[LEAPSECS] drawing the battle lines

[Poul-Henning Kamp]

In message <E1UaGiD-000NIR-DK at stenn.ntp.org>, Harlan Stenn writes:


>Warner, I think your position is only valid form the point of view that

>says a timescale can only be used to count fixed-length seconds.


That is not really what timescales are about.

Timescales, as concept, are for communicating unambiguously the
duration of the time interval between two or more events.

Given the many and varied circumstances of human lives, it follows
that many kinds of timescales are in routine use, from counting to
'safe days' to splitting pico-seconds for science.

Most of these timescales are local, they apply only to one particular
woman or one particular experimental setup in a lab.

The "coordinated" in UTC is all and only about, through international
coordination, providing a timescale which is local to the entire
planet, in order to enable world-wide communication about events
on a global scale.

It should come as no surprise that such a timescale originated from
the TELCO community, when international communications boomed.

It follows pretty obviously, that the important thing about UTC is
that everybody can agree what time it is, with a trivial overhead
spent on "coordination"

Thanks to improvements in timekeeping technology and computer
networking, leap-seconds now impede communication rather than aid
it, because you can't predict them more than 6-8 months ahead, and
they have a unreasonable cost of coordination.

Either the cost and impediment to communication must be reduced
vastly, or leapseconds must go, because the benefit they provide
is utterly marginal.

-- 
Poul-Henning Kamp       | UNIX since Zilog Zeus 3.20
phk at FreeBSD.ORG         | TCP/IP since RFC 956
FreeBSD committer       | BSD since 4.3-tahoe    
Never attribute to malice what can adequately be explained by incompetence.