Instead we should understand that we cannot and should not try to demand or
expect documents that are perfect. We should demand 'good enough' and let
the outside world evaluate and feed the results back to us.
One view is that we already are in that mode. I don’t think any reasonable
person could claim that any specification from a standards body of any sort is
perfect, including from the IETF. More interestingly, the question is whether
our community, directorate, and IESG reviews and associated practices reach the
‘good enough’ level or under- or overshoot. But one person’s egregiously
unnecessary fine-tuning is another person’s major threat to the Internet.
Personal opinion: we overdo it, a lot of the time.
But I think we are agreeing that we actually shoot for ‘good enough’ but do too
little follow-up and revision. I could cite many counter examples where that
follow-up does happen. But in many cases there is no follow-up. Why is that?
Specification turned out to be uninteresting, so need to follow-up? Close
enough, no business need to waste time to get to perfection? Remaining details
hammered out in interops, code already runs, no need to revise? Aside from the
few errata, no need for bigger changes? Worst specs are revised others are good
enough? IETF process too complicated for the update? Probably a mixture of
these reasons.
Jari
signature.asc
Description: Message signed with OpenPGP using GPGMail