Re: [Pesci-discuss] Face-to-face meetings

Jari Arkko wrote:
> Easy. Do both. Its silly to think that 3 x 2 hrs = 6 hrs
> of face time per year will advance any major piece of
> technology. In addition, people will need to have
> design teams, conference calls, interim meetings,
> private meetings, etc.
...
> Having said that, I don't think the focus of the main
> IETF meeting should be just "reporting to a broader
> audience" as Geoff put it. 

I decided to respond to your note, out of the entire thread, because it 
represents some points of view that seem to dominate the IETF today.  
Further, I believe you do not carry the baggage of the IETF's early 
years, so that your "representation" is a bit more pure.

I note that your lengthy list of "In additions" did not include mailing 
list discussion.  Whether that was unintentional or not, we do seem to 
have lost the original model, namely that mailing lists are where the 
primary working group work is done.   Historically, IETF meetings have 
been used for assessing overall wg direction or for intense focus on 
selected issues that have proved difficult to resolve on the mailing 
list.  That made those meetings rather surgical or rather strategic, 
rather than being used for the bulk of the development detail.

The acknowledged advantage of the mailing list as a primary venue is its 
openness.  Requiring travel to a meeting -- whether interim or IETF week 
-- filters out a great deal of participation.  Maybe folks do not want 
the IETF to be as broadly inclusive as it used to be, but that sort of 
change should be explicit, rather than happen as a side-effect.

As others have noted, the model of doing primary work at face-to-face 
meetings is popular with some other standards organizations.  If the 
IETF adopts that model, too, we might want to ask what makes the IETF 
distinctive and worth maintaining as a separate organization.

The ITU is another group that does its primary work with extended 
face-to-face meetings. Watching it in action showed me another benefit 
to the IETF's original process that is easily missed:  It is more stable.

The IETF process historically has produced a sequence of incremental 
developments to a specification, with plenty of time for wide review.  
Hence development tended to be evolutionary and based on extensive 
feedback.  By contrast, the crucible of a face-to-face meeting 
encourages spontaneous -- rather than deliberated -- problem-solving and 
can produce dramatic changes to a specification, without adequate 
understanding of their impact.  I have watched remarkably massive 
strategic changes proposed and adopted over the course of a few days, 
with perhaps 20 people in the room.  Folks won't be surprised to hear 
that the primary problem was that the adopted solution mostly would not 
work...

The mantra that is likely to be used to response to my concerns is "no 
decision is final until it is approved on the list" however this ignores 
some important differences in earlier dynamics versus now:  It is one 
thing to have an active mailing list that is doing the bulk of the work, 
and then bring occasional, discrete issues back to the list for 
post-meeting approval.  It is quite different to move the bulk of the 
work to face-to-face meetings, essentially stripping the mailing list of 
serious content, and then load the list with a large number of decisions 
to "approve".  It is simply not reasonable to expect the latter to do 
more than catch the most egregious problems.

We used to worry about using IETF meeting time effectively.  Not very 
many years ago, using the meeting for "reporting" was very nearly 
banned.  Reporting can be done on the mailing list. 

At the least, we should be clear about the real purposes of IETF 
meetings, and make sure they make sense.

d/

_______________________________________________
Pesci-discuss mailing list
Pesci-discuss@ietf.org
https://www1.ietf.org/mailman/listinfo/pesci-discuss