Document Blocking (Was: I-D ACTION:draft-ietf-problem-process-00.txt)

[Pete Resnick]

On 5/19/03 at 9:56 AM -0400, Thomas Narten wrote:

>I have to say something here, as this issue keeps being brought up. 
>Reading this list, one could easily get the idea that the ADs love 
>to block documents, usually don't have a good reason for doing so 
>(and thus fall back to procedural tricks), and that the ADs actually 
>are hostile towards each others with tit-for-tat happening if one AD 
>blocks a document belonging to someone else.
>
>That is not the IESG I know or that I would want to be a part of.

On the first two points (that ADs "love to block documents" and that 
they "usually don't have a good reason"), I think that you are simply 
caricaturing what's been said. I think everyone who has commented on 
this believes that ADs put discusses on documents almost always with 
all good intentions, and I think they believe that they usually do it 
with good reasons. For the most part, people are complaining about 
the (relatively few) instances where good reasons haven't been given 
or haven't been communicated well enough to the WG.

However, on the last point, that ADs are hostile towards each other, 
I think you will see in the archive of this list several examples of 
former ADs saying exactly that. We have had people say that people 
will put trivial discusses on documents because putting a substantive 
discuss would get retaliation, that one or two ADs must have the 
power to stop a document without the other ADs signing on, etc. This 
isn't coming from folks who've never been on the IESG; it's coming 
from your former colleagues. Perhaps the current IESG doesn't have 
these problems; that would be really good.

But all that aside, what I think you are seeing overall is:

- Past history where it *appears* as if documents have been 
blackholed by a single AD during IESG review (whether or not it has 
actually happened);

- Not enough experience with the document tracker to dissuade that feeling;

- A procedure that the IESG is still using that encourages that feeling

The first point I think I don't need to explain. That's the 
perception, whether true or not.

The second point cannot be addressed by saying "let's see if the 
document tracker solves the problem." The distrust of the IESG 
already exists. The document tracker will, at times, indicate a 
discuss by one AD, yes or no objection by everyone else, and an 
explanation of the discuss that will not be satisfactory to a reader.

The third point is the key one: The current procedure of the IESG 
*reinforces the perception* that one AD can block a document without 
the buy-in of the rest of the IESG, because on paper the rest of the 
IESG never buys in. And with the current perception that an AD might 
(however rarely) vote discuss for spurious reasons, the current 
policy makes for a bad feeling. For example, below you say:

>One of the common "votes" that happens during the telechats is "no 
>further objection" which formally means "no objection" but in 
>practice means "I agree with others, but don't have more to add and 
>don't need to be in the process loop to see that the document gets 
>fixed".
>
>Forcing the IESG to have "consensus" or "unanimity" (those words 
>have been used on this thread) on all discusses procedurally seems 
>like a high overhead approach for dealing with a particular problem.

So maybe the solution here is to formalize "no further objection". 
That is, if there is a discuss that you agree with, you can vote 
"concur with discuss". That's the equivalent of voting "no objection" 
but noting that you actually are on board with the discuss vote. If 
the IESG comes out with one discuss, one yes, and the rest "concur 
with discuss", I'd consider that consensus to discuss. However, if 
you get one discuss, one yes, and the rest "no objection", that's not 
consensus to discuss. That seems to me no more overhead, but now 
you've got the ability to make an IESG rule that says, "If there's 
only one discuss vote and no concur votes, the document passes." And 
from the outside, you'd be able to see what has happened. With the 
current process, there's no way to tell whether there's real 
consensus or it's one of those one-in-a-million "one AD is 
unreasonably holding things up" instances.

>I'd actually like to better understand how much of a REAL problem it 
>is that individual ADs are impropropely blocking documents with the 
>"single AD veto". There are many comments that imply it happens "all 
>the time"

I have heard no comments to the effect that "it happens all the 
time". Again, I think this is a caricature on your part.

>and that "everyone has examples".

I can think of a few, though only one on which I was smack in the 
middle of the fray.

>But I wonder sometimes if we are all thinking of the same document 
>from 3 years ago. We can't
>fix what happened 3 years ago, but we can fix things that are broken _today_.

Even assuming it is extremely rare, leaving a procedure in place that 
allows it to happen in the future is a problem that needs fixing. 
Perhaps the current IESG is a wonderful set of folks who would never 
let such a thing happen. That doesn't mean we should leave the 
procedure as-is.

>>The ADs who think that there is a serious problem with a document 
>>should convince the rest of the IESG that the document is a bad 
>>idea.
>
>This in effect is happening AFAIK. The IESG does support the 
>security ADs when this happens.

It needs to be done publicly.

>>The problem with the current process (as I understand it) is that 
>>it allows documents to be blocked by one or two IESG members 
>>without the consensus of the group.
>
>In theory, yes. But in practice? It would be instructive for the 
>IESG to ask itself where this has happened, and I will bring up the 
>topic. My sense is very rarely, but I could be wrong.
>
>But at least some on the community assume this happens often enough 
>that we need more procedure to prevent it from happening. Concrete 
>examples would be useful to provide context.

Even "very rarely" is enough to change the procedure.

>Note: This note might come across as sounding like I don't think 
>there are any problems that need fixing. There are. But I am 
>unconvinced that the "one AD veto" is one of the real problems. 
>Given that at least some in the community appear to think it is a 
>problem, I'd really appreciate some concrete examples. My suspicion 
>is that if we look at specifics, the reality may be quite a bit 
>different than the appearance.

I've got to tell you that I'm not terribly sanguine about the results 
of you asking for "concrete examples" or your "bringing up explicitly 
within the IESG" the topic of whether other members of the IESG think 
there are examples of discuss votes that are not supported by the 
rest of the ADs. We've already been told that past ADs have held 
their tongues so that there wouldn't be retaliation within the IESG, 
and I for one don't feel like pointing fingers in public about the 
situations I've encountered, if only for lack of good taste and for 
the hurt feelings I suspect it might cause. I can't imagine that 
you're going to get a bunch of very open responses to your query.

Look, we've got a procedure currently that can't publicly distinguish 
between single discuss votes which are supported by the rest of the 
IESG and those that aren't. I think the perceptual and trust problems 
it causes are obvious. I think I have outlined a very low-overhead 
solution to the problem (whether or not we choose that particular 
solution). What's the issue here?

pr
-- 
Pete Resnick <mailto:presnick at qualcomm.com>
QUALCOMM Incorporated - Direct phone: (858)651-4478, Fax: (858)651-1102