[LEAPSECS] Caveat emptor!

[Rob Seaman]

Poul-Henning Kamp wrote:


> Rob Seaman writes:

> 

>> The "day" in time-of-day is the synodic day.

> 

> You keep stating this as a bare fact, without any kind of substantiation,


Please consult 100 other messages.  The days counted by our calendars are synodic days.  Length-of-day is not an arbitrary parameter.  Static offsets or periodic shifts are irrelevant, but you can't just let the clocks run fast forever.  By all means consider the issues associated with pretending that the length-of-day is other than what it actually is, but that is what it is, a pretense.  There are contingent implications of conducting such a pretense.  System engineering best practices exist precisely to address such complications.


> and counter to the plain fact that "a day" can have any relation to earth rotation a legislature wants it to have.


...and:


> Greg Hennessy writes:

>> 

> 

>> My philosophy is that if you want/need TAI, you know where to find it.

>> 

> 

> I would generally agree, but I can see the sense in keeping the fundamental timescale free from legislation.


So which is it?  Do we humbly petition our wise legislators, or do we strain mightily to evade their odious fetters?

There are questions of fact and there are questions of law.  By all means attempt to legislate the value of pi.  Similarly, length-of-day is 86400 SI-seconds + epsilon.  You can legislate that epsilon equals zero, but that doesn't make epsilon go away.  In fact, if you want to mitigate the impact of that epsilon, the first step is to describe it in coherent and common language.

You simply aren't going to get me to agree to a proposition that any legislature (or international bureaucracy) can unilaterally declare the length-of-day to be - say - 50,000 SI-seconds or 86,399 SI-seconds or 86401 SI-seconds or even 86400 SI-seconds.  It is absurd to assert that reality has no say here.

The vote in January isn't just about leap-seconds.  It's about length-of-day.

Rob