My thoughts about the problems of the IETF

[John C Klensin]

--On Thursday, 17 April, 2003 15:46 -0400 Scott Bradner 
<sob at harvard.edu> wrote:

>> But what happens when an AD or IESG member tells a document
>> editor to change x, y & z before a draft can pass IESG
>> review?  Is there any way to appeal this?
>
> to be clear I'd say yes & no  :-)
>
> no - its not written into 2026
> yes - if someone appeals such a thing it would be really bad
> 	if the IESG &/or IAB were to rule it out of order rather than
> 	address the issue
>
> but also - yes
> 	if the AD says that and the WG specifically has consensus
> 	a different way that it would be a process violation to
> 	override the WG consensus
>
> this might sound counter to what I posted earlier today but if
> a AD/IESG pushback to a WG for a security or danger-to-the-net
> issue can not get WG consus to fix then we are in a pile of
> problem

But Scott, it seems to me that this has happened several times 
and, modulo some philosophy and issues about the edge cases, it 
is not only not a process violation, but precisely what we want 
to have happen.  The cases I can remember are all ones in which 
the AD says "this doesn't deal adequately with { security | 
scaling | impact on XYY }" and the WG says "we don't care about 
any of them, just publish the thing".  That creates a 
(usually-desirable, IMO) impasse in which the AD starts getting 
tempted to immediately replace a WG Chair or even to shut the WG 
down for insensitivity to the rest of the Internet.  Usually, 
the thing that breaks the impasse is the WG's understanding (or 
that of WG editors and chairs) that their draft is going nowhere 
until the AD's demands are met".

I would suggest that:

* The AD should have had informal discussions with the rest of 
the IESG (or at least a significant fraction of the ADs) as a 
sanity check before things with the WG reach the "facedown" 
stage.   My perception is that usually that happens.

* If this type of confrontation occurs only when the WG thinks 
it has a complete document ready for IESG processing, the "early 
detection and notification" principle has been badly violated. 
If the problem is understood early in the process and agreement 
cannot be reached, chairs and editors should be changed, or the 
WG shut down, before it goes to the trouble of producing a 
claimed-final draft.

* The AD interactions with the WG if things do reach this point 
should be very public.  And AD who is negotiating secretly with 
editors or chairs about issues adequate to justify holding up a 
document for long periods of time, or until specific changes 
with which the WG might disagree are made, is being 
imperialistic and/or stupid and should be seen as causing a 
process violation.

* And either the WG or the AD should be able to seek a rapid and 
public opinion from the IESG as to whether the AD is behaving 
reasonably.  The only mechanism we have to get such an opinion 
is an appeal.  It isn't completely satisfactory for a number of 
reasons, including some precedent for a "we haven't decided 
anything yet, so there is nothing to appeal" response to an 
appeal.  But this is important enough that, if we need a better 
mechanism, to tuning to the appeals process, we had best be 
working on one.

     john