Wow. This was an interesting thread (that developed quickly). Why aren't you
guys all on airplanes yet?
So, to summarize a couple of points that other people made, but I didn't
want to lose in the forest...
... DISCUSS has no BCP process standing today. I'm not sure that giving it
process standing is the best use of any time we're willing to spend on
... if people would feel better if the IESG reissued the discuss criteria
document as an IESG statement, go for it.
... I understand what Ted is saying about "this needs to be a BCP". If we
were, in any way, capable of making changes to BCP text during the same
decade that we discover a problem, I might even agree. We're not. Pete's
heavy sigh about the amount of stuff that's stacking up for PUFI is real,
but is also the result of the community's decision to ignore process topics
after Montreal. That existing stack does not make me want to put anything in
BCP text that's not already in BCP text.
Look. We can't even come up with new names for three levels of standards
track that aren't stupid, and that really doesn't matter. What makes us
think we can come up with a set of "non-criteria" that will be solid enough
to put in a BCP that might last for more than a decade (please check the
dates on most text in 2026, 2418, etc), especially if the downside is that
an AD can then say "but that's not on the list of non-criteria"?
If the concern really is a rogue AD, I'm pretty sure our ability to predict
and preempt the path of rogue-ivity isn't up to this challenge (it would be
the IETF BCP equivalent of proposing a "Law Prohibiting Cooking Intelligence
to Justify Invading the Country Next to the Country that Hosted the Group
that Attacked Us" for the USA in 2002).
... I'm thinking that unreasonable DISCUSSes will be unreasonable whether
they specifically map onto a "non-criteria" in the current ION, or in some
future BCP, or not. Unreasonable DISCUSSes should always be challenged.
... there is a pre-existing escalation process for dealing with unreasonable
DISCUSSes, but it doesn't sound like we're particularly good at using it.
... there is a pre-existing process for dealing with an IESG that goes
against the current in the community, whether they are actually violating
BCP text or not.
Various folks (who I consider smart friends) have pointed out that we've
never "used" the recall process (also in RFC 3777). That's true, but in
discussions, I've heard about two recall petitions that were filled out and
then shown to the subjects of the petitions, and the recall petition
initiators were satisfied with the results, without "going public". So I'm
not as concerned about a rogue IESG that says "yeah, that's what the ION
says, but we don't care" (or even "yeah, that's what the IESG Statement
says, but we don't care") as other people seem to be. If you're rogue
enough, you can blow off BCP text, too.
On the specific topic of "unfiltered" reviews, (as one of the original
Gen-ART reviewers) I do see authors talking about "satisfying" reviewers.
Please don't. "Satisfy" the COMMENTing/DISCUSSing AD, and "satisfy" the
Always remember that http://www.alvestrand.no/ietf/gen/art/gen-art-FAQ.html
explicitly addresses the question, "What if the Gen-ART reviewer is just
flat wrong?". That is not an accident...
I think it would be nice for the IESG to collect, and maybe even put in one
place, "the way reviews work".
It's worth noting that even for Gen-ART, the process has changed
significantly since we started writing reviews for Harald. Those reviews
were done ONLY for documents on a telechat agenda, and they went ONLY to
Harald, who either laughed, COMMENTed, or DISCUSSed. The reviewers weren't
involved in further resolution, and I'm not even sure authors knew that
Harald's comments came from someone else's review. Eventually, we started
reviewing at IETF Last Call time, and copying shepherding ADs, and then
authors, and then WG chairs, but that's not where we started out.
So, for example, it probably IS worth finding out if the rest of the ADs who
sponsor reviewing bodies also use Russ's division into
- "did you consider the reviewer's comments?", especially when the review
was issued as
part of IETF Last Call, and
- "how did you address this specific comment, which I agreed with?", whether
the AD entered the comment as a non-blocking COMMENT or as a DISCUSS.
IETF mailing list