On May 1, 2013, at 1:59 PM 5/1/13, Dave Crocker <dhc@xxxxxxxxxxxx> wrote: > > The blog nicely classes the problem as being too heavy-weight during final stages. The quick discussion thread seems focused on adding a moment at which the draft specification is considered 'baked'. > > I think that's still too late. ...and not useful unless the diverse review and changes to the document take place much earlier in the process to make sure the document is "baked." Declaring done doesn't make it so. > > Certainly it could be useful, but it's still very late in the process. As Jari wrote, we often bury the heavy tail of the process in a limited discussion among IESG and the document authors. The tail is heavy in two different ways: * significant review and modification takes place in IESG review, after the WG and the IETF have declared the document done * the burden of the review, managing the discussion, making sure any changes fix the problem and don't break anything else often falls on the IESG and even a single AD > > Specification development usually does -- and certainly ought to -- formulate basic design ideas or approaches that motivate the details in a specification. A good specification will include discussion of this, and I think the typical and most significant hashing and re-hashing that happens in later stages is about the design issues. > > While basic design revisions can happen at any time in a process, I'll suggest that a specification effort should be asked to document its interesting design choices early, and circulate /these/ for external review, iterating later if basic choices are changed. > > I suspect that an earlier exercise at summarizing functional goals and design approaches and issues will have a number of benefits, beyond enabling earlier external review. Dave - excellent idea. A summary that explains the "why" first, that promotes early cross-area review or otherwise expands the set of people that look at and understand the nature of the problem and the structure of the specification will allow for earlier review and lighter-weight correction of problems. > > d/ - Ralph > -- > Dave Crocker > Brandenburg InternetWorking > bbiw.net