ietf
[Top] [All Lists]

Re: Content-free Last Call comments

2013-06-11 14:46:23
It's interesting to see that people are interpreting me to mean I want more text. I don't. I want less. Save your breath. There is no reason to send one line of support, and it only encourages the view that we're voting. Details below.

Specifically on Stephen's message:

On 6/10/13 7:36 PM, Stephen Farrell wrote:
I think you err when you say this:

A statement such as the above is almost entirely useless to me as an
IESG member trying to determine consensus. It is content-free.
In fact, you do know Russ. If you did not, then the above would be
far closer to correct. But in reality you do know a lot more than
you claim below. With overwhelming probability, your choice #2
applies and I'm very surprised you don't also think that. If it
made a significant difference and I wasn't sure I'd ask Russ to
clarify. For me, I read Russ' mail and I concluded he meant #2.
And I'm confident in that conclusion.

So I'm really bemused when you say that you don't know how to
interpret Russ' mail.

Let's separate out a few issues. There's what the message means itself, there's its usefulness and how much weight it should be given, and then there's what's problematic about it. I think you've conflated a few things.

As you know, on the earlier message I called out on the IESG list, I guessed wrong that the sender was claiming choice #3 (that is, I thought, incorrectly, that the sender disagreed with a recent argument against publication, but simply didn't explain why). In fact in that case, not only wasn't it choice #4, it was also neither choice #1 (an implementer claiming implementation) nor choice #2 (an expert on the topic), my other two guesses, but rather a generally smart IETFer who thought the work seemed like a good idea. So I'm a little disinclined to guess on motives. In the case of Russ's message, I did guess it was choice #2, but I wasn't really sure.

But let's separate what Russ meant from the weight it should be given. The message was that a person (let's presume with some expertise in the area) had read it. So what? Didn't this document go through a WG? Weren't a good bunch of experts already reading and reviewing this document? If this was during the WG discussion and the chairs solicited some final check reviews, *maybe* the message could have some use. But during IETF Last Call? Even if the message was sent because it was trying to say, "I'm an expert on this stuff and it looks technically sound", is that going to change anything that the IESG should do about the document? I just think at Last Call, these statements of support are either harmless-but-useless or they are nefarious.

Declaring that mail problematic also seems quite purist to me.

I always find it amusing that when Stephen and I disagree, it's almost always because he presumes I want purity (and usually a new process), which I don't, and it seems to me that he wants no rules-of-thumb at all and that everything should be on a case-by-case basis. (I guess that means that I'm into purity, where Stephen has no principles. ;-) ) I think general principles are a fine thing, and we should try to stick to them, but I've got no interest in purity.

The single piece of mail isn't problematic at all. As I said, probably harmless but useless. But the overall impression it gives *is* problematic. It encourages the view that we are voting, that a simple statement of support is important to our process. (What really set my hair on end was the "I support publication" bit. That always sounds like a vote for a new RFC. From Russ, I have a pretty good idea that he didn't mean it that way, but it's a poor formulation and not what we should be encouraging.) It also indicates that during Last Call, we want to hear that a document is getting reviewed. I would like to presume that documents get serious reviews in the WG, and that Last Call time is for people who haven't been participating in the WG to do a final check, letting us know if there's something *wrong*. If the document wasn't getting good review in the WG and needs statements of support at Last Call time, something has seriously failed earlier in the process. The chair or responsible AD should have been saying early on, "Dear Expert, can you please have a read over this document and see if it's sane." Last Call is the wrong time for that to happen. This just encourages more of the tail-heavy process that we discussed earlier.

At the same time I agree we do not want a procession of +1 mails.

Right. And having senior members of the community doing so will encourage such behavior in the future.

But then we won't get that for this draft. And if we did get that
for any draft, some IETF participants (probably incl. you and I)
would notice that and query it or object.

I'd rather have all of us crusty folks model good behavior rather than having to complain about bad behavior later, especially when we'll get responses like, "But the chair did it." We've already seen a couple of comments to my message from non-crusty-folks indicating that any +1 is just dandy. Blech.

Lastly, I think evaluating IETF LC messages only in terms
of how they help the IESG evaluate consensus or not is wrong.
Those messages are for and from the IETF community. So if
e.g. some renowned NFS person who's hardly known to the IESG
were to have sent an equivalent message, that might have been
quite good input that Martin or Spencer could have translated for
you. And that's just fine. And it would be fine if Russ' mail
had been directed not at the IESG but at some other part of the
community.

So this is the most interesting argument. But I would say that requiring this kind of interpretation/translation is problematic as a general rule when the messages are going to the IETF list in response to a Last Call. If a renowned NFS person is doing a review of an NFS document, I think that's great. And I think that's helpful to Martin or Spencer. But if all they're going to say is, "I don't have any concerns", it doesn't seem to me to be that useful a piece of info on the IETF list at Last Call time. Again, at best it encourages the idea that just because Über NFS person endorses the document, that weighs in its favor. The important thing to hear at Last Call time are objections, not endorsements. As a note to the IESG saying, "Just in case you didn't think this got review from the right people before, I, Über NFS person, hereby say that I reviewed it and have no problem with it", that has some content, but it does indicate that we are waiting until too late in the process to be thinking about such things.

So bottom line, I think you're wrong, Russ' mail was not content-free.

Yeah, you haven't convinced me.

PS: Yes, this is a not-very-good academic accusing an employee
of an industrial behemoth of excess purity. Go figure:-)

I'll note that I hide in a small R&D corner of my industrial behemoth, mostly teaching classes to other engineers. You, on the other hand, probably still write code. :-)

pr

--
Pete Resnick<http://www.qualcomm.com/~presnick/>
Qualcomm Technologies, Inc. - +1 (858)651-4478