[LEAPSECS] "China move could call time on GMT"

[Warner Losh]

On Jan 6, 2012, at 1:29 PM, Rob Seaman wrote:


> Tony said:

> 

> 	"I reckon the timezone fudge is workable for rate errors as large as 1e-5, which would imply a timezone change every 11 years."

> 

> The resulting discussion posits the very situation you're spurning.

> 

> More to the point the entire notion of playing musical chairs with the worldwide timezone system ignores the perpetual rate error that matters for astronomical and aerospace applications.  It happens to also not be realistic to expect civilians to shift to UTC+(N+1) every decade but more importantly this would do absolutely nothing to address the technical requirements.  It is purely a rhetorical gimmick to justify redefining UTC without having a plan for mitigating the resulting impacts.


An hour every 10 years is 360 seconds a year, which is 20x faster than UTC can tolerate as it is defined today.  You'd need a leap second every day to keep up.  By that point, either you have to give up synchronization, or you have to transition from a observational calendar (which UTC is today) to a formulaic one.  Or you have to make some other kind of allowance to keep things in sync (redefine second, redefine hh:mm:ss notation into something else, etc).  Trying to force today's UTC into such a future time scape also seems gimmicky.

It all comes down to what time on the clock should tell us: earth angle (eg, where the sun is) or elapsed time since an epoch. This whole issue boils down to that.  Do we continue to take another step away from clocks showing the sun time and completely decouple?  Or do we say that time zones need to be stablish and to take this step is a step too far?

Warner