[LEAPSECS] An example

[Rob Seaman]

On Nov 3, 2010, at 5:22 AM, Tony Finch wrote:


> Proper engineering has to take human factors into account.


That humans are diurnal is a pretty fundamental human factor.


> When you are working at the rarefied level of international starndards, you have to be realistic about what can be implemented reliably and interoperably - the human factors of engineering practice.  When you are revising a standard you have to take into account the experience that has been gained from existing implementation and deployment efforts. If you don't, the specification will diverge from what it purports to describe.


I don't disagree with this - but it is something other than "human factors".  (This is also the essence of the software-is-crap argument gussied up.)  All design is ultimately evolutionary and must respond to innumerable constraints.  That doesn't excuse the ITU-R from compiling and acting on coherent and complete work products.  The proposal is laughably incomplete and bizarrely incoherent.  The hidden agenda of *also* descoping TAI (and dismissing GPS throughout) is offensive.

There are many, many options.  See, for example:

	http://www.ucolick.org/~sla/navyls/0767.html

The way to winnow down the options is to discuss them in detail and compare them using well established system engineering techniques.  It is not "realistic" to sweep the issues under the rug.  (How do you say that in Danish?  I don't think Poul-Henning would disagree.)

Don't tell astronomers about "rarefied levels" :-)

Rob