Re: SUO: Re: Ballot Comment
Matthew makes a very important point:
MW> There are different opinions about how to do standardisation. Some
> are happy to declare a standardisation effort starting with a blank
> piece of paper and a title. Others think that work should reach
> a relatively mature level before being submitted to a more formal
> standardisation process. I have participated in both approaches.
> You may guess which I prefer.
I have also participated in ANSI and ISO standards activities for
the past 10 years. Before that I participated in various IBM
development projects, many of which had a similar loose organization
of committees representing multiple divisions that were supposed to
reach a consensus on a common internal IBM standard.
The two greatest fiascos, one ISO and one IBM, were projects that
started with a blank sheet of paper. (The IBM one was a greater
fiasco, primarily because it was better funded.) The most successful
projects were ones that started with a well-defined specification
that had already been implemented and tested in an earlier, much
smaller project.
Committees are horrible at doing fundamental design, but they are
very good at criticizing an already existing design. (Sometimes
the criticism can be so harsh that they kill the project.) But
when the project is already fairly robust and the participants
in the committee really want to make it better, their criticisms
can be extremely valuable in filling gaps that the original
developers had overlooked.
Bottom line: We need some well developed proposals that have been
implemented and tested, at least at a prototype level, before we
can get into committee-design mode.
John Sowa