Problem: Resolution mechanisms for when working group consensus and IETF consensus or principles are not the same (was Re: what are the real problems)

[John C Klensin]

--On Friday, 23 May, 2003 14:33 -0700 "hardie at qualcomm.com" 
<hardie at qualcomm.com> wrote:

> I think John has made a very important point here, and want to
> pull it up to the named problem level (though I'm not sure
> I've written the right name above).  Here's the critical text
> for me:
>
>> We have both seen situations in which a relatively
>> well-organized,  but very narrowly-focused, group comes to
>> the IETF wanting either  ratification, or an umbrella
>> organization for, standards it wants to  produce.  These
>...
> Another way to put this is that what is a rational decision
> for a working group may not be the right decision for the IETF
> as a whole.  From some perspectives, it may be rational to
> resist change to deployed clients, even though they contribute
> to congestion or lack security. It may be rational to avoid
>...

Ted,

I agree with everything you say and, if my note helped you 
formulate it, I'm pleased.   But I am equally concerned about 
the startup- and mid-process versions of the problem. I see that 
problem as one of the community claiming the "right" to have 
particular things happen if some vague, but important-sounding 
condition is claimed to be met.  You may have missed that; it 
may be a separate problem.   Let me take the part you have 
addressed as a given (i.e., if people want to argue about it, 
I'm happy to have them argue with you :-)).  But I want to give 
an extended, if made-up (or at least deliberately disguised) 
example, of that other part.  I think it represents one of our 
key problems, and one that generates many opportunities for 
heaping abuse on IESG for doing things we _want_ them to do.

--------------

Suppose that we have a large handful of people who have gotten 
very interested in frobitzes.  How they got together isn't 
important -- perhaps they are part of the Frobitz Consortium, 
perhaps they had a discussion in a bar bof and got excited about 
it.  To make the example a bit more interesting, suppose they 
write, and post, draft-bozo-frobitz-requirements-00 and 
draft-fudd-frobitz-implementation-scenarios-00 (we routinely 
kill bad ideas that don't have supporting documents).

Now they ask for a BOF.  "Ask" may not be quite the right word: 
they go to a relevant AD and _demand_ a BOF, based on their 
having interest, critical mass, and documents on the table.  If 
the AD says "this isn't interesting, it isn't relevant to the 
IETF, and no one cares about frobitzes anyway", they say "we 
care, we are part of the IETF, we think it is relevant, and we 
have critical mass".  Then they start looking for over-ripe 
fruit with which to decorate the AD.  At this point, all but the 
most hard-nosed of ADs gets tired of the abuse and need to wear 
a raincoat around the IETF, and gives them the BOF on the theory 
that it is worth finding out what they have to say (and, 
sometimes, in the hope that they will self-destruct).

At the BOF, the AD and a handful of others question whether 
anyone really cares about frobitzes.  A few members of the 
community, possibly including the overworked AD, have read the 
documents and point out that, even if people cared about 
frobitzes, the proposed solution scenarios --and every other 
solution scenario that they can imagine-- will work only on LANs 
containing fewer than 30 hosts and will pose grave security 
risks if that LAN has any connection at all to the public 
network.  The discussion rapidly deteriorates into "you are 
wrong" / "no, you are wrong" and continues until the BOF is 
nearly out of time.  Then the BOF chair asks for a hum about who 
wants to form a WG and the usual suspects, along with a handful 
of other people (a few of them known to scare off any clues 
which see them coming) all hum ...really loudly.

Now approximately the same process repeats itself.  The 
advocates of a frobitz effort point out that they have critical 
mass and that they held a BOF and got a really loud hum.  They 
claim that the scaling and security issues are really irrelevant 
to anyone who would want to use a frobitz.  They note that they 
have documents on the table (maybe at version -01 or -02 by 
then) and are anxious to move forward.  And they tell the AD 
that they are _entitled_ to a WG and that the AD is being 
unresponsive and arrogant by not having set up the WG already.

I suggest that almost everyone who has served as an AD for more 
than a few months has encountered some more or less clear 
version of this story, at least up to that point.

As you nicely put it, we have poor mechanisms to deal with these
tensions.  Most IETF participants would agree that the frobitz 
crew should be encouraged to go away and stop wasting AD and 
IESG time -- unless they are _members_ of the frobitz crew. 
The pushback process works poorly: the AD often gets tired of 
the noise and goes into a charter review process, hoping that 
the effort will disappear --or get constrained into reality-- 
during that process.  Sometimes that works.  It always takes a 
lot of time, with the would-be WG members claiming (often with 
some justification) that the AD is foot-dragging and nit-picking 
-- after all, they have critical mass, documents, and all that 
other stuff.

Often the pattern continues: The noise level may reach the point 
where the AD --and the IESG as a whole-- get worn down enough to 
just approve the WG in the hope that it will self-destruct or 
acquire wisdom...  Well, it wastes a lot more time on education, 
guidance, attempts to inject clues that are not happily 
accepted, etc.   That time has to come from somewhere, and 
other, more relevant, WGs may suffer.  If the AD gets worn down 
enough to be unwilling to close this misguided effort, and they 
produce a document, it either gets Last Called or the AD, who 
concludes that he or she would be too embarrassed to Last Call 
the thing, sits on it (and gets more abuse).  No one responds to 
the Last Call because no one reads it.  And no one reads it 
because no one outside the core of the WG cares enough about 
frobitzes to justify putting in the time... note that we can't, 
in general, even get people to read documents that are obviously 
critical to the infrastructure.

The AD, or the IESG as a whole, contemplate just saying "no", 
but the frobitz team argues that their charter was approved, 
they put in months (or years) of effort, the document conforms 
(more or less) to the charter, they, the real frobitz experts in 
the IETF, have consensus, and they are therefore entitled to 
have their document adopted as a proposed standard.  The IESG 
either gives in and approves it, or tries to kill it through the 
death of a thousand cuts.  Either way, we all lose.

I suggest that is a problem.

     john