Gendispatch @ IETF106, Singapore Monday Nov-18-2019 1810-1910 Chairs: Francesca Palombini, Pete Resnick Session material: https://datatracker.ietf.org/meeting/106/session/gendispatch Recordings: https://www.youtube.com/watch?v=WOGzajgewwI Jabber: Murray Kucherawy Minute takers: Ben Campbell • Intro/Administrivia - chairs, 5 minutes • draft-halpern-gendispatch-consensusinformational - Joel Halpern, 15 minutes Discussion: Barry Leiba: Supports document. This is general policy now, but casting in stone is better. Phillip Hallam-Baker (PHB): Supports this. Described a time where a draft was published where he was not allowed to give input. Jon Peterson: Supports. Should progress this document. Eric Rescorla (Ekr): This should be AD sponsored. Ted Hardie: Broadly in favor. May need to clarify that consensus is needed to publish, which might not represent consensus on the content. AD Sponsorship is appropriate. John Klensin: Principle is correct, but this is wrong way to do it. Better to talk about streams and what is necessary for each stream. Stating in terms of what IETF MUST or MUST NOT do seems wrong. Pete Resnick: Reads that comment to mean after AD sponsorship, he has suggested alternative text. John: Alternate approach. Chairs: Pushing to AD Sponsorship. Does the AD need further guidance? Alissa: It would be useful to get more feedback on alternatives suggested. Would like to know if there are other people/constituancies that need to be involved before IETF last call (other stream managers), I can do that. • draft-moonesamy-recall-rev (outcome from the eligibility-discuss virtual meeting) - Pete Resnick, 15 minutes (re-ordered) Jonathan: Does I* internal recall proposal augment or replace replace the existing process? Ekr: Intended to augment. The original process could still run. Mark Nottingham: What does it mean that the current process is not working, since it hasn't been used for a long time? Pete (et al): That is the problem. It's not used because it's too hard to use/can't be used. Ekr: There were situations where people were dissatisfied with ADs, but chose to wait for term to end rather than recall. Mark: A state of people being unhappy is not surprising. Pete: Not clear if there was no desire for recall and process worked, or that there was a desire and the process failed. Highlights from virtual meeting: - draft-rescorla could cause bodies to become self-selecting. - Equity generally considered an issue worth addressing. - Some desire to solve the broader efficacy issue, but good argument to start with smaller scope. - DoS on IETF raised as an issue. Not clear it's a showstopper. Ekr: Not clear that 2nd bullet was true. Pete: It seemed generally true. Patrick McManus: Not a lot of agreement. The fact that recalls are not used may be due to the cultural impact. Leslie: It might be good to categorize the issues that caused concern and how were they resolved. A change that allows i* to be involved in starting a recall is important. Sometimes they see issues not visible to community. Pete: Patrick's point was important; recalls have negative impacts. All these issues could be separate from what the recall setup is. It may be okay to leave the rest of status quo as is once we deal with the equity issues. Did not hear anyone object to allowing remote participants, it just may not make it easier to use. PHB: Once worked with lawyers to see how to use recall. Conclusion was that it is unworkable because it can take longer than tenure. May blow up organization. May want people to attend meetings to sort out issues directly. This is the last thing he would want to involve virtual participation in. Pete: This was brought up in the DoS discussion. Alissa (as individual): Recall can cause or aggravate shame. The work to try to improve community communication cultural is important. Work to try to more gently ease people out would better align with goal of creating more compassionate environment. Pete: More concerned about people who were doing active harm, rather than just falling down on the job. Michael Toomim (via Jabber) Very greatful for ability to participate virtually. Ted: Returning conversation to how to dispatch this. AD sponsorship not appropriate. BoF? Ted: Virtual participation support might be better started with positive things like nomcom participation. A BoF should be scoped more generally towards support of remote participation. John Klensin: Fear of DoS attack is that we used to allow any single person to start a recall. Experience was that it just didn't happen. Agrees about trying to support people better. Anyone starting a recall should start with supportive efforts first. When draft was first published, discussion was about a BoF, then virtual BoF, then this virtual meeting. If we have a BoF in Vancouver, the effect is to draw thingse Mike ???: Large numbers of bad actors are easy to come by. Cost of remote participation is zero dollars. No impediment to DoSing the process. Barry: Agrees with Ted about addressing broader issues to support remote participants. But it's worth addressing this piece as a way to get there. We have enough to move to WG rather than BoF, way beyond AD sponsoring. Scoping should be done in wg charter. Ekr: A charter won't change the level of support. Stephen Farrel: Agrees to broader scope. Working on making things better for remote participants is worth doing. Not sure recall is the starting point. Pete: Asked in meeting if this would cause active damage. Some thought it would. "We don't want to do this" is not sufficent to establish consensus not to do it. They need to argue that this would do damage. Ekr: There's no consensus on that version of consensus. Pete: Process doesn't say that there can be no objections. Ted: Issues of equity are not things that we can point to the same we as technical issues. The harm here is that if you start the discussion of equity here it starts with a negative bias. We are talking about who gets the right to accuse, rather than how do we empower people to participate. Results will be narrow and unhelpful to larger question and may harm that larger question depending on the result. This scope is wrong, you need a BoF to get a better scope. Pete: That is what I've been hearing. Cullen: Cases in WebRTC where remote participants tried to manipulate consensus process. Would like a better definition of membership for general purposes. Not seeing concrete proposals. Term "equity" to refer to disenfranchisement is grating. Pete: Does that mean we need a BoF or something to do that? Cullen: Would like to see concrete proposals for better definition of membership. Focus on broader thing. Mike ???: Talking about accessibility of organization. Noble goal, but is a different conversation. Starting that conversation experimentally is important. Starting it in a disciplinary process is risky. Patrick McManus: Addressing issue of active harm. One is cultural issues. Also tactical side--disagreement about what qualifies for recall. Recall proposals should probably stop hear, address larger question. Pete: Room seems to be leaning towards taking on the broader issue of remote participation and enfranchisement and BoF that, and possibly virtually BoF that. No disagreement. John Klensins: States disagreement. Real issue with remote participants being able to protect themselves from harmful behavior directed at remote participants. Significant issue independently of how often it comes up. • draft-thomson-gendispatch-no-expiry - Martin Thomson, 15 minutes Proposal to update BCP 9 to remove I-D expiration. How do we make small changes like this? What is consticuancy? How do we decide consensus? Leslie: Small change that leads to larger discussion about how we use this small detail. Answer to broader question is you need a public discussion and general agreement about the size of change. Ekr: Should form this working group. Did that with time machine. Group should recommend AD sponsorship. Ted: Wants to borrow time machine. Other streams use I-D series. Have different opinions about what expiration means. Need to discuss whether this applies to all streams. If not, then decision can be made for a particular stream by the stream manager. Need to be able to identify which drafts belong to which stream. Pete: Chairs did a quick review--more needs to be fixed than is immediately apparent. Martin: State that the expiration goes away, don't "patch" the text in 2026, etc. Stephen Wenger: Even expired drafts can be used for prior art. David Schinozi: Choise of fonts is bad and you should feel bad :-). Supports change because it needs to be changed, and also as a prototype for how to fix certain small issues. Pete: Focus on dispatch question. Running out of time. Alissa: From previous discussion: Other ways to indicate liveness. Expiration can be a forcing function for updates. Stephen Farrel: If we make this change, do it for all streams. PHB: Supports change. If you want to supercede a draft, put in a new draft to replace it. John Klensin: Disagrees. Disagreed with making expired drafts available. Expiration is a primary property of I-Ds. Cullen: Don't spend a huge amount of time on this. I-Ds are effectively used as archival documents. Not important to spend time on this--most epic troll ever :-) Mark Nottingham: Supports. But calling is small and simple may have opened a door to hell. Drafts can change streams. Some have no streams. need to tread carefully. Pete: Just handing to AD is not the right thing. Alissa: Agrees. Did any comments change how Martin would proceed? Martin: Applying to all streams. Pete: May make it a IAB issue. Martin: I'd be happy to talk to IAB. Richard: Can have discussion with stream leads about process changes. Martin: This is IETF consensus document, which is why I brought it here. My opinion is that it cannot go to IAB stream. Will talk to stream managers, then come back. Pete: So you will do the foot work and then maybe go for BoF? Martin: Not worth a BoF. I think this is AD sponsorship. Need to understand/figure out how to make small changes. Jon Peterson: I-Ds drafts are not on a stream. RFCs have streams. We don't know how to target I-Ds. Ekr: 2026 says I-Ds are IETF documents first. John Mattsson: Boilerplate and expiration are separate issues. Which one is it? Martin: Wants to address the process of expiration. IAB owns boilerplate. Conclusion: Martin to talk to stream managers and RSE. If any come back objecting it might stop there. Otherwise, may need a little more discussion, but then AD sponsor. • AOB/Open mic - 10 minutes PHB: Need to change Note Well to add antitrust considerations. Can explain circumstances in private. Pete: Need a draft posted to list.