--On Wednesday, July 22, 2015 11:55 -0400 Eric Burger <eburger@xxxxxxxxxxxxxxxxxx> wrote: > I sort of like this better. The problem we have today was > identified by Dave Cridland: it is already so hard to charter > a work group, people give up and leave. You need an active > mail list, an active BOF, and active BOF, run through the Dave > Crocker and jck filter, and finally get chartered. Easily a > year if not more. I can't speak for Dave (and wouldn't if I could) but, since you called me out more or less by name... I would be quite enthused about setting up WGs on the basis of AD judgment, without BOFs or other heavyweight processes, if the following principles were agreed to and treated as conditions: (1) We did not hesitate to shut them down if they were not productive. I don't want a rigid rule because I think we all know that, in practice, some WGs move at a different pace than others, but it is important that we be able to recognize "not meeting expectations and making reasonable progress" and act on it. In the last several years, I believe we have one successful example of doing so and lots of examples to the contrary. So, whether we are serious at that or not is still uncertain. (2) There be evidence of enough committed people to do the work. We cannot collect signatures in blood or otherwise enforce promises, but there must be more than an enthusiastic individual, company, or organization or two. (3) The initial list of proposed or likely participants or supporters must be sufficiently diverse to show broad community interest. That is key to meaningful community consensus and even more important if there is involvement by one or more IESG members or the companies with which they are associated in the work. (4) The IESG and the community have the be firmly behind the idea that "a WG produced this" is not a guarantee of either standardization or publication if the work turns out to be below the quality standards of the IETF or the RFC Editor. We've seen a lot of "we spent a lot of time on this in a WG and therefore it is entitled to approval even if we will let the community tweak it a bit". That is, IMO, a problem even with carefully designed and vetted WGs; it could be much worse if attached to a lightweight WG setup process. I'd like to see WG-forming BOFs become a thing of the past except when they are actually needed for problem definition, workplan design, or similar activities. It would be, IMO, much more productive to allocate meeting time to a WG to see if it can actually get something useful done than to spend the same time discussing and theorizing about whether getting something useful done is likely or certain. Especially because I think (2) and (3) can only be realized by being sure the IESG is on board with the principles and then exercising discretion and good judgment and because the IESG has to be signed up to make either (1) or (4) work, I don't think this is feasible unless the IESG is committed to it and willing to be held accountable if they fail to come through on those principles. > Even with that heavyweight process, we manage to take on work > that probably belongs elsewhere. And _that_ is my one exception. I'm concerned about the IETF resenting and/or complaining bitterly when some other body takes on work that we believe belongs to us. I'm even more concerned when we take on work that probably belongs elsewhere, especially when there is evidence of forum-shopping. So I think that, when those issues exist, they do need to be aired before WG creation. However, I think that review and any discussions that are needed can be carried out very much more efficiently and with a higher success rate than we have been doing/using. > I like the I-D approach. Provide the resources to get a proto > WG going, let it run for six months, and see if it is getting > any traction. Yes. See above and note that I don't know that "six months" is the right number for every group. However, if it isn't, we ought to be able to figure it out and set a more appropriate topic at startup time or grant one extension and then stop. > I think that will let people relax that don't want to see > the IETF take on work that no one is interested in - if no one > is really interested in the work, it will die in 6 months. > > Conversely, if something that an AD or two thought would not > be interesting takes off in that initial six months, I'm > sure they will claim parentage :-) > > Other thoughts? Because it might prove illuminating, at least about a prior round of this type of conversation but with a slightly different focus, I recommend two long-expired I-Ds to anyone interested in this topic: https://datatracker.ietf.org/doc/draft-huston-ietf-pact/ and https://datatracker.ietf.org/doc/draft-klensin-overload/ --john