[LEAPSECS] alternative to smearing

[Tom Van Baak]

> There's a big difference between these two: February varies in a fixed,
> regular manner, whereas UTC days are unpredictable.  The Gregorian
> calendar is of the arithmetic variety, whereas UTC is an observational
> calendar.  UTC is qualitatively more difficult to handle.
> 
> -zefram

Agreed. The February leap day is a useful analogy when describing leap seconds to non-technical people. But when you get into the details the two are vastly different. For the newcomers to the list...

1) The leap day in February can be handled by any isolated or autonomous clock or timekeeping system. A leap second can only be handled with periodic direct or indirect communication with IERS, or manual intervention with the likes of keyboard input or toggle switches. For secure or embedded systems this is a huge issue.

2) Past dates involving February can be handled with arithmetic. Past leap seconds are handled with leap second tables.

3) Future dates involving February pose no additional computing challenge. Future dates with leap seconds beyond a few months are simply not possible, or possible only with error bars.

4) Code to validate arbitrary time stamps involving February is trivial. Validating time stamps involving leap seconds is possible only if code or tables are updated frequently. And this applies to every layer in the sometimes massive stack from h/w to end-user interface.

5) Proleptic calculations involving February are static. Proleptic calculations involving leap seconds will differ from author to author and will change as new historical research on past earth rotation evolves.

6) February leaps are binary (28 or 29). Leap seconds have three states 59, 60, 61. Actually, it's 4-state monthly: not-known-yet, known to be 59, known to be 60, known to be 61.

7a) The leap in February allows fine tuning of the stable ratio of earth revolution to earth rotation. What is at stake is the alignment of the calendar vs. seasons.

It would take an error in the leap day count of, say, 5 to 10 days before it would be noticeable to people. The worst-case would be 183 days but obviously no one will let that happen.

7b) A leap second allows fine tuning of the unstable ratio of earth rotation to SI second. What is at stake is the alignment of time-of-day vs. noon (in Greenwich). Because most people don't live in Greenwich a system of ~24 hourly time zones exists to move 12 o'clock closer toward noon and thus nominally limit the error to +/- 0.5 hours. However for practical, geographical and political reasons this actual level of error is much larger for many people. Latitude plays a role. Daylight saving time has a role. Also noon wanders by as much as half an hour on its own (equation of time).

All factors considered, it would take an error in the leap second count of, say, 1000 to 3000 seconds before it would be noticeable to people. The worst-case would be 43200 seconds but obviously no one will let that happen.

8) Both of the above ratios are determined by nature. If / when the existing leap day system breaks down, one can simply add or delete N days from a chosen year, as was done by the Gregorian calendar reform. It's not pleasant but it worked. It may be required on the order of several thousand years, depending on N.

If / when the existing leap second system breaks down, one can simply add or delete N seconds from a chosen day. It also would not be pleasant but it would work. Or one can use the existing noon-alignment knob (time zones) to keep noon at 12. It may be required on the order of several thousand years, depending on N.

9) Exposing citizens to leap days is near universal. Printed and computer calendars have no trouble with that extra day. Almost every child learns about leap seconds during their schooling. Some people are even born on a leap day.

Exposing citizens to leap seconds is neither universal nor required. No analog representation of time can cope with 61 seconds, and even most digital clocks do not. Children are not taught about leap seconds. In fact few physicists, engineers, and programmers understand them either.

----

I mention all this not to be pro- or con- leap second, but to explain that the February leap day analogy is both useful and dangerously misleading. One requires a single sentence of explanation and one or three lines of code. The other requires a massive, complex and confusing, world-wide measurement and communication infrastructure.

/tvb