Nelson Minar wrote:
> Median offset of a collection of clocks isn't interesting if error is
> evenly distributed around 0. Ie: if some clocks are -10ms and some are
> +10ms, the median (and average) offset is going to be 0ms.
Median was more of a sanity check - if it wasn't close to zero something
was off in the numbers.
>
> The standard deviation of the offset is very interesting, though, and
> you reported that. What would be even better is a histogram of the
> observed offsets.
>
>
Working on that - probably next week I'll have web pages up with current
an historical data graphed.

John
_______________________________________________
timekeepers mailing list
[email protected]
https://fortytwo.ch/mailman/cgi-bin/listinfo/timekeepers

Reply via email to