Re: AD review delays

[Warren Kumari <warren@xxxxxxxxxx>]

On Wed, Jun 21, 2023 at 11:40 AM, tom petch <daedulus@xxxxxxxxxxxxx> wrote:

On 21/06/2023 14:26, Warren Kumari wrote:

On Wed, Jun 21, 2023 at 4:33 AM, tom petch <daedulus@btconnect.com> wrote:

On 19/06/2023 17:47, Daniel Migault wrote:

I tend to think it is not so much the number of days or months as long as there is a common understanding of that delay. If one estimates X, the other part estimates Y as long as X differs from Y it is likely to generate frustration - chances for X to match Y are likely slim. On the other hand being simply informed it is going to take Z, even Z being greater than X, this is likely to be fine.
While the context of this thread is the delay from AD, this can be easily generalized in my opinion to most of the IETF process.

Generalising to everything a customer expects, a dictum of marketing is that a customer is dissatisfied when reality does not match their expectations so change their expectations. I do not know what drove the start of the thread but dissatisfied customers could be one such in which case the aim of the thread could be to reset expectations.

My expectations are based on several decades of involvement in the process with its ups and downs. The problem used to be the length of time from IESG approval to the publication of the RFC and that has been fixed - I do not know how but it has AFAICT. Perhaps that has now put a spotlight on another part of the process.

For me, though, it is the delays in the WG post adoption of an I-D that are the greatest disappointment, and that could be a reflection on the time an AD has to affect that.

Erm, could you elaborate on the above? Are you meaning: 1: the time after a WG formally adopts a document until the draft-ietf-wg-foo-bar-00 comes out?
2: the time from when a WG adopts a document until it enters WGLC? 3: the time from when a WG adopts a document until it leaves WGLC? 4: the time from entering WGLC until consensus is declared? 5: the time from consensus is declared until it is sent to the IESG? 6: something else?
7: all of the above?

Many of these times are (largely) outside the ADs control (other than choosing more active chairs, cajoling the WG, stomping their feet, contributing text, etc).

Precisely, the times when the WG chairs can make the most difference, which is a bit of a tangent to the original question about ADs (perhaps I should have tweaked the subject line). Mostly I am satisfied with the time spent in AD Evaluation. The time that seems excessive is the time before that, from the Chair declaring that there is consensus to adopt an I-D as a WG I-D to the time that an AD Evaluation is requested, which I think is 'I-D Exists' in the Datatracker plus in some cases a portion of 'Expired' and is 2, 3, 4, and some part of 5 in your list.

Okey dokey, thanks for the clarification.

Yup, this can be one of the areas where IETF can be slow (and can also be very frustrating to participants, authors, chairs, ADs, and everyone else :-)). 

Obviously there are many causes of this, but two of the main ones that stick out for me are:

1: it's often unclear to everyone who exactly is responsible for shuffling a document along at a particular point ("whose court the ball is in").  I've often seen cases where the authors are becoming frustrated because they are waiting on the chairs to do $something, and the chairs are becoming frustrated because they are waiting on the authors to do $something. Everyone is waiting on everyone else. 

Once a document enters the IESG ("Publication Requested"), a new field "Action Holder" shows up in the Datatracker. This makes it clear who exactly is holding the ball. For example, for https://datatracker.ietf.org/doc/draft-ietf-opsec-probe-attribution/ , it's me. Once I've done whatever I need to do[0] I can set it to the authors (or chairs, or shepherd, or…) so that they know that they have the next action item. There is a feature request to add this to WG documents as well, but it's a bit trickier as we have to figure out who all can adjust the status. 

Different working groups also have different styles / cultures. In many cases, it's  authors that are responsible for driving the document progression, keeping discussions going, figuring out (uncontentious) consensus, etc. In other styles of WGs the chairs run much more of the document progression and chasing after authors. Not knowing, or worse yet, misunderstanding the expectations can cause frustration. If it'd unclear who exactly is supposed to be doing the next step, it's often best just to ask — this saves stewing and frustration. I realize that this might sound condescending / patronizing - it's not intended to; I've often been stuck in this exact situation (as an author), and am reluctant to ask because I'm assuming that everyone else knows what's going on and I'm the clueless one….

2: in many cases, chairs are reluctant to initiate calls for consensus on difficult topics and / or starting working group last calls. Chairs are humans too, and if they know that a particular discussion is going to be very shouty it's only natural that they put off / procrastinate getting into drama. Chairs are also often busy, and so it's entirely possible that a chair just missed an email / assumed that their co-chair will respond / do the next step. 

I am hazy about the states that an I-D goes through after consensus is declared for WGLC by the Chair and the start of AD Evaluation.

Yes, this could be better documented. Largely it's:

WG: 

In WGLC → WG Consensus: Waiting for Write-Up → Publication Requested (this is where it transfers from the chairs to the IESG).

IESG: 

Publication Requested → AD Evaluation →  Last Call Requested → In Last Call → Waiting for Writeup / Waiting for AD Go-Ahead → IESG Evaluation → Approved-announcement to be sent → Approved-announcement sent → RFC Ed Queue

(I've left out some of the lesser used / exception cases, like "AD is watching" or "Do not publish"). Most of these can have additional tags / substates, like: "Revised I-D Needed" or "AD Followup"). There are also additional extra steps, like "Ballot issued", etc. 

W

Tom Petch

W

Tom Petch

Yours,
Daniel

On Mon, Jun 19, 2023 at 12:01 PM Salz, Rich <rsalz= 40akamai.com@dmarc. ietf.org> wrote:

I do not know where you get the month from. I was thinking of teh

freeze on I-D submission as being the start of the run up and perhaps a fortnight after to get over jet lag, complete the commitments made during the meeting, catch up on the day job and so on.

My misreading of your comment. Thanks for the explanation.