On Mon, Jan 23, 2012 at 11:58 AM, Dave CROCKER <[email protected]> wrote: > The premise of the open IETF processes is that open collaboration means > broad-based review, diverse perspectives and broad-based support for> > whatever is decided.
Which doesn't mean that every word is or needs to be written in real-time on a public WIKI with everyone watching. The editors usually go off and make changes, and the working group reviews those changes. I'm suggesting nothing more nor less here. > Closed processes make it more likely that there will be narrower > perspectives -- and therefore some aspect of the issue missed -- and less> > support. It actually makes an AD less accountable for their Discuss. I very much disagree with you. What a side discussion does is allow the editor(s) and the AD(s) to understand each other. I have similarly had side conversations with GenART reviewers, SecDir reviewers, and other working group participants. That those discussions didn't happen on the mailing list didn't compromise the process. It allowed us to work out an understanding, and allowed me, as editor, to propose some text, which the working group could review. > Working group discussion and other reviews are open and collaborative. All > of the details, major and minor, for handling concerns is public and on the> > wg mailing list. And you NEVER, in any of your editing in, say, DKIM, went off and discussed things off the list? Every detail, major and minor, was on the list at every moment of its development? > directly to the working group? Why should ADs be immune from the requirement > to take their concerns to the chartered forum that is established for> > conducting exactly such discussions? No one has said that an AD shouldn't participate in a working-group discussion of the issues she brought up. All I've said is that the chairs, not the ADs, should manage the working group, its process, and its discussions. I've had quite a few DISCUSS comments on my documents, which (the comments) were resolved with one or two email messages, sometimes with no text changes at all (there was something the AD didn't understand). While I have no *objection* to the DISCUSS being posted to the mailing list, as a management point I see no reason to remove the judgment from the chairs' hands of how to handle the conversation. > Further, AD Discusses are now publicly documented. So the step you are > objecting to is a matter of convenience, not availability. Exactly. Nothing hidden at all. > streamlines the resolution process by eliminating the information > gatekeeping by whoever is mediating between the AD and the working group: Nothing is "streamlined" when a minor clarification turns into (perhaps) a large number of messages from the working group copied to the entire IESG. > It puts the AD directly in front of the folks whose work is being> > challenged. Ah, here we get to a key point: the assumption that a DISCUSS is adversarial. A DISCUSS does block progress of a document, but it's often NOT a "challenge". It's often simply a request to, well, discuss something briefly. It's also often an administrative mechanism -- as Pete said, a way of tracking something that needs to happen before the document goes to the RFC Editor. The working group can see them any time it wants to look; not all of them need to be explicitly brought to the WG for discussion. > "The entire IESG"? I don't understand how this suddenly puts the entire > IESG into the main flow of interaction between the AD who is blocking things> > and the wg that is being blocked. Pete has answered this: the messages we're talking about are copied to the IESG mailing list. When the WG participants respond, they can remove that CC. We can all guess how often that will happen. > This is a classic error in logic: The current mechanism often works > adequately, so let's ignore the times it doesn't and let's ignore its> > inefficiencies an inequities. Did I ever say anything about ignoring anything? I don't recall that. In fact, I said that when in the chair's judgment the discussion should go to (or at least be copied to) the WG, the chair can and should take it there. We've "hired" the chairs as managers, and they should be left to manage. > My core point is that the /design/ of the current process is inherently > flawed: it violates the IETF's model for handling objections that pervades> > every other aspect of IETF work. There's no question that the DISCUSS process is flawed. The question is whether automatically copying every DISCUSS and COMMENT to the WG mailing list is the (or "a") right way to fix it. I think it's not. >> In this case, in fact, Murray did bring the discussion to the WG, without >> your needing to force it.> > This places an undue burden on chairs and/or authors. It's an "undue burden" on the chairs to manage their working groups? I can't agree. On Mon, Jan 23, 2012 at 12:35 PM, Pete Resnick <[email protected]> wrote: > I'll ask for the chairs guidance, and if they want this > moved off somewhere else we can do so. I'm happy to continue this discussion here. I'm also happy if Pete or Dave wants to take it to a more open place, such as [email protected]. Or, perhaps before that, to [email protected]. > Right. One alternate means for doing this is moderating discussion. We are > generally loathe to do moderated mailing list postings in the IETF unless > things have gotten entirely out of hand in a WG, but we do allow it. When I talk about the chairs moderating the discussion in this sense, I do NOT mean using a moderation queue in the list software. I'm talking about management. This, of course, relies on the WG participants' cooperation, and banning them for being uncooperative is rather extreme. > Right. But the additional worry I have is that responses to it are *not* > publicly documented (i.e., when chairs or editors reply to the AD for > clarifications, potential resolutions, etc.) unless the AD takes the > additional step to update their comments in the tracker. That certainly > needs to change. So, even if it is a moderated discussion, we need to do > something different to document that discussion. This is a key point, which we should deal with in some way. I don't think that putting all of the discussions on the WG mailing list is the right way to handle it. Perhaps having them copied to a discussion archive is the right answer. > This is dependent on the AD *and* the WG involved. An AD might bring up an > issue that was clearly resolved by a WG many months (years!) ago. Without a > bit of restraint by the WG, a lot of sturm und drong can be generated Exactly one example. Scenario 1: DISCUSS message goes to editors and chairs. AD brings up issue X. Editor or shepherd replies, "WG discussed this at length, and closed the issue." AD says, "Ah, OK, thanks," and clears. Scenario 2: DISCUSS message goes to WG mailing list. AD brings up issue X. Five angry WG participants, sick to death of issue X, send angry responses, which go to the IESG. Two holdouts, who had prolonged the discussion endlessly before, take this as an invitation to rant about issue X again, sending it to the WG and the IESG. AD gets the message and clears, but lots of time was wasted. Scenario 3: DISCUSS message goes to editors and chairs. AD brings up issue X. Editor or shepherd replies, "WG discussed this at length, and closed the issue." AD says, "No, I don't think that's the right answer, and I won't clear." Shepherd posts the discussion to the WG mailing list. I'm sure you can see that I think scenarios 1 and 3 are the way things should work, and 2 is the one I want to avoid. The chairs are the ones most able to judge what will happen, and when it should be posted to the WG list. Barry _______________________________________________ marf mailing list [email protected] https://www.ietf.org/mailman/listinfo/marf
