Inline ....
On Jan 18, 2008, at 10:56 AM, John C Klensin wrote:
Hi.
The current cutoff schedule for Internet Drafts dates from my
time on the IESG (i.e., is ancient history). It was conditioned
on the pre-IETF rush and the observation that the Secretariat,
at the time, required a sufficiently long time to get drafts
posted in the pre-meeting rush that, unless there was a two-week
cutoff, we couldn't reliably have all expected documents in hand
prior to the start of the meetings.
Splitting the "new" and "revised" drafts was a further attempt
to compensate when the load built up enough that the choices
were between such a split and moving the submission deadline for
_all_ I-Ds back even further. The conclusion was that a split
was desirable because a three-week cutoff for revisions would
seriously interfere with WGs getting work done in the run-up to
IETF meetings.
With the automated posting tools typically getting I-Ds posted
in well under an hour and a tiny fraction of the documents being
handled manually, the original reasons for the submission
cutoffs no longer apply. It is still reasonable, IMO, to have a
cutoff early enough to permit people to receive and read
documents before departing for the meetings, but it seems to me
that criterion would require a cutoff a week (or even less)
prior to the meeting, not two or three weeks. Other models
about giving people time to read might suggest leaving the "new
document" cutoff at three weeks before the meeting, but seeing
if we could move the "revision" cutoff considerably closer to
the meetings.
Makes sense - also the IESG agree to something along lines of moving
the deadline for drafts that are moving from individual to 00 WG
drafts to be the same as deadline for non 00 drafts but I don't know
that has been implemented yet or not.
I don't necessarily object to retaining the current two and
three week posting deadlines, but I'd like to know that the IESG
has done a careful review of those deadlines and their
applicability to the current environment and concluded that they
are still appropriate, rather than having the secretariat retain
them simply on tradition and autopilot.
I don't think the IESG has done a review of these times - and I agree
the new tools certainly change some of the reasons that caused the old
times to be chosen. At least one of the reason they IESG has not
looked at them is just the IESG is working on other things. As a
result, the status quo ends up being the default decision.
Speaking only as an individual contributor here, I 100% agree with you
that we want these times to be as short as possible yet still leave
enough time that people have a reasonable chance to read the material
before the meetings so we don't have to run tutorials in the valuable
WG meeting time. It may be that the right time is actually different
for some WG or areas though that would be complicated to have in
practice. Often the time right before an IETF meeting lands on a major
holiday further reducing the time and we have travel time so I think
theses needs to be taken into to account when thinking about how much
time is needed. For the past several meetings I have tried to read or
at least skim all the documents that are the agenda of any RAI WG
meeting. I would expect that most ADs would be reading at least all
the documents in the WG they were the responsible AD for. Looking at
the authors and key contributors of WG documents in RAI, I see that
there are many that are very active in over 2/3 of the RAI WG +
Behave. When I look at the number of WGs that some of the IAB members
participate it in, it is also very large.
I can send you my reading lists for IETF 68 and 70 and it is a bit
hard to count "real pages" that don't include all the boiler plate
stuff. For ietf 68, which was a very heavy reading list for me, it
had about 250 drafts with a "wc -w" word count of a bit under 2
million words. While IETF 70 which was very light had around 120
drafts with a word count of around to 0.9 million words. A large
percentage of these were revisions of drafts I had previously read
which greatly speeds things up - clearly no one is going to read this
many drafts where it is the first time to looked at them. At ietf 68,
about 70 of the drafts out of the 250 were 00 drafts but some of those
were just renames so hard to count without a more detailed look.
Some drafts I can read in 15 minutes if I am not sending any comments
and they are short and familiar. A complicated draft where I might
have to go and reread some other background draft or RFC and I send
significant comments will take four hours. Now I have the luxury of
being able to use most of my time to read drafts right before an IETF
but I think the time also needs to work for people who participate in
several WGs and are reading during their "own time" vs. their
"employers time".
If you asked me today if I would make the time longer or shorter, I
have no idea which I would say. I do think that looking at various
groups of participants, who much they need to read, and how much time
they have to do it, is the right way to figure this out.
Cullen <with my individual contributor hat on>
thanks,
john
_______________________________________________
Ietf mailing list
Ietf(_at_)ietf(_dot_)org
http://www.ietf.org/mailman/listinfo/ietf