On Aug 13, 2005, at 2:05 PM, Warner Losh wrote:

If we had a 10 year schedule, every 2 years or so the
next two years would be scheduled.  Eg:

        Jan 1, 2006     announce leap seconds through 2016.
        Jan 1, 2008     announce leap seconds through 2018.

etc.  This would still allow for the sychronization of the civilian
time scale, while still allowing for more predictability.  Over time,
we could likely expand the time horizon for the predictions (eg do it
out 15 years instead of 10, then 20, then 25, etc).

I'll add this to the tree as something like "sliding" or perhaps
"adaptive" scheduling.

Such a scheme would optimize multiple variables, with no single one
dominating.

I'm going to refrain from commenting on the "best" choices from the
decision tree until it nears completion.  However, certainly one must
suspect that any optimum solution will relax a wide selection of the
variables jointly.  It is also quite likely that more than several
solutions will be identified that are near optimum in such a fashion.

Rob Seaman
National Optical Astronomy Observatory

Reply via email to