DTGWG Human Trust Experience Task Force Meeting - 2026/09/15

Hello, Nikki. >> Oh, hi Margie. How are you? >> Good. How are you doing? >> Yeah, all good, thanks. I'm just trying to uh log into the wiki because I I cleared all my cookies a little while ago. So, I'll be straight back as soon as I'm logged in. Okay. Did you have a nice break? >> Yeah, I had a very nice break. It's almost was too I mean it feels like it was such a break because it was so long that [laughter] I'm I'm having re-entry problems. >> Yeah, [laughter] re-entry problems. >> Stay at home uh camping week. >> Yeah. No, it was good. It was uh it was restful and a change. >> Obviously, not as much as usual. It wasn't the kind complete detox that we normally have, but it was it was really good to just uh rest my brain for a little while. >> That's good. So, you need we need to recharge especially especially this season. >> I think it's it's it's space. It's about space and just letting in a bit of light and air and perspective. [laughter] Yeah, we took a hike in um we took a really challenging hike that my kids uh dreamed up in um and wow that was that was both space and effort and endurance. It was crazy about killed us. Where did you >> hike? >> We went to the Doommites and did trail called the Alta Villa one and it's just >> Wow. >> straight uphill for a lot of it. It was really, really hard. >> Well, congratulations on completing the hike and returning. >> Thank you. Yes. felt like I just needed to lay down for a week after that [laughter] >> in a darkened room. Yeah, >> exactly. Or like in I went straight to uh swimming like swimming in the ocean and that was my remedy. >> Hi others. >> Hello. >> Hi. Morning. Okay. So, we'll just uh do the normal antitrust policy notice. Um, [clears throat] so just be mindful of antitrust and competition law. Um, and it's important that we adhere to the agenda and only kind of participate if we're members of the relevant organizations. So I don't think there's anyone new here today. So we'll scoot through that bit. Uh, just quickly whiz through the agenda and then we'll come back to general updates. So, um Daniel will introduce uh yet another enormous and uh rich piece of work um for us to have a think about um and then provide an update on trustlets and the best pract practice repository and then we'll take a look at the uh persona management that Glenn asked us to take a look at uh last week which is in this GitHub discussion um and uh some updates. So I'll I'll leave that uh um so that uh I can just uh do my normal kind of guiding the discussion [clears throat] to come to some outcomes and then we'll wrap up. Is there anything else that anyone would like to add to the agenda or change on the agenda? >> Only that I'm unfortunately have to leave 10 minutes earlier to catch a train. So, sorry about that already. [laughter] >> No worries. Um, well, your stuff's up front in the discussion, so that's good. Uh, great. Any general updates from folks. Okay. Uh over to you Daniel. Do you want to share screens? Yes. Or >> Yes, please if possible. And >> of course, >> I really tried to make that short. Although I could talk for hours about it. Um, let me quickly share the right thing. Yeah. So, uh, I hope you can Can you see uh the screen I'm sharing? >> Yeah, >> I guess that works. Good. [clears throat] So uh what what I want to do is just to introduce something that is very close to my heart and has been a topic of research for at least the last 15 years. And and it it happens now that I had this discussion with this other researcher of the GDC uh which um realized that they have been looking from a security perspective on a similar topic uh and the topic that is very relevant and it basically means now given the fact that everyone will have a wallet very soon. Um so what is actually missing and there are a lot of pieces missing and I really recommend you to uh look at their paper just released in June u that uh assesses uh this uh risk of oversharing uh in the uh future EU digital identity wallet uh and they have a an empirical analysis on how that's going to happen and what is needed and the interesting part is what is needed and that's actually what I'm also been looking into for years and this is this layer for agency. So a little bit something that helps the user also to uh make this disclosures and not just press consent because yet another consent button is not digital agency and doesn't really help anyone because it's more or less the status quo just in an automated way. So the VTA runs into a risk just to be a delegated constant machine which is not what we actually need to design because it doesn't solve the real issue. And the real issue and that's for for many a little bit hard to understand but quite simple is that we are not really rational when it comes up to sharing data. So we have this horistic behavior. We have these biases and we we are bounded in our rationality and secondly we our mental model will be completely overwhelmed uh with selective disclosure and the technology behind so as the chop no one wants to know what it is and and also shouldn't because we wouldn't understand our mental model doesn't work like that. So why not think of a layer in between that uses uh all available information and maybe a little bit more to actually support the user in this process. And I'm now sharing also with you already in in our HDX page here uh the uh the paper more or less. It's not an official paper. It's Rodri a concept paper that um looks uh at both findings from my research and also from the research of this ETH team and derives actually then a few blocks of requirements that we should consider and I know it's it's a big topic it's probably not priority number one and it's not nearly close to priorit prioritized compared to the pumpers conference but may if you opt to that if if we have a little bit of capac capacity, we really should use the opportunity to uh to to level up this VTA that we're planning to build with this kind of capability. Uh that would then really help us um in supporting I guess the user. So I I won't really want to go very quickly through the basic points so that you hopefully see what I mean. So uh we are talking about a uh a layer and I called it now it's all under development but I would find a good name is this trust of disclosure agency layer because it's exactly that. So our our support layer should uh enforce an or enable agency and it needs to be trust aware because this is based on research for uh under in environments of low trust. Uh and that is basically what changes the game and I will come to that. That's a very important thing. It's not only a nice sentence. It's important that we talk about games here. So the evidence is quite sure now. We all know about the privacy paradox. We all know about the data collection dilemma that companies want and also need to at some point collect data. But we also need to minimize data from the legal perspective. So there is a huge uh warfield there and there are also numbers. So this uh effect of oversharing is going to happen in a status quo with these wallets. So that's what uh these research colleagues try to find out in experiments and with mockups and in interaction with users. Uh and they found that this number is very high as we assumed from the privacy paradox. They think that about 20% would even provide an ID, an official ID to a website, a news website. And that is classical oversharing because no one should have to do that. But they would. And they also found out that with this kind of notching layer, we can reduce this oversharing and that by by quite a high factor. And now my part of research is is rather from the information as symmetry perspective and a little bit of game theory. And the good news is that uh if you look at analytical model for something like that so mathematical model then the uh the effect that we need to introduce doesn't need to be really big but the effect is clear. We need to have something in there that enables cooperation or supports cooperation. And this is then being done by adding a certain value for future interactions. So we we should stop looking it at a one-off game. It should be uh it is definitely an iterative game. You will not share data once. you will will maybe share data in different iterations a different amount and you will maybe change the view of the world. So once you have seen that a partner is not cooperating you shouldn't share anymore. So you have to exit the game and cooperation as I said of course is enforced if uh the future value of the interaction is there and you might not think ah it must be huge value so a re a huge discounted cash flow like for companies going public [clears throat] and that's not true because the model already shows that a discounts factor of one/ird meaning 0.33 or 33% is already enough to motivate everyone to cooperate also the future. So imagine the discount factors like an interest rate. So you can discount future profits by 33%. That's a huge interest rate. You you barely look at future revenues, but it's still worth cooperating. And that is a good finding. So these models uh or or what we do has a great chance to succeed. I will show you a little bit more how we get to these numbers. But then now first what is it? So we have this identity and trust infrastructure of course um this will be all these kind of wallets uh and and also together with trust registry. So the the whole VTI structure probably and the human being of course that's why we call it the first person project and this VTA is now the key to for keys credential and proof. So it's taking away the work that human beings should do. It's a technical thing right now for key management and so on. And all it also decides more or less uh how or to share data but now the missing part is this layer this decision layer that supports the user in making decision how much to share and how to behave in a future interactions and this is a bit the mind shift. So it's a dynamic component and so there are a lot of capabilities that we define now some of them origin actually from the paper of the colleagues but I have also my own and together I think the picture is quite nice because they also mentioned the missing part exactly that that we can cover that we have identified this these are here for example the native principles so it's quite a list of capabilities but the good is we can derive requirements from there. It's still a big thing to do and not everything is clear how to do it but there are a lot of good ideas around and we can operationalize this uh very well into code I guess but it takes time but I guess it's worth the effort. So basically it's a huge important to understand the situation. So our VTA needs to be able to reason about the situation otherwise it doesn't make sense. It must observe the request. It must know the principle and with the principle I mean the human being the one that the agent is care caring about. It needs also to verify the rying party as good as it can evaluate commitments and also understand and explanation and explain what what's happening and then the decision part. So there are strategies. So the we need to be able to assess a minimal best situation where we already know mathematically where it is about. So this can also be programmable. Uh the least revealing proof. Um then select the strategy state. That's a very good one I guess because it's about strategy. It's not one strategy I will show you. There are more parts of strategies that you need to apply to play the game and so on. So there is of course more and the interesting part is the feedback loop as usual. So if you can learn from an interaction and adapt the state of your world afterwards you can make this system even better and in the end or in the middle even here it's still the human that decides. So I for the sake of time I will skip a little bit the game theoretical things but game theory is a wonderful thing to u build mathematical models and illustrate the the mechanisms of the game and maybe there is one important thing to understand. We we always tend to think about the prisoners dilemma when it is about data and this is it's just also a bias. If you look what uh Mr. Trump has been posting yesterday there is has been a an ice post about this discussion about AI and whether it's a a threat for humanity and everything he can think about is that they only need one guidance guardrail and that is an intelligent president and of course they have that so we are all fine and it also mentions that he needs to be ahead of the Chinese and so on the usual game and and unfort it's also a position that Palanteer shows that's easy it's understandable because if you have adversaries you have to move but you are playing the wrong game you're thinking of the prisoners dilemma and this is uh where uh everyone doesn't know how the other one reacts and you move uh in the same time uh and the the bad thing with the prisoners dilemma coming from the inmates if you confess or get out and so on is that it will not lead to a good uh utility. So a good outcome because there is the Nash equilibrium as we all know and that will force us to a superior result because in the end of a prison dilemma no one cooperates and we are missing the opportunity of everything. But the thing that maybe Trump makes a mistake is it's not a prisoner's dilemma because our game is a different one and it's a trust game and the trust game is also very popular in game theater for years. The difference is that we have the iterations we have vulnerability once data is shared you cannot take it back it's gone. So this damage is there there is reputation and so on. So we have a lot of more signals than Trump is thinking about. So it's not about the Chinese are bad. It's not the one signal. There is a lot of social signals in there that we need to use in a trust game. And if you model that as a trust game and that's here only an abstraction, then you will find out how these payoff matrixes can really look like. And the good thing is also you can derive uh equations and see how you can steer the system. And you can't steer the system. You can lower a GB cheer the extraction gain. You can raise Q when Q is uh the uh the observability maybe or the audit throughist. You can raise different uh amounts or also different I say levers in the equation also alpha for example and that is the noise. There is always noise. You have to account for that. But the worst thing to do is just to take noise and then exit or uh or take that too seriously. You have also to be able to have something like forgiveness in a real play in a trust game and that will increase your expected utility for both sides dramatically. So the last picture here, there's not one strategy when you start the game. uh you don't just share everything you need but you have the opportunity to share just something and hold you have the opportunity to expand. So once you have more trust in the relying party, you can share maybe a little bit more for the benefit of both. You have or need to be able to reduce once you see that the likelihood or the risk to for a defect from the other party is too big or also if if if just the uh the the system tries to get information that is not accurate. And you also need to be able to recover or to forgive so that you can expand again in order to reach the utility maximum. And these are the just the mathematical equations for all these strategies. And they're pretty easy look look very bad but actually they're gentle because we know that the determinants we know these what we have to change. we only need to find a way how to programmatically steer and change them. But with these uh mathematical model and the right mindset and the right capabilities in our VTA, we might be even able to implement such a system and that would be something really new where I think we could make really a difference then. So that's it in a nutshell. What do you think? >> Well, this is a a very short section. Um, [clears throat] sort of trailblazing. It's like an advert for a longer session. I've got a list of stuff I know that I uh from reading the document and uh >> kind of connections and um interesting factors. I I you know it's it's great work I guess for this group the the human trust experience task force uh what would you imagine the deliverable might be that we would produce um or is is there recommend you know what I guess what are what are you looking for from this group would be my >> primary question but it is uh you know it picks up on the key human problems. Um there are many other factors involved in it. Um [clears throat] but I you know I think we could give it a go. the the VTA. I noticed you said that you'd been working on it for a number of years and I wondered what you know when was that kind of light bulb moment of aha it's the VTA that can enable this stuff to happen? [laughter] >> Yes indeed. So because it has always been a very formal uh kind of work because there there was never really a chance to implement that because you never have critical masses. You you can only do what actually the colleagues did and this is experiments but experiments are not interesting I would say because of all these this round bounded rationality issues and also the the privacy part. So people will will not tell you the truth if he asked him actually. So that's why it was very theoretical. uh but I now see the opportunity that something can actually be built or it should now actually be built and that's why it's it it has some new energy I would say and it's also the outcome of of the study they did is that that there is really something unsure and unresolved before also the AU wallet goes out because that's that's a real risk and also the Switzerland has raised this risk that that we now have then an infrastructure and wallets but we have not solved the problem of oversharing yet. So >> yeah I mean it is an actual issue. Yeah. When we wrote the harms paper, one of the key risks that we identified was um like oversharing on the consumer side but over asking and over verification um which is where the requirement to verify the verifier comes from the mutual authentication verification and that's part one of the kind core driving requirements behind the trust um registry query protocol and the whole trust registry infrastructure. >> Um >> any anyone else got any thoughts or comments for for Daniel? And is this something that we'd like to dedicate more time in a future task force meeting to discuss? Margie Yeah, I think we definitely should discuss it more. Um, I have two uh thoughts. One is well um one is that the the diagram um O2 one loop 15 capabilities. I think that would really benefit from walking through a real world scenario. you know, just making that very concrete as like a yeah, >> in a story effectively that that would make it maybe more digestible. I think I think I got it, but it would be helpful. >> There see something similar like that in in the the paper. It's only it's 50 pages, so [laughter] don't you don't have to read everything, but there is kind of a scenario in there as well. Yeah. To make it a bit understandable. Yeah. >> Okay. So maybe that I mean it could be that that's a deliverable from this. Our first deliverable is that it's kind of uh storyboard >> the the thing out so it's everybody is sure they understand what what is happening. And then the second thought um uh if you go down to 04 well actually if you 03 and 04 um one question is is this purely mechanist or are these formulas mechanistic or are they AI inference driven because I'm asking because um if they're purely mechanistic. That's in my opinion that's a good answer because you're not introducing um the the trust variables around the AI models themselves into this thing that's resolving trust, right? >> Um but yeah, so I um to me I think you would if we were to pursue this, it would seem to me to make more sense to pursue it as a mechanistic thing. I don't know if anyone else does. >> No, I I think that is a very good point because what you say is is a real risk. So if you if you automate it too much or put own model risk in it, then we might make it even worse than it is or introduce a new weakness that we also have to mitigate and and you're absolutely right. We we we don't we shouldn't do that because there is a I think no need for it. uh because we can make that rule based rather uh so and and and secondly the risk is just too high and it doesn't make sense. Yes. So, but that that's a very a good point to pick up, but it would make it even easier to do I guess because there there is no AI reasoning behind except maybe the evaluation or understand the counterpart the lying party because at at some point we need to find out how much evidence we have. What's the assurance level? How do we look at this alpha? The alpha is the noise. How clear are trust signals? Can we use them or should we not use them? So this is where we have kind of a classification but I think we can make or should make that rule based. Yeah. Uh but I see that actually possible because yeah it's it's it's the the main question is rather where do we get the signals from and how do we read the signals and that's then a little bit more complicated but also the good thing is we also have answers here because people have thought of this situation. We should do that. But we should also know how confidence we are about the alpha and the signals because the highest the biggest risk is also for example to show a a a big uh I say proof of evidence that someone needs this and this and these parts of data because the longer and the more explicit this proof is be cryptographic or not but if the customer sees it he tends to believe it even the content is still wrong. So, um this is a difficult thing to do. Yeah. >> But not too much AI. Yeah. >> Great. So, uh Lynn asks, "Is this just an EU wallet or wallets in general issue?" I I think it's it's both. is just the EU ecosystem is being bootstrapped by the national governments, >> right? Yes. Well, it it it is a natural uh thing unless someone else loves it already in a wallet, but I don't see someone really care about it because they they are also missing this VTA ID, I guess. So, something that actually has some capacity to work for you or to assist the user and most wallets are hardly storage, aren't they? Yeah. So they would not care about it. And and the EU and Switzerland, they only look at that now because they set the timeline. So they want to be ready end of year or at least the EU wallet will be ready end of year. Uh so they feel already the pressure by knowing they have unsolved issues. But I think it's a general issue. Yeah. But that's also a good thing for us because otherwise we wouldn't have a mandate. Great. >> Yeah. >> Okay. So, do you want to give us a quick update on trustlets if there's any >> and on the best practice repository? >> I can. Yes. And we touch for it again. So, there is not too much of an update. I have further tried to uh stress this pipeline a little bit and train it a little bit more and there is another video available for you to look at if you want. So there is now one uh I think also about the trust graph as such and this second reader has now been able to be produced in in a short iterations so it improves but I'm not there yet and of course alignment with trust bites uh is not yet done and it is kind of a little bit of a different world yet but we need to of course still do that or or find a way how we can have the a combined maybe approach at one point or or just not to to use such a pipeline. I don't know. Uh and then the next thing is also maybe the the the graphical uh design guidelines are of course very important. So once we have a clear picture on on how our communication should look like what the uh not only color and fonts but uh also what the messages should be what the tone should be then we can of course fine-tune such a pipeline and see whether we can produce uh good quality with a little bit less effort or just go to the created approach back which is definitely more quality but also much for uh effort properly. >> Fantastic. Well, you're a one-man task force. [laughter] >> Yes. >> It's like >> so just for the byite-size trust. So, we just met on Monday and we're going to meet again in two weeks. So if you want to join for that then we can >> have the and and the other thing that we talked about that uh um at by size trust is that we >> didn't want to start making videos until and I think this is a good idea until there's a version 1.0 no release of decentralized trust graph otherwise we're constantly chasing a an everchanging spec because I I've seen lots of changes in the specs >> uh the names of things what parts they need and so on. So so right now by size trust is working on making uh um because there is a 1.0 upcoming release of TSP. So, we're going to work on that first. >> Now, there's valid point, of course. Mhm. Okay. So, the persona management stuff. Um, here's the discussion. Open a new tab. Here we go. So, I don't know how many of you had a chance to look at this. What Glenn's done is provide a series of screens that starts with your attributes and kind of collates them based on the discussion we had last week into these worlds and faces. Um, and then how you manage those worlds and faces against contexts. Um, how you add an attribute, how you add a world and a list view. So, Margie, um I think you've uh added some comments. Um do you want to kind of step through your stuff? Editing your face. Sorry. >> Yeah. Sorry. Um well, I just I just wanted to I I may be behind because you guys maybe have already discussed this. So I was just trying to understand the relationship between the concept you know elements of the conceptual model. So the first comment was just about um whether context is a subset of world. So if if I have a world that is banking and a context which is mortgage and another context which is uh philanthropy or something or you know is is that how it works that I have a a larger the money bucket has different content I I don't know it it seems like that is what he's saying here and I think I concluded that was. >> So I think there was a bit of a misunderstanding about that because he's kind of inserted a layer which is these worlds and really the kind of uh work life play model has money at the center. It's not a separate persona. It's just that your work life and play persona interact always with money and that's where you know the rubber hits the road basically. Um and I think really contexts are applications. They're places you're using this. So um that there's no hierarchy um between worlds and faces and contexts. Um, you can have a face. My understanding is you can have a face that exists in multiple worlds. Um, and actually my follow on comments were just take it out. Either you can use it to bootstrap to help people get thinking about it. Oh yeah, I've got a different email address for work versus home or something like that. But um uh I I'm not convinced. There's certainly no hierarchy and you could take worlds out and not lose anything. He only put it in after a comment from us during the last uh meeting. >> Okay, that's I have to get my head around there's no hierarchy then. Um, but I like that better and I like eliminating a component if possible to simplify. >> Yeah. >> I mean, the human side of this is like it's a natural human thing to have multiple persona >> and or faces and and that's actually grounded in research. This idea of faces. >> Yeah. >> You know, when you introduce yourself to someone at a party, it depends on the context, doesn't it? Is it a work party or [snorts] you know something at a little festival or an after you know there how we introduce ourselves to each other depends on where we are and what we're doing. >> Yeah. >> Mhm. >> So you might be a chess champion but that's not what you lead with when you're on a date maybe or or when you're applying for a job. you know, you you lead with your qualifications for the job or um you lead with a question about your date in a dating environment. But this idea that I have multiple persona and that's perfectly healthy even opposing persona um or identities um and that is a normal identity pathology. Go ahead, Margie. >> I'm wondering if it all of this struck me as pretty abstract like of of course this is con it's it's conceptually rigorous, right? But it's not grounded in a and maybe this is dangerous territory. it's not grounded in a metaphor that is already available in people's lives. So, you just gave a really great example of, you know, you you introduce yourself um according to the context that you're um meeting somebody in. And we had been I mean it's getting into the R cards thing. car like a a business card as a metaphor I think is helpful because it's or a baseball card or you know something that is like um um a nugget tidized set of descriptors that are relevant to a particular context and that you have as many of them as you need for the complicated life you lead or the uncomplicated life you lead. um including um you know a sud sud I hate to say these words pseudonmous uh card where which is you know basically when you want to be entirely um you know you want to show up but you don't want to share anything. I don't know. I'm just wondering if if we could look at it through look at this the set of screens through a metaphorical lens to see how if we could make it more understandable or simpler >> Daniel go ahead Yeah. Yeah. I know. I I also see these points and I agree. I find this view of the these worlds um quite good because it helps maybe a the user to articulate this different personas. Also, I see maybe the the difference to our cards because with the worlds, it's it seems like we then reduce it to only a few perspectives whereas our cards probably can have more, but maybe less is also better. And and and also in this concept also with context, I see already a lot of things that I've been talking about before that that Glenn here needs to build and especially the context. it's quite difficult to uh to build and you probably cannot ask the user what the cont context is. So it's something derived as we talked. So maybe the agent that helps to define the suitable context and the right strategy and compare it to these personas or pictures or worlds. Um but uh I think the we we're on a a good good track there and it certainly helps the the users to articulate these different worlds and this then again informs the profile that my agent would need to have or he needs to know a little bit about his principle. So know the principle can be very well informed by looking at these different worldviews. So I take it as a opportunity that also supports my model although it looks still complicated and maybe can be made a bit easier. Yeah, I had um [clears throat] a few comments also. Uh so I um I agree with you Margie that we're kind of starting at the wrong end cuz that's not naturally how we introduce ourselves by thinking of all the attributes we could share. And so I've suggested a few things and um so I'll just step you through that and see what you guys think. So I first of all I clarifi I clarified what my assumptions were looking at these screens. Um, so my assumption is that most users get to this series of screens based on an invitation to join a VTC and only a small subset of users would need to set up a VTA in order to create a VTC. Um and then my second assumption is that for the initial implementation these users would be highly skilled developers and software engineers. Um I I I don't know what you think about how valid those um oh by hi hi and bye Kevin. [laughter] Um so I I just kind of started with uh those two assumptions. So I just clarified this point on on worlds because um I mean this is a model. Go ahead Margie. >> Oh no, you finish. I just >> Yeah. So it's a model I've been using with clients for like getting on for 20 years when it comes to identity that we all have these kind of three core personas. And if you think about stuff and and money is at the center. So they're not different worlds. they're different faces um different ways of presenting ourselves and not obviously you might have multiple work profiles you know for example I you know I'm a gig worker basically and so I've got like a million and one email addresses and diaries and sometimes I present myself as Nikki and sometimes it's uh something else um uh so it's just a way of giving people a starting point, a way into it. Um, and I would also I'd start with the context. Um, and I I'm not sure it's that difficult to create your context. It's part of discovery in a sense because it's just the applications where you're using this stuff. Um, I'd also just, you know, we've all designed sign up experiences, um, or signed up ourselves for something and the more stuff we've got to type and think about, even if it is asking questions about us, the less likely we are to complete the process. So you could kind of um use derived attributes for example connect your GitHub profile or upload a CV. Uh the attributes could be derived from that and then you can just kind of cross check them and add in anything else that you think is relevant. Also bearing in mind that the applications themselves will have attribute requirements like you're a human for example. Um and I think also all these attributes would have degrees of assurance. Um you might be able to verify some attributes like I'm an EU citizen or um and and within the decentralized trust graph model and the kernel requirements they have this vouching double vouching in phase four. Um uh so that that needs to exist alongside the self assertion. Um and then you know the the screens are just I I'm no UX designer but everyone gets cognitive overload and they're just a little bit too full. Um, so I gave a a prompt to Claude Code and um, you know, it it it's just an illustration of how you could turn it round and have more of a journey rather than just a whoa, I've got worlds, I've got faces, I've got contexts, I've got attributes. What's the point? Why should I spend time on this? Um, so I imagine you'd have an invitation that might arrive in your mobile phone wallet. You open it, it takes you into kind of your context, which are your communities essentially as far as I can understand. Um, and you can see uh the kernel invitation there. It asks you for certain details. you can upload your CV or connect gear or GitHub and it it it draws out some of those details. Obviously, they won't need their GPG key. That's a the whole point of decentralized trust graph is my understanding is to replace that. Um and then they they go on to choose what they want to share which is curating their face. They could change the name of it. they can select what what's seen, what isn't seen um and then um they end up with their face and they they request the vouching. So, and then you can just this is kind of an optional you could you can imagine how you could go into the sort of gardening exercise curating your faces seeing the communities they're accepted in. So, um and my I guess my key comments are make it a journey. Let's understand who it's for. Um, and by the way, my prompt to Claude was that it was a developer, but they were quite a junior person. They weren't and English was not their first language. So, just clearing up a few of the screens and simplifying things. Right, Margie? Then, Daniel. >> Yeah, I just the one com I love these all these assumptions. I think they're the great a great set to start with and I agree that the most common uh use case is really that you're going to be invited which which I think is the way to start that journey as you've as you've shown here. Um I um when we were doing that the user journey before though we had we had started with the person whose job it was to create the VT the VTC from scratch. So I do think we have to >> you know we do have to probably tackle that as well but it's not the dominant use case as you're as you're mentioning. Yeah, I mean actually as part of the prompt I gave them your user flows Margie and said we're starting at the invitation bit here. So I I did build on the work the extensive work that you've already done. Um but as I say I'm not a UX designer. I've just like got years in the trenches with these kind of products and I know that it it it should be the the complexity should be abstracted completely. It's just an enabler and what they want to do is be able to do their jobs with the kernel code. They're also a very special user group. a really difficult and special user group were difficult for me because like I'm not a developer. So I I look at those screens and I think woo where do I start with this lot? Um but it might be the right thing for developers. They might want a little bit more uh detail. Daniel. >> Yeah. Yes. Yes. before I have to leave only what I I like the the idea absolutely because I also think it's not very likely that someone takes a lot of time to just build these potential faces just to have them takes too much effort probably and people don't see the reason why and then to flip it around is natural idea and also this data extraction proposition is very good the only question I have here is what is our position regarding data minimization so from GDPR point of you because it's actually also something we should not do just to make or build out these rich profiles because some of them might not even be needed but we then still >> kind of built this data and collect it and make us ourselves vulnerable but yeah it's open question >> but that's the whole point of the VTA isn't it and the selective disclosure and the zero knowledge proofs >> also Glenn gave me some really important information that I need to pass on to you all. So, uh he said one thing I haven't explained very well is that we've designed the VTI stack on top of trust tasks which means the UX is actually a thin shim on top of on on top on top of it. So that makes it very easy for any UX designer to create their own version of the UX. >> And um he's got a bit of a a vision of us running designathons. I call them design jams, but um and there are many different UXs because different people, cultures, generations treat UX so differently. >> Let's encourage people to build their beautiful front ends that work for them. all the heavy lifting is done underneath the UX over trust task which is what guarantees the safety aspect. Um he then went on to point at this article let me copy that article. Can you see this? The agent is the state. And basically what this is saying is it's presented on the screen and then it's discarded. You're not storing it. You're not building these rich profiles and um it's exchanged in the interaction which which is what the trust tasks you know build up to. Um but it can be presented in any way. It could be audio for example, you know, and for many user groups an audio and um verbal interface is ideal, you know. Um reading and and and navigating screens and and and so forth is more difficult. So, let me uh put this in the chat for you guys. Uh, Margie. >> Yeah, I it's I um I have to think about it, but I think I might slightly disagree with Glenn because um an example of um people will build different interfaces. Um we have seen that over the history of the web. People have built every conceivable form of a website. Um, and they're not all equally uh delightful, usable, um, I don't know, uh, understandable. And so I feel like what this group should be doing is to um distill out the the critical mental models like maybe not be prescriptive about UI the UI layer but be decisive confident about the patterns that any UI layer might have to follow in order to promote human comprehension and utility. Good. Yeah. So, in I characterize these as interaction patterns and I think you can be quite prescriptive about them. Um, I just I I'm always a little bit wary in saying what works for human comprehension because it's invariably with the best will in the world is what works for us from our perspective in terms of human comprehension. I mean the the the construct of the wallet itself is western you know north northern hemisphere. >> You know uh stores of value and status in other types of communities include things like jewelry or ritual scarring and of course clothing. And you know my my wealth and status is projected through the size size of my cattle herd and which grazing land you know so [clears throat] it's always all we can do is the best from our researchbacked and evidencebacked perspective. Mhm. >> Um and and then provide hooks and en ensure counter um in ensure critical friends in other types of community. I think that that's the best we can do in in our situation. Um so I'm not saying you're wrong. I'm just saying that we I'm always ultra cautious about saying this is what works for humans because invariably we can only speak for a small group of humans or you know a group of humans. I wonder like in in a normal product development process, you would you would come to patterns and you would um you know prototype the sequences that embody the patterns and then you would show them to uh different audiences to to uh confirm or to validate or invalidate whether things made sense, what does or doesn't make sense to people. Um, and ideally you do it, you would do it in more than one cultural context or as many cultural contexts as you could uh feasibly um do the same kind of test. Um although that's not Carly Heida commenting on that >> in >> yeah I think the the challenge here is well the opportunity with this that Glenn's given us is it's a real world application isn't it so we don't have to worry to we need to understand the culture of kernel developers so I I don't know who's going to the plumbers conference But a little ethnographic study would be brilliant. [laughter] Uh, okay. We're one minute over time. Thank you all very much for your contributions. I think next week we perhaps need to continue to work on this and Margie maybe if you can lead us on on you're the UI expert. Is that suitable? >> So, what are what are we trying to get to for Glenn and by when? >> Um, he wants advice on the UI for this persona and the Linux plumbers conference is 12 or something. >> Yeah. Okay. So basically we we want to do in um iterations on maybe the expiration done. >> Yeah. And I think uh so we can text him asynchronously to kind of be a bit more specific about what he wants from us. Um but I think it's just some advice. But I, you know, when I was looking at it, I I've obviously got my perspectives, but I kept going back to the work we'd already started and the requirement set that we received in the first place and just thinking, yeah, we need to go back to that. I need to design who who's using this, where have they come from, how do they land on this screen, you know? um and and thinking about the explanation that's needed that we'd spent some time discussing and so forth. So, >> I think that work is it's a good opportunity to kind of string it together. Um but I agree we need more specifics from Glenn and Jeff as to what they want by when. Um so I'll take that as an action. >> Okay, great. Thank you, Nikki. >> Yeah, take care, guys. Have a good uh week and see hopefully see you guys tomorrow at the uh working group call. Okay, bye.