ietf
[Top] [All Lists]

Re: Clarification of my comment on giving up on process issues

2006-03-25 11:05:39


--On Wednesday, 22 March, 2006 13:43 -0500 Sam Hartman
<hartmans-ietf(_at_)mit(_dot_)edu> wrote:

...
However I don't think we're building the sort of community
consensus behind RFC 3933 as an approach to breaking process
reform deadlock that it will actually be useful to us.  What
happens when John submits his nomcom proposal as an RFC 3933
experiment?  Would there be any plausable way he  could move
forward on that proposal using RFC 3933?  

Not that I can find, which, if you are referring to the Nomcom
proposal I think you are, is why I haven't done it. For those
who don't remember, it suggested using a different model for
first-time appointments to the IESG or IAB than for renewals and
building an assumption into the renewal model that two terms
were normally good but that more than that probably indicated a
problem in progress or likely to occur down the line --
draft-klensin-nomcom-term-00.txt (expired) for the curious.    

draft-klensin-recall-rev-00.txt is another example of the same
thing.  There have been an informal in-the-hall discussion or
two, but no focused discussion that could be used to move the
proposal forward.  Unlike the nomcom one, it could presumably be
implemented as a process experiment although, extrapolating from
the number of recall efforts we've had, the experiment would
need to run a rather long time.  But, especially without clear
community support, the IESG would need to be mildly crazy to
adopt it as a process experiment -- it would just seem too
self-serving.

But, more generally, what those types of proposal need,
independent of the process-change model we use, is enough
community discussion to permit making a determination that
people care and that there is sufficient consensus to move
forward.

What concerns me most of all these days is that, if one can get
a process proposal onto the agenda of some WG, or can introduce
it via a meeting convened by some senior member of "the
Leadership", then it draws a handful of process-weenies and
generates comments (perhaps many comments) _from them_.  The
broader community shows little signs of caring, which leaves us
with three alternatives:

        (i) Do nothing, on the ground that, if enough people
        don't care, nothing is severely enough broken.
        
        (ii) Go ahead and made the change, partially on the
        theory that, if it turns out to be significantly worse,
        _that_ will bring the community out to comment.
        
        (iii) Continue thrashing.  Thrashing differs from (i) in
        that (i) doesn't use up community cycles and raise the
        frustration level.  Thrashing does both.

In that light, 3933 was intended to make (ii) less painful and
risky.  But it is not useful for changes we don't know how to
back out.  And, even for those we can back out, the question of
whether we should try an experiment to fix a problem that almost
no one seems to care about is a challenging one.

...
So, I'm close to concluding that we don't have mechanisms for
getting consensus on larger process changes and that perhaps
the right approach is to just move on with our existing
processes.  They mostly work after all.

This, of course, is a variation on (i) above.  It is certainly
more efficient than (iii) and, if we can't even move forward
smoothly with 3933 experiments where there seems to be some
interest, may be better than (ii).    But it is, IMO, a fairly
sad state of affairs.


My argument is that proliferation of competing process change
proposals may well be an appropriate mechanism for RFC 3933
experiments--even when these are significant process
experiments.  I think recruiting the stakeholders will provide
enough of a gate.

But this is only true if the community buys into the approach .

Agree completely.

     john


_______________________________________________
Ietf mailing list
Ietf(_at_)ietf(_dot_)org
https://www1.ietf.org/mailman/listinfo/ietf