In defense of calendrical irregularity

I’ve been getting deeper into timekeeping and calendar-related software the last few years. Besides my work on GPSD, I’m now the tech lead of NTPsec. Accordingly, I have learned a great deal about time mensuration and the many odd problems that beset calendricists. I could tell you more about the flakiness of timezones, leap seconds, and the error budget of UTC than you probably want to know.

Paradoxically, I find that studying the glitches in the system (some of which are quite maddening from a software engineer’s point of view) has left me more opposed to efforts to simplify them out of existence. I am against, as a major example, the efforts to abolish leap seconds.

My reason is that studying this mess has made me more aware than I used to be of the actual function of civil timekeeping. It is to allow humans to have consistent and useful intuitions about how clock time relates to the solar day, and in particular to how human circadian rhythms are entrained by the solar day. Secondarily to maintain knowledge of how human biological rhythms connect to the seasonal round (a weaker effect but not a trivial one).

Yes, in theory we could abolish calendars and timestamp everything by atomic-clock kiloseconds since an epoch. And if there ever comes a day when we all live in completely controlled environments like space habs or dome colonies that might actually work for us.

Until then, the trouble with that sort of computer-optimized timestamp is that while it tells us what time it is, it doesn’t tell us what kind of time it is – how the time relates to human living. Day? Night? Season?

Those sideband meanings are an important component of how humans use and interpret time references. Yes, I know January in Australia doesn’t mean the same thing as January in the U.S. – the point is that people in both places have stable intuitions about what the weather will be like then, what sorts of holidays will be celebrated, what kind of mood is prevalent.

I judge that all the crap I go though reconciling scientific absolute time to human-centered solar time is worth it. Because when all is said and done, clocks and calendars are human instruments to serve human needs. We should be glad when they add texture and meaning to human life, and beware lest in our attempts to make software easier to write we inadvertently bulldoze away entire structures of delicate meaning.

UPDATE: There is one context, however, in which I would cheerfully junk timezones. I think timestamps on things like file modifications and version-control commits should always be kept, and represented, in UTC, and I’m a big fan of RFC3339 format as the way to do that.

The reason I say this is that these times almost never have a human-body-clock meaning, while on the other hand it is often useful to be able to compare them unambiguously across timezones. Their usage pattern is more like scientific than civil time.

136 comments

  1. An interesting application of Chesterton’s Gate. Something that is applied far too rarely.

    Personally, I would like to see *some* simplification. 13 months of 28 days each, plus one or two “World Peace Days” or some such at the turning of the year to round it out. 10 hour days of 100 minutes of 100 seconds. Everyone using GMT/UTC instead of time zones… For me, it really doesn’t matter what time the Sun comes up and goes down, as long as I know that it’ll come at relatively the same time every day.

    Still, I can see your point – the Sun coming up at 4am in Scotland always strikes me as odd, but as a reminder of just how far *north* Scotland really is.

  2. I tend to agree. For another example, I was long in favor of using UTC everywhere and abolishing time zones, but that just means we need a new solution to a problem that doesn’t abolish: people in very different longitudes still have to be able to work together. Sure, we can be sure that we both agree when 1600 UTC is even if I’m in the US and you’re in Hong Kong, but anyone planning a meeting for that time still needs to know somehow that it’s lunchtime for me and midnight for you. And we still have UTC as a common reference to make sure we know what we mean, even if we don’t have our wristwatches set to it (though I know some folks who do).

    1. >I tend to agree. For another example, I was long in favor of using UTC everywhere and abolishing time zones,

      There is one context in which I still am. Timestamps on things like tile modifications and version-control commits. I think these should alwaya be kept, and represented, in UTC.

      The reason I say this is that these times almost never have a human-body-clock meaning, while on the other hand it is often useful to be able to compare them unambiguously across timezones. Their usage pattern is more like scientific than civil time.

  3. Much of this work’s been done already:

    infiniteundo.com/post/25326999628/falsehoods-programmers-believe-about-time

  4. Tara, redefining the second would overturn pretty much every other unit of measurement. The day is going to have 86400 seconds in it, unless you redefine the second to be 86.4% of its present length (as your proposal would do), and that in turn changes how many, many other units are defined. The effects on circadian rhythm understanding and human intuition would pale by comparison.

  5. @Jay:

    I myself would keep the second defined as-is, but stop using it for anything but scientific timekeeping. For civil timekeeping I’d rescale hours and minutes as Tara proposes, then give 1/100th of a minute a name that disambiguates it from the second and define it as an angle of planetary rotation with respect to the Sun, not as a unit of time, as that’s what matters for civil time.

  6. The problem with abolishing seconds from civil timekeeping is that we need both time and duration, both local and absolute time, and it would be much harder to get rid of second as duration. Also there is a place for overlap between civil and scientific time measure. .. and computers are included in it.

  7. I myself would keep the second defined as-is, but stop using it for anything but scientific timekeeping. For civil timekeeping I’d rescale hours and minutes as Tara proposes, then give 1/100th of a minute a name that disambiguates it from the second and define it as an angle of planetary rotation with respect to the Sun, not as a unit of time, as that’s what matters for civil time.

    But … why?

    To “simplify”?

    While leaving it un-simplified for scientific use, where simplicity is so profitable?

    I mean, apart from when doing technical work (programming around date/time handling) nobody ever worries about how many seconds there are in an hour; there’s no need to have metric time – it’s not making anything people actually do harder, is it*?

    (And none of the manifold programming tasks I’ve done around time manipulation would be any easier in the 10/100/100 scheme than the 24/60/60 one, since I am sane and use standardized time manipulation functions to do my time manipulation and comparison.)

    (If it’s defined as angles of rotation, does that mean the not-second gets … longer, as the planet’s spin slows? At least vibrations of whatsis in given conditions are consistent.)

    (* I mean, even simpler would be undivided days, with instead of 10 hours and 100 minutes and 100 not-seconds in each, merely division into milli-days and micro-days. That’s proper SI “simplification”.

    But nobody even suggests that, because it’s even less tenable than HH:MM:SS division – and acknowledging the human utility of time measurement removes almost all the impetus behind a simplification.

    Who needs it?)

  8. “and beware lest in our attempts to […] we inadvertently bulldoze away entire structures of delicate meaning.”

    This is a brilliant warning that applies across many fields of human endeavor, as I am sure you are well aware. Thanks for the thoughtful essay, as always.

  9. Yes, as Sigivald suggests, what problem are you trying to solve by redefining seconds or naming a new scale for civil timekeeping? Those looking to remake the calendar or the clock are proposing a solution in search of a problem.

  10. “We should be glad when they add texture and meaning to human life, and beware lest in our attempts to make software easier to write we inadvertently bulldoze away entire structures of delicate meaning.”

    This has a neat Chesterton’s Fence vibe to it.

  11. The only reform I see as possibly worth doing is one to sync dates with days of the week from year to year. This would involve picking one day each year (two days, in leap years) that don’t belong to any day of the week. E.g. December 30 is always a Saturday, January 1 is always a Sunday, and December 31 is a holiday that doesn’t belong to any day of the week. Then do the same for February 29, or else expand the end-of-year holiday to two days on leap years.

    But even this much I see as only possibly worth doing.

  12. My internal timekeeping is coffee, food, and sex, indexed by the number of songs played in between. Aside from that I don’t notice much about the time in the first place.

  13. I might be more sympathetic to these calendrical warts if circadian synchronisation weren’t already the bane of my life. I already can’t fit in to the 24-hour cycle everyone around me expects; shimming the day with leap seconds to maintain the equation of time gains me nothing.
    Besides, if we abolished leap seconds, the equation would only change by about a minute in one’s lifetime; if words can change their meaning over that kind of timespan, I’m sure the meaning of “9AM” can shift to “just after lunch” in the millennia it would take its referent to make the equivalent shift.

    So: keep calendars, keep February the 29th, keep sexagesimal division – but dump the stupid leap second. It’s worthless.

  14. Instead of moving to a 10/100/100 time scheme, how about moving to duodecimal numbers? It brings things such as degrees and time in line, while yielding numbers with many factors.

  15. As has already been pointed out, leap seconds *specifically* shift the clock slower than cultures can accomodate it, not faster.

    Everything else is fine, but leap seconds, and only leap seconds, are not worth it. Calendar reform could happen, has happened in history, but doesn’t really need to happen.

  16. What about Daylight Savings Time? Where do you think it falls on the human-body-clock spectrum? I mean, some places don’t use it, and they seem to get along perfectly well.

    To me, if you’re looking at clock reform/simplification, Daylight Savings Time seems to be an easier target, rather than changing the days in a month or whatever.

    1. >To me, if you’re looking at clock reform/simplification, Daylight Savings Time seems to be an easier target, rather than changing the days in a month or whatever.

      I agree. The software-engineering pain from DST is large, and the benefits elsewhere comparatively small.

      The real problem with DST isn’t the basic concept, it’s that the offset rules are so variable by location and political whim. It’s like the problem with timezones, only much worse.

      I think DST is an experiment that has failed, and would support scrapping it.

      1. I incline to agreement, but note that the EU has more-or-less solved the offset rules problem within its borders.

        If you’re an EU member, you can pick whatever timezone you like, but you must have DST and you must change on the standard dates chosen by the EU.

        The only exceptions is for extra-European territories south of the Tropic of Cancer, which have the option of skipping DST, or (if south of Capricorn) adopting it in the reverse months – but they still have to adopt the same dates.

        There’s also a minimum notice period for any changes, which makes programmers’ lives much easier.

        I’d still rather not bother, but at least this is a system that removes most of the worst problems.

  17. Agreed–we need time coupled to the seasons and year on Earth, at least while we’re still here. OTOH, do we have to be stuck with honoring Roman emperors? There’s got to be a way to make the calendar more regular but still tied to the seasons.

  18. I second the DST removal. The disruption it causes as people have to re-evaluate “wait, it’s suddenly dark before dinner” and “I lost an hour of sleep”, combines with the mishmash of DST rules in place as it is, are just not worth it. They also break the expectations we have. The seasonal changes in daylight hours are slow enough we can acclimatize.

  19. “OTOH, do we have to be stuck with honoring Roman emperors? There’s got to be a way to make the calendar more regular but still tied to the seasons.”

    Leave the emperors alone! Can you imagine the politically correct clusterf*ck that will ensue when people try to settle on who to name those mid-year months after? Worse, can you imagine some kid in the 25th century asking, “Who was that ‘Trump’ guy they named the month for?” July and August are sleeping dogs. Don’t start poking them with a stick.

    1. >July and August are sleeping dogs. Don’t start poking them with a stick.

      /me wants to plus this about a thousand.

  20. I used to be in favor of abolishing leap seconds for safety reasons: that the complexity they add to timekeeping code creates the potential for very dangerous bugs. Remember the Linux kernel bug from a few years ago that was causing Java applications to hang if they acquired a futex during the leap second? Imagine if that bug had been slightly worse and cause every Linux kernel around the world to panic simultaneously.

    However, having spent the past few months thinking deeply about the problem in the context of NTP, I’ve moderated this position slightly, because I’ve realized that the inherent complexity of leap seconds isn’t bad at all. The problem is mostly that the POSIX time API sucks (and so does Windows’, and so does almost every language’s standard library, with the exception of Haskell’s). Most applications want one of two different kinds of time: they either want an absolute UTC clock, for which timestamps can be succinctly expressed as whole calendar days since a given epoch plus SI[1] seconds since midnight of the current day, or they want a difference clock, which just counts seconds without caring what epoch the count is relative to. The problem with POSIX is that it provides a single clock[2] which serves both use cases poorly. POSIX time is defined as the number of non-leap seconds since midnight UTC, January 1, 1970. This makes a bad absolute clock because it’s incapable of unambiguously representing what time it is whenever a leap second might be occurring, and it makes a bad difference clock because if you subtract two timestamps separated by a leap second, you’ll be off by a second.

    Pragmatically, maybe POSIX is far enough entrenched that abolishing leap seconds remains the only practical solution, and so the argument I made in my first paragraph remains correct. Ideally, though, I’d rather leave leap seconds alone and fix POSIX instead.

    [1] i.e., seconds as measured by an atomic clock, as opposed to mean solar seconds as measured by observing the rotation of the earth.

    [2] POSIX-1.2001 fixes half the problem, providing clock_gettime(CLOCK_MONOTONIC) which makes a reasonable difference clock. But its absolute clock (CLOCK_REALTIME) remains as broken as ever. And OS X doesn’t support clock_gettime.

  21. Calendars are history and culture as well. Asking a foreigner about their new year is a great way to start a conversation about how things work back home, and the answer often has some insight as to how things operate there.

    For example, when is the Hindu New Year?

    A: There’s at least 16 of them, spread throughout the year, but roughly clustered around mid April.

    From a fact like that you might deduce that India is in fact an amalgam of many different somewhat related cultures that are very diverse yet close enough to make common cause and forge a nation…

  22. Even with atomic clocks, you eventually run up against the limitations of relativity itself… a subtle assurance that any two clocks–no matter how accurate–will always lose sync within some calculable margin of error (thus, still useful).

    I read the following article when it first appeared in 2010 and it still provides a fine basic description of the phenomena:

    * http://www.nist.gov/public_affairs/releases/aluminum-atomic-clock_092310.cfm

    For what it’s worth, none of the clocks in my house have a second hand; when I’m aware of timekeeping at that scale, I become uncomfortably fixated on it. Apart from cases of profiling software or inertial navigation, I’d really rather not know the exact time. Keeping it coarsely granular allows me to sleep at night. :-)

    Congrats on your good work with NTPSec, I look forward to its eventual maturity (as do legions of others, I’m sure).

    –M

  23. Death to DST! Too much of my career has been wasted dealing with that politicians’ political nightmare.

  24. > Can you imagine the politically correct clusterf*ck that will ensue when people try to settle on who to name those mid-year months after?

    Anyone want to start a pool on how long it takes before renaming July and August becomes some social justice activist’s hobby horse?

  25. > OTOH, do we have to be stuck with honoring Roman emperors? There’s got to be a way to make the calendar more regular but still tied to the seasons.

    The French came up with a fun one back in the day!

  26. To me the perfect system would consist of 5 months measuring 73 days each. We also end up with 73 five-day weeks, which works really well because the year always starts on the same day. Naturally, we’ll work three days and rest for two, and the religiously inclined can celebrate the Sabbath every five days. Every four years we’ll celebrate Troutwaxer Day as the leap day in this calendar.

    The year will start on the Spring Equinox, and the five days of the week will be named Why, What, Where, When, and Who. The five seasons will be named Spirit, Earth, Air, Water and Fire. Troutwaxer Day (Leap Day) doesn’t count as part of any week, so the 1st of every year is always on “Why Spirit” and the last day of the year is “Who Fire.”

    Eric, since you’re now in charge of time, you may implement this at your convenience. Name a holiday after yourself.

  27. @Sigivald:
    While leaving it un-simplified for scientific use, where simplicity is so profitable?

    Im not leaving the scientific second unsimplified, it would have only the standard power-of-thousand prefixes for scaling and scientific timekeeping would measure a straight second count from an epoch.

    If it’s defined as angles of rotation, does that mean the not-second gets … longer, as the planet’s spin slows?

    Yes, because that is a desirable property for civil time for as long as our circadian rhythms remain entrained to the rotation of the planet.

    The civil not-second could be 1/86400 of a day, but defining a 10/100/100 split keeps the length of hours, minutes, and not-seconds close to the units we’re used to while keeping the length of the not-second distinct from that of the second, which would help drive home that they’re separate units doing separate things.

  28. Time zones are useful. They are a simplification; prior to their adoption, train schedules and the like relied on local solar time, which could vary significantly even within the area swept by one modern time zone.

    But DST can GTFO. Japan doesn’t use DST, and they seem to get along fine without it. It’s a pain in my ass to have to figure out Japan’s time relative to me, and that’s when I only have to deal with American DST; when you add another country’s DST rules into the mix, which may not align with hours, then I have to Google for “time in Brisbane” or whatever and I don’t want to do that. From a software standpoint, being able to reckon time correctly over a long period of time requires not only an updated database of the current arbitrary DST start and end dates, but a record of their entire history over the period in question.

    We should just burn DST with fire. Forget about that happening in ‘Murica though; eliminating DST would change when the football games start! (Yes, a proposal to eliminate DST was voted down in Texas, and that was one of the reasons why.)

  29. @Tara
    “10 hour days of 100 minutes of 100 seconds.”

    60 = 3x4x5
    24=3×8
    As a consequence, it is fairly easy to divede up hours and days in equal periods. That is lost if you switch to 10:100:1000

    @Deep Lurker
    “The only reform I see as possibly worth doing is one to sync dates with days of the week from year to year.”

    That would require to defeat all the world major religions. They have decided to label every seventh day sacred, forever and ever til the end of time. This idea has been around for a long time and the Vatican cs have blocked any suggestion that would desync Sunday.

    1. >the Vatican cs have blocked any suggestion that would desync Sunday.

      At this point I have to put in a supporting word for the Gregorian calendar. It’s actually pretty effective at the job it was designed to do, which is keep civil (and ecclesiastical) time synced to the solar year. There are good reasons it superseded the Julian and has been adopted by many nations with strong calendrical traditions of their own (notably in Asia) that didn’t do as good a job of synchronization.

      Yes, we could in theory change 7-day weeks, but a quarter-lunation seems to be a unit human beings actually like; ours goes clear back to ancient Sumer through Greece. It was adopted independently in the Persian Empire, in Judaism and in Hellenistic astrology. Later it successfully naturalized in Tang-dynasty China and Gupta-period India. No competing calendar division of under a month’s length has ever achieved anything near the same spread.

  30. @esr
    Right. I think most proposed time and calendar revissions are solving non problems at substatial costs. Also, the calendar embodies a lot of knowledge about human life.

    I was already quite old when I finally understood that the lenghts of the months had to incorporate the fact that the Northern winter is shorter than the summer, i.e., the time between the autumn and spring evenings. And that that was the reason February had only 28 days and July and August both have 31.

  31. What bothers me a lot more than months named after dead emperors, is that September-December aren’t month 7-10.

    1. >What bothers me a lot more than months named after dead emperors, is that September-December aren’t month 7-10.

      You too?

      Yeah. But this only bugs people who can read the latin roots in the words.

      Sigh…as Lee Stoller has pointed out, probably best not to try to fix this.

  32. Re: the 7 day week, we could sync DoW and DoM by adding intercalary days (e.g. the Shire calendar) but the Abrahamists will never go for it since they think it would cause them to celebrate the Sabbath on the wrong day.

    Fun fact; Some Muslims used to use a luni-solar calendar similar to the Hebrew calendar, but went pure lunar so that the holy months (Ramadan in particular) could be every nth Moon just like the book said.

  33. @Winter

    That’s a point I hadn’t considered. I am a bit surprised that I haven’t heard that objection before since, as you say, it’s not a new idea. I knew it was part of the more radical calendar reforms out there, with “World Peace Day” or “UN Day” or the like, and that Tolkien used it in his Shire Calendar (among others) with Midsummer Day and Overlith not falling on any day of the week.

    @esr

    It sounds to me like a seven day cycle without warts is what human beings like, even when it isn’t exactly a quarter lune, and even when keeping this cycle wart-free means letting it drift out of sync with lunar and solar cycles.

    So a calendar reform of the sort I suggested wouldn’t just be a matter of defeating the major world religions, but of overcoming normal human preferences that value a wart-free weekly cycle over synchronization.

  34. Joshua: “Anyone want to start a pool on how long it takes before renaming July and August becomes some social justice activist’s hobby horse?”

    Hush yo mouth, boy.

    Jeff: “Forget about that happening in ‘Murica though; eliminating DST would change when the football games start! (Yes, a proposal to eliminate DST was voted down in Texas, and that was one of the reasons why.)”

    The distinction you miss here is that that affects high school football. High school football is a religion in Texas. You mess with that, and you’ll get everyone rising up against you. Remember, high school football is played Friday nights…

  35. Personally, I would like to see *some* simplification. 13 months of 28 days each, plus one or two “World Peace Days” or some such at the turning of the year to round it out.

    In concept, albeit with different numbers, you just described the Badí calendar: 19 months of 19 days (19 being the integer nearest the square root of the number of days in the year), plus 4-5 “Days Outside of Time”. The number of intercalary days is determined so that the spring equinox coincides with the first day of the year, for the local date at a designated spot on the Earth’s surface.

  36. @Joshua
    “Anyone want to start a pool on how long it takes before renaming July and August becomes some social justice activist’s hobby horse?”

    When the Chinese have taken over this will be “yi yue” through “shi er yue”. Or, month one through month twelve. No need to rename them. We can all count to twelve nowadays.

  37. I’m going to weigh in as the heretical IT guy that likes DST, or at least the personal benefits of DST (if there’s an alternate system which would provide the personal benefits of DST with none of the costs, I’d jump to supporting it in a heartbeat).

    The problem is that, for logical reasons, most people’s day goes: morning routine -> work -> dinner / recreation -> evening routine -> sleep; we place the fixed-to-the-clock schedule first. I think, for most people, sunlight is valuable during the dinner / recreation and work phases. An ideal schedule would be one that started the morning routine -> work process at a time fixed to the solar cycle (ie, always wake up and begin the routine at dawn), meaning when extra sunlight is available because the day is longer it’s available in the dinner / recreation period. However, because work needs to start at a time fixed on the clock for practical reasons, and the morning routine is usually fixed in length, we don’t have this option. A non-DST workaround would be to agree to open and close businesses earlier in the summer when the day is longer, but this requires some agreement to changing times.

  38. I used to think DST was silly, but then I read this by Dr. Drang. I’m still against it, but I can see why it’s a decent idea elsewhere in the country. Sample:

    If we stayed on Standard Time throughout the year, sunrise here in the Chicago area would be between 4:15 and 4:30 am from the middle of May through the middle of July. And if you check the times for civil twilight, which is when it’s bright enough to see without artificial light, you’ll find that that starts half an hour earlier.
    This is insane and a complete waste of sunlight. Good for a nation of farmers, I suppose, but of no value to anyone in our current urban/suburban society except those people who get up and go running before work. And I see no reason to encourage them.

  39. Irregularities in the time systems are much like warts and complications in an old software system. They’re often there for a reason. In many cases, though, that reason was either misguided to begin with, or no longer applies. If you don’t need to support Irix, Minix, and other old OSes, you can rip out the complicated build system and series of ifdefs that were there to support them.

    The month sizes and the leap days seem to work out pretty reasonably, I will grant.

    It’s hard to see a benefit to the 12-hour clock, though. The 24-hour clock is working great in multiple parts of the world. People with serious business who *must* get their times correct will by and large already use a 24-hour clock. People using the 12-hour clock seem like people using traditional notation for chess games. Pawn to Knight’s 4… Which Knight column? The obvious one, duh! Um…. How about just say which column you mean instead of being so cute about it?

    It’s also hard to see any true benefits to leap seconds. Certainly they don’t matter to the general public, which is who you’d think the target of a general time system is for. They don’t matter for seasonal drift, because unlike with leap days, the drift is too slow to matter on a scale of human societies. Even astronomers who favor leap seconds seem to only do so because it’s currently the standard. If that’s the only argument, then it seems like the standard could be changed on a scale of, say, 50 years. Start now and do it gradually. We use a “Z” to mean UTC time; so use a “Y” to mean timezone-less UTC time.

    It’s also hard to believe in a large benefit to daylight savings. Many parts of the world don’t use it, and the people there don’t seem to really miss it. The origins of it were in farming and no longer matter for the general public. Meanwhile, if you are one of the few remaining farmers, you can adjust your own schedule. It’s never been easier.

    1. >It’s hard to see a benefit to the 12-hour clock, though.

      I agree. I tend to write times in the 24-hour format the U.S. military uses to reduce ambiguity.

      We should distinguish, though, between superficial UI changes like this one and reforms that would actually involve changes to calendrical and elapsed-time calculations, like doing away with leap seconds. They involve different sets of issues. The merely cosmetic ones are far easier.

  40. “The origins of it were in farming and no longer matter for the general public.” How in the hell does this myth persist? Farmers hate DST and always have. The ag industry fought DST tooth and nail. Cows can’t read clocks. They need to be milked at the same *real* time every day, morning and evening. DST is only of use to people who work inside.

  41. @Lex Spoon
    “It’s also hard to believe in a large benefit to daylight savings. Many parts of the world don’t use it,”

    In Amsterdam (North West Europe)
    21 Dec Rise 08:48 Set 16:29
    21 Jun Rise 05:18 Set 22:06 (DST)

    So, in Winter I leave home in the dark for work and return home in the dark. In summer, there is 9 hours more daylight than on Christmas. We are marginally north of Londen and south of Berlin.

    That hour DST in Summer means I do get an hour more of daylight in the evening.

  42. Yes, because that is a desirable property for civil time for as long as our circadian rhythms remain entrained to the rotation of the planet.

    Fair, in its way.

    But I think it’s okay if we don’t worry about solving a problem (“significant planetary spin slowing vs. our time measurements”) that won’t be relevant for, well, long enough that we’ll either have left the planet or be extinct, most likely – it’ll take well over a hundred million years to hit a “25 hour day”.

    I don’t think we need to be planning even one million years in advance for that kind of thing, you know?

  43. On DST – most “abolish DST” proposals actually involve abolishing “standard” time and staying on DST year-round. This makes winter mornings a bit darker, and winter afternoons not end before 5PM…

  44. > We use a “Z” to mean UTC time; so use a “Y” to mean timezone-less UTC time.
    Unfortunately, Yankee is taken, for UTC-12 (Baker and Howland Is., plus about a 48th of the world at sea).
    Juliet, however, is used for local time.
    Of course, Z is already timezone-less. Maybe you meant leap-second-free?

  45. Does anyone really use those single-letter timezone designators?

    Z is ubiquitous, and I’ve never seen any of the others; in fact, I usually see the letter Z hardcoded in time patterns rather than parameterized.

  46. > Does anyone really use those single-letter timezone designators?
    Navy does, I think. Other than that, only Z. (And they’re only used at sea. On land, you use local time — and there are a full 40 time zones on land, counting only outside DST. Can’t letter ’em all, if you don’t speak Korean or Japanese.)

    1. >Navy does, I think. Other than that, only Z

      Maybe not quite. I’ve read that the other military TZ designators are still in limited use in international aviation. Sorry, no source to hand.

  47. > What bothers me a lot more than months named after dead emperors, is that September-December aren’t month 7-10.

    Somehow, I didn’t even notice until you pointed it out.

    This bothers me… not very much at all, really, but a lot more than it should.

  48. > No competing calendar division of under a month’s length has ever achieved anything near the same spread.

    Er, we’re not talking about doing away with the 7-day week as the primary unit, but merely having an occasional (once or twice a year) 8-day week or “outside any week” day. Your argument doesn’t really seem to apply to that.

    1. >Your argument doesn’t really seem to apply to that.

      No, it doesn’t, not specifically. I’m making a more general case that the Gregorian calendar has a kind of Unix-vs.-Plan-9 effect going for it. Like Unix, it has flaws, but the benefits of incremental improvements are so much lower than the transition costs that they’ll never get traction.

  49. “…the five days of the week will be named Why, What, Where, When, and Who.”

    @Troutwaxer: Change the order. Who’s on first, What’s on second, and consider I Don’t Know for third place.

    @Everyone: We can solve the Roman emperor naming problem and gain more precision if we simply adopt the Mayan calendar. Its 81 year cycle keeps the earth, sun and moon synched up better than any of the others.

  50. Does hackerdom have a standard of user-manageable date calculation formulas? Like that this contract expires at the end of the month which is 90 days from the day of signing the contract? Mainstreamdom has a problem with this, it is kind of being done but not elegantly.

  51. “the benefits of incremental improvements are so much lower than the transition costs that they’ll never get traction.”

    This is the same argument as to why the US should adopt A4 paper instead of 8-1/2×11. Yeah, yeah, paper sizes that are the next large one cut in half, blah blah, yadda yadda flambé. The Europeans who keep pushing this idea for the US never seem to understand that massive costs that their small, incremental advantages incur.

  52. “UTC without leap seconds” exists already – it’s called TAI. (TAI is actually 36 seconds ahead of UTC at present, but you’re going to have _some_ amount of drift anyway). Arguably, POSIX and other time standards should just add support for TAI time in addition to UTC, since the former is the best fit for many applications.

  53. None of us, including the best minds, know the precise length of any single one second period. Because the term itself is an arbitrary construct that helps us humans (lowly but always prone to hubris) to divide and measure to the needed level of practicality. All other considerations aside, Jupiter, Saturn and Neptune have been disturbing the length of our so-called second for 100’s of millions of years and will continue to do so long after we’ve all been and gone many times over. Perhaps our entire race and all creatures with us will have gone extinct and several times over and yet, the major planets will still be disturbing our precious definition of a second. And theminor, oh how they love to mess with our second too, just not to the same extent. And then there is galactic…

  54. The French went to metric/decimal time during one of their revolutions. 10 hours of 100 minutes per day.

    Vinge wrote a novel called “Outcasts of the Heaven Belt”, where time units were based on seconds. You had kilosecs, megasecs, gigasecs…

    Personally, I think adopting the Mayan tzolkin 260-day calendar would be just fine… at least, it would make everyone equally unhappy.

  55. The real problem with leap seconds – that is, the reason they tend to attract efforts to abolish them – is that they interfere with the ability to treat quantities with units of days and seconds (and more generally any other named unit up to a fortnight) as interchangeable scalar values.

    People already know, at an intuitive level, that they can’t use months and years that way (though months and years themselves can be, with each other and other units such as quarters and decades). But they expect to be able to do it for minutes and seconds, because the error introduced by doing so is so small (and historically did not exist before 1972). So running into problems caused by doing so is offensive to this idea of a halfway elegant timekeeping system.

  56. This thread’s title is “In defense of calendrical irregularily”. Shouldn’t that be “-ity”?

    1. >This thread’s title is “In defense of calendrical irregularily”. Shouldn’t that be “-ity”?

      Yes. Hilarious that it got by not just me but several dozen commenters.

  57. > Jupiter, Saturn and Neptune have been disturbing the length of our so-called second for 100’s of millions of years and will continue to do so long after we’ve all been and gone many times over

    Um. No. The second is precisely 9192631770h/?E, where ?E is the energy difference between the two hyperfine substates of the ground state of ¹³³Cs at 0K.
    The only way to change this is to modify m_u, m_d, m_c, m_s, m_t, m_b, the KM matrix, m_e, m_{ν_e}, m_μ, m_{ν_μ}, m_τ, m_{ν_τ}, the PMNS matrix, m_H, E(H), α_{U1}, α_{SU2}, α_{SU3}, or Λ, most of which would make very little difference. (Exceptions: a specific combination of α_{U1}, α_{SU2}, and E(H); m_e; a complex function of that first combination, m_u, m_d, and α_{SU3}.)

    1. >Also, why doesn’t the forum software do encoding? As is, it seems vulnerable to exploits.

      Sometimes it does. The circumstances that trigger it are unclear to me.

  58. > >This thread’s title is “In defense of calendrical irregularily”. Shouldn’t that be “-ity”?

    > Yes. Hilarious that it got by not just me but several dozen commenters.

    “We see, but we do not perceive, Master Foo!”

    ;-}

  59. @ LS

    You’ve definitely got a point. If the first day of every year fell on Who, it would be inarguable that “Who’s on the First.” Leap seconds could be called “I Don’t Know,” because of the obvious question: Why are we adding a second to the calendar this year?”

    So we’ll switch Why and Who, with the first day of the year being “Who Spirit” and the last day of the year being “Why Fire.” This will symbolize the eternal dynamic between science and religion.

  60. @Random832:
    But they expect to be able to do it for minutes and seconds, because the error introduced by doing so is so small (and historically did not exist before 1972).

    I think the bolded bit is the most important thing here. Historically, the Earth’s rotation was the reference clock for every time span less than a month.

    @esr:
    Yes. Hilarious that it got by not just me but several dozen commenters.

    Not at all surprising, though. I’m pretty sure that that’s exactly the kind of error that a nalive speaker will miss with high probability, but that second language speakers are likely to notice.

  61. I’m pretty sure that that’s exactly the kind of error that a nalive speaker will miss with high probability

  62. I have similar feelings about the drive to replace Imperial/English units in everyday life with metric units (I have no problem using metric in scientific matters). English units evolved out of day-to-day human experiences; this includes not just the units themselves, but their relationships; it’s much easier to estimate in real life, on the fly, 2 or 3 times something (or 1/2 or 1/3 of something) than to multiply or divide by 10. Thus, the cup/pint/quart/gallon system is (IMHO) far more human-adapted and intuitive than centiliters, deciliters, liters, and decaliters (plus bad naming choices, given potential confusion between 1/0th liter and 10x liters).

  63. > Personally, I would like to see *some* simplification. 13 months of 28 days each

    This will never fly, because business people like to have quarters of full months. Maybe we could take back some of the days stolen from February and given to honor Roman Emperors, just to even things back out a bit, but I don’t even think that’s doable at this point.

    The algorithm for calculating MMDD from day-of-year is not that complicated, due to the nice regular pattern of 31/30/31/30/31 days per month starting with March. (If we could do any one thing to make sense out of the calendar, it would be to make March 1 New Year’s Day, coinciding with meteorological Spring in the Northern Hemisphere, and restoring the number-ber months to their original meanings, and leaving the irregularity for the last month. But I don’t see any way that’ll ever happen, either.) Maybe we can do it right when we colonize other planets (with different day/year length anyway), but I’m afraid this one’s stuck where it is.

    The reason why DST has to exist in the first place is because we use clocks rather than just getting up with the sun as we did before technology allowed us to efficiently light our homes and workplaces at arbitrary times of the day. As sunrise falls a few minutes earlier each day in the spring, the adjustment is so little as to be unnoticeable. But once we got clocks into the picture, and decided that a certain time of day was the right time to get up, we set ourselves up to waste daylight by sleeping in. DST adjusts for the changing time of sunrise with horrible granularity, but it achieves its stated purpose of saving what would otherwise be a wasted hour of daylight for when we get off work in the afternoon. Having the extra hour for outdoor activities is a Good Thing.

    Could we live with more frequent adjustments of less than an hour instead of doing it all in one swell foop? I doubt it. People would freak out having to spring forward 15 minutes four times and fall back four others. Every 6-7 Saturday nights would be too much. How about two half-hour adjustments each way? Nope, still too many. The current scheme is the least bad way to do it, at least until all clocks are digital and can be programmed to automatically adjust for us a few minutes every Saturday night or something.

    Anyone who wants to operate solely on standard time is of course free to do so, but good luck trying to explain yourself. I see expressions like “CST” when it seems “CDT” is far more likely the intent, making it impossible to communicate this distinction. There are times I ask people which they meant, and their responses indicate they genuinely don’t understand how, say, most of Arizona can be on PST (matching the CDT clocks) while people in neighboring California are on PDT, and calling the latter “PST” means someone’s going to be an hour early for the conference call.

    And the correct solution to the leap second kerfuffle is for people who don’t like them to use TAI as @guest points out. Just give us the API call to get time in TAI, UTC, or any of the many timezones, and I’m happy. I won’t force you to use leap seconds, and you won’t force me to stop using them.

    ObNitPick: It is “Daylight Saving” or even “Daylight-Saving”, not “Daylight Savings“. There is no savings account into which daylight is deposited, to be withdrawn at a later date; the saving of daylight each day is an immediate consequence of getting your butt to work an hour earlier in the morning. Calling it “Daylight Savings” earns consignment to the same group as people who cite eschatology from “Revelations” or go shopping at “Walmart’s” (because so many other books of the Bible have names ending in “s” and so many businesses take the form “${surname}’s” that people feel compelled to add them even when they do not belong). AND GET OFF MY LAWN!

  64. Someone’s circulating a ballot initiative to abolish DST in California; unlike other proposals, it will align us with Arizona, leaving us in PST all year.

    In the other direction, sunrise at 5:30 when twilight is an hour long is still wasteful, so Ray on Car Talk proposed Double-Dog Daylight Saving Time for the northern states (and presumably Canada) – shift another hour from mid-May to mid-July or so.

    The one real advantage of DST is solving a coordination problem – a lot of places would like to shift their hours between summer and winter because daylight matters, and having the government set a specific day for that change is much easier to deal with than remembering that your work shifts on March 15, the kid’s school on April 1, and the Catholic hospital nearby switches on Easter.

  65. > There are times I ask people which they meant, and their responses indicate they genuinely don’t understand how, say, most of Arizona can be on PST (matching the CDT clocks)

    Well, in fairness to those people, most of Arizona is in fact on MST, which happens to match PDT (not CDT) clocks. The usual mode of confusion I’ve observed is Arizonans thinking they’re on “Pacific time” during the part of the year PDT is in effect to the west.

    It’s instructive to note that explaining how people are confused by time zones/DST is itself confusing.

  66. > Well, in fairness to those people, most of Arizona is in fact on MST
    Yes, I brainfarted that one, trying to say MST is the same as PDT, and it came out backwards.

    TL; DR: xST and xDT are two different things, and most people just don’t know that there’s a difference.

  67. @The Monster
    “Maybe we could take back some of the days stolen from February and given to honor Roman Emperors, just to even things back out a bit, but I don’t even think that’s doable at this point.”

    Its more complicated:
    http://mentalfloss.com/article/55327/why-are-there-only-28-days-february

    But that is not the whole story. Due to the shape of the earth’s orbit, northern winters are shorter than the summers. The difference is 5 days.
    http://blog.chron.com/sciguy/2013/01/happy-perihelion-and-why-winter-is-shorter-than-summer/

    This is “corrected” to some extend in the calendar by moving days from winter to summer.

  68. Winter: “Due to the shape of the earth’s orbit, northern winters are shorter than the summers.”

    They only feel like they’re a lot longer…

  69. Oh cripes. I missed the two capital deltas in my post.
    Edited version:

    precisely 9192631770h/ΔE, where ΔE is the

  70. @Winter

    For business purposes, having four quarters (which are not aligned along seasonal boundaries) of size as equal as possible is desirable, so three of 91 days and the fourth of 92, with leap day boosting another to 92, would be best.

    Due to the precession of the poles, the perihelion has not always been in NH winter. It takes roughly 21,000 years for the full cycle to complete, so it’s possible we could shorten some 31-day months and give those days to shorter months to readjust to the new norm every few millennia, but I find it unlikely that such changes would ever actually be accepted.

  71. I think of Daylight Saving Time like tax withholding. We give the government an hour in the spring, and they give it back in the fall. God knows what they do with that hour in between, but at least they give it all back eventually.

  72. If it were actually like tax withholding, you’d get 31 minutes back in the fall, which isn’t much more than you get from Sweden or Canada, but crime rates are much higher and you still can’t afford health care :(

    Plus Sweden and Canada have self-setting clocks, while in the USA, the clock manufacturers lobby the government to make sure everyone has to manually set their clocks.

  73. @Jeff:
    My watch is self-setting, but has old DST rules and no way to update firmware. Everything else I use as a time source is set from NTP.

  74. > My watch is self-setting, but has old DST rules and no way to update firmware.

    I think that was an overextended metaphor.
    Also, my watch and alarm clock are the only things that don’t use NTP as a time source.
    (Fortunately, they don’t bother with DST, just leap years, and only my watch. Though my alarm clock runs fast by about five minutes an hour if the power’s out.)

  75. My watch is radio controlled. I would prefer NTP, but then I would rather use my phone.

    Watches are so last century anyway.

  76. > We give the government an hour in the spring, and they give it back in the fall.

    If you seriously believe that, you have no clue why DST exists. This is exactly the sort of attitude expressed by the misspelling “Daylight Savings Time”, suggesting that an hour of daylight has been put into a savings account, to be withdrawn at a later date. The daylight being saved is an hour every day throughout the entire DST period, which oddly is longer than the “Standard Time” portion of the year.

  77. I mostly agree with you, but I still think daylight savings time should be done away with. It is harmful.

  78. How is it harmful? And if we were to do away with it, would we go to “Standard Time” year round, or to the time we use the majority of the year, currently known as “Daylight Saving Time”?

  79. If we get rid of DST, I’d want to go to Standard Time year round. If we don’t get rid of DST, I’d like to shorten the period where it applies.

    Currently, in the US, DST applies from the second Sunday in March to the first Sunday in November – starting before the spring equinox and ending after the fall equinox. I’d much prefer it start after the spring equinox and end before the fall equinox. Say: Starting on the Sunday after the first Saturday in April, and ending on the second Sunday in September.

  80. @Deep Lurker

    What is the significance of the equinoxes?

    Taking Kansas City as fairly representative of the contiguous 48 states (being pretty damned close to the latitude of the geographic center thereof), consulting http://www.timeanddate.com/sun/usa/kansas-city we find the extreme sunrise times here in CST are:
    Jan 5 7:38 am
    Jun 13 4:52 am (Note that the website gives this as 5:52 CDT)
    The maximum comes more than a week after the southern solstice, and the minimum more than a week before the northern solstice. The casual student of the mechanics of varying sunrise times would assume the solstices would mark the maxima, naively equating “length of day” variations to “sunrise time” variations. This imbalance is due to the fact to which Winter alluded up-thread: perihelion falls between northward equinox and northern solstice, causing the Earth to revolve about the Sun faster (while it rotates at a constant rate). This is also the cause of the figure-8 shape of the Analemma depicting variations of apparent solar motion from the mean.

    This year, DST begins within a week of Mar 9, when sunrise falls at 6:38 CST, and
    and ends Nov 6, nearly two weeks after Oct 24, when the sun again rises at 6:38 CST. This tells me the start time of DST is spot-on, but the end time ought to be two weeks earlier, at least for the sunrise variance typical here. To me, the goal of DST is entirely based on the notion that we set “start of the day” based on sunrise, for instance scheduling school so that we won’t have children standing at the side of the road waiting for buses before sunrise, but that once the sunrise has shifted to at least an hour before that time, we recover the daylight we’d otherwise be squandering by shifting our clocks an hour to match.

    Let’s take Miami, the southernmost major city in CONUS, as a point of comparison:
    The latest sunrise is 7:09 EST, and the earliest 6:29. Sunrises of 6:09 come on Apr 3 and Sep 20, which fits pretty damned well with what you’re calling for.

    Now let’s go the other way, and choose Seattle. 7:58 PST is the latest sunrise. 6:58 sunrises fall on Feb 25 and Nov 3. The former is a few weeks before our current DST rule; the latter right on it. (I could look up Anchorage, but you’d see it’s far worse. And let’s not even think about Barrow.)

    I don’t think we should make DST rules for the whole of the country based on the extremes of Miami and Seattle or Anchorage. And I really doubt people who oppose our current system would appreciate having multiple, staggered DST start/stop dates for different latitudes; it pretty much has to be an all-or-nothing proposition for us to have a chance to keep track of it all.

    So one might make the case that Florida (or perhaps just South Florida) ought to join Hawaii, Arizona (with the exception of Navajo Nation land), Puerto Rico, US Virgin Islands, American Samoa, and Guam, in eschewing DST altogether. Florida’s connection to the Caribbean (most of which does not use DST) through the cruise ship industry provides a good reason for them to want some consistency there. However, I think the ties to the rest of the country are far stronger, and having oddball DST rules could be seen as bad for the tourism industry.

  81. @The Monster:
    My preference is that daylight savings time should only start after the point when sunrise comes before 6 am standard time, (shifting it up towards, but not past, 7 am when daylight savings time kicks in) and that it should end before sunrise starts coming later than 6 am standard time.

    For Kansas City, I would not find Mar 9, when sunrise falls at 6:38 am CST to be “spot on” for the start of daylight savings time, but rather too early a date. What I consider “spot on” would be April 3, when sunrise is 5:59 am CST / 6:59 am CDT. Likewise, if the switch back to standard time came on Sept 11 in 2016, then sunrise in Kansas City would be at 5:57 am CST, which again suits what I want to see.

    For Seattle, sunrise on Apr 3 is 6:43 am PDT, which I’d much prefer as a starting point to the official March 13 daylight savings time that gives a sunrise of 7:25 am PDT. For Miami, sunrise on Apr 3 is 7:09 am EDT which is late from my point of view but still better than the official March 13 sunrise of 7:32 am EDT.

    So I agree with you that

    the goal of DST is entirely based on the notion that we set “start of the day” based on sunrise, for instance scheduling school so that we won’t have children standing at the side of the road waiting for buses before sunrise, but that once the sunrise has shifted to at least an hour before that time

    – but I disagree with you about when “that time” is. I see “that time” as 7 am, and want a daylight savings rule that mostly avoids creating post-7-am sunrises.

  82. Timezones and daylight saving: once upon a time, France and Benelux were in the same timezone as the UK. During WW2 occupation Germans have put them in their own timezone for ease of coordination. After WW2 they decided to stay there, because there were benefits of being in the westernmost part of a timezone. Essentially, it is better to endure darkness in the morning before work than in the evening after work, people are more likely to want to read or spend time outside in the evening, it is both more comfortable and ends up using less electricity. While in the morning people usually just have a quick breakfast and go to work, and on weekends they tend to sleep in, which is ideal in a darker morning. All in all, an ideal life rhythm would optimize on putting sunlit hours to the evening after work, not the morning before work. The only issue is driving in the morning, but AFAIK it is not a very accident-prone time of the day even in darkness.

    There are probably lessons in it. For example, why don’t we just modify working hours instead of the DST and all this? Probably because that is more difficult. But why? Coordination problems? Schelling stickiness?

  83. My watch sets itself off my smartphone, which sets itself off NTP, geolocation/GPS, and sometimes cellular network time, the exact proportions of which are a bit of a (self-induced) mystery to me; but it works.

    Civil Time is a servant (and creation) of man, not the other way ’round. Trying to make human activity conform to an unvarying civil time is an exercise in futility and failure.

    As for metric time, don’t go there. 12 and 60 are both much more convenient bases than 10. I rue that the fingercounters who only counted on fingertips rather than knuckles “won” for determining our written numerals. (You can count on your fingers to 144 without particularly tying your fingers in knots the way you have to do to count to 1024. Though I’d have rather have had that then base 10 as well)

  84. TheDividualist
    “Essentially, it is better to endure darkness in the morning before work than in the evening after work, people are more likely to want to read or spend time outside in the evening, it is both more comfortable and ends up using less electricity. While in the morning people usually just have a quick breakfast and go to work, and on weekends they tend to sleep in, which is ideal in a darker morning. All in all, an ideal life rhythm would optimize on putting sunlit hours to the evening after work, not the morning before work.”

    This.

    I’d like to say good riddance to DST. But all the US timezones, IMHO, need to be shifted to the east. I’m in CST but prefer the summer hours which is CDT but that’s really EST.

    And ESR’s main point that the calendar serves human needs and quirks is spot on.

  85. Count to 144 on your fingers: your thumb is the placeholder, your knuckles the digit indicators. Touch your thumb to the proper knuckle to indicate the digit. One hand is the “ones place,” the other the “twelves place.”

    I count with the right index finger tip knuckle being 1, the pinky finger tip being 4, the index finger middle knuckle being 5, and so on to the pinky finger base knuckle being 12. The the left index finger indicates 12, and repeat on the right, until you get to 144 on the left pinky base knuckle. I have read of people who prefer to count by 3s rotating the scheme 90 degrees such that you count down the index finger, then down the middle, &c.

  86. bfwebster on 2016-03-03 at 13:19:54 said:

    English units evolved out of day-to-day human experiences; this includes not just the units themselves, but their relationships; it’s much easier to estimate in real life, on the fly, 2 or 3 times something (or 1/2 or 1/3 of something) than to multiply or divide by 10.

    The obvious solution is to replace the entire metric system with one based on duodecimal power multiples of Planck units: 12³²×ℓᴘ ≈ 1.79 feet or 54.7 cm, a reasonable measure of length; 12⁴⁰×tᴘ ≈ 0.79 seconds; and so on.

  87. The obvious solution is to replace the entire metric system with one based on duodecimal power multiples of Planck units: 12³²×?? ? 1.79 feet or 54.7 cm, a reasonable measure of length; 12??×t? ? 0.79 seconds; and so on.

    I like it. We’ll combine your system of Planck units with my calendar based on five months/seasons of 73 days each. We now have a perfect calendrical system, which also includes physical measurements of length as a by-product. Since we can break things down very finely, we’ll be able to handle all those little issues about how a day is actually 23 hours, 56 minutes and 4 seconds…

    Unfortunately, a year is about 365 days and 6 hours, so we can’t do away with the leap year without some kind of major orbital engineering (The destruction of humanity is a small price to pay for the perfect calendar.)

    Just to expand a little:

    12^31 = .179 feet, or approximately two inches
    12^30 = .0179 feet, or approximately a quarter inch

    Going upward we have lengths of approximately 18 feet, 180, 1800, etc., so I’d judge this system of measurement as being very workable. I think we should name the length unit of 1.79 feet a Salomon and then we an define other things as exponents of a Salomon.

    This is important. We need to write up an RFC.

  88. This is important. We need to write up an RFC.

    You have about two old-system weeks.

  89. You have about two old-system weeks.

    That wasn’t very nice. Besides, I’m waiting for someone to work out how to apply Planck units to the issue of mass, and I need to find out whether Planck units form the foreground or background to relativity so I can understand how the system works when approaching c, because that will be important for future calculations.

  90. @EMF – “The second is precisely 9192631770h/?E”

    Um no, that is based upon the average of what was observed between 1750 and 1892. It may be an accurate representation of the ephemeris second but it is by no means even close to the thing we call a second.

  91. >Um no, that is based upon the average of what was observed between 1750 and 1892. It may be an accurate representation of the ephemeris second but it is by no means even close to the thing we call a second.

    No, it is exactly one second. That is the definition of a second. Look it up! (And besides, it is very close to one part in 86400 of a mean solar day, to within one part in a million. Sure, when you say ‘just a second’ you mean about ten, but that’s metaphorical.)

  92. “No, it is exactly one second. That is the definition of a second.”

    For now it is…it will change soon. Recent developments in atomic clock technology have rendered the current definition ‘too crude for fine work’. As it is, the latest lab models detect relativistic effects if they are raised less than one meter, or moved at a walking pace.

  93. Apologies for this being terribly OT but this seems like the best community I frequent (or more recently simply lurk) to ask…

    Has anyone used Braid to manage vendor trees/branches in git or is there a better way or canonical method of doing this?

    On Topic: Us westerners have seriously missed the boat. We should all move to a lunar calendar. Or the Iranian solar calendar.

    I kid. I agree with Michael and others. Abolish DST by making it the standard all the time so everyone gets whatever little more daylight at the end of the day.

  94. > Abolish DST by making it the standard all the time so everyone gets whatever little more daylight at the end of the day.

    So, instead of 7:38 being the latest sunrise, it would be 8:38 am? That just means we’ll start school an hour later than we do now, and most businesses will go to later hours, and we’ll be back to wasting the daylight when the sun rises earlier. Everything will be the same, but with larger numbers on our clocks.

  95. > 12^31 = .179 feet, or approximately two inches
    > 12^30 = .0179 feet, or approximately a quarter inch

    > Going upward we have lengths of approximately 18 feet, 180, 1800, etc., so I’d judge this
    > system of measurement as being very workable.

    If you’re using powers of twelve, why do your approximate lengths rise by factors of ten? Based on the above, 12^31 = 1.79 inches, and 12^33 =21.5 inches, 12^34 = 258 inches, a bit over 20 feet, etc.

  96. @ The Monster

    I was still working on my coffee at the time. It works fairly well either way, I think. Anyone who disagrees can walk the Planck.

  97. >So, instead of 7:38 being the latest sunrise, it would be 8:38 am? That just means we’ll start school an hour later than we do now, and most businesses will go to later hours, and we’ll be back to wasting the daylight when the sun rises earlier. Everything will be the same, but with larger numbers on our clocks.

    Why would we start school an hour later? We have these things called lights. When the sun rises earlier most of the year we are already using DST so nothing changes.

    The question is do you want to use lights in the morning during the winter when you are going to school/work or lights in the evening when you typically want do more outdoor leisure activities.

    And frankly 7:30 still means most high school kids are still going to school before sun up and even many middle school kids. My kids have a late start for middle school and are on the bus by 7:30. Most of their compatriots in other schools in the district have been in school a half hour already and got on their bus before sunrise.

    What sucks is when they get home at 3:30 sunset is an hour later. If sunrise is 8:30 it makes nearly zero difference to the morning activities since it’s pretty dark anyway when they get on the bus. Sunset at 5:30pm means they get an hour more of play time outdoors.

    What really sucks is trying to do soccer or other practice after DST ends in the fall when the field isn’t lit. You can’t move the start time much since parents work and it’s hard to get home before 5 to get kids to practice by 5. That extra hour of daylight at the end of the day would be a big help to many parents, coaches and kids.

    Everything certainly would NOT be the same.

  98. > Why would we start school an hour later? We have these things called lights. When the sun rises earlier most of the year we are already using DST so nothing changes.

    Because parents don’t like their children standing at the side of the road waiting for buses before sunrise. It’s dangerous. Therefore, sane school administrators consult a sunrise time table, and set school hours based on the latest sunrise (or perhaps the latest onset of twilight, based on one of the three different definitions thereof), so that they don’t have to explain why anyone’s kids got killed.

    If DST is made permanent, schools will start their days later. There is no way they’ll just accept having the kids out in the dark like that. If DST is ended the other way, by leaving existing Standard Time year round, then the start times will remain what they are now. The alternative is to change the scheduled times two or more times during the school year, and no one is going to do that.

    >What really sucks is trying to do soccer or other practice after DST ends in the fall when the field isn’t lit
    To quote your own words, ” We have these things called lights”. So which is it, because we have lights, it’s not important, or it’s crucially important to sports practices?

  99. Streetlights vs field lights – natural light is better for sports, and artificial lighting is EXPENSIVE both in electric bills and maintenance. For street lighting, the timeshift doesn’t matter – you still need to run the lights for x hours regardless of wallclock time. For sports lighting, time DOES matter.

  100. > Streetlights vs field lights

    My grandchildren have to stand at the side of a rural road to wait for the bus. They don’t exactly have “streetlights” on that road, unlike the nice bright lights here their mother grew up with. This is what school administrators have to take into account when deciding what the right time to start classes each morning is.

  101. Intuiting metric:

    Meter: not possible. I dislike meter. If you want to measure a distance walking, even a tall man like me needs to take really long steps. Or practice the 75cm step. Most women are 1.6, most men 1.8. Not ideal.

    Kilometer: again not a very good one, if I walk for leisure, I do at least 2-3, if I am trying to get somewhere and not waste my time, over 500m I’m am going to look for a bus or get in the car. A mile wouldn’t be better though.

    Liter: you just get used to the beverage sizes if you live in metric land, half liter beer, bottled water, liter milk etc.

    Kg: the liter-kg conversion works well for thinner liquids. Weight of a liter of milk if the container is light. Ultimately, going to a metric gym makes one intuit weight well after a while.

    100 kg: a good bench press. Also the body weight of a taller man with a good bench press and bad diet. It is fairly common in my circles.

    Celsius: ugh. Not really possible. 30 perhaps where comfortably warm becomes uncomfortably hot. The lower threshold varies, I get uncomfortable at 5, I know Scandinavians who are comfortable until -7 or so. However 0C has one tremendous intuitive utility: ice on the roads, drive carefully. See a minus sign in the weather report, take the bus, or drive like a snail. That one is handy.

  102. >Celsius: ugh. Not really possible. 30 perhaps where comfortably warm becomes uncomfortably hot.

    I have the opposite problem. When I hear a temperature in Fahrenheit (that isn’t a body temperature) I can’t grok it without converting to Celsius.

  103. Apart from teaspoons, English volumes would almost form a base-2 system (yes, I’m mixing wet/dry/imperial as if 1 quart is always the same):
    4 drams/tablespoon, 2 tablespoons/ounce, 2 ounces/shot,
    4 shots/cup, 2 cups/pint, 2 pints/quart,
    2 quarts/pottle (obsolete), 2 pottles/gallon,
    2 gallons/peck, 4 pecks/bushel.

    Lengths are much more confusing, though up to the fathom they correspond to body proportions reasonably:
    1 inch= a short finger joint, a hand you know (this is roughly one decimeter), a foot is the length of the forarm, a yard is from the end of the outstretched arm to the nose, a fathom is the farthest a tall man could spread his hands apart (making it quite sensible for measuring depth).
    I can’t pace off a yard, but it’s not for the usual reason–I have a 3.7 foot (or longer) step. But a yard is 2 cubits for me (elbow to fingertip).
    A rod has no rational relationship to anything smaller; it’s 11 cubits (half-yards). But it’s a handy size for meauring land, being about as long as a stick could be without becoming awkward. A furlong is 40 rods, and a mile is 8 furlongs.
    So if I were to rework that system, I’d keep inches and hands, and work up from there.

  104. > Celsius: ugh. Not really possible.

    Well, of course; the obvious temperature scale is Kelvin. Who wants to set their stove to -269 to get their helium to boil?

  105. >My grandchildren have to stand at the side of a rural road to wait for the bus. They don’t exactly have “streetlights” on that road, unlike the nice bright lights here their mother grew up with. This is what school administrators have to take into account when deciding what the right time to start classes each morning is.

    Kids are getting on the bus in the dark anyway. They don’t delay start times to prevent this in the winter because they can’t and still run a reasonable schedule.

    We aren’t a very rural country any more and lighting a bus stop can be done with a solar light and the cost for the light can be offset by advertising or corporate sponsorship.

    http://www.cityoftaylor.com/node/28260

    The cost for the city of london to do solar lighting cost about $2K per stop in their bus shelters.

    “However, we have upgraded 3,660 existing stops so that they are enabled for solar power at an average cost per stop of £1,349.00.”

    https://www.whatdotheyknow.com/request/bus_stopshelter_replacement_cost

    These will work even when sunrise is after your bus pick up time during Standard Time in the winter.

  106. >To quote your own words, ” We have these things called lights”. So which is it, because we have lights, it’s not important, or it’s crucially important to sports practices?

    Lighting sports fields is far more expensive than lighting of school bus stops and burns a lot more energy.

  107. Hmmm… yes, timezones in file timestamps are like mentally deficient attempts to save your location when you modded the file or sent the email. If you want to do that, that’s another system. ty, esr.

  108. > Lighting sports fields is far more expensive than lighting of school bus stops and burns a lot more energy.

    You need a lot more bus stops than sports fields to support a given population, and you don’t get to choose where to build them.

  109. Random off topic thought prompted by the mention of RFC3339.

    I am a bit surprised that nobody has published a edited anthology of select IETF RFCs.

    I imagine this to be a nice printed book with good typesetting containing a select group of perhaps 100 RFCs (or whatever number is practical). The RFCs would be a mixture of the historically significant, technically curious, humorous, and those “in-force” RFCs that are of technical importance today.

    The selection would be made by somebody with the ability to see the documents in their correct cultural, historical, and technical contexts (*cough*). The editor would also write an introduction, and perhaps a few words to preface each RFC to help the reader understand it.

    You could even weave a narrative of the history of the internet and hacker culture through the judicious selection and arrangement of the documents.

    Tom

  110. I’ve recently written a calendar ymdhms to julian day subroutine for my astrodynamic code. What a giant tree of crazy modular arithmetic! It wasn’t fun to debug.

    On the other hand, I now know why in the hypothetical Star-Trek-esque future, everyone uses an unwieldy “stardate” integer to denote time, instead of a more natural mnemonic system. (Besides the obvious fictional reason that the producers didn’t want to keep track of a future history calendar): In space, everything is moving. If you want to navigate, it requires knowing where everything is now, and where everything will be when you get there. That requires keeping your time in (days since ).

    In our spacecraft operational calculations we use days since J2000 or J2000.5, or julian days.

  111. Of course, in some hypothetical interstellar sci-fi future, you’d also have to have a standard reference frame you are measuring time in.

    If you introduce FTL, space-like and backwards time travel into the mix with ship time having no relation with departure and arrival time, forget it. :-P

  112. @Edward Cree
    >Besides, if we abolished leap seconds, the equation would only change by about a minute in one’s lifetime; if words can change their meaning over that kind of timespan, I’m sure the meaning of “9AM” can shift to “just after lunch” in the millennia it would take its referent to make the equivalent shift.

    The tz_data is already capable of handling granularity of a minute. If we aligned UTC with TAI you adjust by a minute when the solar time is off by more than 40 or 45 seconds (be prevent to much bouncing around). Such a system should be quite robust for a while, the main issue would that you would have to move the international date line via the same rule, and eventually it’d move for the middle of one or more continents.

  113. “I’m sure the meaning of “9AM” can shift to “just after lunch” in the millennia it would take its referent to make the equivalent shift.” And in case anyone doubts that time word can shift meaning, it apparently only took “noon” a few centuries to shift from ~3PM (etymologically it comes from nine [hours after sunrise]) to 12.

  114. “The tz_data is already capable of handling granularity of a minute. ”

    I don’t know what exactly you mean by “tz_data” – the data in the [i]tz project[/i] has, as far as I can tell since it was created, had granularity of a second. The only things I can think of with granularity of a minute are far older kernel-based mechanisms (which needed to be able to handle every possible offset within a 16-bit signed integer).

    And, no, there’s no reason you would have to move the international dateline. There’s nothing, after all, requiring the dateline to divide timezones for +12 and -12 rather than (say) +13 and -11.

  115. Eric starts a thread on calendar irreghilarity and I missed it until now? I picked the wrong month for work to explode…

    Troutwaxer wrote: …the five days of the week will be named Why, What, Where, When, and Who.

    LS wrote: Change the order. Who’s on first, What’s on second, and consider I Don’t Know for third place.

    Next thing you’ll be proposing is extending the metric prefixes past atto-, zepto- and yocto- with chico- and harpo-.

    (And we’d still be 14 orders of magnitude away from representing single multiples of Planck times…)

  116. > so does almost every language’s standard library, with the exception of Haskell’s

    Thanks!!

    I’m also in favour of keeping leap-seconds, though in the end I feel like it’s for essentially (pagan) religious reasons, that is, to keep the connection with the day, the most important natural cycle.

Leave a comment

Your email address will not be published. Required fields are marked *