[lsb-discuss] What id driving the current ISO hype?

[Theodore Ts'o]

On Wed, Aug 08, 2012 at 12:37:00PM +0200, keld at keldix.com wrote:
> 
> Wrt. consensus-building, all standards organizations I know of have a process for
> approving their standard, and this is the consensus-building part of
> the organisation. I am sure Linux Foundations also have rules for
> how to approve a standard like LSB.

Sure.  There is, however, a large amount of variation between
different standards organization in the distance between those who
approve the standard and the technical experts.  In the IETF, for
example, consensus is achieved by the participating engineers, and it
is deliberately not by company (although the companies who sponsor
more engineers and who do more work do have more influence in how the
standards will come out --- just as the companies who hire more open
source engineers tend to have more influence in the priorities of what
gets developed in various open source projects).

In contrast, with ISO, it's one vote per national body.  An Oracle
marketing manager in Trinidad or Saint Lucia can apply to be the
representative of their country, and have as much of a vote as all of
the United States, and where engineers in the US need to pay $$$ to
Incites or ANSI, and we need to deal with politics at the corporate
level to determine a single vote cast by the US National Body.  It's
for that reason that I, personally, don't believe ISO has much
credibility as a standards body that I trust.  It's approval mechanism
is far too easy to manipulate, as everyone saw with OOXML, and I saw
when I saw how the ISO experts at various multinational corporations
worked together to get LSB fast tracked as a PAS standard in the first
place.  There was a huge amount of strategizing and political
calculations that had very little to do with technical excellence, but
in making sure that companies like Sun and Microsoft wouldn't
interfere.

> In my environment, which is not American, but Danish, and European Union 
> it really does give status that products are conforming to standards.
> Both Denmark and EU do give special treatment to ISO standards, e.g. for
> public procurement. This recognintion is not given for standards
> form Linux Foundation. As chair of the Professional Linux Association in
> Denmark it is my firm belief that and ISO LSB standard will help my
> professional members getting their products and services sold
> and used  in Denmark and the EU. 

Who do you think the ISO stamp would help against in terms of the
competition?  Solaris?  Windows?  I see two possibilities; one is that
competitive advantage will be negligible; in which case, is it worth
the effort?  If the competitive advantage is very large, then almost
certainly Oracle and Microsoft will get involved with various National
Bodies just as they did with the OOXML case, and what's your plan with
dealing with that opposition then?

More importantly, are your professional members willing to fund the
necessary work so that the LSB can be formatted in a way that ISO is
happy with?  If it was just a matter of taking an LSB release and
rubber stamping it with the ISO rubber stamp, we might differ as to
the benefit, but there is very little cost or harm (although some
might argue that anything that gives more credibility to ISO is a bad
thing, given how fatally flawed its voting mechanism can be, as
demonstrated by the OOXML debacle).

Furthermore, what if various National Bodies vote yes, but only if the
LSB workgroup makes certain changes?  Even if someone else is willing
to pay for the technical writing work, all of the LSB workgroup will
have to involved with any demands for change from the results of the
ISO balloting process.  We could agree ahead of time that the
automatic answer for any substantive change demanded by an ISO
National Body voting on the LSB is an automatic "no", I suppose.

But ultimately, the question is whether people outside of Keld's
Professional Linux Association will be asked to pay the costs
associated with trying to interact with ISO.

Regards,

						- Ted