[LEAPSECS] Consensus has value

Warner Losh imp at bsdimp.com
Mon Feb 7 18:13:39 EST 2011

On 02/07/2011 15:15, Rob Seaman wrote:

> Warner Losh wrote:


>> On 02/07/2011 14:03, Rob Seaman wrote:

>>> Tony Finch wrote:


>>>> the whole point of universal time is that it's the default timscale

>>>> for civil use and only specialists should need anything else.

>>> Stephen should add this to the consensus building list.

>> Yes. Along with the point that civil time keepers use whatever they are told to use... (eg, the shift from UT1 to UTC and China's very wide timezone).

> Again - the point is to build consensus, not to divide and conquer.


> An assertion that "civil time keepers use whatever they are told to use" is not something I would back when engineering a system, no matter what my opinion on other issues. It is a complex value judgement and likely irrelevant (and probably wrong on its face in this instance). Perhaps you can reword it?

Civilian time has traditionally followed whatever was easiest to follow,
or what most people were following. This tradition has been formalized
into timezones and the slight differences between differing details have
been ignored when it was inconvenient to follow them.

Might be a better way to put it. Civilian time users just need to agree
on what time it is amongst the various parties. Everything else is a
second order effect. My statement wasn't so much to say that "we can
dictate and the civilian users will follow" but rather "once a time
convention becomes established, civilian users tend to not worry too
much about the details and accept whatever greater authorities tell them
the time is." They call the time hotline and/or lookup time via ntp
rather than building their own sundial, surveying it and using it to get
second-accurate time.

> As far as "eg" examples, these also seem orthogonal to the consensus building exercise. Examples are great, but have to be interpreted. I don't know that I understand what you are trying to imply by "the shift from UT1 to UTC".

The official time is becoming UTC. It used to be UT1, or rather a more
strict mean solar time than UTC strictly speaking is (as it is just an
approximation of mean solar time). The US was on UTC years before the
law was officially amended to recognize this fact (instead, relying on
"Mean Solar Time, as interpreted by the Department of Commerce" weasel

> Whatever it is, I suspect it is on shaky ground historically.

I don't see how. More countries are adopting UTC as their official time
than are moving the opposite direction. Even those that don't tend to
use UTC as a de-facto official time, judging from what their time
stations/ntp servers produce.

> And while we've discussed timezones any number of times over the years, we've often disagreed on the interpretation of the issues involved. (Note I'm forgoing my usual explication of my own position here - I think the consensus exercise is of value.)

I just point out that timezones and DST suggest that a strict, to the
second, synchronization with local sun time is unnecessary.

> It appears that Warner does agree with Tony and me that:


> "[T]he whole point of universal time is that it's the default timscale for civil use and only specialists should need anything else."


> Perhaps this can be split in two:


> Universal Time is the default timescale for civil use.


> Specialists may need other timescales.


> That we each may have different ideas about the "meaning" of such assertions is not pertinent to building a consensus on the ideas we do share.

My point here is that "Universal" time is used because it was widely
available, not necessarily because of any other intrinsic property of
"Universal" time. It would be unwise to assume that all the
characteristics of UT1 are required in any successor. UTC approximates
UT1 in many ways, but fundamentally differs from UT1 by ticking in fixed
seconds, while UT1 ticks in seconds that vary by 1e-10 or so. UTC also
has some minutes that have a second labeled :60, but UT1 doesn't have
this. Likewise, UT1 never had days that were 86401 seconds long. These
differences turned out to be not a huge deal for most people. That's
why I said above that tweaks to the system that are somehow promulgated
or become defacto standards are adopted when they are easy. You can see
it in UTC even: everybody does the easy parts, many botch leap seconds
and hope that ntpd and/or manual intervention will paper-over any mistakes.

That's why I said that civilian time follows whatever the recognized
authority for time says to use. The authorities have been tweaking both
the underlying basis of the time (GMT (lots of ways) -> UT1 -> UTC) as
well as which timezones different pieces of dirt observe for the past
century and a half. Some of these changes are minor and matter not at
all, while other changes matter a lot (eg, changing the rate that TT
ticks by 1e-15 didn't matter, while changing GMT from starting at noon
to starting at midnight matter so much that a new name for the new GMT
was invented).


> Rob



> _______________________________________________

> LEAPSECS mailing list

> LEAPSECS at leapsecond.com

> http://six.pairlist.net/mailman/listinfo/leapsecs




More information about the LEAPSECS mailing list