18 months

[Joel M. Halpern]

The assumption that long-winded working groups inevitably produce useless 
results does not always match observation.  I know of one working group 
that was, successsively:
Stuck for 2 years
Getting sorted out for 1 year
Producing results for 2.5 years
And then has been doing followon work
The Proposed Standard that resulted from that 5.5 year ramble is widely 
implemented, and is the basis for multipe useful services.  This in spite 
of the fact that many folks feel that with the benefit of even longer term 
hind sight the working group probably solved the wrong problem.

No, the amount of time it took is not good.
No, I am not saying this is the way we want working groups to function.
I am saying that the assumption that long duration = non-productive is not 
a valid equation.

I also know of working groups that produced timely results for problems hat 
appeared to be significant, and which turned out to be of marginal interest.
Yours,
Joel M. Halpern

At 11:14 AM 12/23/2002 -0800, Dave Crocker wrote:
>Brian,
>
>Thursday, December 12, 2002, 2:05:52 AM, you wrote:
>Brian> I don't get how a rigid time limit will make WGs do things better.
>
>Yes, that is a key question.  So here is a rambling attempt to respond:
>
>Do working groups that take a very long time to produce anything do well
>within the IETF? Do they eventually produce something useful? Does it
>eventually become popular in the Internet. I think the answer is clearly no.
>Working groups that go for a couple of years without being productive do not
>tend to get productive later. They lose focus. They lose energy. They lose
>participants. They lose market window. They make bloated, problematic
>specifications.
>
>This suggests rather strongly that the IETF is best for activities that are
>more near-term. If an activity is not yet ready to do near-term work, the
>IETF has demonstrated that it is a very poor venue.  Hence, such activities
>should be pursued elsewhere, and come to the IETF only when they are ready
>to do near-term work efficiently.
>
>The IETF has been having difficulty producing charters that are sufficiently
>crisp for easy guidance to achieve near-term productivity. ("Easy guidance"
>means that it is relatively clear what is in scope and what is out, as well
>as clear what the useful output from the working group should be.) We can
>all wish that this were different, but folks who work on charters and who
>review them are all bright, experienced and well-intentioned. Entreating
>ourselves to 'do better' is not likely to produce better charters.
>
>We are constantly seeing working group discussions that are energetic,
>interesting and unproductive.  This strongly suggests a lack of focus among
>the participants.  In particular, it suggests that participants are not
>feeling a sense of urgency to produce something useful and they often are
>not sharing a common sense of what that useful something should be.  Again,
>we can all wish this were not so, but we do not yet have any other
>suggestions for fixing this bit of entropy.
>
>Deadlines provide focus... if they are enforced.  We have a great deal of
>discussion, now, that suggests we merely need to enforce the milestones in a
>charter.  Such a "change" sounds appealing.  The question is will it work?
>Remember that we tend to produce charters that lack crispness -- often
>including the milestones, and that we have an extremely varied pattern of
>"enforcing" those milestones.
>
>In the face of diversity, variability, uncertainty, etc., it is often
>extremely helpful to find a simple, crystalizing rule that everyone can
>understand and adjust to.  This is, of course, Procrustean.  The nastiness
>of Procrustean rules is clear.  However, that does not mean they are
>inappropriate.  It is remarkably Procrustean to require that everyone making
>IETF standards speak and write English, if they are to be effective.  Yet we
>have found it essential to productivity.
>
>The idea of a universal time limit, for all IETF efforts is intended as just
>such a simple, focusing rule.  The particular number that has been proposed
>tries to allow enough time to fit into the pattern of IETF productivity, and
>to be short enough to exclude activities (or terminate activities) that have
>shown a pattern of non-productivity.
>
>Since the idea of a rule to be productive in 18 months is too onerous, how
>about a rule that says all working groups must undergo zero-based chartering
>every 24 months?
>
>That is, a working group ceases to exist at 24 months.
>
>It may submit a new charter, and that new charter will be viewed on its own
>merits, using current assessments of participant availability, likely
>productivity, and market pressures -- the same as we do for any new working
>group. The previous 2 years of history for the "old" working group would of
>course be taken into account.
>
>(Lest anyone think that the aggregate overhead of such a process would be
>onerous, we should remember just how onerous it already is to have wasteful,
>protracted working groups.)
>
>d/
>--
>  Dave <mailto:dhc@dcrocker.net>
>  Brandenburg InternetWorking <http://www.brandenburg.com>
>  t +1.408.246.8253; f +1.408.850.1850