Re: Specific suggestions to improve San Diego agenda [Was RE: San Diego Agenda issues]

["Adrian Farrel" <adrian@olddog.co.uk>]

Hi,

> I am very disappointed to see your responses to my two previous
> emails, and take serious issue with the rather dismissive tone
> of your responses.

I'm sorry that you interpret the tone of my response as rather dismissive.

I understand that you are disappointed both that your survey draft has not been allocated
a slot on the agenda and that I have not leapt to embrace your suggested agenda
modifications.

> It would be nice to put aside rhetoric and clever remarks to
> focus constructively on the feedback at hand.
>
> The reality is that the feedback I provided is a very
> serious attempt to advance the WG's goals and milestones.

Then I will go through your mail and respond to each point in detail, below.

> Not having an overcrowded agenda is one of the keys to
> advancing WG work.

Agree.

> And, as an active and responsible contributing member of this WG and
> the IETF, I believe it is incumbent to help structure what appears
> to be a needlessly overcrowded CCAMP agenda.
>
> I would have provided this feedback regardless of whether or
> not the you asked for it. I don't think the Chairs have to
> ask for feedback from WG members, for us to provide it.
>
> (Note:
> I have certainly done my share in contributing to progressing
> several recent WG items, including TE requirements, inter-region TE,
> P&R, ASON aspects, node ID and hellos,  and the like -- one need only
> go to the ML archives and minutes for that -- 
> so it is not as if my input is without basis or without the interests
> of the CCAMP WG in mind.
>
> Of course, it is neither possible nor practical for everyone to
> provide inputs on every WG item; nor, in my view, is it beneficial
> for someone to attempt to do that.)

All input from all sources is valuable. There is no need for anyone to stop providing
input just because they have already made comments in many fields. In particular, people
implementing or deploying the technology will have sound opinions across most of the
workspace and should be encouraged to comment and contribute.

> To address your specific points:
>
> i) It is never possible to finish all discussions during a WG meeting,
> (so, of necessisty, many long discussions are transferred to the ML,
> many people are left standing at the mike, etc.)
> I think we all understand that, and I do not see anything
> particularly unsual in it at Seoul.

Well, my memory of Seoul is that we rarely got beyond the first three in line and that
there was no space for debate or discussion. This certinaly gels with the comments I
received afterwards.

I know from experience that for someone to be moved to go to the mic takes a fair amount
of concern, and that it is hugely frustrating for the queue to be cut off.

> But, if that is an issue, then freeing up 30-odd minutes of the
> agenda would seem _even more_ important than I initially thought. It
> would ensure that whatever is discussed, is, as far as possible,
> done so fully.

Indeed. If we could free up agenda time, that would make sense.
The debate (below) is about whether we can free up the time.

> ii) Towards that end, I provided _six_ specific suggestions (with
> reasoning and justification, based on the criteria specified in
> your email), on how the agenda can be made less crowded.
>
> http://ops.ietf.org/lists/ccamp/ccamp.2004/msg00761.html

And I shall respond to each of the six below.

> And, I did not see the WG advance any objections or give any
> reasons why my suggestions are not sensible and reasonable.

Silence is taken as consent in this matter?
I hope you will reconsider this. The WG participants are surely all too busy with many
other things to provide instant responses (or indeed any response) to a debate on the
agenda. But I would be happy to be proved wrong.

> I see that you have simply <snipped> them in your reply, I'm
> afraid that is not good enough.

Oh.
Well, sorry.
The snipping was because I was responding to a sub-set of your points.
I simply did not have time to respond to all of the points that you raised at the time and
thought that you would prefer to receive some response quickly, rather than none. I am now
pursuing every point in your orriginal mail and hope this will satisfy you.

> If the Chairs, as representatives of the WG body politic, draft the agenda,
> and a member seriously provides inputs on it, the Chairs are obligated to
> respond, as long as the inputs are not "frivolous", which I
> certainly do not think anyone can point to as being the case here.

Yes, indeed.
And hopefully the chairs are also allowed to go to bed sometimes? And do other work? And
review all of the drafts submitted in advance of the meeting? And review the relevant and
related drafts from other WGs and BOFs?

The aim must surely be to be a little reasonable here. Allow us all a little space to
breathe and write considered and lengthy responses as time allows.

> iii) The "sense of the room" is neither for my benefit nor
> that of the Chairs.
> It is for the benefit of the WG as a whole, to help advance its work.

Wrong.
If the chairs take the sense of the room it is for their benefit. It helps them guide the
working group that carries out all of its meaningful business on the mailing list, but
allows a quick and unscientific gauge for where the WG is with an issue.

> In that context, if the WG is polled in a rushed way, I re-iterate
> that I don't think that it benefits the WG to do so, and that time can
> be saved.

So, let me give you a concrete example. Is a draft ready for last call? When we want to
know that we send mail to the list (not for a vote, but to get a consensus) and if
consensus is judged (by the chairs) a last call is held.

Authors continually (and I mean continually) ask the chairs for last call. Sometimes it is
obvious that the draft is not ready, sometimes the chairs review the draft and find many
issues, sometimes it looks like the draft is ready. But typically we don't simply send
mail saying "is this draft ready for last call": what we do is get a feeling for whether
the WG is in a position to make the judgement. For example, how many people have read the
draft? who feels that this is an important area? who has implemented? do you feel
confident that this is stable?

We could ask these questions on the mailing list, but it is neither quick nor friendly.
Most people on the CCAMP list are very busy and do not get to read and respond to emails
in a timely manner. It is clearly far more helpful to use a WG gathering to gather this
information and use that as a guide for how to continue.

> Furthermore, I did not see any communications stating that a sense
> of _all_ of these drafts will be taken, only some of them.
> I suggested instead that it would be better for the Chairs to
> highlight all these drafts on the list, and take a sense there.

As above.

But you make a very good point. It might be really helpful if the chairs were to provide a
status (say monthly) to the WG on the progress of drafts and milestones. This would
certainly focus the chairs' activities, might help to drive the editors, and would mean
that a little less time was needed for this type of work at the meetings.

I would be interested to hear from people whether this would be
- useful
- something they would commit to read.

> (So, I am not sure what you are driving at by suggesting
> dropping "my" draft from the list. In any case, there is none named
> after me, and two on which I am only a co-contributor.)

The context of the discussion was very clearly one draft for which you are listed as one
of three authors.

> iv) If there are milestone items that no one in the WG is commenting on,
> despite the WG taking them on, then it is perhaps time to seriously re-
> examine them, or encourage and urge those interested in those items to
> contribute to them.

I could not agree more.
There are two items that cause me grave concern. These are the GMPLS MIBs and Tunnel
Tracing. Both items have milestones which have passed their sell-by dates, yet the WG
shows very little interest in advancing the work.
The MIB authors (those who still remain working in the area) have made frequent requests
for assitance and implementation experience, but none has been forthcoming.
Yet we (as chairs) are told that these two items *are* important and fundamental to the
success of GMPLS. The chairs have given their time to work on these drafts, but help is
needed or the items must be dropped.

> However, I don't see how putting them on the agenda
> achieves the cause of advancing those milestones, except taking
> up time.
> People listen to the presentation, and nothing happens thereafter.

Then we need another suggestion.
Clearly in the desparation to get something to happen, the chairs wish to raise these
items (no other action seems to have any effect, perhaps raising them at the meeting will
work).
Personally, I would welcome time on the agenda (perhaps 15 minutes for each item) to
debate whether we should abandon them, or how we can achieve the milestones which we as a
working group have committed to yet have failed to meet.
However, I suspect that if I allocated time for such a debate, I would receive even more
of a lashing.

> v) Finally, the request to have the AD guidelines be published
> on the ML, was simply to ensure that the WG learnt of them in
> good time, to allow people to plan their work accordingly.

Well, they are out in the open now. I guess they were mainly intended to focus the way
that the chairs work, but I see no reason why the rest of the WG shouldn't be aware of
them.

= = =

Now your original email with responses...

> I have examined the agenda in the light of the above questions, and
> have looked at the drafts currently on it.
>
> The following is meant to be a neutral assessment, based on closely
> following the development of a good majority of the work on the agenda.
> [Disclaimer: It _should not_ be interpreted as reducing the importance
> of any of the work.
> Rather, it is a serious attempt to take up the Chairs' request to
> solve the agenda crunch (that we seem to be getting into more and more
> of late), so that the WG as a whole benefits.]

No such request was made.
Nevertheless, your input is most valuable.

> i) First, I do not believe a mention by the Chairs, in less than
> 30-odd seconds each, of over 15 drafts buys the WG anything.

This is discussed at some length above, but further comments are provided below for
completeness.

It is worth noting that many of the items in this list of 15 received a request for 5, 10
or 15 minute slots. The list, therefore serves two purposes:
- it flags certain drafts about which the chairs have something very brief to say
- it mentions certain drafts that would otherwise be entirely left off the agenda.

This last point is helpful to and welcomed by some people.

> By removing this, the WG saves a full 5 minutes right away.
>
> -- If the Chairs, in fact, want to draw attention to these documents, why
> not post your comments/urgings/admonitions with the pointer to these drafts
> directly to the ML?
> That is the right place to make the majority of the WG aware of these
> important documents anyway.

Agreed. And you may have noticed that I have been working through the drafts, reviewing
the new ones and sending comments to the list.
Editors and authors have also been receiving private emails to help advance their work.

But this is not an either/or. Bringing the items to the agenda still focuses the authors
and provides a brief *coherent* summary for all WG participants and for those many who
come to the meeting to find out what is going on, but who are not subscribed to the mail
list.

As mentioned above, a WG status email might help to alleviate this problem
*in*the*future*.

> -- Also, I noticed in some of your emails that you intended to "take a sense
> of the room" for some of the drafts on this huge list.
>
> I think that _should not_ be done.
>
> It is very difficult, if not impossible, to get a sense of multiple drafts
> in under 30 seconds each, and surely no fair sense can be got that way.

As mentioned in a previous response and above, this is done by the chairs, for the chairs'
benefit. It is not intended to be fair or easy. It is intended to help the chairs gauge
the feeling of the WG in a very rough but most importantly speedy way.

> It it much better for the sense to be taken on the ML (which is the best
> place anyhow) where people have time to think and reflect, and then respond.

No-one is suggesting that such a "sense of the room" replaces the mailing list as the
correct place to take consensus. It is just a tool - one that the chairs find useful.

** Thus, I reject this 5 minute saving.

> ii) Second, given that the ASON Routing Design team has just been
> constituted, and has only just begun its work, it is difficult to see why
> 10 minutes are being given to discuss this.
>
> Their composition and charter have already been posted to the list and
> examined by all, so none of those bear repeating.
>
> It should suffice for the Chairs to spend 2 minutes or so (if that)
> updating the WG on it.

The DT is carrying out a very important piece of work. It is important on four fronts:
- it is providing important function
- it is working on routing protocols that might be easily destabilised
  with dire results
- it requires input from multiple WGs
- it is building bridges with two external bodies with which we have a
  very bad history.

The DT has already started work and produced an outline draft.
They need to report on their achievements and outline the biggest issues and challenges.
They should be asking for any help that they need.

It is possible that they will need more that the ten minutes as I expect some questions
from members of the IGP WGs that are members of the CCAMP mailing list.

** Thus I reject this 8 (or possibly 10) minute saving.

> So we are now at +18 minutes (possibly 20).

5 + 8 = 13

> iii) The end-to-end recovery document was already mature a while back,
> and has been discussed at practically every IETF for the past several
> IETF's.
>
> The issue of misconnections (for the purposes of this draft) has also been
> adequately addressed to the satisfaction of the people who debated this
> issue (which includes me). AFAIK, that was the only substantive addition
> made.
>
> What other open issues are still left that cannot now be satisfactorily
> resolved on the mailing list and that require 5 minutes of the agenda?

I am glad to hear that the changes to this draft address your concerns.
The slot is to allow the authors to say what they have done in response to the concerns,
to check that those who were concerned are now happy, to check that everyone else is happy
that the changes are not detrimental, and to see if there are any other issues.

You may recall that the whole misconnection issue was only nailed and committed for
inclusion in this draft as a result of comments made at a similar meeting.

> It would seem that the last call can easily be done on the list (and is
> probably best done there),

There is no intention to do last call at the meeting.
The intention is to discover whether the decks are now clear to ask for last call.

Five minutes does not seem extravagant for this item. In fact I might expect Dimitri,
experienced as he is to skim through this in only 3 minutes, but if there are issues or
questions...

** Thus I reject this saving of five minutes, but might expect to save 2 minutes along the
way.

> bringing us to +23 minutes (possibly 25).

5 + 8 + 5 = 18

> iv) The segment recovery document was made a WG document after Seoul,
> but I don't recall seeing any discussion on it since.

There were at least 7 emails about this draft since Seoul.
There was a debate about the use of the Association object that had some value.

> Clearly, either the WG has not found time to focus its attention on this
> work between Seoul and now, or sufficient discussion has not been
> initiated on it by the authors.

Or the draft is cooked?
Or perhaps the authors have some additions that they want to propose?
Which is it? How are the chairs to decide?

> Either way, without any debate/discussion on the ML, this draft does not
> satisfy multiple criteria listed above, and it seems fair that the authors
> should attempt to generate discussion on the list, or summarize
> things so it can move forward.

This is a WG draft.
It is our responsibility to drive WG drafts to completion and get them shunted off to the
IESG. This makes room for new work in the future. Note that the authors are not
responsible for WG drafts: they are guardians of the draft for the WG. The WG is
responsible for getting them finished.

I agree that this draft is the most marginal for a mention so far, and if a cut had to be
made in favour of something else then this might be a candidate. It would be useful, but
not essential to cover this draft.

** Thus, hold this five minutes in abeyance pending finding something more important.

> We are now at +28 minutes (possibly 30).

5 + 5 + 8 + 5 = 23

> v) I am not sure I saw the graceful shutdown draft discussed (maybe its
> a split off of some related work), but if it's new work with no discussion
> yet, I would say we do not need 5 minutes for it.

See Dimitri's mail for references to the discussion.

The point here is that we are looking at a bunch of changes to the Hello mechanism (and
other protocol techniques) related to graceful restart. This draft proposes other methods
to take a link out of service and in a graceful way, and since the area is closely related
(and some might argue should be solved in the same way) there is value in bringing the
work together with the previous drafts.

However, I agree that this draft is also a candidate to be shunted out by something more
important.

** Thus, hold this five minutes in abeyance pending finding something more important.

> vi) The comments for the L2 GMPLS draft say "what it's about?".
>
> It seems that any draft for which this question needs to be answered at a
> WG meeting probably first needs to be discussed on the list,  and
> automatically does not meet multiple criteria from those you listed earlier.
>
> Maybe I missed it, but can anyone point out where on the ML this draft
> was discussed? And, what the outcome of those discussions was?

This draft has been around for ever!
I want to clear up once and for all whether this is in scope or out of scope. Whether the
WG believes it is plugging an important hole or simply following a completeness proof.
Will anyone implement it? Would anyone deploy it?
We need to close on this now and either get the draft on board, finished, implemented and
shipped, or else axe it.

I see a brief slot on the agenda as a very good way to force the issue.

For me this slot is more important than many others on the agenda.

** Thus, I reject this five minute saving.

> So it seems we can save a full 33 minutes (possibly 35, or 38 without
> graceful shutdown) from the agenda with substantial consequent benefit
> to the WG, by freeing up time for various other important activities
> (debates, discussions of other drafts, issues, etc.)

Correcting the math, this reads "28 minutes (possibly 30, or 35...)"

I accept that of this sum, 10 minutes are available to make space for something more
important. I am always glad to hear suggestions as to what that might be, but I do not
want to get into a "his draft is less valuable than my draft" debate with anyone.

> It would be good to hear if the WG believes that the above suggestions do
> not benefit the WG and/or should not be implemented (and why).

A well-phrased question, but since you are asking for / proposing change, it is
conventional to ask the question the other way round.

So...

I would be interested to hear if anyone in the WG believes that Vishal's suggestions would
benefit the WG.
Which of the items listed should we abandon, and in favour of what?

> Looking forward to seeing a revised agenda.

I will not be providing a revised agenda until it is clear what we are proposing to put in
place of things we cut, and until it is clear that that is a beneficial act.

The deadline for agenda submission is Monday, but I will be travelling from Sunday so I
will submit the agenda on Sunday evening UK time.

Please note that the agenda will still be open for revision in the meeting. I would
suggest that it is by far the best idea to get all agenda debates in the open *before* the
meeting. If we spend any significant time within the meeting bashing the agenda, we will
waste the meeting.

Thanks,
Adrian