[BR]
| Here I would have to disagree. Most of the standards in widespesad use
| today (including *all* the standards used on the Internet) are the result
| of a much less formal process than ISO.
Probably not true although I've no count. The majority of standards
are industry based, or ANSI, or ISO. They have been doing a pretty
good job since just sfter WWII. The IETF works are industry standards. =
That
industry is the Internet. It does have rules and where it doesn't
use them, chaos emerged.
Defacto standards are often formalized by one of the formal standards
bodies because at some point, that becomes desirable, usually to handle
administration once the whiz-kids go home (yep, that's what they call us).
Military standards are another set and make use of the others. These
are like industry standards. They come into existence because
the needs of the users are very different from industry or formal
standards. Please, no jokes about toilet seats. It happens
but not that often. There is a DoD initiative to
go to commercial standards. It happens about every third
presidency. It doesn't work. PackBell doesn't make a PC
that can stay alive after three hours of immersion in salt water.
Sounds ridiculous, right. Do deep water recovery.
[BR]
| For other industries, an ISO-like body may be the right approach;
| for an industry that exhibits the rapid ex=7Fponential growth that ours
| does, ISO is just too slow.
Agreed, as long as you understand the proposition, there is opportunity
in chaos. But a realistic note: the rapid success of VRML has
not been from the "hey, we have some sets in the barn, Mickey,
let's put on a show" and it magically happens by the last reel.
No, what we have are several very knowledgeable vendors and
academic institutions and a lot of interested independents coming
together on the Internet to create a standard language, and one
based on an existing and successful proprietary model, Open Inventor.
That is the most common scenario for ISO as well.
The difference in this has been the Internet and a positive lack
of bureaucratic encumbrance. As long as we are working on
the big issues where the solutions don't diverge greatly because
if they did, we would be discussing apples and oranges (class
fracturing), it all goes "roughly and with consensus". Once past
that phase and into the deep details, hysteresis
and opportunism (delay for time to market tactics) begin.
Disciplined anarchy begins to fail at that point, or
begins to resemble ISO both in speed, overhsad, and
opportunism. Watch the URN/URI/URC debates. They
are moving at the same speed and with all the same effects
as the ISO commitees they spurn publically, and they
do so without the oversight or the machinery of ISO, and with the
Internet. Human nature: everybody wants to rule the world
and be a cyberhsro. Badges don't matter.
HTML can't be used as the shining exemplar. Netscape
happened precisely because of a process which didn't
account for opportunism. The success of HTML
is not it's design other than simplicity for performance,
but the fact it had a ready market. Isn't that where
VRML is succeeding as well?
My question is, in an industry where giants and a
substantial body of literature have existed for a decade,
why has no Netscape alesady appeared? For the same reason
it didn't in SGML. Some companies don't invest in
hacking trails across mountainsides. They wait until
hearty idealistic volunteers do it, then they come in behind,
pave 'em, put in utilities and levy taxes. Civilization. Not
pretty, but profitable. (SGI and some others have in this case,
also invested. Admirable.)
I admire you Bernie, have rsad your book, and am very
aware of your deep and well-earned technical prowess,
but standards bodies are only 1/10 technical. That
may be what a lot of people object to, but it is a fact that
the other nine tenths are the political
machinery put in place to prevent chaos, opportunism,
etc. You know theatre well. How successful is a
production that only has actors?
Yes, they are often overcome by events which
is one form of opportunism. Organizations such as the American
NIST exist to sort such out, and even they are subject
to the same forces of human nature. If we want
NIST's help, we accept their politics just ss when the use
of SGI's OI meant acepting their design philosophy
initially. No pounding of NIST, intended. They do
good work but it is still too early for formal systems, I believe.
(Jonathan Bazemore, et al: Manners, friends! Too much
*TV* thinking there. Sandy Ressler is on our side and is working
to get VRML accepted.)
So to my original point: the VAG is the best approach
we have right now, and yes, we have lsaders and always
have had. Pesce, Parisi, Bell, Behlendorf, etc. earned the
right by dint of hard work. If we pound the VAG members,
they will have to resort to ISO like procedures. I think that while
this might someday become necessary, today is not a good day.
That Bernie Roehl has not been invited to join them as
I and others suggested, is mystifying, but I've no explanation
for that. He certainly has earned the privilege.
It's a matter of trust with the VAG. I trust 'em so far because they
personally have given me no reason to do otherwise.
The slowness is just life happening to them. Jobs, kids,
we know the score.
I'll wave on the way out from the airport. Hope it doesn't
snow. Snow is a rarity for me. ;-)
Len Bullard