[LEAPSECS] Metrologia on time

[Ian Batten]

> 

> So giving us 3 years notice of leap seconds instead of six months

> should be a total no-brainer.


As I think we've discussed, there are some systems which cannot handle |DUT1|>0.9 (UK broadcast time, for example).

If there is reasonable three year confidence in predicting DUT1, then there is a high probability that |DUT1| according to the predictions will remain within 0.9s, but if by some ill-chance it goes outside that interval then there is a small chance it will be 1.0s, a yet smaller chance it will be 1.1s, etc.  

DUT1 is disseminated in broadcast systems with a resolution of 0.1s, so systems that have to consume it already have to accept that if they need precision of better than +/- 0.05s, they need to get it from somewhere else.  For systems that _can_ handle +/- 0.05s uncertainty within DUT1, would they accept a small chance that for short periods within a three-year cycle, they might instead get +/- 0.15?  What sort of precision on DUT1 is required for the purposes that it's used for?

ian