Pete,
I'm probably going to regret being drawn into this, but I think
you (and Dave) are missing two additional cases, ones that
complicate your analysis considerably even if they don't change
your key conclusion.
As the Internet and IETF evolve, I see increasing frequency of
very homogenous WGs, ones whose active participants consist
entirely, or almost entirely, of people who represent one
perspective and who are narrowly focused on that WG's topic
area. I addition to all of the usual possibilities, including
the ones you mentioned, such WGs run multiple risks. To keep
this from becoming over-long, the one that probably represents
the case we see most often goes like this...
The WG reaches fairly strong consensus on a particular set of
issues, with only one or two people in the minority, taking
positions the WG has decided are not interesting or not
relevant. In that case, unless we require omniscience as well
as fairness of WG Chairs, there is no error: the WG has reached
rough consensus on a topic and those one or two participants are
in the rough _within_ the WG. For that case, if IETF Last Call
and the idea of cross-area review are to be more than rituals to
support our mythology, it is absolutely necessary that someone
who participated in the WG be able to come to IETF Last Call and
say "look, the perspective of the WG was too narrow and, as a
consequence, these issues that affect other things didn't get a
reasonable hearing and the IETF as a whole should really
reexamine them independent of the conclusions the WG reached
(and/or the community should reexamine whether the WG was, in
retrospect, properly chartered, constituted, and composed and,
if not, whether its consensus and request for document
publication should be taken seriously,
FWIW, I believe that "strong consensus, but still wrong" was one
of the cases you addressed with RFC 7282.
I think that "too homogeneous a WG and too narrow a perspective"
argument, if plausibly explained and justified, has to be taken
serious even if there are no (other) new arguments or additional
"homework". I also think it is a particularly difficult type
of problem because, if the consensus recommendation from such a
WG to publish a document is accepted, it reflects badly on the
AD who authorized the Last Call as well as on the WG and its
leadership. While it doesn't make "I lost the argument in the
WG" is not helpful as a means of ending the conversation, "I
lost the argument in the WG for what I believe were the
following reasons ... and the IETF should review those issues"
may be exactly what we need to hear as part of document review.
That argument is very different from either "the WG CHair made
an error" or "I was on the losing side and I want to keep
whining about it and rehash the argument".
Whether IETF Last Call is the right way to deal what that type
of issue is a separate question. You and others have heard me
say from time to time that we don't have nearly enough appeals.
I think ADs should be monitoring WGs for sufficient technical
diversity and range of perspectives and that someone
participating in a WG who believes that the profile of
participants in the WG is too narrow to encourage or allow
arguing against WG conclusions on the basis of broader Internet
considerations (whether decisions about WG scope get involved or
not) should be encouraged to discuss that issue with the
relevant AD and, if necessary carry the appeals process further,
rather than quibbling about specific technical issues on Last
Call. However, success in such an appeal would probably require
shutting down a WG that has an active mailing list and meetings
and is producing results. Our track record for such shutdowns
is fairly close to non-existent.
That same argument for appeals applies when people who are not
part of the homogeneous majority are treated dismissively or
otherwise discouraged from continuing to participate. We ought
to be able to deal with those situations as instances of
bullying, but if that has ever worked successfully, I'm not
aware of it -- maybe you know of examples that I don't.
Where I think we agree is that, while there should be a
presumption that a WG that produces a draft and claims consensus
for it has done its work and gotten it right, that presumption
should not be very strong, should be easy to challenge, and that
the position of someone who disagreed with the WG decision
should not be dismissed on that basis.
best,
john