Re: Ballot on P1788/M0001.01_StandardizedNotation
Dear Van
I agree with almost all of the specific criticisms you make of the
"Standardized notation in interval analysis" paper, but disagree with
the conclusion you draw.
- Yes, the descriptions of notations are imprecise. (But as
definitions-by-example they are mostly precise enough to use.)
- Yes, "e.g., upright normal for matrices, ... conventions are to be
used in the standard need to be specified" for us to develop the
standard text consistently.
- Yes, the notation [..,..] could be better introduced, and I feel
the relation between "pair of column vectors", "column vector of
scalar intervals", and "box as a set in R^n" needs clarifying:
important when some components of a vector of scalar intervals may be
the empty interval.
- etc.
But you write
"It is deficient as normative text in a standard, or as
instructions for later developing normative text. It might usefully
serve as informative material, but presumably there are parts of it that
ought to be normative."
My case is: in its proposed status as an appendix to the standard, it
is informative. None of it is normative. It inherently cannot be. For
one thing, there is nothing in our Project Authorisation Request
(PAR) about standardising notation. And what would it *mean* for it
to be a normative part of the eventual standard? That we could ask,
of any article written about intervals, whether it were "P1788
compliant"? We do not aim to be that sort of police.
Whether we accept it as binding on us while creating the standard
text, is irrelevant. That merely means we take it as normative for
our *process*, which is a quite different thing.
Yes, "It is deficient as ... instructions for later developing
normative text". I hope we produce in due course a crystalline
definition of interval notation that shines as a beacon to future
generations. But, as part of P1788's text it won't, and can't,
contain a single normative sentence. If we were creating a standard
for software that automatically writes papers on interval analysis,
that would be a different matter.
Personally, as M0001's Rationale suggests, I would rather accept the
notation paper as it is for now, harden up the notation "as we go",
and formalise an improved version later. But if
- you are not persuaded by my argument,
- or you feel, say, that from your long experience of standards
work, getting the notation issue sorted up-front will save us
time in the long term,
then I am happy, following our rules, to have an amended motion
discussed and voted on.
Best wishes
John Pryce