[Phys-l] leap second limbo

[John Denker <jsd@av8n.com>]

> Timekeepers at a meeting in Geneva have failed to agree on a proposal
> to abolish the 40-year-old practice of adding the occasional second
> to atomic time, in order to bring it in alignment with world time as
> told by the rotation of the Earth.
>
> The International Telecommunication Union put off the decision at
> Thursday's meeting, saying more study was needed into whether to
> scrap the leap second.

  http://www.aljazeera.com/news/europe/2012/01/2012119184517324138.html

Let me ask the obvious dumb question:  Given that there are conflicting
objectives, why not just have *two* time systems:
 -- civil time (locked to the earth's rotation, with leap seconds)
 -- nerd time  (no leap seconds)

In particular, consider the analogy to time zones and daylight savings
time.  All well-behaved computer operating systems keep time in UTC 
internally, and use an elaborate ephemeris to convert to the local time 
zone, if/when needed.
      (The exception is Microsoft Windows, which *still* tries to keep 
      the hardware clock in local time, believe it or not.  You might 
      think this would be a disaster for anybody who travels across time 
      zones, and you would be right.)

So, why not let computers and GPS satellites etc. keep nerd time, and
then publish an ephemeris that documents how to convert nerd time to
civil time if/when needed?

The two time systems *will* drift.  Wishing will not make the drift go 
away.

===============

Switching now to spacelike directions, people have been dealing with an
analogous issue for many years:  Due to continental drift and whatnot,
the lat/lon coordinates of geologic features *must* drift over time.
You can choose a datum to minimize the drift of things you thing "should"
be considered fixed points, but you can't get rid of all drift.  As a
result, when plotting stuff in North America, you have to decide whether
to use the North America Datum (NAD 83) or the World Geodetic System
(WGS 84).  They differ by a meter or so.

Also note that both of them differ by about 100 meters from the classical
prime meridian as defined by the Royal Greenwich Observatory.

If the geodesy guys can handle drift, why can't the timekeeping guys do
the same?