[rfc-i] RFC Editor structure

John C Klensin john+rfc at jck.com
Wed Sep 10 06:34:39 PDT 2008


Thomas,

I have been wait for promised comments from Olaf before
responding to him and to Russ, and had already responded to Andy
privately, but I think the issues are a little different than
you apparently do...

--On Wednesday, 10 September, 2008 08:46 -0400 Thomas Narten
<narten at us.ibm.com> wrote:

> "Andrew G. Malis" <agmalis at gmail.com> writes:
> 
>> Actually, I DO remember how it used to be, first with Jon and
>> then with Joyce. I think Jon was a unique individual, who had
>> the ability to act as an largely autonomous expert, and
>> Joyce, to a lesser extent, also had an excellent
>> understanding of "what was right" for the RFC series.
>> However, I really do think that those days are behind us, and
>> the community really DOES want process in place where RFC
>> content updates and publication decisions (no matter how
>> minor) are completely transparent and open to public comment.
> 
> I think this is a large part of it. The world has changed
> significantly. And there is plenty of evidence that nowadays
> when someone doesn't like a decision that has been made, they
> complain and/or file appeals. That has a way of changing the
> dynamics of what people are willing to do "just because it is
> obviously right", since "obvious" is in the eye of the
> beholder.

I think there is a difference between what the community claims
it wants (it is really hard to be against "more transparent
decision-making") and what it wants in practice.  In practice,
these drives for transparency and community comment produce more
and more Last Calls with fewer and fewer meaningful responses.
Worse, the responses (and the way questions are asked) seem to
mostly be about time-consuming nit-picking, rather than making
sure that the larger and more substantive questions, such as
"should we be doing this?", are addressed. 

For the standards track, the time that process of delays with
few benefits (other than nit-picking that leads to even more
delays and a lot of frustration) produces calls for formal, "you
can safely implement from this and we may not bother going
further" proposed-Proposed Standards (excuse me, "Stable
Specifications") with WG approval only and no practical
opportunity for the type of open, transparent, cross-area review
that we claim we want in principle.   

For these procedural and administrative things, it leads to
documents that are reviewed on the details, but not on the
principles... principles that are based on the approval of three
documents slightly over a year ago (RFCs 4844 - 4846) and the
assumptions that (i) they were what the community wanted, rather
than the result of an exhausting and long-delayed process that
left too many people too burned out to comment and (ii) that
they have been proven successful over time.  I believe that the
latter assumption is false; that there are procedural
assumptions and boundary safeguards in those documents that have
been ineffective in practice.  My opinion of the first is
probably clear from the way that I state it above.

Perhaps it is time to borrow some ideas back from organizations
we used to criticize as "too slow" but that are now often much
faster than we are.  That might mean a stronger secretariat,
more rather than fewer independent agents who can make
decisions, and stronger assumptions that, if a WG ejects a
document, it understands and believes what it has written.  That
would imply that there need to be good mechanisms for opening
and reviewing things that really need to be opened, but that
arrangements that encourage, e.g., the IESG to start rewriting
text or to quibble for weeks about trivial points (_especially_
at Proposed Standard).

One of the corollaries of that is that we have too few, not too
many, appeals and other calls for reconsideration.   There are
few enough that we consider them a big deal and the IESG feels
attacked when one shows up.  There are few enough that we have
no mechanisms to prevent those who file appeal after appeal,
none of which are upheld, from using the appeals model as a DoS
attack.

But, unless members of the community are going to couple the
principle of "more transparent and open to public comment" with
practical and substantive personal commitments to do timely and
extensive reviews, it is hard to take those preferences as
anything but warm and fuzzy comments.

Finally, if we believe that Jon and/or Joyce were uniquely
skilled or had unique perspectives, designing a system that puts
single people (and requires no more than a single person) into
key decision-making roles (even narrower ones than before), then
we had better be sure that we have a supply of candidates with
at least an equal level of skill to perform those jobs.   If we
don't believe we have those candidates, then designing for
single-person decision-maker slots is just bad engineering.

best,
     john



More information about the rfc-interest mailing list