Lists Home |
Date Index |
As long as people want specification and standard to mean
the same thing, they'll be in a chinese finger puzzle of
their own making and chumps to the ambitions of self-selected
consortia for standards manufacturing.
A spec for a technology that does
not yet exist is research becoming product. A standard
to converge technologies that do exist because such
convergence increases their value to some asset holder
(proprietary, open, it doesn't matter to the definition)
has benefits which have to be reckoned in terms of that
value type to those asset holders.
Netscape was/is a product; not a spec. It
was never a standard either. It was the beneficiary
of some decades of research and the opening of the
internet distribution system. It was only the first,
not the best or the least, but by no means, standard.
But it set an example the market believed was a product
of the forces of standardization instead of a temporal
convergence of the forces of cheap memory and processors,
the opening of the Internet to commercialization, and
the free IP. Trying to emulate its success is like
trying to be the next Beatles. Hard to do because the
same conditions might not exist. Don't buy the
mythInformation of Killer Apps. It's a crock.
Web services are a redux of earlier efforts at creating
networked applications that balance workload. The
visions of 'seamless computing' or "frictionless economy"
are crocks. That network resources can be made available
programatically is not. It seems to me that the web
services picture falls apart depending on which problem
one wants to solve with a web service interface.
If the only web service a system exposes is a query
processor constrained to a schema, that works ok under
common practices. Security? Not so far. So I think
we agree here. Choose wisely the problem to be solved.
From: Hunsberger, Peter [mailto:Peter.Hunsberger@STJUDE.ORG]
Sent: Monday, April 05, 2004 12:45 PM
To: Bullard, Claude L (Len); email@example.com; firstname.lastname@example.org
Subject: RE: [xml-dev] Competing Specifications - A Good or Bad Thing?
Bullard, Claude L (Len) <email@example.com> writes:
> It depends on the local issues. A single spec for a
> new technology might indicate a proprietary development
> ripe for IP exploitation. If it works, why not?
Seems to me you're trying to have your cake and eat it to: on the one
hand any spec. that is "ripe for IP exploitation" is ok, but OTOH, any
spec. that prematurely codifies practice is bad. How are you supposed
to know the difference?
> everything is in the commons and that assumption of
> community property for all things web is one way to
> sort out the naive and inexperienced including those
> who offer up single spec/single use specifications
> labeled or processed as standards. Respect for IP
> is the way forward. IP keiretsu in the form of consortia
> managed royalty free contributions will work both
> for ensuring that submissions are vetted under
> participation agreements, and for keeping as much
> IP as is workable in the commons of jointly indemified
Sure, but that seems only peripheral to the issue is when to spec. and
how much? A lot of the spec. world isn't about IP, it's about trying to
standardize business practices. Trying to use specs. to leverage first
mover advantages is a tad like bit like, <bad-metaphor> trying to keep
the wilderness pristine by building a theme park in it</bad-metaphor>
(though when someone does manage to sneak -- Netscape like -- under the
radar, the advantage can be enormous for a while).
> People make assumptions. That is how they learn. If
> they don't, they fail. Life and death in the ecosystem.
> We got here because too many stopped focusing on developing
> software and started playing the standards game. I blame
> the W3C squarely for that. This community made its own
> problems and this community will have to face up to the
> job of fixing the mythInformation it created.
Given the ambitions of "web services", the ill defined scope, the
previous history of universal interoperability specs of the same nature
(CORBA or the OSI stack anyone?) and the ever increasing pressure to get
to market I can't imagine why anyone would expect any of this to emerge
cleanly or rationally?
Every time I preach "how to do a system architecture" it usually
includes a phrase something to the effect of "use standards where there
is a significant body of best practices that clearly match the business
problems you are dealing with". IOW, pick you risks wisely...