Re: PROTO Process

[Andrew Sullivan <ajs at shinkuro.com>]

On Thu, May 14, 2009 at 10:15:40AM -0700, Fred Baker wrote:
>
> To be honest, getting solid reviews in the WGs can be a pain, and I  
> think that's what you're saying is needed.

Right, it sure _sounds_ like that's the problem.  I guess I'm
sceptical that this problem can really be resolved by more process
rules.

There are three possibilities here.  

The first is that the documents really are getting to the IESG in need
of additional work -- sometimes because the WG didn't tackle document
clarity issues (Fred suggested one such case yesterday), sometimes
because of inter-WG issues (like Peter mentioned), or sometimes
because WG review was just not good enough (as Fred suggests above).
In this case, I have doubts that additional process rules will help.
For in this case, the problem is less the turn-around time, I think,
and more the substantive disagreements that need to be addressed.
It's a bad thing that they come so late, and I don't think new IESG
processing rules will help this core issue of the disagreements
turning up when the document comes out of the WG.

The second is that some IESG members decide to use the position to
insist upon positions contrary to WG consensus (Fred mentioned being
aware of this case yesterday).  In this case, I also doubt process
rules will help.  The rules would of course govern the way that the
dispute would play out, but it wouldn't actually change the result:
either someone would back down, or else the document would go into
limbo forever.  We can't fix this problem with a process rule, in
fact, because sometimes we want IESG members to be able to use their
judgement and to be able to insist on changes in exactly this way.
Unfortunately, that means that individuals with a very strongly-held
belief and who are not amenable to compromise can hold up the works.
This is a problem of human judgement, and I have yet to see a system
where such positions of trust cannot be abused.  If this is the
problem we want to solve, then I think we need to give direction to
the NomCom to the effect that positive evidence of a willingness to
compromise is a requirement of serving on the IESG.  (Yes, I can think
of about a million objections to that, too.)

The third is that almost all of the delay is due to the process,
because people aren't responding fast enough.  Given the number of
documents that actually do get changed, however, it seems possible
that any delays that are just from slow response times are themselves
attributable to the number of documents that need to be changed: a
single person can only process so much, and the volume might simply be
too great if 80% of them really do need work.

All of that sends me back to the problem of getting documents from WGs
that are almost always ready for publication, so that IETF- and
IESG-period changes become the exception rather than the rule.  I
wonder whether there might in fact be a procedural change we can make.

Would better scheduling of document flow from WGs be helpful?  If we
had clear, author/editor-generated dates at which things are projected
to be ready, then we could announce IETF-wide "Document _n_ will be
available on $date, and it deals with {$topic1..$topic_t_}."  Would
such announcements do anything more for review than the current RFC
Editor postings on the arrival of new I-Ds?

Note that none of the above is to suggest automatically that things
like two-week guidelines are obviously bad.  But if 80% of documents
coming from WGs really do need to be fixed on their way through to
publication, no amount of tinkering will cure that bad illness.  We
need to address the basic problem.

A

-- 
Andrew Sullivan
ajs at shinkuro.com
Shinkuro, Inc.