Standards? Conventions? Design Patterns? Whatever Works!
I see Rick Jelliffe finds the previous post
amusing. Glad to
entertain, but please note my
comment.
I definitely agree that "The whole point of a standard is to prevent
one party from having control". That's what the imprimatur of a
standards organization offers. The question I was posing is
how much work refining and field-testing a spec should done before
submitting it to the standards organization. The dominant
paradigm has been to do fairly little and leave most of the work to the
committee. The alternative is submit fairly mature specs for
standardization. This doesn't have to be done by some unholy alliance
of the major companies; the W3C itself now has a mechanism to form
"XGs" (X signifying something like "experimental" I guess) that can
develop specs that are not on the Recommendation track, that can be
picked up by a regular WG if they prove themselves. But when the
standard ultimately comes out, the organization, not the originator(s),
own it in any case, whether it was developed by a WG, XG, or industry
cabal. (My attempted comment
on his blog hasn't shown up for whatever reason).
Another interesting post on this general topic is Kurt Cagle's
XML and the Long Tail.
"if, as I suspect, standards adoption tends to occur first within the
tail, the growing adoption of that standard creates pressure on those
closer to the head to conform to the standard as well, which in turn
makes the momentum stronger even closer to the head. Eventually, this
forces the market leader into a position where they either adopt the
standards or risk losing their market dominance in the face of
overwhelming opposition from both competitors and clients." Kurt
doesn't give specific examples, but let's think of some.
Presumably the model for this would be HTML, which definitely did
emerge out of the long tail and displace the proprietary technologies
of market leaders (e.g. MS's ill-fated Blackbird format). RSS
also comes to mind. But most would argue that HTML and RSS moved
out of the long tail not because they were Open Standards but because
they were (originally)
simple sloppy
formats that absolutely nailed the 80/20 point of functionality at a
vastly lower price/complexity than the stuff at the "head." More
recent examples of technologies emerging from the long tail and getting
somewhat grudging support from the head might include REST/POX and
AJAX/JSON. In these cases, it's again a matter of fundamental
standards (HTTP,XML, ECMAScript/Javasript having been universally
implemented by the head) and sufficing where the head-favored
alternatives added little value (or had not yet been widely deployed).
What drives these things is
not standardization in the sense of
their being a formal, approved spec that developers scrupulously adhere
to. Valid HTML is not particularly common on the Web, many
so-called "REST" apps violate the HTTP spec in fairly dangerous ways,
RSS is actually a constellation of informal conventions rather than a
standard (the real-world success of the Atom 1.0 IETF standard is yet
to be determined), and AJAX/JSON is more of a design pattern than
anything resembling a standard. What does drive this is
evolution -- a
diverse pool of ideas, lots of experimentation / recombination,
real-world success to the most practical ideas, and a weeding out of
the bad ones. At some point things stabilize to the point that
certain ideas get enshrined as "standards" and the evolutionary process
slows down for awhile, until the equilibrium gets punctuated by new
challenges and ideas. That's very consistent with the
"standardize the stuff that has been field-tested, don't standardize it
direct from the lab" approach I was advocating yesterday.
The larger context of many of these discussions seems to be the
controversy surrounding
the OASIS OpenDocument format and the Massachusetts CIO's effort to
mandate it in that state's agencies. As someone who hopes that MS
Office's profitability continues to fund our little non-profit center
in the bowls of the borg, I'm a bit biased, but c'mon folks: This
isn't about an Open Standard getting traction in the long tail and
forcing the "head" to reluctantly do the Right Thing, this is about
using government power to counter market realities. That seldom works
in the IT industry, as the fact that
Ada is
a niche programming language today should attest :-) OASIS ODF will
succeed if it meets an un-met need at a significantly lower real cost
than the alternatives, the way HTML, RSS, etc. have
succeeded. Historically, it has been more or less irrelevant
whether something hitting that sweet spot has been an Open Standard, bunch
of conventions shared by a developer community, or a de facto standard
derived from a proprietary technology.
