 Hello, hello, and welcome to the latest episode of Idea Flow, the monthly discussion of MIT's computational law report, which you can find more about at law.mit.edu. And this month, we are following up with one of our star speakers from the most recent MIT computational law course, Julia Panfill from New America, who gave an overview and led a discussion on the future of property and how personal data can be utilized to ensure people can exercise their rights with respect to their property ownership rights in the post-disaster and in other crisis situations. So I don't want to spoil the headline here. And so without further ado, I would like to hand it to Julia, who is joined by Dimitri and Liam to walk us through a very engaged, almost like a creative working session on this topic. So thank you very much for joining us again, Julia. And the floor is yours. Thanks so much, Daza, and welcome, everyone. So I'm going to share my screen and just walk you through a really short PowerPoint that introduces the idea that we'll be talking about today. And then I'll hand it over to my colleagues Dimitri and Liam, who can sort of take us through the meat of today's session, which will be brainstorm on how to take this from the theoretical to the practical. So let me just move into present mode. Give me a thumbs up if you can see my PowerPoint and it's big. Awesome. OK, so a little background. I work at a bank tank called New America. And there I run a program called the Future of Land and Housing. And that program looks at both existing and new ways for people to secure the rights to their land and their homes. And one of the problems that we see across the developing world, but also here in the US, is that nearly half the world actually lacks any sort of a title or deed or another document that proves that they are the owner or the occupant of their home or even live there. And because of that, they're locked out of many of the benefits that those documents confer that being a homeowner confers. I've listed three examples here. But the one that we're going to focus on today is this first one in Puerto Rico after Hurricane Maria in 2017. About a quarter million people couldn't access FEMA disaster aid because they couldn't prove to FEMA where they lived. So either they didn't have titles to begin with or they were destroyed in the hurricane for one reason or another. So they just didn't have the administrative proof to show this very salient fact about their lives. And something that we're thinking about is how do we make it easier for people to prove this thing that they own their home or that they live there? And an idea we've been increasingly exploring recently has been whether we can actually use our digital trails, the information that we generate just by living our social and economic lives online to help prove this fact. So for example, I have my smartphone and I have my Google Maps turned on. And if I were to look at my Google Maps location history and see the location of my phone between 10 PM and 5 AM every night, it would show a big red bullseye over my house, right? That in itself is not dispositive. You can fake that. But then what if I combine receipts from Amazon that show I've been doing a lot of online shopping over the last year with COVID, packages being delivered to the same location? What if I then overlay my Facebook or Instagram or Twitter data that, again, may be showing that the metadata from photographs comes back to this location? So the question that we're left with is can people be empowered to harness their own trove of digital evidence to prove where they live and access to services that they have before been locked out of? So this is just a quick slide that shows some of the potential sources of data, right? Location history I touched on, digital receipts, social media, things like delivery logs, Uber Eats, Lyft, right? Even your online history of listing places where you've lived. So the question becomes, how do you, how does somebody go about gathering this data, particularly if they're not very tech savvy? And what would it take for an agency like FEMA to accept it as proof? So today, before I hand it over to Liam and Dimitri, I just want to introduce our case study. Hurricane Maria, the Stafford Act, the Stafford Disaster Act, allows FEMA to provide money for repairing what are called owner-occupied private residences. But in order for FEMA to provide you money, you have to prove that you are the owner-occupant. And the main ways to do this are to present either a title or a deed, or if you don't have that, some sort of a tax receipt or will, potentially a utility bill. And what we saw was that after Hurricane Maria, FEMA denied 75,000 applications for the reason, they call it ownership not verified, for the reason that people couldn't prove that they were the owner-occupant. So that comes to about 250,000 people. So the question is, could those people instead use some of these, what we call tapestry credentials? And we call them tapestry credentials because as I mentioned, each credential by itself is not dispositive. But when you start layering them on top of one another, you start to build a tapestry that becomes much more convincing, right? You start layering these proofs. So could we build a tapestry credential composed of people's digital trails to expand the pool of evidence that FEMA accepts? And with that, I will turn it over to either Dmitry or Liam. And Dmitry and Liam, I'm happy to drive if this is helpful. This is very helpful, thank you. Yeah, let's just talk over, sounds great. So I'll jump in. So we imagine this data credential process as being nine steps and some of them easier than others. So imagine the hurricane has just happened. There is a member of FEMA or an advocate coming up to a homeowner and essentially, they would start by educating them on what this tapestry credential is. Hey, you don't have your deed. We're gonna try to come up with the next best thing as soon as possible. We're gonna build a tapestry of your information, physical and digital. You have a right to a copy of it. We are going to keep a copy of it and you have a rights to complete control of what happens. So if you want it deleted, it's deleted for everybody but you, we'll destroy it. Whoever's building this tapestry credential is your advocate and it's here to help you and only you. That's really the way that we envision it. And there's several ways that that advocate can get to you and where they come from. So then becomes the actual discovery process, which is where we build the foundations for making a tapestry. So it's an interview, essentially a set of questions along with tools. So the questions would be just who do you live with? Tell me about the house, the who, what, when, where, right? Tell me about the people who live near you, who you know. And then we'd start using digital tools. We'd probably hand you a laptop and say, hey, could you actually sign in to all the different accounts that you have? We feel comfortable with sharing so we can scrape that data. Can we grab your cell phone? One of the things we've been thinking is grab, take someone's cell phone, shut it off, put it in a plastic bag and hand it to a forensic engineer who can then grab all the cell phone data off that. Like the MAC address of the cell phone and correlate with that, with self-tower logs from say a company like Verizon in Puerto Rico, right? And we can start to get some forensic evidence of where that cell phone's been. All these things contribute to the tapestry and there's a whole set of tools there we go into. So then after we've collected everything, we build a case file. In this case would be for FEMA and the advocate. And then that case file would depend heavily on this database we are building of all of this tapestry of information. And that's really the raw information as it's getting collected. Then there's a step of organization. We normalize all the data. So if we have receipts and digital tweets and blog photos and your power bill and all these different things, we'd start to normalize, which means, hey, this name John and John Smith and Smith, John, those are all the same person. This latitude and longitude on the photo is the same as this address, right? Normalize the location. This dates are, this date and this date time are approximately the same event. So you start just lining up and organizing a giant index of all this information. That way it can be searched. That way it can be analyzed, right? Around this time, we actually start to secure the data. Everything should get multiple copies, should be stored in multiple places. And there's a whole balance of making this data secure in the sense that it's been backed up and isn't gonna get lost, but it's secure in the sense it's not exposed. And then once all that is done, we go through this analysis phase, which is we actually correlate digital trails to build the data proof that comes from the tapestry. And this is really around step six through eight is really when we start to get a lawyer involved. We kind of put more of our legal mind cap on. We're writing, we've now proved what we're trying to say, which is like, many things can be done with this tapestry, but right now we're focused on proof of occupancy, right? We've built, we're now telling the story of how all this data correlates and proves occupancy. We write an actual, like couple page document, whatever it needs to, whatever it would take to tell the story of how all this data comes together. And this would be a report for review with the subject. And then with this report, we can submit it as proof to FEMA, they can go in front of somebody or FEMA or a judge, and it's either accepted or rejected. And we then, if it's rejected, we go back and do more discovery and analysis, right? So that's really on a high level of where that process is. Just to move quickly, this is kind of some examples of just how broad the different ideas we've had of the data types that could go into the tapestry. And it's really about judging how relevant these pieces of data are, how easy they are to get at, and how robust they are, how verifiable, non-falsifiable they are. And we can talk in details about what those are, but we really wanted to make a workshop out of this. We'll probably flip between this slide and the next one on really getting your guys' idea on how to rank all of this. So if we go to slide 10, I guess, Yulia, we can just show where this is all going. So we picked out some kind of interesting examples and five different types of data that we thought of about 30, but these are five that I think are really good examples of the variety. And then we wanted to gather, figure out, discuss as a group how relevant, accessible, and verifiable each of these are and what the common opinions are. We have our own opinions. We'd like to hear more. So we're very close to the problem space. So it's good to get outside and inside. I'll stop here. We can go back to the previous slide in a minute. Yulia, Dimitri, how are you feeling? Is that covered most everything? Thank you so much, Liam. Yes, I'm wondering, do people have any sort of clarifying questions? Does this make sense? Let's see. So I think you could raise your hand or actually let me ask the computational law team. How shall we say just chat? Maybe the best thing is to chat, put your comments into the chat and that would probably be, I guess that's how we've done it before. There's one thing from Nadia, which is what have we got here? No hand function. Okay, legal, I'm just gonna read it. Legal assistance has to come very early on because proof of occupancy may or may not equal proof of residence plus or minus proof of property ownership. Define advocate, a person with authority and legal standing to represent the person. How is authority granted? Authority to collect proprietary information granted to whom? Soul Tower, ping information, for example. And those are some good ones there. No, actually I'm not going to, I have some as well, but I'd like to hear from everybody on the line first. So if other people have questions or comments or any combination, go ahead and use that chat tool. And if you're having any issue with a chat tool, then go ahead and come off mute and just let us know and we'll find a way to make sure you're heard. But would you like to get started with Nadia's inputs? Sure, I can just quickly address the legal one and then maybe kick it over to Liam and Demetri to address the rest. Nadia, you're absolutely right. And I think that looking ahead, I think that the big challenge with this sort of proof is gonna be differentiating, for example, a renter from an owner, because it's quite easy to prove that you live in a certain place. Proving that you have an ownership right versus a tenancy right is a little bit more complicated and involves, for example, having copies of financial records or showing that you've paid for repairs or other things that a renter wouldn't do that an owner would do. So that's a great point. Yeah, going into the authority to collect proprietary information, like from a cell phone tower. There's actually a few ways to slice that. Places like European Union in California, you actually have a right to your data. So we would actually enable individuals either through a tool or otherwise through a couple avenues. You have a legal right as a backup service to use a backup service on your behalf. There's also certain avenues for utilities and telecoms to actually take information requests. And depending on how they're worded and for what purpose and who's asking, they'll respond promptly or otherwise. It's just easier and easier to get a lot of this information. And working with a lawyer and other things, these are all to be found out, but these are the great right questions. Yulia, you wanna keep going? I can, if I can step in real quick. I think it might be helpful to say a couple of words of introduction for Liam and I just to set context because, so for example, Brendan in chat asked a really good question that all of these things are privacy issues. And right now there's a lot of really interesting engineering and standards work being done in the standards community around decentralized identifiers, digital signatures, all sorts of Lego blocks out of which in the ideal world we would be able to build fairly incontrovertible proof of these things. So Liam and I, we're engineers. We got in, we got connected with Yulia because we're building the technology side of it, the privacy side of it. So we're building encrypted decentralized storage, self-administered and self-sovereign identity, all that stuff. And Yulia is actually like hands on and on the ground with the legal and the social aspect of it. So it was a great, great match. Yeah. Over to you. In five to 10 years, we're very much trying to get out these digital credential technologies, digital identifier technologies. So this is, this is all very, very common and actually like this type of, this type of proofing becomes kind of standard in many aspects from the credit score to your loan application to the legal proceedings to everything. Yeah, what's interesting about one, I realize that time is short, but I wanna say one more thing for context. What's interesting about this digital credential tapestry approach is the necessity to work with imperfect information and imperfect administrative and technological infrastructure because like when all this technology succeeds, when everybody will have effortless digital credentials, digital signatures, essentially as identifiers that are accepted by legal authorities, much of this problem goes away, but it's gonna take years and years to do that. What do we do meanwhile is the question that we're talking about here. Indeed, awesome. So I wanna inject a couple of more questions and contributions from the group. One of them comes from Andre who was curious whether there has been any litigation on this so far where perhaps someone's attempted to show some of this type of data to prove ownership rights and maybe it was rejected and it went to court and he's wondering how did the court look at it? And I might just, with a friendly amendment broaden that a little bit, do has there been any dispute even if it's just at the administrative level not necessarily metastasizing to litigation and a final judgment, but has there been any back and forth with government where they've had an opportunity to push back on this to say, yeah, we'd accept this but we wouldn't accept that or any definition like that? Not that we're aware of. This is so new that we're actually not aware of almost any instances where this has even been presented. So the idea would be to get this to a point where it could be attractive enough for either a government or there are other contexts like post-conflict where UN regulations apply that we won't get into in this particular workshop but a different decision-making body to take these on board. Great, so thank you for that. And I wanna add, so to frame this for everybody this is in a sense our sweet spot in the sense that this is a, I guess I'd call it an upstream design oriented new way to encapsulate a kind of a service type. And so the idea is we can, so there's a freedom there to come up with something and then the intersection point eventually if New America sees fit and thinks you've got something that's ready would be to kind of present it to agencies like FEMA to say, here's an idea of how we think you should go forward maybe you could do that with allies. So right now we have the enviable opportunity to sort of imagine here what would be the business operational, the legal and the technical shape of something and how does it fit together in a way that could be a good proposal that they might actually take on. And so there's two other things. One of them was Nadia has asked on that, given the fact that in some areas people may not have the mobile phone, the smartphone access and many of the categories depend on that. She was wondering if kind of a so-called web of trust approach. I'm looking at the wrong thing. Here is a web of trust approach might be helpful on that basically identity proofing part that Demetrius just going into where maybe you could, for example I'll just add to this a little like add something where there's like valid affidavits or something like that in the workflow where people are vouching for yeah this person's been my neighbor for 15 years and yes their kids go to school and have for whatever it is. So basically a web of trust whereby you have a community that's vouching for the truth of the assertion of property ownership. We very much thought of that. Go to your neighbors, see if they have any photos of you at the next door neighbor birthday party and given affidavit and all that. I wanted to just say on the legal side of things I actually talked to a friend who's a lawyer and works with private investigators and even horrible ones like the Pinkerton's and things like that. And actually from what I've been told PIs do a lot of this and they trade a lot of like LMS style tools like this around but it's incredibly bespoke like as far as they're just making stuff put in front of a judge. It's not formalized that has no name like a tapestry credential has no process. So yeah and it runs the complete gambit right exactly what you guys said from affidavit and some physical stuff. We've even thought of exotic things like public data you know my Google Earth image of my house that shows my car up front and my favorite umbrella in my backyard and things like that that I can there's really many, many dimensions to think about. Yeah, that's a really good point. The web of trust and the affidavit is gonna be an important tool in the toolbox. One thing that I wanted to highlight that was brought up in chat by Brian is all right, because this is such a diverse information, all these data streams and some of them easily falsifiable. How do we deal with that? And as Yulia replied in chat one of the most interesting things about this tapestry approach is it's probabilistic in the sense of each one of the data streams by itself is easily falsifiable and does not carry a lot of weight. But the moment you start putting more than one stream together, the nature changes. A great example is in biometrics. For example, can you use just voice prints like on Star Trek? Hey computer admit me into this room. Well, no, just voice prints alone can be falsified but the moment you put voice prints and fingerprints together because fingerprints too can be falsified. We've seen Mission Impossible and all that stuff but the moment you put the two together it becomes orders of magnitude not just a little bit harder orders of magnitude harder. And the moment you have three biometric sources where you have voice print and fingerprint and your retina it becomes orders and orders of magnitude to the point of you can start making assurances and that's the approach we're taking here as well. Very good. All of the trust score would be something that we could explore here. I was just gonna jump in to suggest it's 12.30 so we have a decision we can either continue with Q&A which is great, this is really productive or we can move into this workshopping depending on what the group wants to. I'd like to channel the group's spirit for the moment and say let us move forward to the workshopping because I think that would be an even better funnel for the engagement to focus us in on things that could actually help you go from here to launch. And also it's a format that we wanna experiment with in this monthly session. We haven't yet tried this workshop format so we're very excited to work with you on it. Cool, well, Uli, I'm gonna just gonna grab the reins and see if we can start filling in this chart. How does that sound? That sounds great. Do you want me to keep sharing or? Yeah, yeah, keep sharing. Okay. Somebody else will have to type it in as well. Okay. So one of the ones that we were going to start with, it's really simple, is a digitized traditional credential. So your passport and your wallet got swept away in the rainwater, but you have a copy of Google photos or Apple photos and I don't know if you guys have ever done this with Google photos, but if you actually type in driver's license or passport, it already has a computer vision model that recognizes like birth certificates, driver's licenses, passports, like top 30 most relevant documents. And if you go searching in your Apple photos, it'll type in driver's license, it'll find your driver's license if you ever took a photo of it. And like a good percentage of people actually have for one reason or another had to take a photo of the driver's license and send it to someone, right? So that was one that we wanted to start with. You lost your actual driver's license, but I have a photo of it. So going through relevant, I assume this is fairly relevant, right? Again, yeah, that's a, you know. Can I start yet? Yeah, yeah. Okay, I wanna contribute. So I'm gonna, this is, so I have no idea if this is the right way to do a workshop in an idea flow format. So I'm just gonna try something and if it works, we can do it this way. And if it doesn't work, let's try other ways. So here's my idea, everybody. Under records, bills and receipts, the question is, is it relevant? Accessible meaning like, can we get it? And how would you get that data, I think is the question. And then can, is it verifiable? Okay, so on another project, I'm working with Uber API. And I would, and so I just had the idea that maybe the part of the, so Uber does send receipts by email, right? But that's going to fall down on the verifiable side a little bit unless you have, unless you go back and verify it with Uber, which is a lot of, you know, cost and it may make it infeasible. So it's not that it's unverifiable, it just may not be a good fit for this because of the cost and coordination and everything. So, but the API is something I wanted to throw out there. So the Uber API does have, if I remember correctly, trips and payments. And so I wonder if it might be possible, and maybe it also looking at things like Lyft and maybe Zipcar to see which ones have APIs to show that this person's like trips had continuously originated and terminated or even if they at a certain location over a span of time. And so is it relevant? I think that would be relevant to saying that they were, you know, that they may occupy that place as a home. Is it accessible? API is accessible. Some questions about, you know, would there, who would be running the app on our side? But let's just say it's technically accessible. And is it verifiable? Well, I like that part best because it's an OAuth 2 implementation, which means if there was, for example, a mobile app that the person were using. So we're just in that category. And they had an Uber account, which they would if they were taking Uber, they could authorize a new America app or whatever you would call this program with OAuth 2 saying like, yes, I give permission for you to go to Uber and to, you know, get this information. And now we've authenticated them as that account user and we provided valid authorization for this app to gain that information about their trip history. And so that's my thought for, yeah. So just really quickly, this is a lot of the stuff that Dmitri and I have been playing with is we've been wanting to use the LightScope crawling toolchain where we'd have the forensic guy come in, install this app on their phone. We have an app called LightScope. It essentially just makes a list of everywhere you go, everyone you interact with, everything physical and digital you own or interact with. It builds that automatic database, right? And the way it works is we have a Windows Mac, iPhone, Android app and all those, you download the app, it just crawls the device for all the logs that are standard in that operating system, right? We have like 20 connectors, one of them's the Uber, right? We have a data scraper, it's a browser extension. The thing that we really want to do is the second we actually scrape the information right off the phone or from Uber, we're stamping it with verifiable credentials saying this app, and it probably wouldn't be the LightScope app, it would be the digital credential maker, right, app, we rebuild it. The digital credential maker was signed in by the advocate and the advocate is now scraping that app, was given access by the subject of the credential and now it's grabbing and immediately verifying at this date, at this time, off the Uber API, we saw this, you know? Cause it's, you know, Dawzi, you're like exactly in the mindset of where we're trying to go with this. I just thought I'd bring that together, it's like we're trying to add these credentials on top of uncredentialed information, the second we actually pick it up and the tool chain that's gonna be required to do that. I'm actually updating this document, Julie, I don't know if you have to like refresh or something, but it just- Oh, cause it might not refresh in present mode is what it is. So you may want to share screen in- Yeah, yeah, that's exactly right, yeah. Do I ask before, and then I want to turn it over to others that may want to help to try to fill out the slide, which I think is the, this is the method and the mechanism for engagement in the workshop, which is so far having, I can tell everybody, I just tried it, it was great. I recommend it to you. Like it's fun, especially once we see something, yeah, there we go, okay. And so I've just added the URL to the API that I think is the right one. I'd been working with the driver API, but I think this is the user API that we would want. And could you just speak to one thing before we hand it to another person to try to fill out some more of these Hollywood squares? When you talked about the life scope application, which is actually how I know you, Liam, and it's a really interesting thing, it sounded like you were saying you would kind of scrape the phone for a bunch of stuff, but I wonder, isn't that like a different data source than getting it from the source itself through an API from Uber? Yeah, the phone is essentially like, like people don't know this, but your phone has a rolling log that's like a six months to a year old and it's like every cell phone tower, every wifi access point, every time it's been turned on and off, there's just, there's an insane amount of data just on your physical logs. And I'm kind of a crazy person. I just back up literally everything, terabytes and terabytes into my own search engine. And I'd say the phone has about 30% of your personal information you can't get anywhere else. But you're absolutely right, Daza, that it is just one source. It would be that source combined with other authority sources like Uber's API. Yeah, just for example, when I get location off my phone, I'm getting location from the IP address, which is like provisioned by my internet service provider. I'm getting it from my cell phone, like my beacon strength from my cell phone tower, the strength of my wireless router and my GPS and a bunch of magnetic and gyroscopic information. And together that all puts, that all can be put together as like a massive portal. This is really hard to fake all of those at once, right? So if my phone log has all of that, plus my Uber Eats log, you know, GeoState log has all of that, just with the phone and Uber Eats, we've now correlated like nine points of data. And four or five of these can be verifiable on my router's log and my, in the cell phone tower log, just right there before we connect like 38 yards, you know, so I kind of get you an idea, so. Can I try, I want to try to be more helpful in the format. So now that I see what's happening, I see, I want to make a suggestion. I'm just going to go in here and get a little pushy. Like my suggestion is that we take the API and put that under accessible. So this is how, so what is the thing? It's Uber rides. And then just in case we have like more things under this, I'll just label this number one also. And I'll just put that there. And so now that if we have a two under records and bills, we can say number two is whatever my utility bill. And then is it verifiable? I guess I'll say number one, which is like Uber would be, yeah, like the OAuth two token is, you know, essentially. Well, actually, let me ask you guys, like how would you, I feel like I could, if there was like, if it went to court or if I had to like, you know, go to a magistrated FEMA, I could, I feel like I could lay the foundation to prove from the token, from the authorization token and the access token and the handshake with the API, if I had those logs, that this was it. And then if we subpoenaed Uber, they could show the corresponding, you know, ones. But like that's a, I mean, where would you say, how would you say this would be verifiable in the ordinary course? Like what would we put in that cell? So I can jump in. I want to be mindful of the time in that we have a lot of other rows to go through, but so basically it depends on the company. Most of these things, like the OAuth two tokens, plus combined with a subpoena process, definitely verifiable. And if we partner, if the company wants to make this process easier, if they want to take a couple of steps towards this, they can make it much, much easier just by signing a couple of things or by making the subpoena process smoother. Yeah. I have friends at Google. They like Gmail gets dragged into this stuff all the time. Like Bob sent a death threat from his Gmail. And now Google is being dragged in the court to prove that he actually sent it from his Gmail account. So there's like, it very much does, half the time these companies actually literally have a response team that their only job is to write affidavits that like, you know, we got the database admin, the database admin says this is in the database. And it's real, you know, things like that. So let me just jump in and say, I know Brendan and Andre have had their hands up for a while. I just wanted you to get Brendan and then let's do Andre next. And thank you for bringing that to our attention, Yulia. Brendan, you go first and then we'll do Andre. Oh, hi all. Sorry about the glare. I can't deal with it. But yes, very interesting conversation. I would just like to strongly suggest that the idea that we're going to use all orth going forward for anything I think is not actually going to be the case because that is part of the web two architecture and it has no place at all in web three. So I wouldn't be, you know, looking too strongly at it because it's going to go away with all the centralized entities. And that's my thought. I hope that- From your lips to God's ears, right? That's right. That's right. Thank you so much. That's a really good point. Thanks. Andre? Andre, yeah. You're up. Hello. And my doubt is, when we are searching for all this information and classifying relevant, accessible and verifiable, we could add another classification of the importance of the information in order to select the better information to use as proof before FEMA or before another government agency. Do you have this preoccupation on classifying this source and classifying the type of information? Yeah. I mean, we kind of like tent that all under the relevance side of things but it's exactly, you're exactly right. And really it's an exploration in like the granularity of how we're going to rank this stuff. You know, this is where we're like going from qualifiable to quantifiable, you know, so. Okay. And I'm sure that probably the information that's very relevant for some person is not very relevant for another person depending on the situation, the personal situation and all the data that could be greater. This really comes down to like the Herculean task that Dmitri and I have put on our shoulders, which is how do we build that whole questionnaire system to get at the right questions? Well, still only being like an hour or two or three that we really need from the subject where they're out in front of a laptop with an advocate, logging in the stuff, answering questions where it's not, and how much of this can we automate while also getting the right evidence. So yeah, you're totally right because we're trying to make one form that is going to be very different for a 16 year old, you know, in South Florida than it is for a grandma in Puerto Rico or the Middle East or somewhere, right? So like that, it's almost like at the very beginning. Do you have a smartphone? Well, that radically changes the next 20 questions, right? So we very much had that type of thinking, you know, so. Okay, perfect. Yeah, thank you. I see Walter has his hand up to you. Yeah, I just had a quick question. If we can, if application data is relevant, accessible, verifiable, how relevant consequently do we think the GIS and Maps data would actually be? That seems like it's more subordinate than application data than anything else. Well, there's a slide we really didn't show which is like when it comes to the data, there's really the data owned by the subject, the public information and information owned by others, right? So like your neighbor or the city or Verizon or something like that, right? And I'm not sure if they're like really subordinate as much as it is, it's more of like a question of relevance. I don't know. There's a lot of variables that play here, but I hope that kind of keys on what's the point you were getting at, Walter. So. I would also say that, you know, this gets back to the idea of layering. So you may not have application data if you don't have your location data on your smartphone turned on, right? So in that case, potentially GIS data would be a fallback. So trying to understand, you know, or if you don't have, to Liam's earlier point, if you don't have a smartphone, just kind of trying to understand the full range of options. Yeah, layering makes a lot of sense. Thank you. There's a lot of like crazy weird stuff that works out well. Like people don't know this, but there's about 80 companies that have scraped all the public photos on the internet and either found their location data inside the photo or they've used photogrammetry to place these photos. And I've searched places I've lived and I've popped up in the background of public photos and not just Street View, you know? And I played this game in Africa, you know, with a friend of mine who owns like a matcha farm in Kenya and she's in the photos, you know, from random tourists and the passersby. So we can get really creative with this. Like, you know, so. Okay, so we've got another candidate. I want to put into the workshop machinery, which is from Nadia who was talking about bills, which he thinks are relevant definitely because they're physical artifacts that can be used as exemplars. And so I want to take that as a starting point and ask, are there examples of bills or receipts, perhaps, where there's an authoritative, verifiable version that's digital and we're maybe at least in a U.S. context where we in some contexts are able to afford digital records, you know, authenticity as like the authoritative source of the record, even if it's not paper, are there some bills that we could have access to and that we can verify? From what I've been told, bills are great. They're used, I'm sure you guys have submitted bills already as proof. So there's like, that's one of the ones where it's like the practical framework, the legal framework, all the utilities are used to taking those types of inquests. So yeah, bills are like a slam dunk. And I think there's actually regulation in America for companies to keep all those logs for years and years. So yeah, that's all great. So we've got utility, tax, and what else did she say? Utility, tax, school records. I never even thought of school records. Great. This is why we're brainstorming because like it's too much to think, just think of everything in your life. Also repair bills, like, you know, the bills of putting a new roof on your house or putting a fence around your yard, home operating. This is like, I guess it's a receipt, huh? This is the thing about the questionnaire that's gonna be kind of crazy is you have to look like kind of sparked everybody's memory. Like what's your water bill? What's your memory? What did you do last month? What did you do the month before? The month before, you know, tell me where everybody works. Like this is gonna, this building this tapestry might be a very, very involved, like biographical process for an individual. So... Which is I think where the relevance question comes in because it could be that, let's say, three or four records that are highly relevant have the same kind of proof strength as 15 records that are less relevant, right? So how do we get to that, you know, that sweet spot? So one thing on utility bills, for me to get my visitor parking permit in Cambridge, I bring utility bill and sometimes like a lease, but I actually just bring the physical paper bill and hand it over the counter. It's like incredibly antiquated, but I know that the new CIO there, Pat McCormick, good buddy, is looking at, you know, saying like, can we take a picture of it and put it through a form? And then in theory, you know, I don't know if they would, but there is a ever source API where they could check the number against if it is at the same name of the person and whatever, or they could spot check it. But what about like, so for bills from utilities, what would the concept be? Like, how would it be accessible? Would we want to get the record like a screen? Would we want something that's electronic? Or we want a photo of it or how would we do it? It, you know, the real fast one is to take a photo of it, right? And then the screenshot's kind of secondary. A third one is like an authoritative letter from the actual, you know, doing an inquest at the utility. This is where, you know, the equation changes really quickly, right? Because if like the third part, is there a way to make like the something better than a screenshot? Like if we write an open source, you know, digital credential scraper and it's scraping the page on like the advocate's machine and adding a digital credential, I scraped the bill at this time. It's not just a PNG screenshot that anybody can fake, but it actually is signed with their digital credential. Have we now elevated what was otherwise a screenshot to something much closer to, you know, an authoritative, you know, letter from the utility, right? And, you know, exactly, you know, Daz, as you were saying, I would love to have like the ability to speed this stuff up where we're taking a photo of the bill and then using optical character recognition, which is standard everywhere, you know, photo, you know, photo to searchable text and grabs all the name states and addresses right out of that so we can scale this stuff very, very quickly. So yeah, there's like just the bill itself, that's it's like a five dimensional problem. So. Right, right. If I can jump in. Walter said something really interesting in chat about, if we think about this from a verifier's perspective, what kind of tools can we give them to measure the cumulative records? Either statistical distribution or like what do they need? So one of the things that we were hoping to kickstart with this workshop specifically is to gather ideas from all of you on two fronts, on the scholarly and scientific, right? So somebody really smart needs to sit down and write that paper on the types of sources and how much statistical assurity each one presents, right? So one, this needs a lot of study. And two, possibly even more relevantly, we need whatever the FEMA examples of mock trials, right? We're looking for connections with administrative authorities to get a sense of what, how much legal assurance would they need? What kind of sources do they accept of this table here? Which is the most helpful to them? So Walter's a really good question that we're just starting the conversation to start to answer. We love to plan that mock trial soon and have as many of you who wants to be a part of it, like join us. So like as it was throwing out a call to action there, we are very much interested in collaborators. Excellent, okay. So just one thing is I think we might lose some Brian. So thanks for joining Brian and we'll have the next update on the MIT computational law report at the next idea flow session. So we'll just need to spend more time on it because we're accumulating a lot of great announcements that he has. And so why don't we pull up the level now and just go where you were starting to head, I think, Liam, which is, I hope that this was helpful input on the chart. And at a higher level, what are some of the next steps and what if people wanted to continue to be engaged? Liam mentioned perhaps a mock trial or something like that. Or could you just help us look ahead? Yeah, we want it. Yulia, you want to jump in? Your show. Sure. So I think that as you can see, this needs to be significantly more fleshed out both from the tech angle. So talking about, you know, I think somebody raised portability just a moment ago in the chat and so forth. And from the legal angle, right? Or legal administrative angle, what would it take for an administrative agency to accept this? And it's a little bit, you know, I think that one way to go about it is to develop these ideas further and then present them to somebody like a FEMA. Another would be to develop a rudimentary proof of concept and go ahead and run with it and try it out and see what we can actually gather. But regardless of which of those paths we pursue and we may pursue both simultaneously, I think the most immediate next step is if anybody is interested in continuing to think about this, shoot one of the three of us an email. We've all put our emails in this PowerPoint and I'll drop mine in the chat as well. And we can see how we can get you involved. We also are a shameless plug, New America is currently hiring for summer internships. So if anybody is interested in that and wants to continue working on this, then just contact me. Over to you, Liam and Dimitri for any other ideas you might have. Oh, that's it. We have more writings on the subject to share. If anybody's interested and wants to reach out on that, even to a lower bar of just send us criticisms and comments. Anything you guys can think of, because time went fast and there's a lot to talk about. So I want to just thank you all so much for coming together and for preparing this so well, for helping us innovate the format with like, literally how would you do a workshop on Zoom and in an hour at least get the head start on it? This was really great. I hope that some of the contributions were helpful. I also hope at a higher level, it was helpful to in a more practical way to get this before a larger audience, to see who might want to play at the next level. We'll do our part to take the video of this and to publish it and to blast it so more people can have the opportunity to experience this with us than we're here today and to take you up on your invitation. And I would just like to suggest at a high level, I would think the cruising altitude for this, if it were fully successful, would be one where maybe there isn't a need to have like the Sherpas involved and like individuals that are advocates and that are helping people. You would want to find a way for it to be more of like programmatic and mass scale. But I think this is a really great way to look at how do you do the initial prototyping and testing to see how is this really going to work and to see what's relevant and what FEMA would buy that and all of that stuff. So I think this is a very well specified first step and I'm so excited to see it moving forward and I want to encourage everybody to follow up with this team and to help out. I fantasize about, you know, digital, you know, tapestrytradential.com and you just go there and get one and it's just like a foundation of something, you know, or just another tool and the tool box. So that's great. Thank you so much. Thank you, everyone. And thank you, Daza and others for organizing this. This is incredibly helpful. And if you want us to present this to any other party at MIT or Boston or otherwise, somebody reach out. We're loving, loving to put this on tour. So, yeah. But standing tall. We might take you up on that. So thanks again, everybody and have a great weekend. Then we look forward to seeing you next month, the last Friday on Idea Flow. Bye bye. Bye.