Hello folks,
In my note, I said that the IETF Last Call process was somewhat broken.
I really meant to say that the WG Last Call process was somewhat
broken. Sorry about that!
Regards,
Charlie P.
On 7/31/2018 10:03 AM, Charlie Perkins wrote:
Hello folks,
Please excuse if this is not relevant, or has already been suggested
-- I didn't read the whole thread! The heart of my suggestion could
be summarized as reducing AD workload, but along the way making some
improvements to the process of document review and progress. The idea
is motivated by comparison to IEEE task group management, which seems
to go more smoothly.
First, I think that the IETF Last Call process is somewhat broken.
It's not really Last Call; instead, it is an urgent plea for people to
read the document. It could be compared to a Letter Ballot in the
IEEE. But the implied urgency in the IETF context leads to a sense
that the document is almost "done", when in fact the document may have
major issues that need a lot of attention. Letter Ballot isn't like
that. All of the comments are tabulated, and subsequent task group
meetings are dedicated to resolving the comments. Sometimes, there are
*thousands* of comments (for documents in large task groups like
802.11ax).
Moreover, it sometimes happens in the IETF that WG Last Call ends for
a document and no one in the working group has read it. In contrast,
at least in 802.15 Letter Ballots, it is very very likely that quite a
few people will read and comment on the document. That's because,
essentially, they have agreed to make document reviews as part of
their willingness to participate in 802.15, which could be compared to
an Area in the IETF (please note this is a very rough comparison).
How could the IETF find more assurance that a WG document would get
sufficient review?
Perhaps the pool of readers should be identified before the WG adopts
the document. If that happened, then when the document is ready for
wider review, the reviewers are known and have already expressed a
willingness to review the document. In the IEEE, the reviewers are
partially motivated to retain their voting rights, but in the IETF we
probably wouldn't want to institute such a system as that.
Nevertheless, having a solid constituency for document review and
comment resolution in the IETF would help a lot.
My hope would be that, with a better process for document review and
comment resolution, the ADs would find their job to be a lot easier.
Lastly, I know that some working groups already try to do some of
these things. I suggest that we will know when the review process has
been sufficiently improved, when a document almost never goes through
Last Call without sufficient comment.
Regards,
Charlie P.
