> the moon could add a leap back 2 seconds every century to reconcile with Earth
This seems like a system built to catastrophically break the moment it has to scale even a little bit.
Why not start with a solid foundation? A native lunar time based on facts on its ground. One would also reference UTC, the same way many people have multiple clocks. But there's no need to be doing major relativistic correction between two internally-consistent blobs on the fly over the entire edge.
This is a step back to the bad old days when every town had its own definition of noon. Communication and anything that required scheduling was a mess because there were too many local definitions of time.
Specifically: telegraph and railroads killed local time.
Until then, things happened far enough apart, and sufficiently loosely coupled (shipping or riders on horseback being the principle means of long-distance travel) that coordination at a finer grain than a few minutes (and quite often hours or days) was unnecessary.
Financial transactions are now precise to the microsecond, amongst other processes. Scientific / measurement precision can be far higher still.
Or take a step back and look at the larger picture - devise a universe time that we can all adhere to where ever we are. Why remain earth centric when we're making more and more steps away from earth?
Then how about using the pulses of a distant pulsar to measure time? That should get us through to the next few centuries at least.
The fine article mentions that Lunar atomic clocks could be placed in Lunar orbit. However, there are very few stable lunar orbits due to the numerous mascons near the surface. And station-keeping would affect the clocks' accuracy - not to mention that the tidal variations on the Earth side and far side will affect the clocks as well.
And any Lunar-specific solution would be unlikely to scale to Mars, Mars' moons, asteroids, and interplanetary spacecraft. Thus reinforcing the distant pulsars' signal as the optimum solution for the time being that humans remain in the solar system.
Because it's impossible. There's no such thing as a universal time; it's all relative to wherever you are at that moment. Earth is small enough that relativistic effects between any two points are insignificant, but the Moon is far enough away that it'll be noticeable, though not that much. Anything farther, however, and it all breaks down. Within the system, it'll probably be "close enough" to have an offset between different places. If we expand to different star systems, though, it'll be impossible because they're moving in different directions.
There can be no absolute time, because time is relative to the observer. And if the observer is near to a big mass or moving very fast time passes differently for them.
I suggest that we should use a universal date system, and a local clock system simultaneously. The computers can change between them. Other solutions (using a universal time for local things, or using local time but adding leap microseconds here and there) all seem bad to me.
You don't have to worry about scaling- Earth has one moon. And there are just eight planets in the solar system. It seems obvious that any off-earth colony governed by humans on Earth will have at least one time system that attempts to reference Earth time. There may also be a local time scale in sync with the rising or the setting of the sun. But that probably won't be the case on the moon, because a "day" is weeks long. So the humans will use a clock that references UTC. Perhaps a fixed offset from UTC if the Moon base is governed by a single country.
The hard science part is dealing with gravity wells and keeping stuff in sync. But the top level answer will be that clocks will follow UTC, and any communications systems between the Earth and the Moon will be built to tolerate 2.5 seconds of lag.
Good luck with that one. It sounds like one of these problems without any good solution. You know something is going to break because someone will screw up the conversion.
I feel this is a case where it'd be better to have something totally different. Like time written in letters rather than digits. So that there would be no way to mistake one for the other.
I work in finance, and it's always a lot easier to spot rescaling problems when dealing with jpy than any other currency. The fact that 100 yens ~ 1 usd makes any mistake glaringly obvious.
> Why build a timescale that ONLY works for the Moon? Then, we need atomic clocks on all spacecraft ... or a Martian timescale. Scalability is key here, and we are lucky as we can build a GALACTIC timescale through ... you guessed it .... pulsars! A pulsar-based-timescale using the timing of pulsars around the galaxy, so all locations and receivers can tap into it universally. This also has the benefit of not having to build GNSS on the far side of the Moon, protecting its pristine radio heritage for #astronomy. Here's some further reading: https://academic.oup.com/mnras/article/427/4/2780/971096?log...https://academic.oup.com/mnras/article/491/4/5951/5612203https://www.esa.int/kids/en/news/ESA_test_a_new_pulsar_clock
GNSS could just turn itself off on the far side of the moon? You could even use a mechanical timer to turn the whole satellite back on — Battlestar Galactica style low tech solutions.
Sufficiently significant, influential, and/or widely-adopted concepts can have very real impacts. Dismissal or belittling on this basis alone really adds nothing to this (or most other) discussions.
Surely this is a problem that has already been well and truly solved by the earth-based GPS systems. They are based on a collection of different satellites that are all in different positions with independent clocks that tick at slightly different rates, yet the overall outcome is a reasonably coherent notion of what the current time is. Couldn't the lunar positioning system be formed as an extension of that?
The funny thing is that GPS satellites operate in the context of special relativity, where events can occur in reverse order depending on the observer's velocity, so the whole notion of an "overall outcome" is pointless...
Events can't occur in a "reverse" order. At sufficient distances, the evolution of two regions is entirely causally independent, so there is no sense of simultaneity.
Otherwise, the causal order of events is always defined. Different observers may have "more information" about some events over others, but over ordinary (even multi-galactic) distances, events are well-ordered and sequential.
> Events can't occur in a "reverse" order. At sufficient distances, the evolution of two regions is entirely causally independent, so there is no sense of simultaneity.
This discussion is essentially about whether "events can occur in reverse order for different observers" is the right way to say "the order of time coordinates of two spacelike-separated events is reversed for different observers".
This was in reply to:
> the overall outcome is a reasonably coherent notion of what the current time is
The notion of an observer-independent "current time" implies simultaneity which doesn't exist in SR, and if you mistakenly assume an observer-independent "current time" which you'd likely assume to be the time axis of your own frame of reference, then two spacelike-separated events would be reversed for some observer who did the same but moves with almost speed-of-light relative to you.
Please keep in mind that I wanted to point out the irony in using GPS for a universal reference time without giving an intro into special relativity.
Nitpick: GPS is affected by general relativity effects (different gravitational field) rather than special relativity effects (where they would have to be traveling a substantial proportion of the speed of light).
GPS is affected by both general and special relativity, in opposite ways. The GR effect is larger, but if you don’t include the SR effect, you’ll get it wrong.
I've worked in the field of high precision surveying since the 1980s or so, ever since being intrigued by reverse engineering NAVSTAR signals prior to their form being fully public.
I absolutely agree that GPS is affected by both general and special relativity, however I do take issue with the close of your linked article.
Relativity is not just some abstract mathematical theory: understanding it is absolutely essential for our global navigation system to work properly!
I'll insist that while the corrections must be made for GPS to be useful, understanding the why and the model of the cause for those corrections isn't essential as is insisted.
Suppose we have no knowledge of relativity but a general understanding of trigonometry and from there surveying and point fixing against mountain tops and base stations; we extend that to doing the same against stations in motion (our satellites) and we find that we are in error by some unknown function.
We might posit that the cause is the effect of solar wind and the magnetic field against our satellites, we might guess that there's an unknown additional small force (or forces) that excites us to experiment further and expand our grasp of physics .. however ...
In the meantime we can generate accurate corrections by developing the error f(t,parameters) against a series of fixed ground points that should return the same fixed location but instead inform us of our error.
We can then solve for a Kalman filter that corrects in the near past and near future, and upload those corrections to our satellite fleet for inclusion in the broadcast data packets.
In fact, this is what is already done, despite our understanding of relativity as there are indeed other factors that cause errors.
In a similar manner aircraft performing geomagnetic surveying develop heading correction Kalman filters by flying figure eight patterns over quiet zones so that they can then fly patterns East-West | North-South and measure the reaction of the earth structure against the geomagnetic field in a manner independent of the aircraft heading and its own response (as a lump of metal moving through a field) (these being factored out by the correction filter).
To simplify what others have said, it's only possible to reorder two events A and B in Special Relativity if A can't possibly have caused B and vice-versa. Otherwise, SR maintains ordering. Look up timelike versus spacelike separation.
As a noob, I recently went down the rabbit hole of time-setting for astronomy and space applications. It's an absolutely fascinating topic.
Having the same time in two places which are very distant apart, are in different inertial frames, or are in any combination of inertial and non-inertial, well, it's very difficult. The rhythms of life, and special and general relativity get in the way.
But the need to convert from one time system to the other arises all the time.
For example, if you want to to know if somebody looking from Newcastle, Australia, will be able to have direct line of sight to the asteroid 253 Mathilde on the night of the 23rd of September of 2177, then you need to work with UTC (which accounts for Earth changes in rotation speed which are unpredictable) and with coordinate frames which are inertial to the solar system (to apply Kepler laws). So you not only need to keep track of unpredictable things like Earth's rotation altering the precision of time, but also the drift caused on clocks by relativity.
Note that problem here is that people live on Earth, and they want their time to reflect sunrises and sunsets, and they want it to track other natural cycles like the seasons. So we have UTC and UTC+n . Similarly, people living on the Moon may want to have a timesystem that reflects their sunraises and sunsets. But none of that matters to laws of physics that use a more fundamental notion of time, e.g. orbits.
JPL has somewhere a set of routines that convert between different time-systems, and some software packages like Mathematica also come with the functionality. It does make sense to add a new timesystem for each body with strong enough gravity where people is going to live.
Further down that rabbit hole, consider timekeeping and data aquisition for the SKA (Square Kilometre Array) project with radio telescopes in both Australia and South Africa.
Consider a common target of interest over 48 hours .. one side of the planet is rotating toward the target, the other side away from the target, both are sampling the same signal spectrum with varying doppler effects .. oh, and the planet is moving toward or away or at some angle in its orbit about the sun, so there is another relative motion.
Now, for better resolution, the same target is viewed again some six months later - earth is now the other side of the sun, the planets relative motion has changed, and locations in Australia and South Africa are now reversed in their motions toward and away.
This all poses some issues when data signal stacking.
if that is the idea then why not think forward and setup a clock system based e.g. on Lagrange points, away from the gravitational field peculiarities of this or that body?
In Andy Weir's /Artemis/, the Moon in on Kenyan time, because that's where all the rockets take off from. This is in turn because they had land on the equator and created something like a Shenzhen (low or no tariffs, friendly regulation for foreign capital, etc.) for space launches.
If the moon uses a different time than UTC, it will also be necessary to define the switchover procedure and a boundary for spacecraft from Earth, which presumably launch using a terrestrial clock.
The relativity problem is already solved. If it weren't GPS wouldn't work. It's somewhat complex, but it basically boils down to defining the Earth as the preferred reference frame (Earth Centered Inertial System, or ECI) and doing the necessary transforms[1].
The lunar "day" is actually nearly a solar month and it makes sense to call that a month.
However on a shorter timescale, it makes sense for the "day" to be the interval for one rotation of the earth relative to the moon. So at noon each day, the same point on the earth is facing the moon, say the Greenwich meridian. So this lunar day would be slightly longer than the earth solar day, roughly 30.53/29.53 earth days or 24 hrs 49 minutes approx. Would make sense to call such lunar days something different like "earths." So you would say something like "5 earths until the sun sets." Makes sense to also keep day hour minute and second the same as earth synced with UTC.
45 comments
[ 919 ms ] story [ 1179 ms ] threadAs the clocks very, very slowly drift apart due to relativity, the moon could add a leap back 2 seconds every century to reconcile with Earth.
Or add a leap back hour in 200,000 years.
This seems like a system built to catastrophically break the moment it has to scale even a little bit.
Why not start with a solid foundation? A native lunar time based on facts on its ground. One would also reference UTC, the same way many people have multiple clocks. But there's no need to be doing major relativistic correction between two internally-consistent blobs on the fly over the entire edge.
It worked until there were tens of thousands of towns. Even then, we didn’t port Greenwich time to Alabama; we aggregated the commonsense approach.
We could inhabit every planet and solid moon in the system and not come close to they breaking point.
Until then, things happened far enough apart, and sufficiently loosely coupled (shipping or riders on horseback being the principle means of long-distance travel) that coordination at a finer grain than a few minutes (and quite often hours or days) was unnecessary.
Financial transactions are now precise to the microsecond, amongst other processes. Scientific / measurement precision can be far higher still.
The fine article mentions that Lunar atomic clocks could be placed in Lunar orbit. However, there are very few stable lunar orbits due to the numerous mascons near the surface. And station-keeping would affect the clocks' accuracy - not to mention that the tidal variations on the Earth side and far side will affect the clocks as well.
And any Lunar-specific solution would be unlikely to scale to Mars, Mars' moons, asteroids, and interplanetary spacecraft. Thus reinforcing the distant pulsars' signal as the optimum solution for the time being that humans remain in the solar system.
There can be no absolute time, because time is relative to the observer. And if the observer is near to a big mass or moving very fast time passes differently for them.
The hard science part is dealing with gravity wells and keeping stuff in sync. But the top level answer will be that clocks will follow UTC, and any communications systems between the Earth and the Moon will be built to tolerate 2.5 seconds of lag.
I feel this is a case where it'd be better to have something totally different. Like time written in letters rather than digits. So that there would be no way to mistake one for the other.
I work in finance, and it's always a lot easier to spot rescaling problems when dealing with jpy than any other currency. The fact that 100 yens ~ 1 usd makes any mistake glaringly obvious.
https://mastodon.social/@CosmicRami@aus.social/1097534349342...
> Why build a timescale that ONLY works for the Moon? Then, we need atomic clocks on all spacecraft ... or a Martian timescale. Scalability is key here, and we are lucky as we can build a GALACTIC timescale through ... you guessed it .... pulsars! A pulsar-based-timescale using the timing of pulsars around the galaxy, so all locations and receivers can tap into it universally. This also has the benefit of not having to build GNSS on the far side of the Moon, protecting its pristine radio heritage for #astronomy. Here's some further reading: https://academic.oup.com/mnras/article/427/4/2780/971096?log... https://academic.oup.com/mnras/article/491/4/5951/5612203 https://www.esa.int/kids/en/news/ESA_test_a_new_pulsar_clock
Sufficiently significant, influential, and/or widely-adopted concepts can have very real impacts. Dismissal or belittling on this basis alone really adds nothing to this (or most other) discussions.
Otherwise, the causal order of events is always defined. Different observers may have "more information" about some events over others, but over ordinary (even multi-galactic) distances, events are well-ordered and sequential.
This discussion is essentially about whether "events can occur in reverse order for different observers" is the right way to say "the order of time coordinates of two spacelike-separated events is reversed for different observers".
This was in reply to: > the overall outcome is a reasonably coherent notion of what the current time is
The notion of an observer-independent "current time" implies simultaneity which doesn't exist in SR, and if you mistakenly assume an observer-independent "current time" which you'd likely assume to be the time axis of your own frame of reference, then two spacelike-separated events would be reversed for some observer who did the same but moves with almost speed-of-light relative to you.
Please keep in mind that I wanted to point out the irony in using GPS for a universal reference time without giving an intro into special relativity.
https://www.astronomy.ohio-state.edu/pogge.1/Ast162/Unit5/gp...
I absolutely agree that GPS is affected by both general and special relativity, however I do take issue with the close of your linked article.
I'll insist that while the corrections must be made for GPS to be useful, understanding the why and the model of the cause for those corrections isn't essential as is insisted.Suppose we have no knowledge of relativity but a general understanding of trigonometry and from there surveying and point fixing against mountain tops and base stations; we extend that to doing the same against stations in motion (our satellites) and we find that we are in error by some unknown function.
We might posit that the cause is the effect of solar wind and the magnetic field against our satellites, we might guess that there's an unknown additional small force (or forces) that excites us to experiment further and expand our grasp of physics .. however ...
In the meantime we can generate accurate corrections by developing the error f(t,parameters) against a series of fixed ground points that should return the same fixed location but instead inform us of our error.
We can then solve for a Kalman filter that corrects in the near past and near future, and upload those corrections to our satellite fleet for inclusion in the broadcast data packets.
In fact, this is what is already done, despite our understanding of relativity as there are indeed other factors that cause errors.
In a similar manner aircraft performing geomagnetic surveying develop heading correction Kalman filters by flying figure eight patterns over quiet zones so that they can then fly patterns East-West | North-South and measure the reaction of the earth structure against the geomagnetic field in a manner independent of the aircraft heading and its own response (as a lump of metal moving through a field) (these being factored out by the correction filter).
Having the same time in two places which are very distant apart, are in different inertial frames, or are in any combination of inertial and non-inertial, well, it's very difficult. The rhythms of life, and special and general relativity get in the way.
But the need to convert from one time system to the other arises all the time.
For example, if you want to to know if somebody looking from Newcastle, Australia, will be able to have direct line of sight to the asteroid 253 Mathilde on the night of the 23rd of September of 2177, then you need to work with UTC (which accounts for Earth changes in rotation speed which are unpredictable) and with coordinate frames which are inertial to the solar system (to apply Kepler laws). So you not only need to keep track of unpredictable things like Earth's rotation altering the precision of time, but also the drift caused on clocks by relativity.
Note that problem here is that people live on Earth, and they want their time to reflect sunrises and sunsets, and they want it to track other natural cycles like the seasons. So we have UTC and UTC+n . Similarly, people living on the Moon may want to have a timesystem that reflects their sunraises and sunsets. But none of that matters to laws of physics that use a more fundamental notion of time, e.g. orbits.
JPL has somewhere a set of routines that convert between different time-systems, and some software packages like Mathematica also come with the functionality. It does make sense to add a new timesystem for each body with strong enough gravity where people is going to live.
Consider a common target of interest over 48 hours .. one side of the planet is rotating toward the target, the other side away from the target, both are sampling the same signal spectrum with varying doppler effects .. oh, and the planet is moving toward or away or at some angle in its orbit about the sun, so there is another relative motion.
Now, for better resolution, the same target is viewed again some six months later - earth is now the other side of the sun, the planets relative motion has changed, and locations in Australia and South Africa are now reversed in their motions toward and away.
This all poses some issues when data signal stacking.
Something about the Year of the Rabbit . . .
IIRC, that accounting will stop if leap seconds are eliminated.
if that is the idea then why not think forward and setup a clock system based e.g. on Lagrange points, away from the gravitational field peculiarities of this or that body?
[1] https://apps.dtic.mil/sti/pdfs/ADA516975.pdf