[LEAPSECS] drift of TAI

[Zefram]

M. Warner Losh wrote:

>Cool...  A gigasecond is around 31 years, a megasecond is ~11 days and

>a kilosecond is about 17 minutes.


Yes.  The kilosecond is a convenient unit of time for intra-day planning.
(Metric day folks endorse the centiday, 864 s, for the same purpose.)
100 ks is a reasonable diurnal cycle.  I hypothesise that a human
civilisation entirely disconnected from natural diurnal cycles would
use an artificial 100 ks cycle for this purpose, and a 1 Ms cycle as
the analogue of our week.


>                                   How would you keep such divergent

>times strait?


Not sure what you mean here.  Relating the SI time units to conventional
time of day is messy; determining which calendar day my birthday falls on
is non-trivial.  In the absence of days and years, though, calculations
that involve only metric units are a lot easier than what we presently
put up with.


>Let's meet for dinner in about 200 kiloseconds?


The linear TAI time now, in my preferred notation, is 0m026.  (This means
26 ks after a 10 Ms rollover.  I'm dropping high- and low-order digits
that don't help for the task.)  Dinner will therefore be at 0m226.
I'll be having some other meals before that, of course, since I don't
want to starve.  If we're living in an asteroid belt, the 200 ks interval
is simply two sleep cycles.

-zefram