[LEAPSECS] Leap smear

[Rob Seaman]

It remains stationary to some bounds and over some long enough interval.  This was described, for instance, in the 1999 GPS World article.  Both the bounds and interval are currently comfortably small.  Arguing to relax these is quite a bit different than arguing to remove them completely.

Anyway, I don't believe I made an argument in that message.  I made various assertions about engineering processes.  System engineering is how one achieves a cogently stated goal.  My message didn't touch on DUT1, on limits on DUT1, on leap seconds at all, on historical precedents.

The ITU (or rather, parties pushing this proposal through the ITU) have apparently never considered any options from the GPS World article other than ceasing leap seconds.  I am stating that "ceasing leap seconds" is not a coherent position precisely because it ignores the 1s/10s/100s/1000s issue that you raise.  If you want to pursue a goal of ceasing leap seconds there are inherent implications that will always come back.  A coherent engineering plan would describe how those implications will be handled.

Leap seconds are a mean to an end - they are an aspect of one possible class of solutions to the underlying problem space of civil timekeeping.  Desiring a solution without leap seconds is a viable goal (although it seems more work than it's worth to some of us :-)  Asserting that one can simply stop issuing leap seconds is incoherent and incomplete.  Asserting that there are no implications worth discussing - besides being insulting and paternalistic to those of us who will have to rewrite a large amount of expensive software - is rather inane.

Rob
---

On Sep 26, 2011, at 1:11 PM, Ian Batten wrote:


>> Another way to state the underlying requirement is that calendars count integral days.  Points awarded for anybody who can make this work for some definition of day that does not remain stationary with respect to mean solar time. 

> 

> It doesn't currently remain stationary with respect to mean solar time: that's why we have leap seconds.  Your argument seems to be that there's something magical about max |DUT1| being ~1s, which wouldn't be satisfied if it were ~10s or ~100s or ~1000s.   Why is 1s acceptable and 10s not?  When UTC was first confected, the bounds weren't defined like that.

> 

> ian