Re: [tz] New Yorker article on David Mills and NTP
I follow numerous astronomy lists, and there is serious discussion at the IAU (International Astronomical Union) to get rid of leap seconds altogether, and possibly replace them by something else, perhaps leap minutes (or even leap hours) - making them far less common. I believe they expect to make a decision soon, though they have been discussing it for years and had planned to make a decision 2 or 3 years ago. In this case we would add or drop a minute (or hour), when UT1 and UTC were more than a half-minute (or half-hour) out of sync. The concept with hours is that it would work more like a day-light savings transition, and named UTC1, UTC2, etc. Had this been done originally, there would have been no leap-second transitions up to now at all, in fact. ----- Original Message ----- From: Fred Gleason via tz [mailto:tz@iana.org] To: "Time Zone Database discussion" <tz@iana.org> Sent: Tue, 4 Oct 2022 13:05:49 -0400 Subject: Re: [tz] New Yorker article on David Mills and NTP On Oct 4, 2022, at 12:07, Steve Allen via tz <tz@iana.org> wrote: The implementation of time in computing systems is fragile because itis based on an unrealistic notion that clocks are always right andnever need to be reset. I suspect that one of the primary reasons that such time resets have been so fraught (and hence bitterly opposed by many ‘operations’ type folks) is their relative rarity. There have been only 27 leap seconds introduced since 1972, which means that deployed systems rarely have to cope with one and thus get the inevitable bugs identified and shaken out. It would almost be beneficial if such an event would occur at least once per annum, to allow the ‘reset’ logic to be regularly exercised. Cheers! |---------------------------------------------------------------------| | Frederick F. Gleason, Jr. | Chief Developer | | | Paravel Systems | |---------------------------------------------------------------------| | There cannot be a crisis next week. My schedule is already full. | | | | -- Henry Kissinger | |---------------------------------------------------------------------|
On 10/4/22 13:01:31, dpatte via tz wrote:
I follow numerous astronomy lists, and there is serious discussion at the IAU (International Astronomical Union) to get rid of leap seconds altogether, and possibly replace them by something else, perhaps leap minutes (or even leap hours) - making them far less common. I believe they expect to make a decision soon, though they have been discussing it for years and had planned to make a decision 2 or 3 years ago.
In this case we would add or drop a minute (or hour), when UT1 and UTC were more than a half-minute (or half-hour) out of sync. The concept with hours is that it would work more like a day-light savings transition, and named UTC1, UTC2, etc.
Had this been done originally, there would have been no leap-second transitions up to now at all, in fact.
With a "nearest minute" convention for leap minutes, there would have been a leap minute in 1997, when TAI-UT1 exceeded 30 seconds. I'd vote for UT1 for civil time and IT, and TAI otherwise. Some IT organizations have adopted leap smearing. For information on a UT1 NTP server, see: <https://www.nist.gov/pml/time-and-frequency-division/time-services/ut1-ntp-t...>. -- gil
On Oct 4, 2022, at 12:55 PM, Paul Gilmartin via tz <tz@iana.org> wrote:
I'd vote for UT1 for civil time and IT, and TAI otherwise. Some IT organizations have adopted leap smearing.
If "IT" means "information technology", it's a big field, and not all uses of time in it are the same. Time sindications entered by or shown to users are presumably in hours/minutes/seconds form, possibly with a date in year/month/day form. For those, there is a choice between UT1, UTC and TAI, all of which, as far as I know, represent time stamps in that form. In most cases, a system would presumably show, and expect, something in civil time form, although there might be some computing equipment accepting and displaying TAI. However, there are places in IT where neither UT1, nor UTC, nor TAI, all of which, as far as I know, represent time stamps in forms such as YYYY-MM-DD HH:MM:SS{.fraction}, are used. Instead, a count of ticks since some particular instant in the past is used, where, for different representations, different things are done at and near the time a leap second occurs, including "do nothing different from any other second". Those time representations may be used to show civil time (local or UT of some sort) or to show TAI. They might *also* be used to calculate the time that elapsed between two events, as a (possibly non-integral) number of SI seconds.
On 2022-10-04 19:01, dpatte via tz wrote:
I follow numerous astronomy lists, and there is serious discussion at the IAU (International Astronomical Union) to get rid of leap seconds altogether, and possibly replace them by something else, perhaps leap minutes (or even leap hours) - making them far less common. I believe they expect to make a decision soon, though they have been discussing it for years and had planned to make a decision 2 or 3 years ago.
The BIPM is now in charge, and the CGPM on 2022-11-15..18 in Paris will vote on the proposal D on pages 7..8 and 23..24 in [https://www.bipm.org/documents/20126/66742098/Draft-Resolutions-2022.pdf/2e8...] Michael Deckers.
On Tue, 2022-10-04 at 20:15 +0000, Michael H Deckers via tz wrote:
On 2022-10-04 19:01, dpatte via tz wrote:
I follow numerous astronomy lists, and there is serious discussion at the IAU (International Astronomical Union) to get rid of leap seconds altogether, and possibly replace them by something else, perhaps leap minutes (or even leap hours) - making them far less common. I believe they expect to make a decision soon, though they have been discussing it for years and had planned to make a decision 2 or 3 years ago.
The BIPM is now in charge, and the CGPM on 2022-11-15..18 in Paris will vote on the proposal D on pages 7..8 and 23..24 in [ https://www.bipm.org/documents/20126/66742098/Draft-Resolutions-2022.p df/2e8e53df-7a14-3fc8-8a04-42dd47df1a04]
Michael Deckers.
The essence of proposal D is: recognizing that the use of UTC as the unique reference time scale for all applications, including advanced digital networks and satellite systems, calls for its clear and unambiguous specification as a continuous time scale, with a well-understood traceability chain, decides that the maximum value for the difference (UT1-UTC) will be increased in, or before, 2035, requests that the CIPM consult with the ITU, and other organizations that may be impacted by this decision in order to − propose a new maximum value for the difference (UT1-UTC) that will ensure the continuity of UTC for at least a century, − prepare a plan to implement by, or before, 2035 the proposed new maximum value for the difference (UT1-UTC), − propose a time period for the review by the CGPM of the new maximum value following its implementation, so that it can maintain control on the applicability and acceptability of the value implemented, − draft a resolution including these proposals for agreement at the 28th meeting of the CGPM (2026), encourages the BIPM to work with relevant organizations to identify the need for updates in the different services that disseminate the value of the difference (UT1-UTC) and to ensure the correct understanding and use of the new maximum value Considering that there has been about one leap second every 18 months, the maximum value for abs(UT1-UTC) would have to be around 60, so this proposal is contemplating leap minutes. Assuming agreement on that value is reached by 2026, there would then be nine years to update everything that depends on abs(UT1-UTC) < 1. One example that comes quickly to mind is the time signals broadcast by WWV, which include the value of UT1-UTC to the nearest 1/10 second, but doesn't have enough space for abs(UT1-UTC) > 1. See https://www.nist.gov/time-distribution/radio-station-wwv/wwv-and-wwvh-digita... I think it would take much longer than nine years to update the WWV transmission format, since that requires updating all consumers of this data. John Sauter (John_Sauter@systemeyescomputerstore.com) -- get my PGP public key with gpg --locate-external-keys John_Sauter@systemeyescomputerstore.com
On 10/4/22 13:15, Michael H Deckers via tz wrote:
in Paris will vote on the proposal D on pages 7..8 and 23..24 in
For those who haven't seen it, this proposal would abolish leap seconds. Although it officially proposes only that we not worry about |UT1-UTC| for at least 100 years (i.e., that we kick the can down the road to our great-great-great-grandchildren), our descendants will almost surely kick the can down the road to *their* descendants, which means leap seconds will be abolished indefinitely. Although this obviously won't work in the long term, leap seconds wouldn't work either because days will become too long for them too. Predicting length of day (LOD) is a tricky business with no real consensus. That being said, Tyler predicted that for the next 3 billion years LOD will increase roughly linearly, to 30 hours. So assuming our timekeeping civilization lasts long enough, eventually we'll need something, even if it's not leap seconds. -- Tyler RH. On the tidal history and future of the Earth–Moon orbital system. Planet Sci J. 2021;2(2):70. https://doi.org/10.3847/PSJ/abe53f
On Tue 2022-10-04T14:20:46-0700 Paul Eggert via tz hath writ:
Predicting length of day (LOD) is a tricky business with no real consensus. That being said, Tyler predicted that for the next 3 billion years LOD will increase roughly linearly, to 30 hours. So assuming our timekeeping civilization lasts long enough, eventually we'll need something, even if it's not leap seconds.
A decade ago two state of the art atomic clocks would agree with other within 1 second after 300 million years. In 300 million years a calendar based on purely atomic time will have counted 3 billion more days than have been witnessed by people living on the surface of Earth. -- Steve Allen <sla@ucolick.org> WGS-84 (GPS) UCO/Lick Observatory--ISB 260 Natural Sciences II, Room 165 Lat +36.99855 1156 High Street Voice: +1 831 459 3046 Lng -122.06015 Santa Cruz, CA 95064 https://www.ucolick.org/~sla/ Hgt +250 m
On 10/4/22 15:24:49, Steve Allen via tz wrote:
A decade ago two state of the art atomic clocks would agree with other within 1 second after 300 million years. In 300 million years a calendar based on purely atomic time will have counted 3 billion more days than have been witnessed by people living on the surface of Earth.
On such time scales and much sooner, current astronomical data bases will have been obsoleted by chaos in: o ΔT o n-body celestial mechanics o precession of the equinoxes o stellar proper motions. Necessary empirical corrections will be made. Smearing has proven less disruptive to computer systems than leap seconds. I still favor UT1 as a generalization of smearing. Are pulsar frequencies measured with TAI or UTC? When Standard Time was established in the 19th Century, the width of a time zone was about a day's travel distance. This is obsolete. People cling desperately to the alignment of clocks with solar time. Consider Uyghur civil disobedience in Xinjiang. And yet, historically a six-hour adjustment was made abruptly, since: <https://www.biblegateway.com/passage/?search=Matthew%2020%3A1-16&version=KJV>.
-- Steve Allen <sla@ucolick.org> WGS-84 (GPS) UCO/Lick Observatory--ISB 260 Natural Sciences II, Room 165 Lat +36.99855 1156 High Street Voice: +1 831 459 3046 Lng -122.06015 Santa Cruz, CA 95064 https://www.ucolick.org/~sla/ Hgt +250 m
-- gil
On 2022-10-05 06:02, Paul Gilmartin via tz wrote:
Smearing has proven less disruptive to computer systems than leap seconds. I still favor UT1 as a generalization of smearing.
Unfortunately the precision of UT1 is only about ms, limited by its predictability: too low for current commercial requirements, and insufficently accurate for engineering uses. https://www.nist.gov/pml/time-and-frequency-division/time-services/ut1-ntp-t...
Are pulsar frequencies measured with TAI or UTC?
All three use SI seconds currently.
When Standard Time was established in the 19th Century, the width of a time zone was about a day's travel distance.
More like a week for most. An hourly zone is about 1600km/1000mi or about 40hr travel at 40km/hr 25mi/hr (which is still about the average NA freight train speed, due to length and weight of loads, over variable quality trackbed, and lack of isolated or dedicated routes). The Pony Express took 10 days to go about 3000km 1900mi - nearly two zones. It would nowadays take over a day by NA long distance rail passenger service: Toronto-Vancouver express takes 96hr for about 4400km 2700mi - 45km/hr 28mi/hr.
People cling desperately to the alignment of clocks with solar time.
Most legal systems are still based on solar time, used as civil time: why leap seconds are required if you base time keeping on caesium atomic microwave emission frequency derived seconds. Many countries wish to avoid the legal wrangling, duration, and costs of rewriting a lot of legislation, and then political wrangling to try to pass it. You may imagine the irrational bases of many arguments! ;^> Previous posters have mentioned countries likely to wield veto power over any changes to the status quo like dropping it. -- Take care. Thanks, Brian Inglis, Calgary, Alberta, Canada This email may be disturbing to some readers as it contains too much technical detail. Reader discretion is advised. [Data in binary units and prefixes, physical quantities in SI.]
Paul wrote:
On 10/4/22 13:15, Michael H Deckers via tz wrote:
in Paris will vote on the proposal D on pages 7..8 and 23..24
For those who haven't seen it, this proposal would abolish leap seconds... Although this obviously won't work in the long term, leap seconds wouldn't work either because days will become too long for them too.
As I understand it, one way of looking at the new, leap-secondless plan is that if you wait long enough, the multi-minute jump that you then need isn't a "leap" anything, it's a political change to the definition of a time zone, readily handled by the well-tried mechanism that we on this list all know and love.
On 10/4/22 19:51, Steve Summit via tz wrote:
if you wait long enough, the multi-minute jump that you then need isn't a "leap" anything, it's a political change to the definition of a time zone, readily handled by the well-tried mechanism that we on this list all know and love.
This should be fun once we start changing time zones to to be more than 24 hours away from where they are now. Espenak & Meeus 2006[1] predict that this will be around the year 7000. It will be like "Around the World in Eighty Days", but in slow motion and involving the whole planet rather than just Phileas Fogg and Passepartout. [1] according to RH van Gent's ΔT Calculator <https://webspace.science.uu.nl/~gent0113/deltat/deltat_main.htm>.
On Tue 2022-10-04T21:11:51-0700 Paul Eggert via tz hath writ:
This should be fun once we start changing time zones to to be more than 24 hours away from where they are now. Espenak & Meeus 2006[1] predict that this will be around the year 7000. It will be like "Around the World in Eighty Days", but in slow motion and involving the whole planet rather than just Phileas Fogg and Passepartout.
[1] according to RH van Gent's ΔT Calculator <https://webspace.science.uu.nl/~gent0113/deltat/deltat_main.htm>.
Morrison, Stephenson, Hohenkerk, Zawilski Addendum 2020 to 'Measurement of the Earth's rotation: 720 BC to AD 2015' https://royalsocietypublishing.org/doi/abs/10.1098/rspa.2020.0776 their extrapolation (which they will tell you not to believe) also puts Delta T of 24 h a little after year 7000 -- Steve Allen <sla@ucolick.org> WGS-84 (GPS) UCO/Lick Observatory--ISB 260 Natural Sciences II, Room 165 Lat +36.99855 1156 High Street Voice: +1 831 459 3046 Lng -122.06015 Santa Cruz, CA 95064 https://www.ucolick.org/~sla/ Hgt +250 m
Paul Eggert via tz <tz@iana.org> writes:
On 10/4/22 19:51, Steve Summit via tz wrote:
if you wait long enough, the multi-minute jump that you then need isn't a "leap" anything, it's a political change to the definition of a time zone, readily handled by the well-tried mechanism that we on this list all know and love.
This should be fun once we start changing time zones to to be more than 24 hours away from where they are now.
Everyone on this list will be safely dead before any of that matters, but ... how would that work really? I think most people understand time zones as "offsets from UTC". We know how to cope with changes in those offsets, having done it many times before. But it's not clear to me how tzdb could handle step changes in UTC itself in the same way. That looks morally equivalent to leap seconds, which is exactly the under-tested case we're bemoaning in this thread. regards, tom lane
Tom Lane via tz said:
On 10/4/22 19:51, Steve Summit via tz wrote:
if you wait long enough, the multi-minute jump that you then need isn't a "leap" anything, it's a political change to the definition of a time zone, readily handled by the well-tried mechanism that we on this list all know and love.
This should be fun once we start changing time zones to to be more than 24 hours away from where they are now.
Everyone on this list will be safely dead before any of that matters, but ... how would that work really? I think most people understand time zones as "offsets from UTC". We know how to cope with changes in those offsets, having done it many times before. But it's not clear to me how tzdb could handle step changes in UTC itself in the same way.
No, you don't do that. You leave UTC alone and each country shifts its time zone by an hour relative to UTC. For example, by omitting an autumn shift (if I've got the sign right) or, if it doesn't use the despised bi-annual shifts, by having a one-off shift. Eventually you'll have places on zone UTC+30 or UTC+51. -- Clive D.W. Feather | If you lie to the compiler, Email: clive@davros.org | it will get its revenge. Web: http://www.davros.org | - Henry Spencer Mobile: +44 7973 377646
On Tue 2022-10-04T22:51:35-0400 Steve Summit via tz hath writ:
As I understand it, one way of looking at the new, leap-secondless plan is that if you wait long enough, the multi-minute jump that you then need isn't a "leap" anything, it's a political change to the definition of a time zone, readily handled by the well-tried mechanism that we on this list all know and love.
Puttig the onus onto legislators is perhaps payback for the fact that the inception of the leap second was prompted when Germany passed a law that made mean solar seconds illegal. With the telecom technology of 1970 the only viable option for a single time scale that could be broadcast everywhere as legal time was TAI offset by leap seconds. This was adopted despite a report from astronomers which pointed out that the leaps of time would wreak havoc for automated systems when such systems would become commonplace. -- Steve Allen <sla@ucolick.org> WGS-84 (GPS) UCO/Lick Observatory--ISB 260 Natural Sciences II, Room 165 Lat +36.99855 1156 High Street Voice: +1 831 459 3046 Lng -122.06015 Santa Cruz, CA 95064 https://www.ucolick.org/~sla/ Hgt +250 m
On 2022-10-04 5:20 PM, Paul Eggert via tz wrote:
Predicting length of day (LOD) is a tricky business with no real consensus.
I think that statement is a little misleading. While there is no specification for, that is, there no "official product" of, C04, it is internally used by both the USNO, as the timescale maintained, distributed to, and used by GPS, and the IERS, to determine values published in Bulletin A and to determine leap-second values as published in Bulletin C. From what I can tell C04 is *the* 'consensus' of LOD as a matter of practical common use. # EARTH ORIENTATION PARAMETER (EOP) PRODUCT CENTER CENTER (PARIS OBSERVATORY) - INTERNATIONAL EARTH ROTATION AND REFERENCE SYSTEMS SERVICE # EOP (IERS) 20 C04 TIME SERIES consistent with ITRF 2020 - sampled at 0h UTC https://hpiers.obspm.fr/iers/eop/eopc04_20/eopc04.1962-now (This is the version distributed by the IERS. (There are other links at USNO and NIST, and possibly others, to the same information) It is indeed very tricky business, but the good folks at the many contributing observatories and time laboratories together with many data processing centers have made landing an airplane within a few inches possible. How this is accomplished is a topic of a very long essay. -Brooks
On 10/6/22 13:40, Brooks Harris via tz wrote:
Predicting length of day (LOD) is a tricky business with no real consensus. ... From what I can tell C04 is *the* 'consensus' of LOD as a matter of practical common use.
Sure, but C04 describes LOD in the past. I was referring to LOD in the future. C04 is published with 30-day latency which is not enough for some apps, and I think even the IERS Daily Rapid EOP data have one-day latency. So even if you want to know LOD "now" you need to predict based on what it was earlier; and of course if you want to know LOD next week you must predict. Michalczak and Ligas[1] have a brief summary of some competing prediction approaches in their introduction. If you want to predict LOD for the next five millennia, which is what I had been talking about, the error bars get considerably larger and it is indeed a tricky business. [1] Michalczak M, Ligas M. The (ultra) short term prediction of length-of-day using kriging. Adv Space Res. 2022;70(3):610-620. https://doi.org/10.1016/j.asr.2022.05.007
On 2022-10-04 4:15 PM, Michael H Deckers via tz wrote:
On 2022-10-04 19:01, dpatte via tz wrote:
I follow numerous astronomy lists, and there is serious discussion at the IAU (International Astronomical Union) to get rid of leap seconds altogether, and possibly replace them by something else, perhaps leap minutes (or even leap hours) - making them far less common. I believe they expect to make a decision soon, though they have been discussing it for years and had planned to make a decision 2 or 3 years ago.
The BIPM is now in charge,
I think this may be a matter of debate or controversy. UTC as currently specified is controlled by ITU-R. Note the line in Draft Resolution D on page 24 "requests that the CIPM consult with the ITU, and other organizations that may be impacted by this decision in order to ..." and "encourages the BIPM to work with relevant organizations to identify the need for updates in the different services that disseminate the value of the difference (UT1-UTC) and to ensure the correct understanding and use of the new maximum value."
and the CGPM on 2022-11-15..18 in Paris will vote on the proposal D on pages 7..8 and 23..24 in [https://www.bipm.org/documents/20126/66742098/Draft-Resolutions-2022.pdf/2e8...]
Note the line in Draft Resolution D on page 24: "decides that the maximum value for the difference (UT1-UTC) will be increased in, or before, 2035," As I understand this there is a country with veto power that will not accept changes to UTC until at least 2035. I believe this is why the 2035 date is mentioned. Several other countries remain opposed to the change. I think this suggests UTC with leap-seconds will be with us for a very long time. -Brooks
Michael Deckers.
On 2022-10-06 20:53, Brooks Harris via tz wrote:
I think this [who is "in charge of UTC"] may be a matter of debate or controversy. UTC as currently specified is controlled by ITU-R.
Well, TAI is produced by the BIPM, UTC - TAI is decided by the IERS, and the 26th CGPM have approved a definition of UTC (which implicates that it is the business of the BIPM to give the definition). The ITU decide how UTC is used in their standards, and in particular, how UTC is disseminated with radio signals; they do not (re)define UTC. There is a Memorandum of Understanding between the BIPM and the ITU that regulates all this (and promises cooperation) in legalese: [https://www.bipm.org/documents/20126/42177518/BIPM-ITU+MoU.pdf/d5ae968b-28ba...] Michael Deckers.
On Tue, 4 Oct 2022 at 13:06, Fred Gleason via tz <tz@iana.org> wrote:
It would almost be beneficial if such an event would occur at least once per annum, to allow the ‘reset’ logic to be regularly exercised.
…which is certainly a valid argument *against* widening the acceptable delta beyond 1 second and kicking the can down to future generations whose procedures would be less practiced and polished for *any* such change, let alone one of larger magnitude. At least until such time as our current procedures start to fall short and we're going through these motions a few times every year, at which point it would then become better to level-set and widen the acceptable gap by some factor around 5 or 10 so we're back to "every few years" again. There's undoubtedly a "sweet spot" between impact and frequency, and so those of such a view might well argue we're already in it (the last few wobbles of the earth notwithstanding). It seems it would be difficult to maintain what balance currently exists by throwing one or more factors of 60 into the mix. On Tue, 4 Oct 2022 at 15:01, dpatte via tz <tz@iana.org> wrote:
The concept with hours is that it would work more like a day-light savings transition, and named UTC1, UTC2, etc.
Those on this list understand well just how wild (and political!) DST changes on the order of an hour have gotten in places within a single century or even individual years. Imagine being the ones in charge of getting everyone on board to renumber the hours after thousands of individual local customs spent N millennia gradually adapting throughout the generations! Despite technical shortcomings with each change, there are certainly social benefits to keeping these increments small and unnoticeable to most… So, yes, of course we'll need major overhauls or something entirely different eventually. In the meantime, refinement is a worthwhile goal. But it also seems wise to reflect on the fact that, at 50-ish years, the whole practice is just *so* young on the timescales that would really seem to matter here. -- Tim Parenti
participants (13)
-
Brian Inglis -
Brooks Harris -
Clive D.W. Feather -
dpatte -
Guy Harris -
John Sauter -
Michael H Deckers -
Paul Eggert -
Paul Gilmartin -
scs＠eskimo.com -
Steve Allen -
Tim Parenti -
Tom Lane