In an ideal world we might wait longer before initiating
standardization, but I think we must balance the cost of not acting
against the cost of acting. My own view is that there is FAR more
pressure to stabilize DC (communities saying, in effect, if you don't
stabilize, we will proceed down our own path) than there is resistance
to deployment because some of the details are not perfect.
In the IETF world, conformance testing is trivial: the bits get from
point A to point B, C, D, and E, or they don't (of course, I
oversimplify). It is an interesting question, in the metadata world,
when we would say we have interoperability and when we don't. In the
world's largest deployed metadata system (MARC Cataloging), there are
many inconsistencies and shortcomings, but it nonetheless is useful. We
try to deal with the problems, but never with the expectation of
perfection.
We should take comfort from this example of imperfect usefulness and
forge ahead decisively. Not to say that we cannot, in parallel, study
the effectiveness and interoperability of DC and improve it.
On Wednesday, January 14, 1998 5:20 AM, Jon Knight
[SMTP:[log in to unmask]] wrote:
> On Tue, 13 Jan 1998, Weibel,Stu wrote:
> > In related developments, NISO has approached us about standardizing
the
> > DC, and I'm of the opinion this would be a strong asset for our
> > community. While NISO is a national standards organization, rather
than
> > an international one, it is a reasonable place to get our feet wet
in
> > the formal standardization process.
>
> Before DC is formally standardised, shouldn't we have some more
extensive
> deployment and interoperability testing first? I'm from the old IETF
> school of "running code and rough consensus" rather than the old ISO
> "standardise then think about implementing". No point in going
through
> the formal standards process with something that we later find to be
full
> of holes (not that I think DC is you understand). At least wait until
the
> Helsinki break out groups all report back with the various details and
see
> how the whole thing hangs together before passing the DC to NISO, or
any
> other standards body, for consideration.
>
> Tatty bye,
>
> Jim'll
>
>
-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
-=-
> Jon "Jim'll" Knight, Researcher, Sysop and General Dogsbody, Dept.
Computer
> Studies, Loughborough University of Technology, Leics., ENGLAND. LE11
3TU.
> * I've found I now dream in Perl. More worryingly, I enjoy those
dreams. *
|