ietf
[Top] [All Lists]

Re: Why do we have working group charters (was: To "lose the argument in the WG")

2017-02-19 17:50:14
SM,

--On Saturday, February 18, 2017 11:07 PM -0800 S Moonesamy
<sm+ietf(_at_)elandsys(_dot_)com> wrote:

Yes, as long as you remember that they are supposed to be
representing the will of the community and accountable to it.
Neither WG Chairs nor ADs have any other source of authority.
The difficulty with your statement above is that it can be
read to imply that there are no constraints one, or appeals
from, the decisions.  That is not true or, more specifically,
if one wants a healthy IETF that produces credible work, it
better not be.

It is possible to appeal the decision of the WG Chair, and if
that fails, to file an appeal with the Responsible AD.  There
is still an option to file an appeal if the issue is noticed
later in the process.

I think we are in danger of having two separate conversations
here and not quite realizing it.  I think Ted and I have been
discussing what happens, or could happen, in practice.  Your
response has been to tell us what the procedures are and allow.
I've confident that I understand those procedures and almost as
confident that Ted does too.  It is probably helpful to remind
us, but let's not get them confused with what happens in reality
(the difference between the procedures and reality may also be a
useful conversation but it is, again, a different one).

The most extreme example of the difference is, of course, that
the written procedures specify a remedy for an AD who has gone
off in the weeds, seriously underformed, or otherwise lost the
confidence of the community in the form of a recall procedure.
That procedure has never been used up to the point that a recall
committee is appointed, much less beyond that point.  One could
infer from that that we've never had a serious problem with an
AD.  I don't think that is true, YMMD.  One could also infer
that informal discussions and interactions, including
encouraging people to resign rather than face the actual recall
procedure, have worked well, making that actual procedure
unnecessary, but the procedures do not call for those actions.

In the charter case, unless someone is following the work of a
WG very closely (for which see the "I don't have time"
discussion that spawned this thread), the first warning that the
WG has produced work that drifted away from the charter involves
a document showing up for IETF Last Call.  That poses a very
difficult problem for the AD for two reasons.  First,
inconsistency between a draft document and charter provisions
can be the result of differing interpretations of the charter or
a discovery that completing the chartered work required some
excursions into directions not anticipated when the charter was
written.  Deviations from the charter are rarely a
black-and-white matter: what is more typical is "some" or "a
bit" and a judgment call as whether or not the issue is actually
significant. In principle, we probably all agree that it would
be best to revise the charter as soon as such situations are
discovered.  In practice, there are lots of reasons why that
often doesn't happen, many of them rather good reasons.   If the
deviations are significant, one of the not-so-good reasons is
the it requires ADs to admit that they let the WG get out of
control.

And, yes, a decision to not put the document on hold while the
charter is revised can be appealed.  But holding work up,
especially work that appears to be finished, is a big deal and
there is, appropriately, a lot of reluctance to do that.  So the
appeals don't happen and, when they do, they often don't go
anywhere.

There are constraints.  At the WG level, the WG Chair is
accountable to the Area Director.  A competent Area Director
would probably ask for an explanation if the document exceeds
the scope of the WG Charter.

Again, "exceeds" is a judgment call.  Asking is one thing,
getting a useful answer is another.  And telling a WG that
thinks it has finished a document and may be waiting to close
down that they have to start fussing with charter revisions
and/or completely redesign the specification is very hard and
not always a good idea.

best,
   john