Chronological Thread 

< Chronological >      < Thread >

---

AT> P.S. regarding different standards or notation
AT> systems, it will be never irrelevant to have multiple
AT> standards (domination may be mantra of Microsoft), it
AT> gives freedom to choose and that's beauty of open
AT> community based projects.

What follows is personal opinion, not an official position
on behalf of any group I'm involved with.

Agreement on single open standards for single domains or
application areas *is* beneficial and a goal to strive for.
An example of this is today's web. Would the web be what it
is today if people had not settled on one set of standards
(HTML for describing pages, HTTP for the protocol)?
Agreement on single standards in single application areas is
what enables practical interoperability in those areas.
(Sure, you can have interoperability even in the face of
multiple standards, but it's a lot more expensive to
implement, and thus not practical in the long run.)

Note that my claim is specifically about *single* domains.
Multiple standards are definitely needed for non-overlapping
(or mostly non-overlapping) domains or application areas.

Multiple proposed standards for the same single domain arise
(1) in the early stages when people are still grappling with
understanding their needs on the one hand, and implications
of specific format/notation/whatever choices on the other;
(2) when an established standard is found to be deficient
as a result of experience, evolution, and/or innovation.
For #2, if things have been going well, people will want to
update an existing standard (e.g., version 1->2) rather than
replace it entirely, but in other cases, wholesale
replacement may be called for.

I hypothesize that when people say "multiple standards are
good in open community projects", it's because people
actually are thinking about multiple application areas, or
an area has not settle on common standards, or people are
dissatisfied with the existing standards.

People sometimes bring up Microsoft as a counter-example to
single standards. The actual problem is not the adoption of
single standards, it's Microsoft's specific approach, which
involves (i) a single entity making top-down decisions for
its own needs without consultation with other groups, (ii)
not documenting the formats, (iii) changing the formats
roughly every time they update their applications without
prior announcement, and (iv) a belligerent business style.
In other words, people's problem with Microsoft is actually
rooted in Microsoft's lack of an open process. I wager
Microsoft would be viewed much more favorably if they
developed their document formats using a bottom-up community
process.

In summary, I claim that there are reasons why an area might
end up with multiple proposed standards, but it is false
that having multiple standards is better than using one
standard when a single one can be agreed upon.

Agreement on a single standard certainly takes a lot of
effort, exploration, and time, on both the technical aspects
of defining the standard and the social aspects of arranging
community buy-in. Sometimes it takes decades before
quiescence is reached on a given topic. But once a
sufficiently good single standard for a given area can be
found, it is far better to stick with it, and spend your
efforts on something else.

MH

---

• [cellml-discussion] Opinion on multiple standards, Michael Hucka, 10/31/2008

---

Archive powered by MHonArc 2.6.18.