[rfc-i] RFC Editor Model Version 5 and revised RSE SOW

John C Klensin john+rfc at jck.com
Sun Apr 26 16:30:46 PDT 2009

--On Sunday, April 26, 2009 08:08:50 +1200 Brian E Carpenter
<brian.e.carpenter at gmail.com> wrote...

>> However, when disputes happen they are 'registered' with the
>> RSAG (who cannot immediately influence _that_ decision,
>> except with advice). The registered disputes are reported to
>> the IAB who in their oversight role could establish whether
>> there is a pattern of acting in bad faith.
> That is far too weak an approach IMHO. Firstly, what on earth
> does
>    If the Series Editor decision is not satisfactory, then the
> the    matter must be registered with the RFC Series Advisory
> Group.
> actually mean?
> Either an appeal decision is appealed to the
> RSAG, or it isn't. In fact, it can't be, since the RSAG isn't
> superior to the RSE. 


I have to agree with SM and Olaf on this one.  First of all, we
have gotten ourselves into a situation in which there is no time
to really get this right (if, indeed, that is possible... see
the comment about RFC4071 below).  Two things follow from having
a definition of the RSE characterized as having executive
management responsibility for the RFC Editor function.  

The first is that, if the IAB (and IAOC?) cannot make a good
choice of someone to fulfill that role, we are in deep trouble
-- while one might debate how good the document is about dispute
resolution and appeal procedures, it has absolutely no
provisions for how the system works with a vacancy in the RSE
position or if an RSE is paralyzed procedurally.  FWIW, I remind
you (and everyone else involved) that none of the IAOC, the IAB,
the IAD, or the IETF Exec Dir have the skills needed to carry
out the RSE job as part of their job descriptions.

The second is that there better not be a lot of appeals or
internal disputes within the RFC Editor function that cannot be
resolved by reasonable discussion and informal mediation.
Because the job is critical-path and contractual, such disputes,
if unresolved, could lead to just the sort of procedural
paralysis mentioned above.  I note that RFC 4071 has no section
titled "how the IETF is to function if there is no Secretariat".
The RSE is critical-path in the same sense that the Secretariat
is... and that even the IAD is not.

Even if we do have appeals or other debates, they are almost
certainly going to have to be informational, "please don't do it
this way next time", rather than retroactive corrections on
which operations wait while decisions are being made.

If things break down, exactly how the appeals procedure works is
going to be the least of our worries.

> There's no need to reinvent a new form of wheel. Just cut and
> paste from section 3.5 of RFC4071. That defines a very clear
> procedure with effective oversight.

Really?  That appeal mechanism has never been used, despite the
fact that the IOAC has failed to live up to many of its
responsibilities under other sections of that document.  There
have been transparency failures, attempts to approve things
internally (arguably including earlier version of this document,
the RFI materials, the final SOWs, the procedure for appointing
the RSE, etc.) that RFC 4071 clearly requires be exposed for
community review.  While one can debate any given case and
regardless of the intentions, the overall impression is one of
operating in executive session and by executive authority, not
the level of openness that 4071 calls for.

And, given that there have been no appeals, the community has
either concluded that things aren't bad enough or that invoking
the appeals process would be too expensive and probably
pointless... just like the recall process.

I'd rather have a vaguely-defined procedure that might work and
that would allow for lots of informal mediation rather than
copying a precisely-defined and formal one that does not appear
to work.  Wouldn't you?

And, in case it isn't clear, I agree with SM about the cutoff
date for the initial population of the RSAG (or whatever it is
called).  If things work perfectly, then six months (or
whatever) should be no problem, and I have no problem with
setting that as a goal.  On  the other hand, if they go badly
enough that the system, as conceived today, has to be
significantly revised, having to debate whether to shut off the
institutional memory at the same time is not in anyone's best


More information about the rfc-interest mailing list