Reference Platforms and Standards Models

Since it's been quiet here on lsb-discuss for a long time,
I thought I'd stir the pot a little bit.  The rest of this
message is a snip from an Eric Raymond interview with 
EWeek.   The interview context is /entirely/ Java, but there
just might be some applicability to LSB in here. Some of
you know the context that causes me to visit this, it
will just have to remain a hint however :-(

Comments?

-- mats



I can tell you that one of the courses that I have 
recommended to Sun in public in the past-and one of
the things I'm continuing to discuss with them, 
though I'll carefully not tell you anything about
their end of this-one of the things I'm continuing
to discuss with them is a strategy whereby they open
control of the source. That is, they issue the Java
reference implementation under an open-source license,
but they keep control of the brand. They keep control
of the testing, they keep control of the verification,
they keep control of the certification, they retain the
right to say this and only this is Java once it's passed
our compliance tests and it's guaranteed interoperable
with everything else. 

And I can tell you for sure and certain that a model in
which the reference implementation was open-sourced and
was being collaborated on by Sun and the entire open-source
community, but on another level Sun retains the brand and
the exclusive right to call the results Java and collects
its value from that, that is a model that the open-source
community would be totally happy with. We'd be completely
down with that. 

[Interviewer] I've been interested in this shift, 
particularly in terms of government procurements where
there seems to be a subtle move from calling for open-source
support to open-standards support. Have you noticed this
What's your take on the whole issue of open standards versus
open source?

I would say this: If it doesn't have an open-source reference
implementation, the term "standard" is an abuse of the language. 

That's still a very strong position by today's standards.
But it won't be in three to five years. If it doesn't have
an open-source implementation, how do you know what 
standard means? There's a technical level to this, and there's
a political level. The technical level is that standards
always under-specify things. You never learn everything 
you need to know from reading a standards document. There
are always assumptions and stuff that slip in, and you need
to go to a reference implementation to compare your behavior
with that to be sure that you're conformant with reality as
opposed to the theory of the standard. So for that technical
reason, if it doesn't have an open-source reference
implementation, where are you? You're stuck. You're at the
mercy of the vendor again. It might as well not be a standard
at all.