Meeting minutes
VCWG Confidence Method Meeting
Joe Andrieu: Howdy folks.
Phillip Long: Greetings.
Denken_Chen: I just found out that I haven't been assigned as co-host for this meeting link yet,…
Joe Andrieu: Oh, so you couldn't start.
Denken_Chen: right? Uh no I believe it will be uh from my news team.
Joe Andrieu: Yeah. Yeah. I don't know how to change that. I don't think uh I think it's uh we should ask uh the Ebon. I guess we should ask
Denken_Chen: It's okay we can extend a tack.
American Educational Credentials Program Connection
Denken_Chen: So uh sorry Philips I haven't respond to your email. Uh help me introduce some people from the American educational credentials people.
Denken_Chen: Uh I haven't got enough time to go through of this. Uh will you will you go in go through tac next week and…
Joe Andrieu: I will be there.
Denken_Chen: Philip will you go in there?
Joe Andrieu: Phil, do you think you're going to be uh are you joining us at TPAC?
Phillip Long: Sorry, I was on mute.
Phillip Long: Not physically, unfortunately. I'll be joining via um via remote at 5 in the morning.
Denken_Chen: Okay, got it.
Denken_Chen: Yeah, I should have more time. Yeah, I should have more time to arrange some international connections on my side so I can so I can uh with your help to to have more connection with the American educational credentials program things. Yeah. have to to pack.
Phillip Long: Who who is the American credentials program?
Denken_Chen: Um no I mean the people you invited uh introduced to me uh I say recommend me to connect with me marks are uh the American Association of
Denken_Chen: college registrate registr and admission services I right yeah yeah sorry I haven't got enough yeah I mean I haven't connect with them yet yeah sure thanks for your help.
Phillip Long: Oh, Acro.
Phillip Long: Yes, Acro. I I I know them. That's okay. I know them well. And was that uh Mike Simmons or…
Scott Jones: Is
Phillip Long: who was the person? Okay, let me know if there's anything I can do there because I know that group well.
Joe Andrieu: Elena, I understand you're going to be at TAC.
Phillip Long: That's an That's an important meeting. Wow.
Elaine_Wooton: I will be at TAC.
Elaine_Wooton: That's right. I'll be there the whole time.
Guinness Brewery Visit Discussion
Joe Andrieu: Okay. Um, do you want me to um would you like to join the signal chat to figure out when we're going to get to the Guinness Brewery?
Elaine_Wooton: Uh, yes,…
Elaine_Wooton: I would like to join a signal chat. Absolutely. We're going to we're going to get a lot of stuff done while we're there.
Joe Andrieu: All right,…
Joe Andrieu: I'll pull you into there.
Elaine_Wooton: So, I'll just tell you that um my family is Irish.
Joe Andrieu: Well, it's a bucket list item for me. So, I'm really excited that not in the Great Wall of China.
Elaine_Wooton: So, when my two boys each turned 18, I took them to Ireland and took them to Guinness that Guinness to have their first official beer. So,
Joe Andrieu: Oh, wow.
Joe Andrieu: Yeah. Props.
Phillip Long: What? Yeah.
Elaine_Wooton: Family thing like like prime like prime bear.
Joe Andrieu: Now, now I'm jealous.
Scott Jones: Is it um is it similar to Like I don't want to talk smack about American beers,…
Scott Jones: but my understanding is if you go to like the Kors factory and you drink it fresh coming off the line, it's like a religious experience totally different from what you get in the store. Is the Guinness is like going to the source at Guinness similar? Are you getting like God's God's quality of beer? Yeah.
Elaine_Wooton: Well, you're getting you're getting Yeah, you're getting the fresh stuff, but the other thing is that, you know, if you're smart about it, you wait until you're on the top floor, which is a 100 Yeah. 360 degree windows looking out and…
Scott Jones: Wow. Wow.
Elaine_Wooton: you're up there and you're having the beer and it's Yeah. It's an experience. Yeah. or or you can wait and…
Elaine_Wooton: you can have your Guinness, your first Guinness while you're there when the uh Irish dancers are dancing and um that's also an experience. So, it's more about making it an experience, but the beer is pretty good.
Joe Andrieu: Yeah, it also the the beer you get internationally will have been treated differently.
Joe Andrieu: Um although I don't know if that's true for beers in the UK. Um, but I know the beer in the US is processed,…
Scott Jones: H. I'm guessing you don't want to do a pilgrimage there. Yeah.
Joe Andrieu: pasteurized or something before they ship it out. So, I'm I did not know that.
Elaine_Wooton: So Joe, in case you didn't know, there is a Guinness plant in Baltimore. Elaine Wooton:
Phillip Long: Wow. Well,…
Elaine_Wooton: So So yeah,…
Joe Andrieu: So they make they make the American version.
Phillip Long: that's worth that.
Elaine_Wooton: so they make the American version. That's right. That's right. What's really crazy,…
Joe Andrieu: That's right.
Elaine_Wooton: right? Well, I've thought about it…
Scott Jones: Wait, I'm done.
Elaine_Wooton: because it's right here by me. But, uh, what's crazy is I took my kids all that way to have their,…
Joe Andrieu: What the hell?
Elaine_Wooton: you know, their first beer, and both of them prefer, um, Guinness had a cider,…
Phillip Long: You can't shape taste.
Face Vector Progress Update
Elaine_Wooton: and they preferred the cider. Was like, "No, you guys." Crazy. Um I I did want to I'm not on the agenda, but I Yeah, I'm not on the agenda, but I did want to raise that um Manu and I heard from Ilas,…
Elaine_Wooton: who is the the young man that's working on the face vector, and he says he's made pretty good progress, and uh he wants to meet um wants to meet and so Manu and I were like, well, you can meet with us or…
Elaine_Wooton: you can present because of Yeah.
Joe Andrieu: Uh you you were um breaking up the lane.
Phillip Long: You broke up.
Joe Andrieu: If you could repeat your last sentence or so.
Elaine_Wooton: Okay. Hold on. Yeah. Yeah. Hold on. Let me phone. Is that better?
Phillip Long: Oh, yes.
Elaine_Wooton: Yeah. Yeah. I was talking to you through the phone. I mean through the car. Yeah. So, we heard back from Ilas and um he's made good progress on the small face vector thing that would go in the barcode and Manu and I uh connected with him to maybe meet, but Manu suggested he present to this group, but then we still have the same issue company not being able to do something with W3C or something like that. So it may be that I'll get the information from him and bring it back to the But anyway, there's just there's progress with that. If you don't know the story, it's actually his PhD project to create this thing.
Elaine_Wooton: So he's not getting his PhD done.
Phillip Long: What is the problem with the company and…
Phillip Long: the W3C?
Elaine_Wooton: Um I don't know company.
Phillip Long: Okay. Yeah,…
Elaine_Wooton: So we're trying to get them to So the issue is with open sourcing the face that you know the the thing um well that's not really the issue but anyway I guess there's a a a version of W3C that's European and we're trying to get them to talk to them but they would have to talk to us. It's not really clear what the issue is. It just it's a no Yeah.
Phillip Long: I just we seem to make it hard for ourselves to do the work this way.
Elaine_Wooton: Yeah. But the intention is to open source this technology and…
Phillip Long: Of course.
Elaine_Wooton: if then we can use it.
Phillip Long: Right. Yep.
Elaine_Wooton: So it it's I don't want to say it's moot because we want to have a relationship with good. They have really good intention but um if they can't do it you know like we wanted to be an invited expert and that that wasn't going to work. Stop. Um,…
Scott Jones: What's next?
Joe Andrieu: Hey, Lane.
Phillip Long: Do we have an Renaissance.
Joe Andrieu: How big is the is the company involved?
Elaine_Wooton: it's Cabinet Renault. It's a French um French. It's called Cabinet something something Rena. So it they do red teaming for um biometrics. Um but they do it for French government. They're pretty involved in like everything. Um let me while we're talking I'll find the link to the company so you can see what it is. But he works for the company but he's working on his PhD.
Elaine_Wooton: And uh I met one of the directors from the company last year and I said, you know, we all really really need a small face vector um cuz we need to be able to put biometrics in the barcode. And he goes, "Oh, I have the guy who can do it." And I was like, "Okay, nobody's done it till now except SpaceX in their own way, but nobody's done it till now. Are you sure he can do it?" He goes, "Yes, he can do it." And and I'm like,…
Scott Jones: How small do you need the need it to be?
Elaine_Wooton: "Okay, that's fine." So anyway, I'll um I'll send you guys the link to the company while we're on this call covering other things. But uh once I've met Mano and I think we'll try to just meet with them and then I'll come back with some slides know what what progress he's made. But it looks pretty good now. Well, we need to be able to put it in the barcode. Um well, okay.
Elaine_Wooton: So, I think it's almost moot right now because we needed to be able to put it in the PDF 417 on a driver's license, but in order to make that quantum resistant, we're just screwed, right? So, I mean, we're going to develop thing just on the verge of can't use it. Um, but maybe we can. I mean, if the photography is all just hashes, then we can still use it.
Elaine_Wooton: Anyway, I don't know. Yeah. Um,…
Scott Jones: Well, just uh I wonder something to take offline,…
Scott Jones: but I think we'd explored in a conversation how my uh my embeddings are quite small, too. And just to double check that 512 embeddings would fit in a barcode as well. is
Elaine_Wooton: let me find out what the max is that we can use and then yeah, let's take it offline and figure it out. Um, I need to I mean literally I advise the AMVA car design.
Elaine_Wooton: AMP is the Americanation of Motor Vehicle Administrators and I have a a path in where I give them input about what technology to use. Um they're always about three years behind. Um but uh you know I want to be a and I give them options, right? I say this is what's out there. So Scott, if it'll work, I want to tell them and if I sol works,…
Scott Jones: Please. Gotcha.
Elaine_Wooton: I want to tell them and then they can decide what they want to do with the with the information. I'll tell you right now, they're struggling because um FaceTime and the UR code are pushing really really hard for the driver's license market in the US, but that proprietary free, it's proprietary and nobody wants touch it.
Elaine_Wooton: Yeah. So, yeah, let me check a 512.
Elaine_Wooton: But Scott, why don't you and I just chat again. I mean,…
Scott Jones: We have fun,…
Scott Jones: you know. We have fun. It's worth it anyway.
Elaine_Wooton: yeah, might as well. All right, I'll find that link and put it in the chat while you guys go on with the Yeah.
Joe Andrieu: Scott, what's the data size of those 500 um elements you mentioned?
Scott Jones: Uh, 512 bytes is what I screwed up saying.
Joe Andrieu: Oh, yeah,…
Scott Jones: Yeah. Yep. Oops.
Joe Andrieu: that should be able to fit in a barcode. It's much easier than the quantum signatures that which is a more breaking change.
Elaine_Wooton: Yeah. Mist put out some research um last year I think about actually putting really just low resolution images into the barcode.
Elaine_Wooton: So, they've done some research with I think it's uh like 72 dpi, pretty small images. And the the purpose is to just compare to see what the same photo that's on the front. So, you're not doing face match. You're not doing anything what's in the barcode. You're just checking to see that the photo that's on the front of the card is actually the image that's on the back.
Scott Jones: Gotcha.
Elaine_Wooton: And that matters because we can we can sign the barcodes, but we can't do any photos. So people are able to completely counterfeit a license, everything, digital signature, everything, and then just put a different photo in. So if you put a really low resolution version of the photo into that barcode, you can at least tell that it's this says with pretty high accuracy, you can tell it's the same photo. Anyway, I'm taking you all on a tangent. That just say that's my world.
Phillip Long: greetings. Mano
Denken_Chen: Yeah. Money, please. Money, please.
Manu_Sporny: Yeah, I mean not a not a tangent, Elaine, right? Like that's exactly why we're here is to figure out which one of these,…
Elaine_Wooton: Oh,
Manu_Sporny: you know, um mechanisms uh confidence mechanisms uh works in the real world.
Scott Jones: Okay.
Manu_Sporny: And so really appreciate your input on all of this stuff. Um that's it.
Joe Andrieu: Yeah, plus one.
Joe Andrieu: Okay. Um, if we want to turn to our regularly scheduled agenda. Apologies for not getting anything out to the group. I do want to acknowledge that that was asked for and I think we did it once. Um but Dan, let's let's uh just a note to both of us to try and get a notion of our agenda out to folks um before our meetings. Um in this case,…
Scott Jones: It's going to get
Assurance Levels PR #54
Joe Andrieu: uh Denin, I think let's go over your um PR and see how long that takes us. Um after that, we could look at um the emerging diagram that's going to go into the confidence method.
Update Assurance Levels description by denkeni · Pull Request #54 · w3c/vc-confidence-method · GitHub
<Elaine_Wooton> Here’s the company with the employee working on the small face vector Cabinet Louis Reynaud
Joe Andrieu: uh sorry the threat model um and I don't know if we have anything else for the agenda other than uh we could also look at issues I guess and if anyone else has a suggestion for the agenda feel free to raise your hand and we'll see what you want to talk about okay hearing no clamor uh Denin why don't you introduce us this is update assurance level descriptions Poll number 54. I'll drop that into our chat and thank you. And could you lead us off?
Denken_Chen: Yes. So my main point is is I decided to drop ISO things uh because there is an assurance level standards called ISOC uh 2915 and it's being used in some banking systems in in some areas and also in my countries. Um however I I understand that there is a conflict or some um arguments over whether to include ISO uh documents here.
Denken_Chen: So uh it's not our top priority so I just dropped this but and also I have included the EU EI EI does they have their own level of assurance things uh I'm not sure whether it's appropriate to site and…
Scott Jones: I'm sorry.
<Joe Andrieu> Update Assurance Levels description by denkeni · Pull Request #54 · w3c/vc-confidence-method · GitHub
Denken_Chen: yet does assurance devil framework here but I just put it there and so these are two things we NIST uh 8 63-4 and also EI does develop assurance two things to be the inside that and so I would like to have some feedbacks from the community and to see whether we can merge this PR Yeah.
<Scott Jones> I think my VTO would love CLR's sailing imagery from the go, thanks for sharing
Denken_Chen: I move forward to uh understand whether we are ready for horizontal review.
Denken_Chen: Let's see.
<Scott Jones> *CTO
Phillip Long: Just to be clear,…
Phillip Long: the recommendation is just to drop it all together.
Denken_Chen: All right. I didn't catch it. Could you speak it again?
Phillip Long: Sure. the the recommendation is to drop this reference to um that particular um item.
Joe Andrieu: Yeah.
Phillip Long: Hold on, let me get to it again.
Scott Jones: Okay. Just
Phillip Long: Um it's SP800-63-4. Just drop reference to that al together.
Denken_Chen: Yes, we just assign the value of of identity assurance level one two three uh just use lead that value…
Denken_Chen: because lead is uh good enough for many trust framework there outside uh as far as I know several countries are establishing their own regulation based on uh framework. Yeah. Thanks Manu for screen sharing this and also manu your up
Manu_Sporny: Yes. Um just to note that uh I reviewed the PR and it looks good at a high level to me. Um yeah, and apolog I I was just screen sharing uh Phil to kind of help um point out the text that um Denin was uh talking about.
Phillip Long: Mhm. Mhm.
Scott Jones: Wait. That's
Manu_Sporny: Uh I think the this this PR basically um provides two kind of assurance levels or just starts defining two assurance levels. One of them is the NIST SP863-4 one which defines identity assurance level one through three and then the other one is the European EIDS regulation which specifies assurance levels low,…
Scott Jones: Yeah, that's What was it?
Manu_Sporny: substantial and high. Uh and then the PR kind of shows how you would express one of these uh in a VC. Um, so based on all that, you know, I I a plus one to the PR and merging the PR. In my review comments, I did have a question around like I've always been confused about whether or not assurance level should be a part of evidence or a part of something associated with the credential subject. Um, so you know, when it's associated with the credential subject, it's effectively the issuer saying, "Hey, I established this insurance level uh with the credential subject.
Joe Andrieu: Okay.
Manu_Sporny: Um they are at you know an IL you know level two um and I'm wondering how this interplays with things like a driver's license or an identification card.
Scott Jones: Oops.
Manu_Sporny: Um would the assurance level go on the information about the the human being for example um or would it go on the card information?
Manu_Sporny: I would expect it to go on the human like this. But I I'm curious to hear what other people think about like what is the relationship between assurance level and evidence and when should you put one and the other and and so on so forth. Um, director
Joe Andrieu: uh yeah two things including a new topic…
Joe Andrieu: but on this evidence one I I do have a pretty grounded opinion that to me evidence is providing some source data that can
Phillip Long: Mhm.
Joe Andrieu: evaluated by the party who is processing it. Um, in terms of trying to establish why did they believe that? Um, and to me the assurance level is a simple assertion without the evidence. So we met this criteria. Maybe it required blood samples, maybe required iris scans, maybe it required fingerprints, but it is not the blood samples or the fingerprints or the iris skin, right? It's just an attestation. um where the issuer is able to say, "Hey, there's this other standard that's been established and we met that standard." And so we are letting everyone know that for this credential subject in this document, um we have established that the the other aspect of this which is something that that Dave Longley brought up is that maybe this should be bound to a temporality or a time at which that assurance level was achieved.
Joe Andrieu: Um I had always simply said hey the credential is also time bound like the the it is issued at a point in time um typically they have expireies etc. So at the point of issuance that that assurance level was valid um to my mindset was enough. But I do think it's interesting that if you have um a an assurance level whose timeliness may be of primacy maybe we should have an explicit date in here. Um uh and then the the third thing it was a question just for you Dan. Um there isn't any ISO thing in here that you're removing.
Joe Andrieu: So, was it just conceptually you were hoping to put one of the ISO things in? Just this edit doesn't We're not removing any ISO stuff in the text itself.
Scott Jones: two.
Joe Andrieu: That's it.
Denken_Chen: Okay. Okay.
Denken_Chen: Uh so first we are not re removing ISO uh 29. Uh what's it brand name?
Joe Andrieu: Okay. It just sounded that way when you presented. So,
Denken_Chen: Yeah, I I did not remove ISO 29115 out of our spec uh because I just initially I trying to add them because these are one of the three frame uh assurance level framework.
Denken_Chen: work we've been researching on. So, but it's not in heavy use. So, I'm I'm I'm okay with not adding them. That's first thing. And to respond menu questions. So, first uh for example, the biometric image I believe most of the portrait has been included for the credential subject level of attributes. For example, in driver license or many other government issued ID will mostly or say passport will include the biometric image already. And so one thing we can do that is to have a biometric image in the credential subject and site that site it uh with confidence method to have a hint that you can use that image for confidence method.
Denken_Chen: Uh like earlier the eland mentioned that the image could be as simple as very low resolution for comparison…
Scott Jones: Let's see
Denken_Chen: but it could also be as high level of resolution for in person checking like for example in passport the portrait image is often used for border controls for face recognition right so that's one way we could do it
Denken_Chen: And uh another thing is uh we see the biometric image portrait image along with the identity assurance level things is important uh because that will assume uh that will assure that the image is for example with the highest I3 the portrait image should be taken in person or in highly uh controlled virtual environment. So very um that will be very highly assured image to be the subject. Uh that's for I3…
Scott Jones: What's up? Just
Denken_Chen: but if that's as low as I one or two that biometric image probably doesn't matter very much.
Denken_Chen: It could be just like I submitted my photo um image for application but I did not show that in person for the whole credential issuance process. So that says a little bit differently because in many of the countries out there there are some credentials does not really replace the physical for example physical driver license or physical ID card and so that could be lower assurance level. So the verifier could take that into considerations whether they would accept that uh for example I2 of image. Okay, that's it. I think next is Phillips. Phillips is uh raise your hand before manual, right? That's
Phillip Long: I did, but Amanda, if you wanted to go first,…
Manu_Sporny: No, no,…
Phillip Long: that's fine.
Manu_Sporny: please go, go ahead, Phil. I can I can
Phillip Long: Okay. Yeah. I just wanted to uh support Joe's interpretation that evidence I agree with him is is information that is third party that is independent and…
Scott Jones: 86. Good.
Phillip Long: which um provides the detail but under which an assurance can be assessed directly. uh otherwise it is it is an assertion and therefore from manu's description I which I agree with it would be associated with the issuer um the issuer says it's one two or three uh you have to trust the issuers uh to detroating evidence for example as you're describing it um information about the uh the density of the resolution
Phillip Long: ution and uh the the time at which that information was uh determine or the the time at which that uh evidence was was collected that would be in in the evidence field to corroborate it accordingly.
Phillip Long: So I think that that split of rel relationships for what goes in credential subject versus what goes in evidence is the one that we are appropriately following. Thank you.
Joe Andrieu: Go ahead,…
Joe Andrieu: man. I think you also have some engagement in the chat on this issue you're you're raising.
Manu_Sporny: Oh, okay.
Manu_Sporny: Um, let me Okay,…
<Denken_Chen> We have this Verifiable Credential Confidence Methods v1.0
Scott Jones: What is this?
<Joe Andrieu> Verifiable Credential Confidence Methods v1.0
Manu_Sporny: so uh I think I was getting ready to raise a relation to the evidence property. Okay. Okay, perfect. Uh, I did not know we had this in the spec and…
<Joe Andrieu> You beat me, Denken. =)
Joe Andrieu: Ed, edits our welcome.
Joe Andrieu: Like maybe it's not clear, but we do at least have a draft of something.
Manu_Sporny: Okay. No,…
Manu_Sporny: that's great. Um, okay. So, uh, totally agree with the, um, explanation, uh, Joe, that you mentioned. Also agree, Denin, with what you said. I was going to say like, hey, let's write this down in the spec, but, it looks like it already is written down. So, apologies for not realizing that was in the spec. Um, the only things that are left are the, you know, Dave Longley's, uh, concern. Um, so I'll So, so plus one, I think this is the right way to think about it and and that's great. Um uh the to Dave Longley's concern, I know that with at least some of the state organizations that we're uh working with um they can take a photo of the individual and do the identity proofing once at one date and then when they issue that verifiable credential, it's a different date and it is definitely not the date on which they did the the assurance level thing, right?
Manu_Sporny: So, for example, um for a driver's license, they'll do the IIL level whatever um uh on the day the person is like renewing their driver's license uh and then they can issue three or four of them later on if the individual loses their driver's license if they lose their phone if they you know whatever um and that date that the credentials issued is not the date that they establish the the IIL. So, I do think Dave has a good point there. Um, and you know, I did and and and therefore we add a date, right? I I don't feel strongly about when we do that. Um, but I do think that we should I think we should do that because they're not you they would not have a way to say like, hey, we did I you know, two years ago and then we issued this credential two years later. Um, so that's point one.
Manu_Sporny: I think we should dig into that. The other thing um I'm wondering about is if there's a way to connect the assurance level to the evidence field, right? Because, you know, there's a graph of information. They're in different parts of the graph. How do we say that, you know, we achieved an IL2 on this individual based on this evidence and then we point to the evidence thing, right? So, how do we how do we I think just having it in the credential doesn't tie it as directly. Um, and we could and and one way we could do it is literally put evidence colon and…
Manu_Sporny: then uh have it linked to a local fragment identifier in the local graph as an example. I don't know how I think about that. I I don't know how I feel about that, but that would be one way to do it. Um, anyway, just some thoughts. I I think this is all on a really good trajectory. Um there's just some fine-tuning uh we may want to think
Joe Andrieu: So, uh,…
Joe Andrieu: plus one to figuring out a date field. It seems like it could be something in the direction of um date established or something. It is different than the valid from and invalid till which is about the credential, right? This is about establishing for that individual.
Manu_Sporny: Yep.
<Phillip Long> Does that mean the date the image was taken should be in the evidence field, as distinct from date of credential issuance
<Denken_Chen> The verifier could refer to issuance date at top level? Verifiable Credentials Data Model v2.0
Joe Andrieu: Um but I wanted to to push back Manu about the desiraability of evidence in general.
Joe Andrieu: Um, I think it's a big privacy concern to to encourage that people put raw data that could be used to correlate individuals in the credentials. So my preference would be to just not explain that you could do that thing. I think the to me a big part of the point of the confidence method is to provide additional mechanisms that are privacy respecting that nevertheless give some of those benefits of strong identity. Right? I think that's the trade-off that we're negotiating. Um, with that, go ahead, Dan.
Denken_Chen: I'm trying to understand where to put the date for the identity assurance um things. Uh the most intuitive one is the issuance state that has has been already at the top level the valid from uh data field and so we will assume that the people the subject has been verified uh with the identity assurance level framework at the issuing date or at the issue
Denken_Chen: process. So I I was wondering whether we should put another new uh date for the competence for each of the competence method also. Um and also I'm also curious about the VC there there's a refreshing mechanism that that's probably already in production somewhere else and we do have similar issues as well. For example, our national health insurance card. The portrait image on it is the first photo that we've been taken at the insurance date. It's probably at my childhood and never refreshed after the re after all.
Denken_Chen: Um but it's very different from for example driver license or passport that may require you to refresh your portrait at every new refresh period. So yes, so there are very different policy there. So I'm just curious about whether we should add a new date for the competence method.
<Phillip Long> @Joe - it sounds like evidence then is, at least in this case, is a unnecessary inclusion in the credential at all.
Manu_Sporny: Yeah. Um, so two two things. So to respond to you, Joe, yeah, that's a that's a fair point on um we don't want to uh in general, especially for like human use cases, uh suggest that you should be putting evidence and invasive data into these uh credentials as an issuer. So that that resonates with me, Joe. So I retract my desire to figure out a way to link them together. Um uh and plus one I think you know we don't need to because of what you said which is like we're trying to provide more privacy preserving um uh uh holder uh or subject uh empowering mechanisms to you know prove who they are without invading their privacy.
Manu_Sporny: So, plus plus one all that. Um, don't need to figure out a way to link evidence to assurance level. Um, Denin, uh, to your point about, you know, date, I think we basically put date maybe in here. Uh, like an established date like when did you establish the identity assurance level? Um any it can be rough like it could be a year. It doesn't have to be like a a date time stamp which again could you know violate the person's privacy.
Manu_Sporny: Um any yeah so may maybe you know we do that and and and suggest that people make it do not make it fine grained like don't don't do date time uh try to keep it to year um or…
<Phillip Long> Images in drivers' licenses last for 10 years in some states, without a reissued image
Manu_Sporny: maybe month but you know not day I there's some privacy implications there right um but other than that I think We're We're good.
Joe Andrieu: uh that's okay.
<Joe Andrieu> @Phil well, I'm not a fan, but it is something that might be useful for certain parties in certain situations. I've always considered it about the credential as a whole, eg, evidence of earning a degree in a degree credential
Manu_Sporny: I mean, you know, and and these can be done in a future PR. I don't think they need to be done in this PR. Um, as long as we're kind of tracking the issue, that's
Joe Andrieu: I I will add an issue to support that. I I agree. We don't need it in this PR. The the the date thing is intriguing. Um I guess I'm curious if anyone knows what are the date related implications in these different standards. Um like does it matter that it's been within 30 days for example?
Joe Andrieu: Um if we have those sort of temporal notions as people consider I2 or IL3 you know was it done within the last two weeks then we would need I think a date rather than a month. Um but I don't know if those specifications even entangle at that level. Go ahead. Dang it.
<Phillip Long> evidence supporting a degree assertion is public to begin with so that's not a concern I don't think
Denken_Chen: Yeah, I still think the issuance date at the T level is good enough for this. Um, the reason is we usually refer to that issuing date to make sure it's the up to date, the latest attributes for the subject. And in some cases we do issue um some short uh time frame um that is expired very soon for example in one month or say in just one year for most of our credentials. So we will force the user to refresh or to reissuance uh its credentials.
Denken_Chen: So that's one way to have that shortlived credentials. And so um I'm more supporting that we just hint the verifier that it could check the issuance date uh as a reference for the identity assurance level because it is expected to be done at that date the issuance date.
Joe Andrieu: So, at least uh in California, um others can speak more broadly to other uh jurisdictions or situations. Um, but I continually get new driver's license, both physical and digital, from California without re-establishing the identity assurance that they did when I was physically in the office. Um, and this can go on for decades. Um, because I just pay my renewal fee. Um, and uh, especially for the digital ones.
Joe Andrieu: I mean, I think the California DMV, the digital stuff refreshes frequently, intentionally, which further separates it sort of every time from the assurance level moment in time where they actually checked me and took a photo and did a fingerprint or whatever. Um, go ahead, Manny.
Manu_Sporny: Yeah, that's correct. Um, uh, so Denin, I think the problem with suggesting that the assurance level is tied to the validity date of the credential is that that is um not correct in the broad set of use cases. Um, right? Meaning like for your national health ID uh card, that's not correct. For the uh driver's licenses, in most states, it's not correct. And so I don't think we should suggest that. We should just s suggest like um if there's no date then the issuer uh all they're stating is that they establish this identity assurance level at some point that makes sense for them. But if you don't have a date, you don't really know when they did it. And so if the issuer wants to make it clear they should add a date.
<Phillip Long> We can start with something less specific for the time of the image capture, and see what the real world response is to determine how to refine it, if needed,, going forward
Manu_Sporny: Um and if not, you just know that they did it at some point uh before they issued the credential and and they they are convinced that the individual still has this level of identity assurance uh associated with them.
<Ted_Thibodeau_Jr> @Phil - Data being "public" doesn't remove privacy implications. The old White Pages were "public" but that didn't make them useful for tracking activity of the persons listed therein.
Manu_Sporny: Um but that's like, you know, I I don't think we need that in this in this PR. Maybe we have some language that we add to the spec later on. I don't think we need to, you know, slow the PR down because of that. Um, that's
Joe Andrieu: cool. Thanks,…
Joe Andrieu: Manu. Um, so yes, let's not slow this down. I am creating an issue. I'm halfway through typing it up. Um the I wanted to kick around while we were on this topic. Is there something similar with um uh the confidence methods? Um my sense is no because the point of the confidence method is not necessarily to establish that the issuer did something but rather to give a mechanism for the verifier to do something.
Joe Andrieu: Um, but even as I say that out loud, I can imagine that the issuer may want to um uh assert that they tested this mechanism themselves, right? So that they verified this this confidence method. They check that the user has control over the the key or they check that the user actually could satisfy the biometric. Um so it's a slightly different notion. Um, but maybe maybe we should expand the date to to uh both properties that we're talking about here. Thank
Denken_Chen: Yeah. Okay. Now I understand your point manual answer and so I agree with that and then uh I would suggest that that date um should be put into the confidence method rather than the assurance level things. And the reason is we we are now putting the confidence method and assurance level as the same level of data outfield. And the reason is I believe in most of the credentials out there they are they issue a an a credential based on the one single assurance level and uh however the biometric image inside or biometric vector or the the key binding stuff that could be done at some other date right so for example I could take my picture
<Phillip Long> @Ted - good point. I was thinking we don't want to include something one would only find in the credential. But I think you're saying institutions making degree completion evidence public is itself a privacy consideration we should not emulate.
Denken_Chen: yesterday, but I got my driver license next week. So, the biometric image should have a date of yesterday where when I took my photos. So, that could be a reasonable way to put it a date into each of the competence methods.
Denken_Chen: That's it.
Joe Andrieu: Okay, I think that's good.
Joe Andrieu: I um I created this issue uh 55. Um I recommend in there maybe we try assurance date um since right above it we have identity assurance level right as one of those things. So assurance date feels aligned to me. Um, but we can debate that and talk about that. Kick it around a TAC again. Uh, I think we're pretty much good with this PR. Um, I will I will give it my positive review. Um, I see also Manny marked it as good to go. Um, does anyone have any objections to merging this in? Okay.
Phillip Long: None here.
Joe Andrieu: Um then I will uh and I we Wait. You don't do any of the stuff that uh I mean there's no reason not to just squash and merge now.
Denken_Chen: Yeah, you can do that.
<Joe Andrieu> Add date field for Assurance Level · Issue #55 · w3c/vc-confidence-method · GitHub
Joe Andrieu: All right, rebase and merge. Um, I don't think it matters in this case,…
Manu_Sporny: uh rebase if you squash I think it gives other it gives you like credit uh sometimes or it says you co-authored it correct yeah and…
Joe Andrieu: right? Okay. So, so despite squash and merge being the default on GitHub, our default should be rebase.
Manu_Sporny: it GitHub's weird the the default is whatever you just did in another PR.
Joe Andrieu: Oh, really?
Manu_Sporny: Yeah. No.
Joe Andrieu: It's not even organizational repo specific.
Manu_Sporny: Yeah. And also depends on,…
Joe Andrieu: Ah, yeah.
Manu_Sporny: you know, the month. Sometimes they make a design decision and don't tell anyone. So,
Threat Model Diagram And Flows
Joe Andrieu: Annoying. Okay. Well, I've rebased and merged. Um, I think that's good work for today. Um, and we can debate another time whether or not we want to extend the assurance state idea to the other fields. Um, what else might we look at? So, we could look at the diagram. I don't know that that's that we're going to be able to get into that that much. Um, uh, Denin and I have updated the diagram, or rather I updated it and convinced Denin that it looks good. Um, uh, but we do not have the flows, and the flows are the tricky part. So, I started working through the flows. I mean, I guess I can go ahead and share it. I may as well tell you where I'm at. There's no reason not to.
Joe Andrieu: Um um so this is an iteration of diagrams that we've been working on um that we've shared with the group uh a couple of times. See, zoom in a little bit. Hopefully that's a little bit better. Um and I've only started for the biometric templates working through what are the flows and writing them up separately so that we can understand that we're talking about the same thing and part of that will also lead to question or answers to the questions like is this what I'm calling a biometric template request um is that just a VPR or is it some other artifact or is it just an abstract thing that we are describing as you know they provide the metadata
Joe Andrieu: um or the information about the kind of biometric that they need. Um I don't know yet like I don't know if we want to standardize that as a set uh as a a JSON object with certain properties or if we want to keep it abstract. But this walk through will highlight all of that for us and um I'm hoping to go over it um at TAC with folks. I think we've got an hour for for us um to talk about confidence method things. Um but let me take actually that segue and open it. Is there anything else we should be talking at TAC? Um seems to me kicking off this uh threat model in earnest is my goal. Um I'm curious if anyone else has something else we should get to.
Joe Andrieu: I know Denin and I are working towards a checklist to get to horizontal reviews. So our hope is to have so I think we've got the text in place and frankly for horizontal review we don't need the threat model although it would be nice to have. Um, so my hope is to get something documented and into here so that we can submit issues on all the different horizontal review groups that that you know because that's the process is we go to their repos and we say, "Hey, we'd like it review." Um, and that is my goal for um having that ready by the time we're in TAC. And I see Manu, you raise your hand. Go ahead.
Manu_Sporny: Yeah, I mean all all that sounds good as a plan, Joe. Um the I think you know on the question of like are we ready for horizontal review? Um it would be good to have I mean yes we don't need the threat model to kick off horizontal review and…
Manu_Sporny: if you think that you can get it into some kind of you know shape um were were you saying before you were going to try to do it before TAC or…
Joe Andrieu: Yeah, this is part of my travel stuff and…
Manu_Sporny: Okay. Okay.
Joe Andrieu: I'm I've got most of Monday available. So, not to mention the time on the plane. But, uh th this is I feel good about the objects in this diagram is the flows and…
Manu_Sporny: Okay. Okay.
Joe Andrieu: then the write up of those flows. Like that's the remaining work there.
Manu_Sporny: Okay. All right. Um, yeah. I mean, you know, of course, it'd be nice to have a threat model before we went to horizontal review, but I mean, look, might as well kick off horizontal review. Um, and then if this is going to be in place within a week or two, like they're not going to get to it before you you have a first draft, I think. Um uh and plus one, you know, there's a a significant amount of detail here and I'm sure the writeup uh is is going to be fine. So plus one to asking uh to kick off the horizontal review uh at TAC. Uh I think all the sections uh look good uh to me. Uh we do have some things we want to clean up.
Manu_Sporny: This is editorial so it could be commits made on main but there's a section called verifi ver verification confidence that has to be determined that has been replaced by like the cryptographic key and controlled identifier sections and…
Manu_Sporny: then we have a biometric image confidence section that's also empty uh…
Joe Andrieu: Right. Yeah.
Manu_Sporny: but that is uh replaced by section 52 biometric image. So, it's just cleanups like that so it doesn't confuse the the reviewers into thinking that we we have a empty section.
Joe Andrieu: Cool. Thanks,…
Manu_Sporny: Um, but yeah, other than that, I think uh we're good to go for horizontal review. And thank you very much to both of you uh for moving this along to get us to that
Joe Andrieu: man. Yeah. Uh those I think are easy sections to remove. So, uh Dan, let's uh let's get a PR in to address that.
Joe Andrieu: Um, I don't know if you'll have time before I have time, but um, I think we can get that done by next week.
Denken_Chen: Yeah, I can do it now.
Joe Andrieu: Okay, awesome. Um, all right. I think that's all we have for this week. Um, and I do want to know I want to say this out loud. I don't because I haven't talked to Denin about it and it's it's I don't know where the cultural um uh uh I don't even know how to describe it. There was an effort I recall maybe two years ago that went around to try and get calls like this to end five minutes early. And part of that was uh because postcoid we all scheduled all our meetings back to back to back to back to back. Um, and it was exhausting. Um, so I would like to move back into that. Um, uh,…
Joe Andrieu: which means we have, uh, just one more minute left. I'm seeing some good thumbs up. And Ted, you jumped on the queue. Go ahead.
Ted_Thibodeau_Jr: Yeah, I was just going to voice the thumbs up the scribing thing.
Joe Andrieu: Okay, cool.
Joe Andrieu: Um, and did you get your earlier question answered about how do we uh uh because I I saw your comments elsewhere about um uh our our scribing.
Ted_Thibodeau_Jr: Yeah. Um,…
Joe Andrieu: Yeah. Okay. Cool.
Ted_Thibodeau_Jr: I'll start with that um the repo that that Denin has posted uh Denan pointed me to and I'm sure I'll go someplace else because that one's specific to a working group. But that's okay. It goes in the right direction.
Manu_Sporny: Hey Ted, I I hear you. It's almost certainly going to land on my plate unless somebody else wants to write the code. I'm just letting you know. I'm apologizing in advance. I'm not going to get to it for weeks, if not months.
Ted_Thibodeau_Jr: That's okay. If it's on the list, it'll get there eventually. Cool.
Joe Andrieu: It's a good ask.
Joe Andrieu: I mean, I wanted the same thing. So, you know, easily navigable minutes would be welcome. Okay. Um, anyone have anything else to add before we wrap? Okay. Excellent. I look forward to seeing you in person for those of you who are going to make it out. And I hope those of you who can't um if you can find time in your schedule. I realize it's an awkward time for uh many of us in the US. Um but next week is TPAC and then as is um uh uh the norm. We will not be meeting the following week even though that would be our schedule for this group.
Joe Andrieu: Um, so we will be meeting again whatever the rhythm is after that. We'll we'll uh I guess we'll update the calendar and…
Elaine_Wooton: Thanks all. Bye.
Joe Andrieu: let people know. Okay, thanks everyone. Take care. Meeting ended after 00:54:23 👋 This editable transcript was computer generated and might contain errors. People can also change the text after it was created.