There are two kinds of specs: Experimental, and Standard. Standards
are set either by standards bodies or by organizations that somehow
control the intellectual momentum of the name.
I don't think we should try to control or limit the number of
experiments. Experiments are good, give us experience about what users
want or don't want, etc. Names like x-<browser-name>-html are useful
for experiments.
Standards are that: they're change-controlled, agreed to by multiple
parties, or else controlled by a single organization. I think there
should be a limit to the number of standards around.
If an organization wants to register a media-type and widely promote
its use as a standard, we should encourage that. Adobe can own and
change-control pdf, Mosaic Communication can own and change-control
mcoml, and the IETF html working group will manage the development of
"html".
It's really OK for the experiments to last for a year; using
text/x-mcom-html or text/x-ncsa-html doesn't actually slow anybody
down, and will give us good real-world experience with these document
types.
When you do experiments with formats and protocols, there are two
outcomes: the experiment is a success (and you migrate all the old
documents that were labelled text/x-mcom-html to be also labelled as
text/html; version=2.0), or the experiment is a failure (and you leave
all the old documents labelled as they were until you can go through
them and fix all the <CENTER> tags.)
Don't imagine that just because we have a proliferation of experiments
that we also have a proliferation of standards. As more and more
people fool with the web and get into the web tools business, we'll
see more and more experiments. Let's not make all the experimenters
wait before they can deploy their experiments, even to the extent of
launching products! Of course, real world products get launched
before standards are approved. The worst thing that can happen,
though, is when the product developers somehow set their products up
so that you can't tell the pre-release-standard version from the one
that actually showed up in the standard. (This happened, for example,
in the interface to Class 2 fax modems, and made an enormous mess for
anyone writing an application, because the pre-release versions used
the official standard name.)
We already *do* have a mechanism that allows both a formal
standardization process and a rapid prototyping environment for new
browser developers. People just have to use it. It doesn't hurt, isn't
very cumbersome.