Lists Home |
Date Index |
Yeah, standards fail too. But let's ask if the
process will be improved, and the risk made
more evident if we separate specs from standards
and assert claims based on two different processes
with two different sets of inputs. As I recall,
the W3C started out like that. What information
or revelation changed that stance?
It may not fix the problem of a flawed standard,
but it could be better. I realize that the W3C
also wants implementations before it signs off,
but evidently, that hasn't work in this case.
Maybe when we know that a spec and the technology
it requests have global risks, we have
to slow it down. Given vendor domination,
it is hard to do that at the W3C and no
single individual should have to stand up to
that kind of pressure. If we separate these,
then we put the risks right up front.
It isn't a problem of who; it is a perception
of authority over expertise and unknowns.
DSDL appears to be sensible. Why? Because
the experts now have a solid foundation of
mistakes and successes. Factoring these out,
doing the merging is now not a matter of
guesswork. In my opinion, that makes the
technology a ripe candidate for a standard.
What is the problem here? It means a delayed
gratification and a longer risk for early
adopters. We would have to accept that we
might have to toss one or two away; even ones
we paid for. Ok. How is that diffferent than
what we are doing now?
From: Tim Bray [mailto:firstname.lastname@example.org]
Bullard, Claude L (Len) wrote:
> Whatever the merits of James' analysis,
> it is about two years too late.
I really hope you're wrong. Early days yet.
> Let me ask: should we quit claiming to be creating
> standards and admit the role of specifications? That
> is, one implements a spec and assumes the risk that
> it may have flaws. A standard, by contrast, should be
> relatively risk free because the implemented technologies
> and the communities that use them have uncovered them.
Oh "standards" are "risk-free" are they? I know a guy who can tell you
about the greater than a billion dollars he helped a well-known computer
company piss away betting on OSI networking. I could cite the name of a
half-dozen other "standards" that are now industry in-jokes, and so
could you Len. The notion that a standard's origin has much to do with
how good it is is unsupported by the evidence. -Tim