--On Wednesday, December 03, 2014 16:05 +0000 Adrian Farrel
<adrian@xxxxxxxxxxxx> wrote:
> John,
>
> I placed the document on the IESG telechat after a period of
> silence that followed the publication of the -07 version of
> the document and an email summarising the changes from the
> editor.
>
> At the same time I sent an additional note to this list. I
> don't know (or care) whether that is what triggered renewed
> conversations, but I welcome them. I don't think you should
> take such a negative tone and assume that the no-one wants to
> listen to your concerns. Let's see whether we can address them
> before the telechat in just over two weeks' time.
Let me clarify my point of view and the source of that tone. In
terms of value to the community, I welcome documents that
provide experience-based advice on how particular roles should
be carried out --whether via the Informational RFC process, by
updates to the Tao, or via the EDU team. At the same time, I
see documents that seem to impose normative requirements or that
can readily be interpreted that way as a problem. In this
particular case and independent of anything else, I don't think
the quality of work produced by the IETF or the efficiency with
which it can be produced is likely to be significantly enhanced
by the publication of this document _in any form_. YMMD but it
makes me feel that I'm making a bad judgment by postponing IETF
technical work to review and re-review it and write comments.
So, I said in August that I thought this was too normative and
prescriptive and that, while an advisory document would be fine,
this was something else. A few others said much the same thing.
I (and I think we) saw that as a fundamental problem with the
document, not something that could be covered over by a
paragraph or two that amount to "regardless of what the rest of
this document says, it isn't really normative and doesn't really
constrain WG behavior". My belief that comments of that sort
were not being treated in a serious way is based on the
evolution (or lack thereof) between the August versions and now.
It was strongly reinforced by Martin's "it is good if a
secretary does all that" response to Murray this morning:
whatever value the August comments and the new text may have, it
appears clear that he hasn't understood or accepted it.
So, independent of your summary below, I was disappointed that
the Last Call was reopened on the newer version.
However, part of what the community expects the IESG to do is to
make priority judgments about what is important and what isn't.
One has to hope those judgments reflect community consensus but,
if we stop trusting the IESG to be able to make them, we are,
IMO, basically finished. So, when you and your colleagues
issue a Last Call on a document that falls even vaguely into my
area of competence, I need to take that as an instruction to
give it priority over IETF technical work that has not yet
reached Last Call. If Pete and/or Barry don't agree that my
reading and responding to this document takes priority for my
IETF time over, e.g., PRECIS, IDNA, and URNBIS work for which
I'm temporarily in the critical path, then there is, from my
point of view, an intra-IESG discussion that needs to be held.
At the same time, _I_ can think that these priorities are bad
judgment even while I feel that the potential for harm from this
document is high enough that I need to spend the time. And that
combination gives me a case of bad attitude, for which I
somewhat apologize.
> I believe I see two issues raised by you and Melinda in this
> thread which I may summarise as:
I disagree with your summary. See below.
> - This I-D needs to be clearer that it is not normative, not
> defining process, and not directing WG chairs to use
> secretaries.
Yes. I would add some details to that list, but yes.
> - The use of a WG secretary somewhat reflects an abdication of
> responsibility by a WG chair who was appointed to carry out
> these tasks.
Not my point and maybe not quite Melinda's although I won't
presume to speak for her. I think there are many good uses for
WG Secretaries and more than a few bad ones. I think that
anything that reduces WG Chair accountability and responsibility
is a bad idea. For that reason and absent unusual
circumstances, I'd rather not even see co-Chairs even though I
recognize that we are seeing circumstances I'd wish were less
common more often. One result is that I don't want to see a
document or statement that implies "there should be a Secretary
and the Secretary should be doing these things", not because
that is an abdication of responsibility by a WG Chair but
because it makes the Secretary accountable for those tasks (it
is not clear to whom) and not the WG Chair. As long as the
Chair is responsible and accountable, I don't care how work is
spread out or delegated within the WG. If, as Melinda noted,
the WG Chair is not being effective and successful and a
Secretary is appointed to try to work around the problem, then
that is a problem... and more likely an abdication of
responsibility by the responsible AD than by the already-failing
WG Chair.
There is also a third point that, for me, is very much part of
this picture. I spent years of my life working in SDOs which
had a large portfolio of procedures and review states and who
were de facto convinced that, if all of those procedures and
reviews were committed, the result was, by definition, a good
standard. Most of my earlier work wasn't in networking, but I
suggest that model is exactly what brought us the wildly
successful OSI model and protocols. I don't want to see the
IETF go (or go faster) in that direction and that makes me very
reluctant to see us move from the equivalent of "it is ok for
WGs to have Secretaries if the WG Chair(s) decide that is a good
idea on a case by case basis and such Secretaries serve at the
pleasure of the Chair(s) and do what they and the Chair(s)
jointly conclude is useful" toward "there should be Secretaries
for WGs and they should do these things".
Coming back to Melinda's comment (at least as I understood it),
if we think we have an organizational or management problem,
let's figure out what it is and fix it, not slide in another
layer or management or indirection and hope that will solve
something.
Unless it is done _very_ carefully, there is only one step
between a document like this and an appeal that a particular WG
result was not fair because the WG didn't have a Secretary who
had the powers and responsibilities called for in the document.
I think that, and the alternative of every WG having a Secretary
to avoid doubt, would be bad. I don't think the present
document is done well enough to avoid or minimize that risk. It
isn't clear to me that the community investment needed to get it
to the point that it would be safe is worth the effort.
> Now, I think the first of these might be something we can
> address with some careful text in the Abstract and
> Introduction. Would you agree? If so, I' m sure we can work
> something out.
One of your IESG colleagues asked me almost the same question
off list. He is welcome to share the rest of my response and
attempt at a constructive suggestion/offer with you as
appropriate but, briefly, no. I think the authors tried that
with the first paragraphs of Section 2 and that the problem
isn't just placement. The assumption that the document is and
should be normative -- that there should be secretaries and that
they should have certain authority and behave in certain ways --
is far too pervasive for a few sentences that say in essence "if
you read part of this as normative, never mind, it isn't really
intended that way". It needs to be restructured and rewritten
and, fwiw, I don't see much hope for that in an individual
submission unless the authors understand and are signed up for
the difference in program.
> The second is more of a philosophical question. The chair is
> "contracted" to deliver and given certain powers to achieve
> delivery. Can the chair delegate the actions while supervising
> their implementation and carrying responsibility for their
> correctness? This is maybe an issue we can address by making
> it clear who carries responsibility for actions carried out
> by a WG secretary and what the complaint/appeals path is.
Yes, but see above and note that, from at least one point of
view, you are proposing to solve a problem I see as "trying to
create too many rules or too much uniform structure" with "more
rules and more structure".
> But I would note for the second of these points that the WG
> secretary role exists and there is also the concept of a
> "delegate". These ways of handling IETF work are not
> introduced or made concrete by this document. We can't,
> therefore, hold this document to account for the existence of
> things we don't like.
Agreed. On the other hand, if the document is, even in part, a
symptom of a problem, then we should be identifying the problem
and trying to fix it rather than documenting the symptom as
desirable practice.
> We can only ensure that the document
> accurately describes what exists.
If that is your criterion, then the document fails it because it
describes one cluster of ways of using WG Secretaries (and then
strongly suggests it is the best one) but not all of the ways in
which Secretaries and or are not successfully used. So it is
not an accurate description of what exists, only of one model,
and it isn't clear about that.
best,
john