[time-nuts] The future of UTC

[Chuck Harris]

Poul-Henning Kamp wrote:
> In message<4E203B60.6080809 at erols.com>, Chuck Harris writes:
>
>> Nice article.  One thing stands out to me, though:  How do you
>> propose knowing 20 years in advance the schedule of leap seconds?
>
> That is for the geophysical community to figure out.  They still
> get to decide when leap seconds happen, only they have to tell
> the rest of us 20 years in advance instead of 6 months in advance.
>
> How well they can do this (ie: how small can they keep DUT1)
> depends on the quality of their science (and/or coin-flips)

I can see a 20 year prediction being seriously fraught with error.

I tend to think of the machine clock as being something that should
keep ticking along one SI second to the next, keeping count of the
seconds since some epoch.  I am not at all happy with the idea of
having it magically stall, or stutter. That's something for some
library function to keep track of after the fact.

Unfortunately, the leapsecs list never made the threshold of my
free time allocator.  I subscribed for a while, but found it rife
with bickering that seemed intractable.  Everyone of the reasons for,
or against, the leapsecond was valid, and incompatible.

-Chuck Harris