Re: [LEAPSECS] What I do. What do you do?

From: Ken Pizzini <ken_at_HALCYON.COM>
Date: Wed, 29 Jan 2003 00:39:45 -0800

On Tue, Jan 28, 2003 at 04:30:41PM -0700, Rob Seaman wrote:
> Perhaps it would help our discussions to simply describe our professional
> (or otherwise) connections to UTC and other precision timing issues.

Okay, I'll bite.

I have no direct professional connection to UTC. I am a computer
programmer who has, at various times, had to deal with a variety of
different frustrations related to how time is kept, one of which is
reminding other programmers that "because of leap seconds, not all
days are 86400 seconds long, stop building that assumption into your
programs" (a very poorly received warning, I might add). I have a
strong appreciation of the value of a good standard, and likewise a
strong disdain for any standard which I can recognize as poorly put
together.

While I have limited technical background with precision time, I do
have a strong incidental personal interest in its issues. While my
weak background makes me sincerely doubt I will be in a position to
slice fine technical details in discussions of precision time, I do
feel that I have ample background to understand the kinds of issues
involved, and because I lack a certain amount of sophistication, I have
some ability to notice when people are making claims that are based
on "I'm used to thinking this way" because their arguments fail to
enlighten me as to "why this way of thinking is correct and relevant".
(Put another way: if you know you're right and yet a posting from me
seems to miss the point, first check your assumptions, and then try
explaining a different way.)



[samples of Rob Seaman's job task descriptions snipped for brevity]
> The notion of discarding the connection between UTC and GMT is not
> just philosophically distasteful - it clearly would make my job more
> difficult for zero benefit.

Given my position as an intelligent-but-uninformed observer, it seems
like UTC is a "good enough" standard for your uses. I see that UTC, as
defined, is useful to your doing the job in the way that you currently
do it, and certainly more useful than TAI-plus-constant-offset would be;
if those are the only two options then of course you personally will
favor UTC. I do not see that UTC is a maximally useful time reference
for your work, though.

When you look at pictures of stars as they vary over time, do you
really find that a time reference based on the annual mean interval
of earth rotations _relative_to_the_sun_ to be exactly what the
doctor ordered? Ignore "in the real world where I work" for the
moment; ignore "UTC" and "UT1" and "TAI" and dream a little: what
would an *excellent* time standard look like for your specific niche?
Granted there will be a need to compromise eventually in order to
make everyone (almost) happy, but I think the discussion to date of
(crudely) "UTC vs TAI" fails to take significant details into account,
making any hypothetical consensus less optimal than it could be.

Yes, civil time, an important consumer of any time standard, would
much prefer approximation-of-earth-rotation-relative-to-the-sun
than random-arbitrary-interval (e.g., 100,000 SI-second "days")
or approximation-of-earth-rotation-relative-to-"fixed"-stars as its
time standard. But civil time is a different consumer of a potential
time standard than you-as-worker-for-NOAO-Data-Products-Program,
and wouldn't be nice if someone smarter than both of us could,
upon understanding the details of the issues involved, see a better
solution than UTC which addresses both of these (and a vast array
of other) uses? Without the details of what the various "ideal"
standards would look like, such a hypothetical insight seems, to my
mind anyway, impossible to come by. I realize that the astronomical
community has evolved to a consensus that UT1 (approximated by UTC)
is a highly useful way to mark time, with the additional feature that
it is usable as a civil time standard, but there is so much of that
evolution which is based on historical accident rather than purely
technical requirements that I find it hard to believe that there would
be no possible way to improve upon it, even after non-astronomical
constraints are factored in, if only it were possible to start anew
with a clean slate.


As to FITS standards: well of course they are UTC based --- that is
what the current time reference standard is; I wouldn't expect them
to use anything else (even more so since much of this is data whose
collection depends on earth-orientation-in-space, not merely SI-seconds
elapsed between observations). And any proposal for a new time
reference standard would need to have a transition plan which would
enable straightforward translation between it and something resembling
UTC for at least a couple of decades while legacy applications,
standards, and on-line data either exceed their useful lifetime or
get converted (archived off-line data would not get converted, though
a means to make such a conversion on an as-needed should be defined,
even if that conversion is (as would be the best case) a "no-op").

A meter is no longer the interval between two scratches on a specific
platinum-iridium bar; the US federal encryption standard is now
Rijndael (aka AES), not DES. Standards evolve, or are replaced
wholesale, for various reasons such as improvement of accuracy,
reproducibility of results with different equipment (including
"sofware implementation" as a kind of equipment), or suitability
for their ultimate purpose. Legacy applications are important,
as is the economic impact of change, but neither is an adequate
rationale for why things should not change if something which offers
clear advantages can be devised. Is UTC really the best that can be
done to meet the varied needs that currently make use of it (because
it is "the" standard)? Maybe, but I sincerely doubt it.

                --Ken Pizzini
Received on Wed Jan 29 2003 - 00:40:16 PST