Inline <tp>
----- Original Message -----
From: "Thomas D. Nadeau" <tnadeau@xxxxxxxxxxxxxxx>
To: "Berger Lou" <lberger@xxxxxxxx>
Cc: <ietf@xxxxxxxx>
Sent: Thursday, February 26, 2015 10:23 PM
> On Feb 26, 2015:5:12 PM, at 5:12 PM, Lou Berger <lberger@xxxxxxxx>
wrote:
>
> Tom,
> On 2/26/2015 4:48 PM, Thomas D. Nadeau wrote:
>>> On Feb 26, 2015:4:30 PM, at 4:30 PM, Joel M. Halpern
<jmh@xxxxxxxxxxxxxxx> wrote:
>>>
>>> Thomas, if participants who can not make the conference calls are
obliged to listen to the full recordings to get the key points of what
has happened, why, what are the questions, and similar issues that need
to be visible to the WG, then we are not running an inclusive process
that allows for participation by the range of individuals we need.
>> That is the problem. We're really only geared to have full-on, in
person meetings all the time. That does not lend itself to being
flexible/agile. That is not to say that I want to exclude anyone, but
to be fair, if a subset of people want to move a pile of work forward,
we shouldn't ENCOURAGE that behavior not stifle.
>
> It sounds to that you are describing design team meetings (or, in some
> cases, an authors meetings). I'm not sure what makes folks think that
> the only way folks can collaborate is via the mail list or full on
> interim meetings.
Yes.
>
>>> There is a reason that the IETF distinguishes between design teams
meetings, where the design team has to explain their work carefully to
the WG, and working group meetings. There has always been a problem of
not getting as much context as we would like from the WG minutes. But
since we explicitly take all resolutions to the list, this is
ameliorated by folks being able to ask for explanations, by those issues
being taken to the list promptly, and by the fact that we only met 3
times a year. If you have bi-weekly calls and the WG can not tell what
is going on with those calls, then what you have is a design team. And
then the folks involved need to own up to it as a design team,
understand that they need to explain to the list what they have
analyzed, their reasoning, and their conclusions.
>> Design teams should be able to work asynchronously, but with fixed
schedules and not have to have everything explicitly documented at every
step.
> 100% agree.
>
> I've been involved in many of such meetings (conf calls, webex, etc.)
> and the key is that DT/authors' meetings need to be explained and
> discussed on the list *at a time of their choosing*, which in my
> experience is usually when a draft is published/updated.
>
>> If someone else is curious, they can get involved but not in order to
slow things down or throw a monkey wrench into the works. If people want
to keep leaning back on 10 year old process RFCs and arguing "well thats
just the way we've always done things around here" then this
organization is going to continue to slow its progress even more - and
its descent into irrelevancy. There are a lot of people here (myself
included) that want to evolve things because they think the IETF still
has a lot to offer the industry. But if the organization won't evolve,
people will take the path of least resistance and go elsewhere as they
have been doing if you haven't noticed.
>
> This statement just confuses me as you note below we've always had
ways
> for folks to make progress fast -- when there's interest in doing so.
> It just seems to me that many are enamored with interims and think
it's
> the sole/best way of demonstrating progress between full meetings.
> Perhaps you're just saying that they're mistaken...
Its not being enamored as much as it being one of the only
obvious/acceptable vehicles to progress WG-level
work forward - at least if the management is involved. In NETMOD for
example, we've broken the interims into two "themes": Yang 1.1 work and
modeling. The former is like its own design team, and the latter is
like many design teams coming to one place every other week. The former
not only meets every other week, but discusses issues on the list. But
to the latter - that is more like a touch point for those subteams.
Those subteams go off on their own for weeks at a time and iterate as
needed. And they work without all of the overhead of a formal meeting.
They may or may not discuss progress on the list until issues come up.
<tp>
Tom
Take a small group of engineeers, expert in technology, get them to hold
regular meetings focussed on a narrow range of topics and they can make
faster progress, as you cite for 'netmod'.
What is also likely to happen, and I see it with netmod, is that they
will develop their own way of working, their own terminology, their own
technology even, which de facto raises the bar for anyone else who wants
to participate or to understand what happened. They don't mean to
exclude other people, they just do. (small groups, Psychology 101)
I see 'netmod' as a poster child for this with its issue list, state
machine for issues and so on. Even though I was tracking the list when
the 'Ynn' issue list was created, I don't know where its state machine
came from. In recent minutes, I don't know what
" AB: I am not sure YANG 1.0 specifies C1 explicitly somewhere.
JS: Does A3 not follow from A2?
KW: A3 is more a corollary of A2.
AB: The high-level problem is how to create and maintain the
information needed to achieve A4. "
is about; a brief search of mailing list and I-Ds gave me no explanation
for A2 to C1.
And if a different group of engineers works on different topics, then
they will likely, in the absence of any guidance, use different
technology, different terminology and end up with a way of working that
is as alien to the first group.
As I said, changing the way we work.
Tom Petch
</tp>
--Tom
>
> Lou
>
>> If you want a real example of how this can actually work, watch Anees
explain how Open Config has done this with just weekly phone calls and a
bunch of people typing on keyboards. They've done this in less than a
year, and have rough consensus and (production) running code. This is
how the IETF used to operate: people got together, hacked code and got
things working. The goal was not having meetings, but producing code
with rough consensus.
>>
>>
https://code.facebook.com/posts/1421954598097990/networking-scale-recap
>>
>> --Tom
>>
>>
>>
>>> Yours,
>>> Joel
>>>
>>> On 2/26/15 4:21 PM, Thomas D. Nadeau wrote:
>>>>> On Feb 26, 2015:4:16 PM, at 4:16 PM, Brian E Carpenter
<brian.e.carpenter@xxxxxxxxx> wrote:
>>>>>
>>>>> On 27/02/2015 09:08, Thomas D. Nadeau wrote:
>>>>>>> On Feb 26, 2015:2:42 PM, at 2:42 PM, Benson Schliesser
<bensons@xxxxxxxxxxxxx> wrote:
>>>>>>>
>>>>>>> Nico Williams wrote:
>>>>>>>> Yes, but a record that a concall or other interim meeting took
place,
>>>>>>>> and who attended, even if there are incomplete or missing
minutes, is
>>>>>>>> important for IPR reasons. Ensuring that such meetings are
NOTE WELL
>>>>>>>> meetings is (should be) a priority, and that includes ensuring
that a
>>>>>>>> record of that much exists.
>>>>>>>>
>>>>>>>> Ideally the concalls and other interims would be recorded.
>>>>>>> I agree completely. My point was that meeting records (including
minutes) will inevitably be incomplete, or possibly inaccurate, and that
relying on the mailing list as an authoritative record is more
effective.
>>>>>>>
>>>>>>> Of course it is disappointing that we can't meaningfully
translate voice discussions into text, in the minutes or in mailing list
threads. If there were some magic tool e.g. that took better minutes
then I'd be happy to use it. But otherwise, I think we just have to
trust chairs to manage WG collaboration in whatever way is most
effective for their WG's collaborators.
>>>>>> The first step is to agree that an A/V recording is record
enough.
>>>>> It absolutely is not enough. Please see my previous message,
>>>>> and the relevant rules in RFC 2418.
>>>>>
>>>>> Brian
>>>> You are missing my point. RFC or not, the IETF needs to evolve.
>>>>
>>>> --Tom
>>>>
>>>>
>>>>
>>>>>> Perhaps having meetbot/txt notes that at a min include
actions/decisions like we do in the issue tracker we've used for
NETMOD's Yang 1.1's issues.
>>>>>>
>>>>>> --Tom
>>>>>>
>>>>>>
>>>>>>
>>>>>>> This will inevitably be suboptimal for some part of the
population. (For instance, I've never been able to find an interim
meeting time that fits the schedules of all attendees.) But if they (we)
always revert to the mailing list for decision making then I suspect our
work can remain open and transparent.
>>>>>>>
>>>>>>> Cheers,
>>>>>>> -Benson
>>>>>>>
>>>>>>>
>
>
>
>