays) aka 9h... high noon at .5d / 5dd / 50cd. So take your half hour lunch ( 2 centiday lunch). Come back to the office at 52cd Leave work at 71cd (17h). supper at 75cd (18h). etc. The 11oclock news is now on at 958md. They'd probably make it a nice number like .95day. "Late news nine five".
Realistically these would end up being aligned with rounder numbers. Just like no one has an official shift start at 9:13.
1 centiday is roughly 15m (14.4m), so 4 centidays is almost an hour. So instead of meeting someone in an hour or 15m, you meet them in 4 cents, or a cent...
That all seems very weird but I'm sure it would be perfectly natural once adapted to, I would think. Just typing it up is a bit of a mindfuck, but I don't think people born into it would question it any more than they question our current time system.
The real problem is that there are 3600 seconds in a day, so you'd probably want to redefine a second as being 10 microdays (which is 864ms), which would break every other unit pretty much.
Leap seconds and that too. but yeah, that's what I've got right now.
Sent from my PDP-11