Decentralized Trust Graph Working Group Weekly Meeting - AM/EU Time Zone - 2026/08/19

All righty, folks. Sorry, the uh the Zoom gods are being a little slow this morning. Morning, Daniel. I see you there. Hi, I'll pre- I will prepare our very full agenda page. Get ready to share this here. And we'll give it 1 more minute for folks to roll in. And let's see. Share here. And does that. Arrange windows. Chat in there. Okay. All right. Actually, I'll get a little bit more room on that window if I go like that. All right. Can you see my screen okay? Oops. >> Yeah. >> Good. Okay, it's 4 minutes after, so let's go ahead and get going cuz we have a very full agenda today. And I don't believe I'm just looking at We have Martina, but um uh Nikki is out on her annual um, August sabbatical that uh, I swear starting next August I'm going to do the same thing. So, welcome everyone to our August 19th decentralized trust graph working group meeting. We are a joint working group of Trust Over IP and the Decentralized Identity Foundation. And this is our antitrust policy notice that hopefully everyone knows well now. Um, next thing [clears throat] we always check to see do we have any new members that would like to make an introduction? I'm looking up and down the list and they all look like very familiar names. Uh, so but always checking. Um, >> [clears throat] >> you also can uh, add something to our bio page. Uh, D D D bio and have a link to that up in the chat. And uh, you can you can also put something in in the chat. Um, if there's anyone new. All right. So, um, quick review. We have actually awful lot of announcements uh, coming across but that's uh, we'll we'll do that quickly. Um, we have several task force reports um, including incredible set of notes that Brendan just put together this morning. Um, and uh, from four four task force reports here today and then thr- actually m- m- three but we just expanded to 1/4 uh, special topic uh, that uh, Kaliya um, suggested to add this morning. So, we got a lot to cover and so let's dive right into it. Um, so starting with the uh, announcements I'm quickly going to I will actually what I'm going to do is I'm going to put before I start editing this page all the links to everything we're going to be discussing are on the uh meeting notes page. So, that's a link that you can click into um as we're editing the page, you will be looking at the previous version and as soon as we add any more links or anything, we will um refresh it and [clears throat] and to refresh it yourself. So, anyway, I'll I'll quickly cover uh we have been invited uh to help find there should be an help there right there. Help plan uh San Francisco Democracy Week uh which is week of October 5th. There's a bunch of us are going to be at the um LF um Plumbers Conference uh and the Open Source um uh Summit LF Open Source Summit in Prague that week. So, um number of us will not be able to go, but if you're going to be in San Francisco or going to SF Tech Week and you want to be part of that, they have a planning call this coming Thursday and I'll put a link that link into the chat. You're welcome to join that. Um Daniel, do you want to quickly say something about your uh submission for the uh GDC 26 fight of use case promotion? >> Sure, we'd love to, yeah. Uh good to have your attention for this case because I need some votes to get visibility for this case. Uh this is a case that this submitted to the uh uh GDC via Vitis. They're going to celebrate kind of a digital trust award and we built a very nice case. It's actually built in pre-production, so it's tested, it's live. Um and it's about AI fairness and responsible AI. So, the problem with responsible AI is that the machine itself cannot be held responsible and it's not really responsible, but we can use AI responsibly and we can do that by using also our verifiable credential concept and what we do is we evaluate systems on commission or you can do it yourself on a platform and the engine identifies the right mechanisms, the right meshes and the right jurisdiction. So, it's quite complex thing, but you can then seal your readiness and your verdict into a verified credentials. And we don't stop there. We also provide a customer receipt for the one person where the decision has been made for and this person can then see together with some explainable AI capabilities what made the decision, what could have flipped the decision if the owner of the seal wants that. So, that's optional, but it gives the user, the people actually some agency back in the world where more and more systems are making the decisions. So, we'll be great to get your vote just to have some visibility and I will uh also paste link directly in the comments. Thank you. >> Yeah, I think I already did that. So, you got you have it twice. >> Fantastic. >> [laughter] >> Excellent. Thank you uh Daniel. I already voted by the way. So, >> Cool. Thank >> All good. Um all right. Um >> [clears throat] >> we will have a a special topic covering prep for global digital cooperation, but I just wanted to uh note that there are still tickets available if you're interested um in in coming or thinking last minute. Um so, um the Linux Foundation Decentralized Trust still has tickets available. Uh you can uh contact me or any any of the LFDT staff to ask about that. Um and just quick note, um First Person Project and our partners including Infinity will be hosting a first-person meet up in Geneva that afternoon before of Monday before um TV uh the >> [clears throat] >> GDC. And uh we we welcome anyone who wants to come to that. We have We're still figuring out the the venue. I'll have more details next week. And lastly, we now have a a special topic on this uh but I'm going to copy this link out here that Kalea um brought to my attention like last night. A new paper from Singapore AI Safety Hub called Designing Agent IDs. And they have a task force on that they put out this paper. And Kalea will talk more about it on our uh uh and our potential response to that uh in the uh agenda item our special topic number four. So, I'm going to stop and say, are there any other announcements anyone wants to make or we can dive into the rest of the agenda? I mean, just checking. We have a really full house today here. Ah, okay. Let's dive into it. So, we don't did not have any action items from the last meeting. I missed the last two meetings, but I checked the uh very excellent notes by the way and um um so, um and Nikki's uh uh on her sabbatical this week, so no activity uh risk and sexual harm prevention except I think Saikrishna has been doing some work there. Saikrishna, is there anything uh you want to report on that? Do we have Saikrishna today? >> Not specifically, Doman. >> Okay. >> The only The only The only thing I'll highlight is that the task force is now doing things asynchronously using GitHub and Discord. >> Ah, that's that's a good point. Uh so, anyone who who is interested in that, I know Saikrishna, you're doing um uh I like it. That uh work asynchronous in in the GitHub. Um thank you, Drummond. If you happen to have a link, um I can stick in there. I would be happy to do that in the notes if you stick that in the uh in the chat. Okay, uh Brandon, I'm going to turn it over to you uh to give the credentials task force report. And I'm going to get your notes, as many as I can, up here on the on screen. >> Sure. >> [clears throat] >> Uh thanks, Drummond. So, um yeah, just put a bunch of links in here. So, there's been some activity recently. The main thing to know is that there is now a weekly uh call um about the credential spec. Um you know, what the pace has ramped up a little bit as we're moving towards GDC and then the Linux Plumbers conference. Um So, um yes, uh that's on the the Trust over IP calendar. You can find it there. And that's an open meeting. Um if you want to join us. Um so, the the notes, I'll just go through them. They're also just listed here in the document already, but uh we had a conversation about, you know, keeping track of the credential spec versus the Trust Test spec. And they are, um you know, needing to be aligned in various ways. So, we're going to use semantic versioning, semver, uh with like a major, minor, um and a patch uh version. And so, just decided to do that. Um each uh spec will have free-floating versions. So, they're not going to be locked together. So, you know, the credential might be on 0.4 and the Trust Test might be on 0.9. But each one will reference the minimum, um you know, version of the other that is compatible. So, that's um how we're coordinating that. Just a reminder, um these are the links to the specs right now. Uh one other thing that was decided is that it's time to create a trust task. I I'm I might be scooping later later notes from the trust task task force, but they're they're going to be making a spec repo as well. So, this will help us all to to keep things aligned and clear. Uh one of the things that we're doing in Keyring, you know, our project, this is not you know, the official references of of the task force or anything like that, but we are as we're working to integrate the spec and to prove things out, we're making little references that you can see to show how things work and to confirm that they do work. For example, we've recently added one that is implementing our witnessing flow, our witness flow witnessing relationship credential formation using trust tasks. So, that you can find there and others. And so, that's just a resource that will be working as we're extending Keyring and making it more interoperable with the rest of the Open VTC DTG world. Um just linking now some things from the repos that are helpful if you want to dive into details. Um there's an issue about what you might call trust ceremonies. So, trust tasks are a little more atomic. You know, it's like one specific trust task, but what if you have a orchestration of multiple trust tasks and credentials that need to be linked together. You know, the simplest example is if you need to know that a trust task completed successfully to be able to interpret a credential. So, the credentials have a task context field in them. And we are also looking at having a digest that links to the digest of the trust task that is is referenced. So, there's a conversation about this this issue 173. We have a poll that Alberto spearheaded into the credential repo, which is aligning with what's already in place on the trust task spec that Glenn's put in there that allows this coordination to work, you know, this kind of multi-part um trust task flows. And so, it seems to be working pretty well. But again, more eyes on it, more folks reviewing it, more folks implementing it in their own flows is going to be very helpful. So, that's one of the highlights. You know, one of the recent bigger bodies of work is making sure that we can orchestrate these these multi-trust task flows and that if a credential comes out of it that you can verify that the trust task, you know, completed successfully or has the state you would expect it to have. So, I think we've made some good progress there, but your comments are very welcome. This next one here, verifiable delegation credential, that's the special topic one of the special topic Drummonds mentioned that we'll talk about in more detail below. Just wanted to put a quick note that it's important to distinguish delegation from authorization and there's some good notes about that on that pull request and then we might hear more about that today. And just also a final housekeeping thing. So, the credential task force repo, which is different than the spec repo, had an outdated version of the spec and was not pointing to the official spec repo. So, that's been cleared up. So, now hopefully all issues related to the spec are put onto to spec repo and we still have the task force repo for more general conversations. But so just wanted to you know, let folks know where to find things. So I think that's a good high-level overview. Drummond, do you want to say anything more should we take any more questions about this? >> I was going to say any any questions Riza put something in the chat talking about contributing trust task lib P2P. >> Yeah. >> Go ahead and Riza. >> Yeah, so we've been tracking what you've been doing. Thanks so much Brendan and uh being mulling over whether to fork the code. We really don't want to do that. Instead, what we want to do is contribute a rust crate that to to your code base and then we would basically integrate just at the crate level. So that means so let me explain what it is we want to contribute. We want to contribute trust tasks for lib P2P. So P2P peer-to-peer infrastructure like the one that Quai is building does not assume that any node has a public connection. So we want to establish the trust tasks between peers and that seems to be a use case that is a gap in the current code base and we would like to contribute that but then make that part of your project and we would basically just integrate at a crate level. So we would basically get that bundle that that library. Um >> [clears throat] >> And Riza, have you looked at the trust spanning protocol at all yet? >> Um, I've not yet. I think there's a there's there's a massive overlap between what you're doing Steve and so you're you're creating the theory of it. Yeah, I've >> Well, that's actually when James worked with uh Sam over at uh Kerry, so it's it's what we're building on, but it's is not in fact my work. >> Okay. So, if if it So, so I've got a document that I can I'll be able to share shortly with with you all and and maybe we're reinventing a wheel that already exists, but I've looked through the existing documentation at least under the the DT the distributed trust graph working group, and I couldn't find the the sort of functionality that we needed. >> Yeah, uh Yeah, I think that sounds really great and yeah, I I personally, just speaking only for myself, I like the idea of rather than forking, you know, contributing back when there's shared shared elements and I think I think we all agree that this is not transport specific what we're doing, you know, these these trust tasks and credentials should work over any transport and so libp2p, you know, is certainly one of important one of those. So, all of that that makes sense to me and we are another thing that's on my mind is kind of conformance tests. I think we're moving to that world where we're we're going to need some suite, you know, that can test that folks even with different implementations, you know, are producing the, you know, compatible results, but this sounds great to me, Reza, so. >> Thanks. I'll follow up with you and see Glenn's got questions here separately. I'll follow up with Glenn as well on the specifics. >> Yeah, perfect. And again, I wasn't here um last week, but I did listen to, uh, to the meeting where, uh, Eric and and Shannon were presenting their work on the P2P Trust Act that is based on P2P. Uh, so I think you have nice, uh, potential synergy there to, uh, Reza. Um, and, uh, hey, that's what we're here to do. It's It's bring all this, uh, the code and the standards together. So, um, why it's feeling like an only an hour a week for these meetings. Well, we now have The one thing I want to reinforce is we now have, uh, Tuesday morning is now three contiguous, uh, time slots for, uh, for task force, uh, meetings. Um, and and all those, by the way, are on the Trust Over IP calendar. So, the editors meeting, uh, the credential editors meeting is open to anyone who wants to attend. We had several other, uh, folks attend yesterday. Um, and then we have the ZKP meeting followed by the HTX meeting. Um, so, Tuesday mornings are, uh, deep dives on all those topics. Everyone is, uh, is invited. Um, okay. >> Can I say one last thing? Just real quick, uh, Drummond. >> One more thing. >> Okay. Real quick, um, just wanted to flag to the ZKP group, um, there are some overlapping issues that we'll want to discuss with you when we talk about doing digests over credentials. Um, and, you know, are we supporting selective disclosure? Are we not? So, uh, we're not we don't have anything for you to look at specifically, but just letting you know that that's coming up and we'll probably have to cross over there. So, anyway, thank you. >> Fantastic. Excellent. All right. I knew there was a lot to report there. Okay, next up, Trust Task Protocols. Uh, you've already mentioned that they're going to have a separate spec, but I'm going to turn that over to, uh, Glenn and Jeff. Anything they want to report? >> No, I think, uh, all that was covered in Brendan's update. >> Cool. All right, I will put the link to the the new repo that thank you Jeff for he oversaw getting that set up already. So that pattern by the way will apply to all the specs. Any any task force that says we need to do a spec because the way spec up T works is a as utilities easiest to create a dedicated repo for that and all the spec up T machinery that goes with it. So so for every every every spec we have we'll have a dedicated repo and we will keep a an index of all of them on our on our main page in the wiki and also on the on the DTD general um GitHub page so you can use either one as a as a front door to all of what we're doing. Okay, on the VDS task force formerly our cards no activity this week but I will talk a little bit in my uh summary of the New Zealand trip. There's a lot of interest in that function once people understand that BTAs can do peer-to-peer data sharing it's a they just start seeing lots of uses for that. I don't have anything to report on agent names this week if if I don't know Marcus as I don't see Marcus here so I don't think we have anything to report there. So over to our last two task force reports first ZKPs Scott Mitchell. >> Thank you. I will use my auctioneer voice in the interest of time. We had a good call yesterday. We welcomed David Condry from Writers Logic as a a new member to this group and fascinating to learn about what Writers Logic is doing. We are graduating our requirements doc to an official V0.4 that Sankarshan produced and I'm coordinating with him live to get a pull request together. Our verification registry got its full uh uh first full external run from Glenn Gore. So, thank you for that, Glenn. He ran all three circuits himself. They all passed and filed a ratified position. Um, great independent validation. We'd love to get more folks uh digging into it as well as we keep going. Uh Dennis demoed a probabilistic trust prototype, which was really cool to see proving credential connections and issue of verification in about 6 seconds on device across thousands of credentials. Um the the cross task force streams we're tracking. We had one from Jeff Turk about identity linkage. Um, and then the one we just heard from Brendan. We're here for it. We're excited for these cross task force syncs whenever the time is right so we can stay aligned. Uh looks like we're getting closer to getting the Berkeley team involved. Drummond is going to give us a direct intro um myself and Mitchell to Dr. Sanjam and and uh Garg and Hart. Um Yeah, and we are being very intentional. Drummond raised a great point yesterday. This task force is dealing in very esoteric subject matter. So, we're taking it very seriously to uh lean into plain language explainers and an AI first approach. Um and we'll keep ourselves honest on that. So, if anyone says what you're doing is really confusing, I don't get it, we'll try to make it better. >> Absolutely. Thank you for uh for mentioning that. It it I attended that call yesterday and boy, it's it's it's like you know, suddenly waiting into a uh uh uh you know, an ocean beach and realizing, "Whoa, this gets really steep really fast." >> [laughter] >> But, it's it's absolutely necessary and I really really uh am glad that that uh task force is up and going. And I I do have the action item uh Scott to make that introduction. Um I looked at it at midnight last night and said, "Oh, I'm going to do that in the morning." So, I'll get that done shortly after this call. >> Thank you. >> All right. And lastly, uh HTX. Um and I know Margie's on vacation this week and next before she goes to the GDC. So, that's over to you, Daniel. >> Thank you again. Uh can you uh give me the right to share? >> Yes, indeed. >> quickly there. Oh, hell. >> Sharing it now. Go for it. >> you. Try that. >> [gasps] >> Um um um um um um. So, I hope you can see my screen. Yes, I guess you can. >> Yes, indeed. >> So, just want to give you a a quick insight in in what we do and how we do it. And this is also the company's this Margie's work. So, I don't want to take a kudos from her. So, I would rather go into the methodology quickly very quickly. So, what we do here now is actually apply a a standard human-centered experience analysis approach. We derived a human trust experience design framework for that from that. And it's basically a structured approach how we look at at what's happening and how we derive our deliverables from that. And I guess this picture I'm showing currently shows everything on one card. So, it's it's a process that applies different lens uh lenses to the same topic. So, we we choose a scenario and then we we look from a persona perspective on it. We look We look the into the scenario in detail. We look at journey mapping, empathy mapping, and so on. So, there are different lenses uh that there is the person human, the journey, or the interaction itself, which is uh the topic of the analysis mainly. And in the end, that's the main goal we derived in our design requirements from that. And that's also where our trust experience best practices uh collection could help them to map these moments of truth you identify uh to to best practice in this area to find the better solution. So, this is basically what we do and just to give you without going into the content, we don't have that time, but but how this looks like. So, it's really an iterative very heuristic approach and the benefit is that we identify the important and and maybe the the interesting moments while playing through the this whole uh scenarios and um it it starts with uh the number one step here so with the persona itself so that we can be more empathic and and go through this entire step until uh step nine, not step 10 yet. And when we have uh a little bit issues to understand how it really works, then we have to dig deeper and uh play around and just uh uh play through the entire action sequence and that's also where we get or where we can validate our assumptions because we cannot yet observe things, but we can do some interviews and talk with you about that and you're also happy of course to reread all the time this Excalidraw drawing is available uh via link from our page or in the comments. I going to paste that as well. So, uh you can always leave a note in here. It's open. So, that's it from my side. Thank you. >> Excellent. Uh and thank you for the link. Um that uh um that HTX call yesterday was just fabulous. There is going to be so much goodness coming out of that. Uh which I will emphasize when we get to our next Oh, I guess we're there. Um all right, I'm going to share my screen again and we'll go back to um our reports. Uh let's see. Here we go. All right. And oh, Zoom has learned to actually uh replace things in the right place. Oh, this great. Okay, so um perfect timing. We're uh halfway through. We have uh again four special topics to try and cover in the next uh half an hour. So, I will keep in mind as uh as as um concise as I can, although I could go on for hours about the uh the learnings of my trip to uh New Zealand um and and actually the very last day to uh uh I spent 36 hours in Melbourne um and had a fantastic conversation with two long-time Trust over IP contributors there, John Phillips and Joe Spencer um and we recorded a podcast there that um actually I'll I'll find a link to and put uh uh put in there, but it you know, it's sort of summarized all of the work we're doing and uh and the things I learned on my trip to New Zealand. The number one thing I learned was there is tremendous interest in uh verifiable trust infrastructure that we are creating here with the decentralized trust graph. Um I will share the uh the the link in there, too. The uh presentation I gave that for specific reasons because they wanted me to talk about uh they invited me uh uh in the in the keynote of the their annual conference for Digital Identity New Zealand to talk about sovereign AI. Um that's Oh, uh Sai Krishna, fantastic. Thank you. You've already got the link to the uh that podcast. Anyway, they wanted me to talk about sovereign AI, so I uh I I I shared uh uh I I ended up calling the version of the Trust over IP stack that we're creating here with uh decentralized trust graph the sovereign stack and I'm realizing that um there was I wanted to show one specific um I just have to get it up in another window here. And let me close this for a second. I want to show it to you. And so I will actually stop sharing and switch over to show this. Um mm okay, here we go. Start here and I'm going to go into slideshow mode because this is a build. Uh okay, are you seeing this now? >> Yes. >> Okay, good. So um again, I I tried to capture this in one screen and here's the quick narrative that I gave to it. So the sovereign stack the heart of it is of course tree uh keys um um you know, public private keys and algorithms that can be strong enough to course be quantum resistant even though that does affect the storage size and everything else. Uh in any case, the next next key uh uh component verifiable identifiers from the term that we use here at TrustOver IP and in the TSP spec because it it covers both DIDs and um autonomic AIDs which is uh the the term that Keri uses. Um and of course the next thing after that is verifiable credentials. And uh what what explained because I you know, one of the reasons they invited me down there was because I'm known for um the uh uh the book that I did on on these components uh being self-sovereign identity um and really establishing the the the the the essential tool necessary for for wielding uh these things as digital wallet. And my key message uh you know, you know, everyone there were actually a couple people that had the book uh uh down there and it was like, oh yeah, this is great and the group I I first addressed was um, uh, called the, uh, trusted credential adoption working group of Digital Identity New Zealand. And they're bringing six credentials to market. They've been working on it for, um, like 9 months now after they passed, uh, a, uh, a governance framework for New Zealand. Um, that authorizes, you know, all of this stuff. One of them is a digital driver's license. One is an age over 18 credential. One is a government access credential. Um, and there there are a couple other more more specialized licensing credentials, but they're coming from different issuers. They've got a very healthy ecosystem. Um, and so they they really get this paradigm. And they were really interested in uh what I'm what was I bringing? I was saying, hey, we can use all that, leverage all that, but when you add, uh, when you wrap those wallets in a verifiable trust agent, and then you enable those to form verifiable trust communities, and those to form verifiable trust networks, well, then you have verifiable trust infrastructure. Uh, uh, thank you, Glenn, for for for, uh, you know, um, arriving at that at title. And now you actually got infrastructure that can deal with digital agents of all kinds, of course, including AI agents. I was stunned at how they were just like, whoa, this is really, you know, they just it didn't take much to, you know, explain the advantages of doing this. Uh, I followed it with our typical explanation of how the, uh, DTG works. And they were very interested, uh, in establishing, um, not just VTCs, but but potentially separate VTNs in New Zealand. Um, and in particular, I don't have time to go into detail. Um, there's a it it New Zealand is home of um the the Maori people uh um that is uh one of the most um uh There's a good quote for it. Anyway, they're just one of the most influential and and and well-recognized indigenous peoples in the world. Uh they tend to lead out uh on on on issues involving indigenous rights all the way back to their uh Treaty of Waitangi, um which is the basis for New Zealand government. >> [clears throat] >> That's still, I guess, very controversial, more so than the US Constitution in in in the United States. Um, but a key facet of that from the Maori standpoint is uh governance of peers that are all sovereign. And so, they saw here the potential to have multiple VTNs in in New Zealand, which are peers that do not depend on each other, but are totally interoperable by virtue of the standards we're working on. Um, so uh I'm not making that up. There were Maori leaders that were part of all these conversations. I talked to you directly, and they were just uh just way more interested than I had any reason to expect. Um and uh so, we are we're following up with them. I have another call with them this afternoon. Um, it was it was really exciting. Um I then uh in Melbourne uh uh uh uh with was there and and was explaining this again to uh John and Joe at um Sezoo is the name of their uh um uh digital consultancy. It's part of something called 460 Degrees, where we had an afternoon uh um event, and they were saying, "Well, that's fantastic. What about Australia? What about the provinces in Australia? What about Wellington as a city? Um, isn't a city uh doesn't have so many VTCs that it that it that every any any large city could be a VTN? And I was like, well, yeah. So, I have a follow-up call with them later this week as well. So, that's my my main message is there is enormous interest in our infrastructure and uh some folks are already following up at GDC coming up later this week. So, I think that's all I'm happy to take any questions or have follow-on calls with anyone who's interested in this. So, I will stop sharing that and uh just get back to uh our main our main screen. So, uh And well, yeah, I I want to keep my under 10 minutes to get it over to the next thing. So, is there any uh Oh, uh anyone from New Zealand going to GDC? Uh Andy Higgs, who is the executive director of Digital Identity New Zealand, really wanted to go, but I don't think he's got the uh the budget uh to be able to do that. He was going to investigate to see if anyone else from their group. I have a call with him uh later today, so I will get the answer to that. Um Uh I I I really hope that would be the case. But in any case, he he asked me to represent what they were doing um there at uh uh at GDC. Uh how can I move that window? Sorry, I'm going to share uh my screen again, but I don't want to Oh, I see what I'm doing. Uh Yeah, okay, sorry. I had to get back to this. Um okay, here we are. Um All right, I got I have to open the chat to see are there any other questions about that or should we jump on to the next? Uh hello. I can't get the chat up. There's my Zoom is stuck for some reason not wanting to let me see the chat. How's that? I'm going to stop sharing for a second so I can re-share and see if I can get it to work. Oh, fascinating. It's just not letting me see the chat. All right. Well, um Glenn, over to you. Maybe I don't know if you want to share then I could use the chat again. >> Uh, I'll just put the link in. >> I can also uh click over to show the uh um your pull request if you want me to do that. >> Yeah, that might be easier on your screen than mine. So, I know last week we talked a little bit into for the last couple of weeks, I think delegation has come up uh a few different times in different strategies of uh what to do with it, but on the implementation side we've run into limitations of what we can express via the existing uh decentralized trust graph credentials uh around delegation. So, we're proposing that there is a new DTG credential type, which is a verifiable delegation credential. Uh I will say that there is likely to be closely linked with another credential, which uh just for ease will name a verifiable authorization credential, and the two kind of go together uh as you go through this, but um the difference is when you delegate, you're giving permission to the delegated authority to act on your behalf. Um whereas when you give authority, you have permission to uh access is kind of how you think about it. So, um having authority to access my email uh doesn't mean you have the authority the delegation to send an email on my behalf as you go through this. This is important for parent-child relationships, you know, um elderly individuals, those that for whatever reason can't give informed consent. They may be illiterate, semi-literate, um they may be uh in a coma, lots of different reasons in daily life while each of us might not be able to give fully informed consent and may want to give delegation at a very specific level as we go through this. Um so, it's quite a complex topic when you kind of think about it. Uh and there's some notes in here already around you does it need to relate to a relationship credential you kind of make sense in a way that if you're delegating authority, there's probably a relationship there, but should it be expressly called out? Um we're proposing a new credential type instead of overusing one. We did try actually to use the endorsement credential type to uh carry delegation, but we ran straight into the issue of you endorsement saying I don't know uh Scott is awesome at ZKP is not the same as giving Scott delegation rights to uh a GitHub repo on ZKP, for example. Um and so that's where it was coming up in the Linux Foundation work was as we start thinking about maintainers versus contributors, if a maintainer is away, they may actually delegate their maintainer authority to somebody as a trial uh of them progressing from being a contributor to a maintainer, for example, uh as we go through this. So, there's a lot of detail in here. I'm not going to go through it all. Um but, you know, kind of open questions are is it the right word, uh delegation? Uh are there do we need to define the authority credential at the same time? Um how do you define scope of delegation within this? Uh we do believe that the delegation credential in particular has a very strong requirement for ZKP constructions because you may not want to disclose to everyone what exact delegation you're giving uh between you and the recipient uh where there may be others that are witnessing that uh along the way as well. So, this will be part of, you know, a lot of conversations on the DTG working group uh as well as kind of trust tasks as well cuz they do believe together. Um but, if you want to know more, either book time or feel free to respond to the PR or talk to myself or Brendan. Um happy to answer any questions on this one. >> This is a good, big, deep, ripe topic. Um when it came up on the editor's call uh yesterday, I actually said, "Ah, I'm going to note this as a milestone that we finally uh are going to go down the delegation rabbit hole." Um which, frankly, I don't think we ever could have avoided. I'm surprised it's has taken this long. Any uh any questions uh for for uh Glenn just a high-level about this PR? Um go ahead, Erica. >> have what I put in chat. Um >> Oh, Steve. Okay, go ahead, Steve, and then Erica. >> I in uh my proposal, I I uh had the idea for a a multi-hop delegation record. And so basically, I would think that these individual credentials would live inside this sort of multi-hop record, which would then be digested as a whole. So, you would have a verifiable provenance. Have you considered about multi-hop? >> Uh I probably need to understand what exactly what you mean by a multi-hop, but >> whoever you delegated something to delegates to somebody else, and they delegate >> Ah, okay. So, there is there is uh we do have a concept of that, and there's actually a uh like a max depth of how how far away can you delegate from the original person as you go through this. But again, I think if you've got thoughts on that, Steve, happy to dive into it cuz I think that's one of the areas that's uh ripe hazard for sure. >> [laughter] >> It was uh uh Sankaran wrote this piece on it, which was, you know, informed my design. So, anyway, okay, back to you in in just book something then. All right, thanks. >> Thank you. Excellent. Next in the queue is Erica. >> Yeah, thanks. I put a a comment in there, and I I think it kind of touches on what Steve was just talking about. I've been also trying to deal with delegation in our stack, and thinking of it as a uh verifiable data structure, and not a credential. Um but yeah, I guess I I really thought about making it a new credential. Um but I also I'm not ready to give up on it potentially being a just a data structure. Um >> [clears throat] >> is there is there a space for a conversation about comparing those two approaches? >> Uh yeah, I mean, would love to understand you kind of have VDS versus VDC. Um >> Yeah, yeah. I I I mean, I know VDS is Yeah, they're not really graph claims, and that's the thing. Um but yeah, okay. I just want to throw that out there that that we're taking a kind of a different path. We may end up at the same place. >> [laughter] >> Yeah, I mean, it could be that the VDS goes into the VDC, and the VDC is just a representation on the graph. For example. >> Yeah, yeah, exactly. Yeah. >> Cool. I'd love to join that because I do think the two things are are closely uh uh related. Um Uh anyway, over to Grace. >> Yeah, just really quickly, our um we have a task force under the trusted agentic AI uh working group, and they they're actually in um close to draft uh 1.0 of a review of the different types of technologies and how they think about delegation and delegated trust, and you guys were just talking about uh the difference between delegation and and Anyway, it's a really good piece of work. I don't have the link on me cuz I'm not I'm not in the office right now, but um if you look at the trusted agentic AI working group and under the delegated trust task force, it's almost a book. We We're still in the point where we would love anybody to review it, and there are a couple of missing chapters, but basically, we've gone over a lot of the technologies that do this and the confused deputy problem, which is uh what you guys were just referring to. So, that's just something to look at, and uh it's a good reference material. >> Grace, can you uh can you shoot me a link to that when you find it? Cuz >> Um yes, I can shoot you a link to that. >> Yeah, uh just cuz it'd be good to just see if we can link to that. >> Got it. And I And when when that eventually surfaces, let me know, and I will add it to the uh to the notes here on this page because uh I yeah, I've been aware um you know, there's there's a that's a very active working group at DIF and uh that we were going to overlap at at some point uh Andrew Castleman who helped start that um felt the same way. So, that's good stuff. Thanks, Grace, for pointing that out. Uh I'll just add note here. When we get to it, okay. Um any other We have We have 10 minutes left or 9 minutes and two more things to go over here. Um So, yeah, let's let's uh keep going here. We've got uh GDC coming up uh in 2 weeks. Um what I've what you see on your screen is um since they they published the schedule of talks, um there are four directly um you know, that we are directly part of giving that are that are that are tied, you know, um to uh to our work. Um those are the first four listed here. And then we're also included speakers in three more um that are more of general, you know, discussions uh that cover our space and they want us to be a you know, a panelist on those. Um uh I think the the main question Glenn has raised and it's a it's a really good one, which is um a lot a lot of the focus there will be um on the uh it's it's a global conference. So, not the whole thing will be even though it's hosted by Switzerland in Geneva, uh it won't be just about uh eIDAS 2 and the European digital identity wallets uh uh work. Um it really does, you know, have as much focus on on the you know, initiatives uh and and approaches around the world. Um it's one of the reasons Linux Foundation uh as a group is is is so involved. Um uh so, anyway, Glenn, you you brought up with me a really good issue, which is alignment between verifiable trust infrastructure and the ideas. Um, and uh I I do you have any specific ideas for how um we we want to uh you know, follow up on that discussion or particular um um folks or parties that we should uh be trying to open that discussion with? >> Uh ideally trying to find people who either directly involved in the standards of that or can influence or open doors cuz while I does is good VTI while it's aligned and can plug into it it feels like EU has missed some important uh steps in this that the VTI is actually solving and so it'd be good to say, "Hey, you've got something that's almost 80% there, but architecturally you you've kind of skipped over and it's a bit hand-wavy. This could fill in some of the blanks for the EU digital identity uh experience. So, I'm just talking this, you know, a little more bluntly and directly, but we need to frame that a bit better. Um, but for example they have a philosophy of it's per device completely locked down, never get your keys. So, if you ever need to move across devices or you have multiple devices, it's effectively a you know, each one is a walled garden and that does not work in the kind of world that we're thinking of with VTI. Um, definitely doesn't work in a world with AI. Um that, you know, kind of having delegated access to things on your behalf as you go through it. So, it'd be good just to kind of ask them how are they thinking about this? Do they see, you know, I would explain VTI to them uh Drummond the exact slide that you showed before of the the onion layers and then basically are they philosophically opposed to this or do they see that this could work and actually help drive adoption of the EU standards globally into different communities that are going to come up? >> 100% agree and be very happy to you know do everything we can there. Um, I think we'll we we can strategize further on that. in our pre-meetings uh and but I'm already already trying to plant seeds with a few other people that I know are going uh and we can do more of that in in our uh pre-meet on the afternoon of Monday the 31st for anyone who's going to be there then. Some people are arriving you know that evening or or the or the next morning. um um Glenn you also um just just for anyone who who wants to come to that um the you and Jeff have mentioned doing a uh uh a a deep technical rehearsal of uh of the VTI um setup that you you plan you know to be showing at the Linux Plumbers Conference. Um, is that still something that uh uh you the two of you think you want to tackle that that afternoon? >> Uh yes for probably half of the plumbers I would want to one do a user experience test, two would like to do some scale testing. So, the more people we can get and uh the faster you can create persona DIDs at the click of a key on your keyboard and just basically try and break it. Um, we would want to just do some bug smashing cuz uh only 2 weeks after GDC we're going to be in a room full of highly technical people who are we're to do everything they can to throw things at it. So, uh it's an important milestone just to get as many hands, as many different keyboards, operating systems, uh usability in there and just see what the experience is actually like. It's going to be horrendous. I'm looking forward to it. >> [laughter] >> All right. Um I will, too. So, again, we'll get more details out that we'll get the space uh uh reserved to get more details out. Uh we'll probably uh I'll I'll share it as soon as we have it. You don't have to wait till next week, but we'll certainly have it by next week. >> [laughter] >> All right. Uh unless there are any questions about that, we will have in the final 4 minutes uh Kalia, do you want to talk about this? Uh I could I could pop open this PDF if it would help. Uh but do you want to talk about this paper and and what you're thinking would be ideal for us to produce? >> Sure. Um I [clears throat] was contacted by the folks who wrote this paper last week. Um I calmly sat down and read it yesterday and was very impressed with its breadth and the questions it was asking. Um they, you know, sort of asked who has a compre- like, is there anybody else doing a comprehensive map of all the options out there? And I said the answer is no. And then I replied to them and mentioned the agent names working group. >> [clears throat] >> Which is sort of what I had gleaned is happening in in coming out of this cluster a few months [laughter] working on all this stuff. And Drummond was like, "Oh, no, no, there's a whole idea we have about how agents uh and people connect." And I said, "What can I point them at?" Because these folks are clearly with, you know, two you know, reasonably together governments, Singapore and Korea trying to make sense of all this stuff. It would be good to point them in what we're doing, and they And German said, "We don't have a paper." And I said, "Oh, well, how How can I tell them about what you're doing if there isn't a thing to point to that sort of paints the picture about how agents fit in?" So, you know, it's it was you know, is there a short It doesn't have to be a long paper, but something that the folks who wrote this paper could receive and understand what you all are doing relative to agents and humans and how it fits into this landscape that they're doing a reasonable job mapping. >> Yeah, and I want to reinforce the as Kulia pointed me at it, I I quickly read the paper last night. And uh it's not very long. This This one you put the link in the chat. Uh and this is a table it has, which is pretty I've been, you know, keeping track, not not my own list, but um by my I I think there are a few additional protocol um proposals I've seen that don't have the visibility to have made this list. Um but but the solution we effectively are proposing, uh again, the the title of the talk we have a GDC, the the one uh the first one we submitted is decentralized trust graph uh a unified trust infrastructure for the internet of humans and internet of agents. And so, we we do have, uh as I I saw the thumbs up from Glenn, a pretty comprehensive answer, um but we haven't documented it in a in a paper that would be, you know, something this group or anyone else would say, "Oh, we understand what you're proposing and how it would compare to some of these others." Um and and and her point is a good one. Um I knew my initial reaction was, "Ah, I'm struggling with all the other the I'm trying to get written right now." And I thought, wait a minute, there's a whole bunch of us here and and some people here that have some very powerful agents working with them. Uh so, who is there the one this call with a simple is there anyone you know, now well, Glenn's already um um volunteering. So, is there anyone else who would like to join uh Glenn's effort to produce such a paper um you know, in the relatively near future? >> I mean, I I I don't understand really the scope of what they're asking for because uh I'm doing more high-level stuff over in the agents group uh which meets uh tomorrow um but yeah, the uh so, I I I don't know yet, but I'll look into it. >> Cool. All right. Um >> I'll put my hand up for that one as well. Okay. >> Excellent. >> That's some agent major energy into the paper. >> There you go. >> They're they're they're basically just asking, what's what are pragmatic solutions that exist to solve this problem? That's the short of they're asking for. >> I was going to I know we're at time, but but I was going to ask Glenn because this is coming from uh Singapore, were you aware of this uh >> Yes. Yeah, yes, we are. Yeah. Yeah. >> All right. So, so this isn't isn't news to you. All right. That would be fantastic. Thank you, Kalia, for uh bringing this up. Um she has the broadest scope of everything happening in the space. Um so, really appreciate that. Uh and with that, I think we're done. Um amazingly, I actually got through it all. Um just heads up, we'll have a normal meeting next week, but the following week is GDC and I'm proposing that we have a meeting-free week that week because a lot of us will be tied up there. Um so, obviously any task force that decides it still wants to meet that week can do that, but otherwise, if you're notify us if you're planning to do that, otherwise, we're going to uh tell uh um the LFTT folks to uh clear the calendar for that week. Yeah, as as Sankara is is suggesting. So, I'm going to ask you right now, any task force who is still planning to meet that week, let me know, otherwise I'm going to say just uh wipe the meetings for that week so no one's confused. And uh we will see you all next week or at the next next task force meeting on Tuesday next Tuesday morning, my time. Thank you all. >> Mhm.