See in-line.
Dan
-----Original Message-----
From: John C Klensin [mailto:john-ietf(_at_)jck(_dot_)com]
Sent: Sunday, June 17, 2007 9:13 PM
To: Bernard Aboba; ietf(_at_)ietf(_dot_)org
Subject: Re: Reforming the BOF Process (was Declining the
ifare bof for Chicago)
--On Tuesday, 12 June, 2007 09:52 -0700 Bernard Aboba
<bernard_aboba(_at_)hotmail(_dot_)com> wrote:
...
For example, by restricting the function of an initial BOF to a
determination of interest and a decision to form/not form a study
group, the opportunities for unfairness and bias can be reduced.
Once the study group had produced a charter and
documentation of the
formation criteria, the review of these documents could
proceed with
more information than is typically available as the result of a
(potentially delayed) 2nd BOF. Also, the review could
utilize existing
procedures for ensuring transparency and accountability, such as an
open review process and documentation of DISCUSS comments.
Bernard,
Some specific observations on shifting more toward an IEEE model...
(1) Unless things have changed since I was last paying
careful attention, the process you describe is used to adding
a new project to an existing technical committee. The
process for creating a new technical committee is somewhat
more elaborate.
The IETF has no layer between the steering group (IESG, TSC
(?)) and the WGs who actually do the technical work. We also
usually try to charter WGs on a short-term, project basis
rather than assigning new projects to existing WGs. Because
of those differences, we need to be careful about analogies
unless we are willing to rethink process models in much more
fundamental ways than tweaking the BOF process.
Actually I believe that the IEEE have study groups both at the level of
technical committees (Working Groups) which end if successful with new
projects within existing Working Groups as well as at the level of the
whole IEEE which may end with new Working Groups being formed. Now a
IEEE Working Group vary very much in size and scope, but there are a few
like IEEE 802.1 (bridging, security QoS) and IEEE 802.11 (wireless LAN)
which are comparable in size, scope and number of projects with an IETF
area rather than with an IETF WG.
(2) I don't have statistics, but my impression is that most
technical standards work in the IEEE these days (not just in
802, but more broadly) starts with a proposal to standardize a
specific industry technology. Those proposals are debated and
refined, but the assumption is that little fundamental
engineering or design work is done in the standardization
process.
This also differs from case to case. In some IEEE projects work may
start with a given technology on the table, in some other there may be
more than one, or just an idea or several ideas about how to solve the
problem, and much debate happens until an initial proposal is baselined.
I do not believe that overall the distribution of cases is significantly
different to what happens in the IETF.
We behave as if the IETF is still doing engineering
design work. Maybe it is time to drop that as an inefficient
and ineffective fantasy, but, again, considering that
involves much broader issues than reforming the BOF process.
(3) Many of us are concerned about the length of time it
takes to move a well-thought-out idea that has significant
support forward from initial proposal to a functioning
standards development effort. Perhaps we should be
concentrating as much on that question as on the one of how
we cut bad, or inadequately supported, ideas off as cleanly
as possible.
I believe that the lack of some agreed criteria of approval for new work
is a problem. Maybe this is what you mean by 'how we cut bad'.
Adding a study group creation process and a study group
process to the existing opportunities for delay would not
contribute to speeding things up. Indeed, it would do much
the contrary.
(4) I, and others, have said this before, but my belief is
that the single most effective change that could be made to
the BOF process would be for the IESG to stop taking its
ability to project the future so seriously. As you and
others have commented, the track record on that isn't good
anyway. Suppose, instead, we were to permit WGs to be set up
on a provisional basis, with intense review after some
reasonable period of time and cancellation if they were not
performing adequately and producing results the community
seemed to care about. This would combine some of the
advantages of IEEE-like study groups with a streamlined
process that focused on the one true measure of whether a WG
could produce useful work: starting it up and seeing if it
produced useful work.
What would be the differences between 'study groups' and 'WGs set up on
provisional basis'?
Since decisions as to whether to charter a WG are somewhat
subjective and the IESG has broad discretion, this is a
change the IESG could institute --experimentally or
permanently-- on its own initiative without our spending
energy on changing the
processes. Its success would, however, depend on a change of
attitude in the community: today, so much effort goes into
the charter process because it has become almost impossible,
in practice, to kill off a non-performing WG or to put a tight
leach on one that is wandering in the weeds. ADs who try to do
so do not win popularity contests, to put it mildly. If the
IESG saw clear and broad community support for chartering WGs
that were questionable but plausible and then tuning charters
or killing non-performing WGs if things didn't work out as
originally conceived, I believe we could get a great deal of
the nonsense, arbitrariness, and delays out of the early
parts of the process.
But I don't think that support is there, at least yet.
Without it, changes in forms or procedures probably do not
produce better results and, if delay is considered an
important cost, might produce worse ones.
regards,
john
_______________________________________________
Ietf mailing list
Ietf(_at_)ietf(_dot_)org
https://www1.ietf.org/mailman/listinfo/ietf
_______________________________________________
Ietf mailing list
Ietf(_at_)ietf(_dot_)org
https://www1.ietf.org/mailman/listinfo/ietf