[LEAPSECS] The relation between calendars and leap seconds.

[Poul-Henning Kamp]

There is one requirement you do not seem to even think about, much less
mention:

	The limited human intelligence.

As we saw a couple of years ago, the 400 year leap-year role is slightly
above the level of complexity humans can deal with, a significant
fraction of "people who should have known" didn't.

Daylight savings time is borderline, but thanks to the implementation
between two non-workdays, fallout is mostly contained to social
embarrasment.

Another smart feature of DST is the fact that UTC is unaffected so
intelligently designed software only see DST as a representational
problem and not a problem of timekeeping.

Leap seconds fail the IQ compatibility test on so many levels that
it is unnecessary to reiterate them here.

So further requirements for The New Time Order are:

	Ia) Without steps.
or
	Ib) Have steps that are sufficient large to be noticed, and
	   they must fall at times where the fall-out from mistakes
	   is reduced.

and
	II) Not require human intervention for automatic systems to
	    perform timekeeping on the fundamental timescale.

Poul-Henning

-- 
Poul-Henning Kamp       | UNIX since Zilog Zeus 3.20
phk at FreeBSD.ORG         | TCP/IP since RFC 956
FreeBSD committer       | BSD since 4.3-tahoe    
Never attribute to malice what can adequately be explained by incompetence.