I think Dave is exactly right on just about everything here. I have a
couple of issues with the details of some of the logistics, but it is in
the details. For those who want to see my thoughts on sausage making,
continue below.
On 1/22/12 12:51 PM, Dave CROCKER wrote:
I'll suggest that "pushback", per se, is the tactical point.
I should have been more clear that I intended exactly your strategic
point: Engage.
The strategic point is the dual obligation for:
1) the AD to be (more?) clear about their underlying concerns and
not just assert the problem or the solution they think will resolve
it, as well as the requirement that they engage in a /direct/ dialogue
with the working group, and
2) the working group to be (more?) clear about its underlying goals
and requirements, as well as engaging in real and constructive
dialogue with the AD(s) toward a useful resolution, rather than merely
one that pacifies the ADs.
Absolutely agree on both counts. How that direct dialogue occurs is an
interesting question; more on that below. But the AD's concerns must be
clear, and the WG should look to address the concern (even if addressing
it is, "we have documented that we understand the concern and we feel it
is not a problem we will solve), not just appease.
I have more than my share of just getting pissed off at ADs who lodge
vague and even inappropriate Discusses and can attest that while
pushing back is fine, indulging in being pissed off is never helpful,
even when they warrant the reaction. (That's meant merely to mark an
extreme, not claim it applied here.)
+1!
Unfortunately, my review of the current wg's mailing list record shows
quite a bit less of #1 than there should be, IMO, in spite of Pete's
active, follow-on participation.
Yup. And in this case, I think that, until I jumped in, #2 wasn't going
so well either. (That is, Stephen's DISCUSS comment was -- and
unfortunately is still written as -- "H has problems; you want HMAC",
and the WG's response was, "HMAC is overkill", instead of figuring out
that the problem was the lack of clear text on the use case, and the
solution being to clarify the use case, and probably state that neither
H nor HMAC was necessary.)
(And I note that the Discuss is not yet cleared...)
An unfortunate side effect of there not being an appropriate tool. ADs,
both the ones issuing the DISCUSSes and the ones managing the WG getting
the DISCUSSes, have turned the ballot position into an issue tracker.
That's dumb and needs to be fixed. But I'm guilty of this too. A long
and different discussion I'm happy to have in some other forum.
Stephen has not engaged directly. Now it well might be that Pete has
been an adequate surrogate, but forgive me, it is Stephen who holds
the Discuss.
So let me talk a bit about direct engagement. It's not the standard
thing for the IESG now, but you're going to see more of it soon. During
the last WG chair lunch, we had a discussion about copying all ballot
DISCUSSes and COMMENTs to the WG mailing list and I think this is
exactly the right thing to do. My plan is to, on a case by case basis at
first, add the WG list to the magic field in the datatracker to which
such comments go. The only issue is that this has to be reasonably managed.
First, keep in mind that ADs make lots of comments (DISCUSSes and
otherwise) and it would be quite a lot of mail for an AD to get if they
got the full flood from every mailing list they sent a comment to,
especially because so many of us make a slew of comments all during the
telechat week. (Yeah, we're pretty poor planners.) So when we do direct
engagement, I'm going to ask a bit of forbearance from the WG and try to
avoid flooding the poor schlump AD who made a comment with a huge number
of messages. But that's just a matter of being a bit more "conservative
in what they send" when an AD is in the mix. (And I'm only talking about
volume, not tone. :-) ) We can figure out how to do that.
Second, note that currently all DISCUSS and COMMENT messages and all of
their replies are Cc'ed to the IESG list. Just flipping the switch on
would be potentially a *lot* of email to the IESG list. So we may want
to figure out a way to split the "WG is mulling this over" messages from
the "we have a response to this" messages.
Finally, the current model has the AD directly engaging with *only* the
editors and the chairs. Sometimes that's helpful because having only a
few people chatting with the AD to understand the AD's issue is often
easier than having the entire WG have that conversation. But that also
means that sometimes the chairs or the editors come up with solutions to
the AD's issue that the WG wouldn't buy into. (Jeez, I hope that's
rare.) Bringing everyone into the conversation is again going to require
a little mental shift so that people (WG chairs especially) get used to
how to manage these conversations.
All of these are simply (?!) cultural changes about how direct
engagement will work. I think direct engagement is a *much* better
model, but it does have side effects, and I want to make sure that those
side effects don't cause a backlash against doing so. Right now, the
model is "only do direct engagement once there's a problem", which is
the wrong model. But I want to minimize heartburn while getting to the
right model.
So, with all that said, I'm happy to get Stephen directly engaged on
this particular topic, though I think we now have a pretty good handle
on what the issue is and how to deal with it. I'll leave it to the
chairs to tell me what level of involvement they think is good at this
point.
It really is essential that a wg not be forced to guess what will
satisfy ADs who hold Discusses.
+1.
As a matter of due diligence, I'll also ask folks whether they believe
the modifications to the specification retain its previous level of
utility and pedagogy as a specification, for strangers out there in
implementation land who lack the background from participating in the
working group? I ask this because sometimes handling the one point
raised by a Discuss alters the answer for other aspects of the spec...
You bet!
If one reviews Stephen's Discuss, one sees an engineer asserting the
details of an alternative solution, with what is really none of the
underlying concepts or issues that might justify it. However the
concerns in the underlying stuff are the real substance.
Perhaps my reading comprehension is inadequate here, but I simply
don't see the material a working group ought to be given, to permit
serious dialogue.
In retrospect, I completely agree. Stephen's DISCUSS comment is how to
address what he believes the issue to be, not a discussion of what the
issue really is. If his assessment of the issue is correct, that might
be a real timesaver, but if it's wrong, well, we get to where we are now.
Forgive me, but that really /is/ overstepping (or understepping) the
role of the IESG at this stage, IMO, even if the suggestion is the
right one.
I agree with understepping, not overstepping. We've all gotten into bad
patterns with the goal of helping things move faster. (You will note
that the IESG Statement on DISCUSS Criteria talks about always making
DISCUSS comments "actionable".) As I said above, when it works, great,
but when it doesn't, it causes bad feelings. In this case, Stephen
wasn't trying to force a particular technical solution, and therefore I
don't think he was exceeding his authority. But neither he nor the WG
initially figured out what the real problem was, and that's a
shortcoming in what happened.
Thanks for a good analysis, as well as an opportunity to talk about what
I'd like to do to address these problems.
pr
--
Pete Resnick<http://www.qualcomm.com/~presnick/>
Qualcomm Incorporated - Direct phone: (858)651-4478, Fax: (858)651-1102
_______________________________________________
marf mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/marf
