--On Thursday, 23 December, 2004 10:22 +0100 Harald Tveit
Alvestrand <harald@xxxxxxxxxxxxx> wrote:
>>> John,
>>>
>>>...
>>> I like all of those properties, and it should be a small
>>> twist of language (starting from the text in the draft, not
>>> the most recent suggestion) to make it come out that way.
>>> But I'm not sure I'm reading your words correctly, so better
>>> double-check....
>>
>> A fascinating question, actually. I think you are reading my
>> words correctly and that, by happy coincidence, the words that
>> are now in the draft are fairly easily adapted.
>>
>> But the principles are more important than the words, and I
>> think this is a profound change in principles. It is, I
>> think, a significant change to say "if you expect the IAOC
>> model to succeed,
>>
>> * the IETF has got to keep its hands off the day-to-day
>> decisions, even when they seem wrong
>>
>> * the IESG and IAB need to be prohibited structurally
>> from micromanaging, or managing at all, beyond the
>> degree that the IAOC wants to permit. They supply
>> input, they make requests, but decisions rest on the
>> IAOC side of the wall and stay there, with the only
>> _real_ recourse being to fire the IAOC
>>
>> and then to figure out a way to implement those principles.
>
> Actually, I think we have agreement on those principles, and
> have mostly been struggling with how to express them.
>
> Appeals procedures (under whatever name) are procedures that
> should exist so that if someone thinks something's gone really
> wrong, it's possible to get it noticed and talked about - and,
> if it's necessary, corrected.
>
> Trying to manage anything by routine appeals is totally broken.
Well, perhaps we are close enough that it doesn't make sense to
have further discussion without text, and text in context, but...
I think the "something's gone really wrong" and "...corrected"
part of the above _might_ miss the point I'm trying to make.
(1) If you have an appeal procedure, any appeal
procedure at all, then it is up to the judgment of the
individual members of the community whether a decision
is "really wrong". I don't need to worry about denial
of service attacks here, just about people acting in
good faith and with the best of intentions. Remember we
have had debates on the IETF list about the variety of
pastries to be available during meeting coffee breaks as
well as the theory for selecting meeting sites. I may
be too concerned about this, but I think those debates
could easily have led to appeals about meeting locations
when people concluded that the decisions being made were
against the expressed consensus desires of the
community (or against plain good sense). Such appeals
have been prevented by the assumption that the IETF
membership has no ultimate control over _individual_
secretariat or Chair meeting siting decisions and that
contracts are already signed by the time meeting sites
are announced. That has protected us from procedural
entanglements that could easily make meeting scheduling
impossible. But, with the presumed requirements on the
admin structure for openness, that particular protection
disappears.
(2) If there is an mechanism, any mechanism at all, for
the IESG and IAB to second-guess administrative entity
decisions, I think it can be guaranteed, given the sorts
of personalities we put on those bodies, that the
mechanism will sooner or later be used and that, once
used, there will be a tendency for its use to become
routine. We appoint people with strong opinions and
with a tendency to want to manage, sometimes
micromanage, anything for which they feel responsible
and to do so aggressively. If kept under control and in
perspective, those tendencies can be very useful. But I
think any person who has listened to a few IESG
telechats, any WG Chair or Editor who has tried to get a
document through the system in recent years, anyone who
has worked for the Secretariat, and many others can
easily identify incidents that they would consider out
of control or without adequate perspective.
(3) Part of the problem here is that, on the technical
standards side, part of the job of the IESG is to ensure
that the collection of standards the IETF emits are
essentially consistent, that we don't have a standard in
one area that makes a standard in another area
impossible to implement. Sometimes we don't get that
quite right and it requires some sorting-out. But it
works most of the time, and works precisely because that
is a real IESG responsibility which the IESG and the
community take seriously. On the admin side, the IAOC
has to be able to work out a plan and strategy that
works in works in a consistent way. I want to see that
plan and strategy be public, and be exposed to comments
from the community, and so on, but they have to
represent a framework that the IAOC believes is
workable. The possibility for "despite your framework,
we really need bagels flown in daily from Manhattan"
decisions, or even decisions about contracting
organizations or contract terms, overriding the
framework is a recipe for a fast trip down a slippery
slope toward complete disfunction.
We need to prevent that, and we need to prevent that with real
firewalls, not just "it should be appealed only if it is bad
enough; the real risk is 'routine' appeals" language and
intentions. For the IAB or IESG to tell the IAOC "we know
something about this, or have a perspective on this, that we
don't think you considered adequately; here is the information,
please reconsider" is fine. It provides a means of input and
it contains its own safeguard mechanism: were those requests to
appear once a week, the IAOC would presumably start blowing them
off. And then the community would need to decide whether the
issues were actually important enough to replace the IAOC
membership and, if not, whether it was time to aim the
clue-by-four in the direction of the IETF Leadership.
The need for those firewalls is also the reason I've been
reluctant to see the IAB and IETF Chairs even as members of the
IAOC (not even getting to a debate about voting rights),
weighting the IAOC more toward folks with administrative
management/ finance experience than people selected primarily
for their technical skills and, perhaps, technical project
management experience. I think we need to appoint the IAOC,
make sure it has access to adequate information about what is
going on in the IETF and what the IETF needs, and the let it
function. The community has apparently concluded that the best
way to get that information to the IAOC is to put the IAB and
IETF Chairs on it. I disagree, but that is fine. There are
other areas, perhaps including Exec Dir appointments, where it
is completely rational to require IESG and/or IAB advice and
consent before a decision is actually made, but that is
pre-decision, not reversing decisions.
For decisions once made, those options for strong communications
makes other sorts of firewalls even more important. And the
key ones of those are "no appeals procedure" and "no ability of
the IAB or IESG to reverse an IAOC decision". None. No
judgments about what is sufficiently important or sufficiently
wrong. None. If the IAOC doesn't like the behavior of the
IAD, and tries to correct it by advice and negotiation, and
fails, their only recourse is to fire the IAD. And if the IETF
Community (the IAB and IESG included) doesn't like the behavior,
or decision-quality, of the IAOC, and can't fix that through
advice and negotiation, then it needs to have a mechanism for
firing and replacing the IAOC and should use it. But no
second-guessing or reversal of individual decisions. Ever.
john
_______________________________________________
Ietf@xxxxxxxx
https://www1.ietf.org/mailman/listinfo/ietf