DID WG Telco — Minutes

Date: 2021-12-14

See also the Agenda and the IRC Log

Attendees

Present: Daniel Burnett, Brent Zundel, Ryan Grant, Markus Sabadello, Charles Lehner, Justin Richer, Michael Prorock, Manu Sporny, Juan Caballero, Orie Steele, Joe Andrieu, Dmitri Zagidulin, Ted Thibodeau Jr.

Regrets:

Guests:

Chair: Brent Zundel

Scribe(s): Markus Sabadello, Manu Sporny, Charles Lehner

Content:

agendabot: clear agenda.

Brent Zundel: We begin with agenda review..
… Any re-introductions?.
… Agenda today is DID method registration, then DID registries PRs, then if time remains we discuss DID rubric..
… Any comments on the agenda? Any additions?.
… This is our last meeting of the year. We will re-convene in January..
… We hope this will provide a break for everybody..

1. Baseline registration requirement for DID methods (issue did-spec-registries#382)

See github issue did-spec-registries#382.

Manu Sporny: Chairs and editors talked about this, we think we can propose something that addresses what we discussed..
… We are effectively wrapping up in the group, doing anything drastic might be not great at the end of the charter..
… There are a couple of requirements from the group..
… Some have said we’re uncomfortable with some of the DID methods that are getting in..
… Others have said making a higher level of value judgement might be a large amount of work..
… Some people are saying we should make certain value judgments and we shouldn’t allow “bad” methods..
… And then there are variations, e.g. two lists, etc..
… One proposal without strong objections seemed to be: We have a registration process with a checklist. Editors check this on a high level, then allow registrations on a first-come first-serve basis..
… The proposal is to continue to do that in this WG, and then work on something new in the next WG..
… Then we will have a more thorough checklist of things to check..
… We would expand the list into more items..
… Question then is what do we do with the currently registered methods..
… In the new WG, we we tell all DID methods that we are planning to increase the requirements for registrations, to improve the quality. We will give method authors time (e.g. 6 months) to meet the new requirements. Then there will be a warning phase, and then methods will get de-listed from the registry..
… For now, we will continue to do a light check..

Daniel Burnett: I believe you covered everything. This is an initial proposal that will be discussed in a new WG..
… This specific proposal is not a mandatory decision, the group will take this up again..

Ivan Herman: I have no problem with what was said, I just have the practical fear that memories will fade over Christmas. Manu it would be great to do a PR on the new charter..

Manu Sporny: yes, Ivan, I took an action to do that last week and failed to do so – it’s still on my “to do” list..

Ivan Herman: We haven’t touched the charter for a long time, but I think this should be in the new charter text..

Michael Prorock: Alongside this, we would open an issue on the charter that we will re-visit the registries..
… Also, this would give us time to review registries from a W3C perspective..

Ivan Herman: The current charter proposal already refers to the new process, including reference to the registry mechanism..
… We already say it will be a registry per the new process..

Markus Sabadello: I also think this is good, so I wouldn’t add anything. I agree that we shouldn’t make drastic changes or increase burden or workload on anyone, light checking that we’ve been doing makes sense to continue that..
… The question is still with did:id registration, what do we do there… if there is a registration that doesn’t meet normative requirements, would we allow it for now or not?.

Manu Sporny: Taking the case of did:id, I think you asked for specific changes, the assumption is we can then merge it if those changes are addressed..
… More generally speaking, we haven’t asked many other DID methods to make such changes, do we want to be so stringent?.
… Our light checking so far has not been about checking every single normative statement. We believe the vast majority of DID methods don’t meet that..
… In a future process, we will require all normative statements to be met..
… If we pull in did:id now, we will then revisit it in the future process..
… In that case, the process would address the concern in time, just not immediately..

Joe Andrieu: manu were you saying right now the editors shouldn’t apply the normative rules? They did that in did:earth.

Manu Sporny: We’re trying to apply the same rules to everyone when they register..
… I think there was another issues did:earth about it being a meta DID method..

Joe Andrieu: It was a better method than many that are currently listed..

Manu Sporny: In that case we should have registered it..
… The general question you ask, what are the criteria now, and what should they be in the future..
… Today there is a checklist when you open a PR..
… Many registrations are provisional. Many security+privacy sections are single lines..
… We’re saying now we probably should have had a higher bar, and in the future we want a higher bar but also give people plenty of fair notice, and give them several months to improve their method spec..

Markus Sabadello: I agree, in the past we have definitely registered DID Methods that don’t meet requirements. We have DID Methods that have weaker specifications than some other methods that we may have rejected at some point. There is inconsistencies both ways..
… We have allowed DID Methods that we allowed, we rejected DID Methods that are stronger than some of the other DID Methods. We do have a statement in DID Spec Registries – new registrations must meet normative requirements, but as Manu mentioned, we haven’t applied this consistently. In general, +1 to new process with higher bar in the future. The thing is that if we don’t enforce these normative rules, what’s the basis on which the.

Manu Sporny: Editors allow/reject specifications?.

Justin Richer: +1 to enforcing the normative rules.

Joe Andrieu: +1 to enforcing the normative rules, across the board.

Markus Sabadello: If there is a proposed method, like did:id, I don’t think we should allow it with the current process. We shouldn’t increase workload, but if there isn’t a registration that meets the requirements, I don’t think we should allow it..

Justin Richer: It seems the primary issue is that entries that were added to the registry before the registry itself was finished..
… So there is now an expectation that earlier registrations need to be honored in some way and kept in there..
… I think that is an absurd expectation.
… I completely understand it’s a lot of work to clean up; that said, I agree with Markus that unless the registry is actually following some rules, what’s the point.
… The structures are here to show that the rules have been followed by someone (the editors themselves).
… We should have an additional column for “preregistered” methods..
… Saying that we decided to keep this here, but we are not actually saying they passed a structured review. That can be a boolean..

Manu Sporny: On justin_r ‘s point, we had that field called “provisional”. Members of the WG argued to remove it, we had consensus about that..

Justin Richer: those members are wrong and the WG is wrong.

Manu Sporny: People felt strongly about removing it. That was the “provisional” tag..
… I think the Editors have been consistent with those provisional registrations..
… E.g. are 3 bullet points enough on privacy and security considerations?.
… I think there has been consistency with provisional registrations, there was an understanding that such registrations are work-in-progress..
… Everything in the registry today is under the notion that it’s provisional..
… Removing existing registrations may not be received well in the community. Doing it right now at the end of the charter would send a wrong signal to the community..
… The suggestion is to continue processing as we have, and then in the next charter we will be applying all normative statements. The community will need time to internalize that..

Brent Zundel: Realistically, what we have right now (it’s not an official W3C registry), it should be understood that the registry itself is provisional..
… Some registrations were written before some normative requirements existed..
… We plan to transition this note to an official W3C registry. Then all the current, provisional registrations can be transitions, but the time is not now..

Daniel Burnett: This amounts to a +1 to the last 2 comments. The proposal is acknowledging the fact that there will be a future version. We need to declare this done, this doesn’t mean requirements aren’t applied as best as we can, but there is a timing issue..
… It’s better to have this happen in the chartered group, rather than to revisit previous registrations..
… But we will everyone know that this is coming and that they have to clean up their methods or be removed..

Orie Steele: At this point I’m +1 to the proposal..
… The tendency to say that we need registered methods to be flawless is not going to work, and not consistent with other W3C work..
… I agree strongly with Justin+Markus, but the best time to address that is in the new charter..
… As I read the Web Payments specification, this seems to be the same direction they are taking with identifiers..
… Strongly in favor of addressing this as a proposal in the new charter..

Joe Andrieu: I’m really concerned this group thinks is appropriate to apply new rules..
… It’s not appropriate for new entries in the space to be treated differently..

Manu Sporny: Agree with Joe’s last statement..

Justin Richer: it is entirely appropriate to apply new rules. The world changed..

dmitri: I’d like to make a proposal. We were discussing as if we consider that registries have an obligation to be backwards consistent..
… No other registry (IANA etc.) have that requirement. All registries have forward consistency, but no backward consistency.
… My reluctant counter proposal is what if we don’t wait until rechartering and just reenforce the fact that if a WG comes to a new criteria decision, it should be applied to new registrations including those that are in PR right now..

Markus Sabadello: I’m also not favor at this point to revisit any existing registrations and we shouldn’t introduce new columns, tags, or lists, or change the process. I’m supportive of Manu’s proposal, we should do all of that in a new Charter, that sounds really good, so I fully support that..
… I like what Dmitri said, in new charter, in general it could be appropriate to allow what Dmitri said, allow new criteria to come up and apply to new registrations. We did have this discussion about method names, being indicative of what a Method does or how an infrastructure works. New requirements could come up over time that could apply to new registrations..
… I like what Manu said and what Dmitri said..

Brent Zundel: Can we have a decision now, or do we need more time to discuss this?.

Joe Andrieu: I like what dmitri said about forward compatibility..
… Manu do you mean new registrations?.

Manu Sporny: Yes.

dmitri: Just to clarify, did:id is a new registration, or accepted?.

Manu Sporny: New, currently in process.

Orie Steele: Please review did:object as well : https://github.com/w3c/did-spec-registries/pull/383.

Manu Sporny: If this proposal passes, we will process did:id under the same rules as in the past.

Markus Sabadello: I think this proposal is good. I think we should pass that..
… I don’t think it means we will accept did:id as it is right now without the requested changes, so I don’t think it changes anything about whether we should support this proposal..

Joe Andrieu: Is the current process to check conformance of all statements?.

Manu Sporny: No, we have never done that for any other DID method to date. So we would continue to not do that until we have a new WG..
… There is a checklist currently when you register a new method. If you can’t check each one of them, you can’t be registered..
… The problem is that you can look at these, and one person could say you meet the requirements, and another person could say you don’t meet them..
… With did:earth I think there is a misunderstanding..

Joe Andrieu: It definitely got a different review that it would have gotten 6 months ago..
… what’s the checklist and how do we adjust it?.

Ryan Grant: i need a link to the checklist.

Orie Steele: There is a link in IRC to a method that has the checklist checked.

Manu Sporny: That list is what we had consensus on in the group. If we wanted to change that, the group would have to have a discussion and come to consensus.

Joe Andrieu: Are there conformance requirements in the spec that are not in checklist?.

Manu Sporny: Yes.
… Markus is pointing to some older text that says you have to meet all normative requirements in DID Core..

Joe Andrieu: Just trying to understand the gap between what’s in the spec, and what’s in our current process.

Manu Sporny: Dmitri do you want to write down your counter-proposal..

Joe Andrieu: +1 to supporting manu’s registration proposal given the checklist.

Manu Sporny: JoeAndrieu – this section contains the normative statements IN ADDITION TO what’s currently required for registration – https://w3c.github.io/did-core/#methods.

Brent Zundel: Any suggested changes to the first proposal?.

Joe Andrieu: I care, Justin.

Justin Richer: A lot of comments are about “we can’t apply this, because we haven’t done it in the past”. The mistakes of the past should be learned from, not repeated..
… dmitri’s proposal is functionally the same as mine, with slightly different flavor.

Joe Andrieu: Those are slightly different.

dmitri: I think there’s a slight difference in that mine introduces new criteria..

Justin Richer: the difference JoeAndrieu points out is not intentional, I agree with Dmitri’s details.

dmitri: manu’s proposal is apply normative criteria after the new charter..

Joe Andrieu: mprorock’s is good for me too.

Manu Sporny: My understanding of your proposal is we would de-register everything.

dmitri: No, just going forward. We would have more stringent criteria for did:id and future registrations..

Justin Richer: I agree with dmitri’s point that we need to apply the rules about conformance and compliance now, and we cannot sit on this stance “oh well we allowed everything in, so we also have to do it now”.
… Ultimately it’s about being useful to the people reading the registry..

Joe Andrieu: I do think there is general consensus for moving forward with regard to not apply the current rules to new registrations.
… If we were to start applying conformance requirements now, that would be a huge change to the process.

Justin Richer: I am not asking to not do that..

Joe Andrieu: justin_r says use the registry process which will change.
… I don’t think there is support for going to conformance right now.

Daniel Burnett: We need to decide what we do now, not decide what the future group will do.
… I believe we can reach consensus on applying today’s rules.
… The next group may change the rules.

dmitri: I agree with burn and ivan said. Should we suspend adding new methods while in maintenance mode?.

Joe Andrieu: -1 to altering the entire registry process at this point.

Manu Sporny: agree.

Justin Richer: +1 to that – if the group is going to punt everything to the next group anyway we should.

Manu Sporny: -1 to altering the registration process at this point in time… in the future, sure..

Daniel Burnett: -1 for pausing registrations or changing the current process.

Brent Zundel: -1 to altering the process right now.

Joe Andrieu: Operating the registry doesn’t need a WG. Only changes to process need a WG..

Brent Zundel: -1 to suspending.

Ryan Grant: +1 to a consistent punt.

Manu Sporny: Happy holidays folks! :).

Brent Zundel: During the discussion there were several points where it seemed we were close to consensus..

Brent Zundel: Conversation can continue in the Github issues..
… Happy holidays, thanks to all scribes, it’s a pleasure working with all of you, see you next year!.