what are the real problems

[Dave Crocker]

Joel,

JJMH> For example, as worded Dave's note suggests that there is no drawback from
JMH> having too many efforts.  I don't think anyone believes that there are no
JMH> drawbacks.

Including Dave. He certainly did not mean to imply that having multiple
efforts is without its downside. However he is frankly far more
concerned with the downsides of prematurely terminating valid (parallel)
efforts.


I have a simple model of IETF work:

   1. There must be a clear technical need/goal

   2. There must be a core technical competence that can reasonably be
   expected to achieve that goal

   3. There must be a core, committed constituency for developing and
   another for using the result.

Satisfy all 3 and and it is a legitimate IETF effort.

Where the modern IETF gets into trouble is when we try to get clever and
pretend that it can dictate among competing, legitimate efforts.

I claim that all we can or should do is to keep the pressure on to
ensure high technical quality.

Terminate an otherwise-legitimate technical effort early and we are
pretending to have a far better understanding market preferences than we
actually have.

We also will preclude serendipity.

If we had made THE choice of the right way to do international
characters for email, we probably would not have gotten ESMTP. There is
a good chance we would not have gotten SNMP. And so on.

Lest someone feel certain that these various opportunities for shutting
off parallel efforts would have been certain to come out the "right"
way, I'll simply ask why?



JMH> Just in case, let me suggest one of the many drawbacks.  I believe that no
JMH> matter what organizational structure we craft for ourselves, there will
JMH> always be a leadership load based on the number of "things to be lead"
JMH> (working groups, activities, ...)

And I'll repeat that I thoroughly agree that paying attention to
leadership load is an important -- nay, an essential -- concern. But
let's apply that concern diligently, rather than opportunistically.

As a community, we are so grossly inefficiently about this topic, it is
actually hypocritical for us to to apply it in this way. And, no, that
does not mean we should ignore the question of load. Again: it means we
should look at the topic seriously and thoroughly.


JMH> Another major issue with allowing competing efforts is that the results do
JMH> not interoperate.

TCP and UDP do not "interoperate". IS-IS and OSPF do not interoperate.
And so on.

Yes, competing efforts usually do not interoperate.  That is what makes
the phrase "let the market decide" highly non-trivial.  It means that we
do, in fact, let the market choose based on criteria beyond simple
technical evaluation.  The alternative is parental and premature.



d/
--
 Dave Crocker <mailto:dcrocker at brandenburg.com>
 Brandenburg InternetWorking <http://www.brandenburg.com>
 Sunnyvale, CA  USA <tel:+1.408.246.8253>, <fax:+1.866.358.5301>