[ntpwg] KoD 'backoff' BCP/direction

[Rob Seaman]

TS Glassey wrote:

> the NTP we are working on is a technical curiosity since without  
> those corporate client's and their users, most of us have nothing to  
> do.


A pretty successful technical curiosity, serving a wide range of  
purposes here.  Corporate mileage may vary, but again - what keeps the  
Fortune 500 from buying their own clocks?  Oh, that's right - nothing:

	http://www.romainjerome.com/en/les-inatendus/day-and-night-en.aspx

Tag line:  "An exceptional timepiece that does not indicate the time!"


Not only is better the enemy of good enough - sometimes it isn't even  
better :-)

Some people have nothing to do, indeed.

Brian Utterback wrote:

> Auditor's don't pay. Auditor's never pay.

Yeah - I thought somebody would point that out.  But "physicist" was  
an even poorer fit to the true stakeholders.  In any event, if the  
government or corporate world need a timekeeping system with certain  
characteristics, they have copious resources to bring this into  
being.  That they may not recognize their needs in this area is not  
the fault of the kind and clever folks who have toiled thanklessly on  
the excellent NTP.

> Auditor's issue requirements and other people pay to meet them.

Requirements are not issued, they are implicit in the problem being  
solved and must be discovered.

M. Warner Losh wrote:

> A similar situation with leap seconds: They have been promulgated  
> and everybody must cope.

Well - no.

The world turns.  The moon steals its angular momentum.  Some  
mechanism is needed to accommodate.  It may not be leap seconds, but  
the requirement exists whether or not crabby astronomers point it out.

> it is safe to say that NTP implements them because they are a  
> requirement of UTC time.

Again - no.  NTP implements leap seconds because they are a feature of  
UTC.  UTC itself has been selected to satisfy numerous requirements.   
We're constantly confusing the activities of problem definition,  
requirements discovery, and the specification of a conforming solution.

> Wouldn't the need to have a traceable UTC merit consideration?  Or  
> at least an explicit mention as beyond the scope?

These traceability requirements have yet to be discovered.  There is  
simply a hazy assertion that they exist.  I don't disagree, since they  
aren't well enough elaborated to construct an opinion - even of  
whether they are beyond the scope of the system in question.  What  
that says to me is that the current work before the group should be  
wrapped up with a pretty bow before any other action is taken.

Voltaire also said, "Perfection is attained by slow degrees; it  
requires the hand of time."

Rob Seaman
NOAO