[LEAPSECS] Metrologia on time

Rob Seaman seaman at noao.edu
Thu Aug 4 12:16:54 EDT 2011

Hi Tom,

> I'd guess that any person or any system that needs DUT1 has long since switched over to telephone, fax, or the internet to obtain this information.

"I'd guess" is not an inventory.

> The internet is faster, more reliable, and far more global than LF or short-wave timecodes ever were. Further, you now get 4 or 5 digits of precision instead of just 1, as well as history and predictions tens or hundreds of days in advance. All with one line of code and a URL.

"One line of code and a URL" is not an Interface Control Document.

> So regardless of what happens with leap seconds, I vote that the lonely DUT1 digit no longer be updated in time codes.

The only votes that count are the member states of the ITU-R.

> Few tears were shed when GOES went away. Or Loran-C. Or CHU. Or pop-corn. Or when analog TV time/frequency went off the air. It's time for DUT1 at 1 baud to make its exit as well.

"The tire didn't come off with the first four lug nuts, so why should it with number five?"

In contrast, the transition from analog to digital TV was planned out and executed in great detail. Consensus-building occurred before decision-making. Provision was made for backwards compatibility. Funding was provided to end users in the form of coupons for a simple conversion device.

The unwisdom of decommissioning LORAN did receive negative comment.

I deployed and operated an automated feed of GOES data from the 9600 baud UNIDATA stream in the 1990's (as well as a more direct whole-disk mbone feed from the AF via Maui). This was just before they transitioned to the 3-axis stabilized spacecraft. There was a period of time when the U.S. fleet was down to one GOES and they parked it due south of Arizona, very handy for astronomers in the Southwest (though our risk of hurricanes is quite low). Reports of GOES's demise are greatly exaggerated:


On the other hand, there is a nice retrospective of the GOES time code service:


It would be prudent to generate similar reports for DUT1 before shutting it down.

The elimination of these other time signals has acted to increase, not reduce, the importance of getting it right for UTC and DUT1 themselves.

> If you want to play it safe perhaps NIST could freeze DUT1 in WWV/WWVB at its current value of -0.3 until someone calls to ask if there's a problem.

Is the notion described here safe? How would one test such an assertion? Is the operational system the correct sandbox?

Redefining UTC and contingent changes will directly or indirectly impact 7 billion people. Why would the ITU-R seek to "play it" anything other than safe? It is the ITU that will be left holding the bag, none of the other alphabet soup of organizations.

> After that issue is settled, then permitting DUT1 to temporarily spill over 1.0 seconds is the next hurdle.

An incremental deployment plan is indeed likely to be safer than going cold turkey. Having a plan of any sort is even more fundamental.

I remain unclear on how the application of "time tested" systems engineering best practices is a controversial position.

Rob Seaman
National Optical Astronomy Observatory

More information about the LEAPSECS mailing list