Re: USNO leap seconds - a minmum-change approach

From: Rob Seaman <seaman_at_noao.edu>
Date: Sat, 19 Aug 2000 08:55:52 -0700 (MST)

Glen Seeds does an excellent job of describing the status quo. Remember
that doing nothing was one of McCarthy and Klepczynski's five options.
Here is one reason to choose that option - people understand it even if
they don't realize it.

Brian Garrett says:

> I haven't yet commented much on this issue since I'm an interested
> bystander rather than a member of the PTTI community.

Aside from the obvious fact that the members of the PTTI community
don't appear to be interested in commenting either (at least on this
list :-) the reality is that this is not predominantly an issue for
technical experts. We are discussing a change to civil time as used
by six billion (and counting) human beings.

> This whole issue has, from the beginning, fallen into the category
> of "fixing what ain't broke"

Here I don't agree. The good folks in the IERS community could surely
provide us with copious data detailing the continuing slowing down of
the Earth's rotation. Something most certainly will need to be changed
in the future. The question is whether current ephemeral technical
issues will direct both the choice of solution and the pace of its
implementation.

Steve Allen says:

> It would be a Really Good Thing (TM) if the understanding of 64-bit
> Unix time were implemented in a fashion more conformant with the
> reality of leap seconds than the current 32-bit definition is.
> The protocols necessary to do this in a robust fashion need to be
> determined ASAP.

and makes a plea for help:

> It would be nice to see some responses from respected members of
> the timing community.

It seems to me that if this mailing list is anything other than a tool
of some faction of the PTTI community whose purpose is most certainly
opaque to the rest of us, that very pertinent questions like Steve's
would not go unanswered.

John Cowan says:

> The trouble is that many systems cannot "download" the magnitude, as
> they are not on line. Just think about all the different clocks there
> are in the world in various embedded systems ranging from microwaves to
> cars. In general, we do not care about errors in the clocks on these
> systems, as they are used only by humans, but in a future version this
> might not be the case.

I'm struct by the phrase "used only by humans". GPS is used only by
humans. GLONASS is used only by humans. Apparently the spread spectrum
community has problems handling leap seconds. Who cares about spread
spectrum technology beyond humans?

I'm also struct by the notion that no human would ever have personal
need for a clock accurately synchronized to the Earth's rotation. After
just a few years of our projected bright-and-shiny leap-second-free future
(and certainly after a few decades), any student working on a science fair
project anywhere in the world (except perhaps the polar regions) would be
able to discern the discrepancy between clock and Earth. (I suppose an
exposition of the new time-varying sidereal corrections might become a
replacement for the ever popular volcano model.)

Correct me if I'm wrong. M&K offer five possible future choices.
I might combine some and add others, but basically they come down to
continuing to do what we're doing now (at least until it turns into
somebody else's problem), to redefine the second (I challenge anybody
to find a physicist to go along with that), to tinker with the scheduling
algorithm (in which case folks like us wouldn't even be paying attention),
or to just forget the whole thing and allow all the world's clocks to
drift from the standard time grid that our 19th century counterparts so
cleverly devised.

Why is the last option the horse to beat?

Well, it's suggested that emerging technologies have intrinsic
difficulties handling leap seconds. Now apparently it's only "some"
emerging technologies - and they aren't really "emerging" at all, make
that "interim" - but are the difficulties intrinsic? That's the real
question.

What fundamental timing complications occur due to leap seconds?

That is - what fundamental complications that wouldn't be more
appropriately addressed by merely selected an unsegmented time scale
like TAI for that project in the first place? It worked for GPS, why
not for the other guys? After all - these are highly technically trained
(and highly economically advantaged) folks - why should everybody else
(literally "everybody else for the rest of time") be expected to
accommodate them?

Rob Seaman
Received on Sat Aug 19 2000 - 08:56:09 PDT