 and their seats were going to get started. Hello, everyone. My name is Yulia Panfill. I'm the director of the Future of Property Rights program here at New America. I'd like to thank our co-host, Omidyar Network, and also my colleague, Chris Mellon, who was really the driving force behind this gathering. Thank you to everyone here today. And thank you to about 25 of you who are currently tuning in over the live stream. We are excited to be joined by experts, policymakers, and technologists from all over the world. This is the first official week of fall. And here in Washington, fall is apple season. So I will start off our conversation by asking a question. Who knows? Please raise your hand. When is the best time to plant an apple tree? OK. Anyone else? Anyone know when is the best time to plant an apple tree? That's right. According to a Chinese proverb, the best time to plant an apple tree is 10 years ago. I would like for this quote to set the tone for our discussion here today. We will be discussing topics that are cutting edge. And some components of what we will be discussing may not be ready for immediate rollout. That's OK. What we're doing here is R&D. And we all know that R&D does not start when a product is ready to go to market. It starts years and sometimes decades earlier. So what exactly is the problem that we are here to discuss? Most simply, it's the fact that around the world, billions of people are unable to prove where they live. Their inability to prove this simple fact has far-reaching consequences. The climate march is going on just a few blocks from here right now. As we know, weather events are becoming more frequent and severe. And around the world, millions of people are being displaced every year. After their homes are destroyed by hurricanes or cyclones, families who can't prove where they live are often locked out of development aid. Just look at Puerto Rico. 60% of aid claims after Hurricane Maria were rejected. The primary reason? Claimants couldn't prove their occupancy. Similarly, 70 million people are currently displaced from their homes all over the world, many due to conflict. Many of these people, perhaps the majority, have no way of proving where they lived before they were displaced or that they own their homes. That means that they're unable to return to their homes even after a conflict is over. Or if they choose not to return home, they're unable to recoup the value of their house. And perhaps most fundamentally, proving the location of your home or land is a fundamental step to obtaining property documents. And as we know from the research of the Global Property Rights Index, the architect of which is with us in this room a substantial proportion of people around the world lack property documentation. So why don't people have documents to prove where they live or that they own their property? One major reason is that the government processes for registering property are completely inadequate and outdated. We all know this. The processes are intensely bureaucratic. They rely on a small guild of surveyors who go door to door slowly to measure plot coordinates and transmit that information to some centralized authority that combines it with fees, more paperwork, more verification, and eventually, maybe 10 years later, provides a formal title or another document. To give a famous statistic from Uganda, there are 15 million unregistered land parcels in the country. It would take Uganda's 75 surveyors, more than 1,000 years, to provide documents for all these parcels. As a result, billions of people are locked out. Except that survey plans and notarized forms are far from the only evidence of property rights. In fact, our property rights are evidenced by a multitude of small everyday events, where we sleep at night, where our mail is delivered, our relationships with our neighbors, the fact that we paid to put a new roof on our house and put a fence on the yard. This is going to be an easy one. Please raise your hand if you have a smartphone. Come on, don't be shy. Please keep your hand up if your smartphone has even a single location enabled app or service. Google Maps counts, Uber counts, yes, even Tinder counts. Each of your phones produces data that can be used to predict with great certainty where you live. Not only that, your phones carry a digital trail of your purchases, which can serve as evidence of property improvement. They carry your social networks, presenting an easy way for neighbors to back up a claim of where you live. And it's not just you in this room. According to Pew, more than 5 billion people have mobile devices around the world. And over half of those connections are now smartphones, right? So this is coming quickly. Pew found that 45% of people in developing countries have smartphones. That means that very soon, perhaps already, a South Sudanese refugee arriving in northern Uganda can show where she lived when she was in South Sudan and can use this evidence to help prove a property restitution claim. So to recap, we have a problem. Billions of people can't prove where they live right now, and it's locking people out. And we have at least an inkling of a potential solution. The sensors in our pockets can reliably and cheaply produce a variety of data that, together, can show our life an incredible detail. But how do we get from here to there? This is obviously not just a matter of printing out your Google location history and taking it to your local land office. There are a lot of questions that have to be answered first. For example, how are people supposed to gather this information about themselves and store it in a useful way, particularly if they're not tech savvy? Why would administrative agencies trust this data and onboard it? Are there some types of data that are more or less trustworthy? And what does this data actually prove? After all, proof of location does not constitute a property, right? And perhaps most importantly, how could such a system be built in a privacy-preserving way? How do we make this not creepy and not surveillance? We can't answer all of these questions here, but this workshop is an attempt to get a bunch of smart, outside-the-box thinkers in a room to take a first stab at asking, is this viable? And if so, how? This group was carefully chosen. We kept it small, and we kept it multidisciplinary. You'll notice that only one of our speakers is a land rights expert. That's by design. It might not surprise you to know that other fields have gone much further in exploring this topic. So we invited experts from those other spaces to share their insights with us. Our first speaker, Xavier Wallenweider, is a development economist at Flowminder. He will speak specifically about location data, including phone call location data and how this has been used to aid disaster response in Haiti and Nepal. Our second speaker, Matthew Pritchard, a founding director of digitally designed HLP will speak about land and property restitution. He'll describe his work in Iraq, attempting to use new forms of digital evidence to help refugees and IDPs reclaim their homes. Our third speaker, Heather Dahl, is the CEO of Sovereign. She'll speak about self-sovereign identity, a system that can bring this digital evidence together in a way that's privacy-preserving, trusted, and easy to use. And finally, our last speaker, Emmanuel Loutouze, is the director of the DataPOP Alliance, a coalition of big data and development created by the Harvard Humanitarian Initiative and MIT Media Lab. He's going to give us the big picture about the big data revolution and what it can tell us about development. Please hold your questions until the end. We're going to have a group Q&A with all of the speakers on stage together. After that, we'll have lunch, and then we'll get into the fun stuff. Debating and discussing whether this idea is really viable, and if so, how we can make it a reality. I wish you luck and hope that you enjoy what I'm sure will be a provocative discussion. Xavier. Hi, everyone. Thank you very much for the organization and the invitation. I'm Xavier Vrenvaider from the Flomender Foundation. As you hear, I'm not a native English speaker, so in case something is unclear, please ask me to repeat where's your hand. So the aim of this presentation is mostly to give you an overview of this location data. We're going to talk about two types of data, mobile phone metadata, cold detail records, and Google history location data. So just a quick word about who we are. Flomender Foundation is a nonprofit organization. We are mostly based in the UK, so sometimes with a small office in Geneva. And we use, indeed, mobile phone metadata to support decision-makers in humanitarian operation and development intervention. So we look both at population mobility thanks to this mobile phone metadata and also population characteristics. We focus mostly on developing countries. And our typical use case cover displacement, migration, dynamic population mapping, infectious disease, commuting, and urban planning. This is just to give you a quick overview of this cold detail record, this mobile phone metadata. So how the network is structured is that you have Bay station, where on each bay station, each cell tower, you have various cells. And then these cells are going to route the phone call across the network and the territory. Each mobile device is uniquely identified via a Smithden or IMC. And the mobile network operator are going to maintain a database of the cold detail record for billing purposes and to optimize the management of their network. So typically, you have user X, which make a call. And then it's going to be routed by a tower. It's going to go to the server of the operator, go to the next tower, and reach to the next user. Each time you do a phone call, you are going to be recorded in a database, which looks typically like that. You have the calling party, Mr. A calling Mr. B. You have the cell ID to which the call was routed from the A party and the B party. And based on the analysis of this record, you can assess more or less where the people are. Going to go quickly on the algorithm used to assess home location. One of the things to consider is that it's relatively imprecise. We are not talking about an accuracy of one meter, two meters, 10 meters. We're talking about hundreds of meters. And really, the accuracy is going to depend on the spatial distribution of the cell phone tower. So if you are in the urban area, you're going to get a very high, relatively high spatial resolution. Going to talk about probably hundreds of meters. If you go into rural area, particularly in developing countries, you're going to talk about kilometers. So naturally, when you are thinking about using this kind of data to assess or rather to claim a property rights, it's going to be hard. At best, it's going to show you where you have been living more or less. This is the first thing I'm going to skip this little presentation, is just how the home location algorithm works. The interesting message from this slide is that simply the home location is going to be an average of the cell tower that you use at night. And so it's going to be an estimation. You have this point on the map, which are the location of the cell tower. And our algorithm, the Hartigan algorithm, developed by us, was implemented in our Floquids software, assess your home location based on a kind of linear average of the location of the cell tower. So spatial resolution, we'll give a concrete example. For experience in high teeth, you have six base stations of square kilometer on average. Median distance to the base station cell tower is 360 meters. In suburban area, it's 500 meters. And in rural area, you can have one single base station covering many kilometers. Cell coverage can be quite patchy as well. So that's the surprise. You see a map of rural Nepal. And the bottom map is on the same scale as Zoom on Washington, just to give you a kind of idea of the scale. And you see that one cell is covering a very large territory. The little dots are groups of households. So it's really like, yeah, it needs to be considered when thinking about using this type of data. That's really the advantage of mobile phone metadata called data records, CDRs. It's that you don't need to have a smartphone to leave this type of trace. It's just like a TQ phone, so there's like a huge penetration rate. And then there are one-two data just to get a better sense exactly of the precision. I mean, we haven't carried any study ourselves. And we're coming here to a study done by a exact man at O, conducted in New York in the U-Durvish, mostly. And there, we had really large error up to a mile for the median distance between home location, the real one, and the one which was estimated. So this is going to be going to go quick on this one as well, just to give you a flavor a bit about what we do. And because when we use the CDR data, we use them always at an aggregated level. So we don't look at individual data per se. And it's really important for us actually to look really at population movement, both because of data per se, ethics, and also because it's what we want to work on. We really want to talk about population movement. So this is like a map of population with distance from both a print after the high T earthquake. With the same ID, you estimate the home location before the earthquake. You estimate the home location after the earthquake, and you can have an idea of this. Well done also in Nepal, there's a 15 hurricane. And then just to share this result, very interesting to colleagues, Yonik Lopin and the CCE, who looked at the rate of recovery trajectory in the mobility pattern of IDPs after this three disaster, Nepal earthquake, high T earthquake, hurricane Matthew. The result were very interesting because the rate of recovery, recovery being defined as being back to your normal mobility routine and pattern after the earthquake, this rate of recovery was comparable across the three disasters. You had like, on the red galaxy, you have the percentage of IDPs which remain this place. And on the horizontal axis, you have the number of weeks after the event. And you see that in the three cases, we have this convergence. Used also in urban planning, high T again, this is night location, day location. Resolution of the map is 500 meter here. Just to give you again, again, the type of use case. So as a summary, we are really always looking at aggregated fields. So now working with GPS data and Google history location data. So this Google location history data are based on the phone internal GPS, the connected wifi devices and the cell tower. The number of collection of data point collected is correlated with the distance to the nearest cell tower. These data are super easily downloadable. You can really go online, download your own data. Super easily, you can visualize them as well. Really easily in a Google timeline web application. Here, we are going to look at my own data. Again, the aim is a bit, I wouldn't say to set the scene, but at least to give you an exposure to this type of data. So you know what we're talking about. So that's my data. So from 2016 to September, on the left graph, you see on the vertical axis the number of data points collected. And on the horizontal axis is the time. You see that there are gaps. It's times when I stopped Google map, rather the collection of Google history location data. And then on the right-hand side, you see a box plot of the number of points collected on me. And the number are just huge per day. We are pretty much at 500 location data points collected per day. So I'm not doing anything. It's just like, it's the same for you. If we're talking about months, it's around like 11,000. So it's really like a lot, a lot, a lot, a lot. Looking at my data, it was like even a day where they collected like 4,000 points. It was like 31st of December, 2017. I don't know. They wanted to know the place of the party, perhaps. So it's just like crazy. And that's my data in Geneva. And so left map is 2017, right map is 2019. And what is interesting here is that I moved flats between 2017 and 2019. But I kept the same office. And you can really see it super easily. There's like pretty much no analysis here. It's just like plotting the data. And the little circles are just like, I don't see too much around the data. So it's just like showing where there are more points. And top bottom, I was living, help me if you're going to hear me. Now, should I stay close to the mic? Or is it okay? I mean, yeah. Okay. So this is where I used to live. Yahrouz, that's my work place. Now I live here. Still work there. And what is quite amazing is that here at the hospital in Geneva, I made like a little two-week stay there because of an injury. And it's also recorded. That's my mother's place flat. And so really in two seconds, you get like precise, meaningful location with this type of data. Here it's like a zoom on a Google Earth. Not really clear on the picture, but that's the kind of cloud, the white cloud is my location data at my office. And it's amazing how well it fits the window of my office. It's just like super precise. Then, so in terms of accuracy, there are two ways of measuring accuracy. In the Google history location data, you have an estimation of the accuracy. So if you look at just the September data, the blue line here, 99% pretty much of the data are below 100 meter precision. Most are around like 20, 50 meter precision. So it's really good. If we take the entire history of my data, 2016 to present, you still have like 25% of the points which are not precise, like more than one kilometer. Most of them are when I was traveling, actually, on the car, et cetera. But the important thing is that you have this estimation of the accuracy. So you can really filter out your points so that you can just focus on the data points which are precise. And then a much more robust analysis has been done by a colleague at WALPOP. He's been on trial with 21 participants over 367 days. And they got like typically a precision of less than 100 meters between the Google history location data. And using only the GPS of the phone. So just a few considerations on data privacy. So how we use that reminder of CDR? First, we sort of anonymize the CDR. We don't use the word anonymize because anonymize would mean that you have found a way to anonymize the data so that no one ever with any type of technology and using third party data could recover the true information. In our case, it's not what we do. We don't claim doing this. What we do is that we just hash the identifier. You remember the MISDN and IMC, which are the unique identifier of the phone. So those one are hashed by the telecom staff. Which means naturally that we are not in a position technically to answer a request from an individual asking us to compute for them their home location. We don't have the technology to do so first. And also it's totally outside our practice area. It's really things that we don't want to do. Because what we want to do is really set up a system which is really robust, limit the risk of any data leakage and limit the risk of the identification. So all the infrastructure is maintained in a secure way behind the telecom operator firewall. So what we do is that we install a server behind the firewall of the telecom and then the staff dump the data having previously hashed. And we never take out the server and the data and we aggregate the results. And again, the use cases are really on population level stats, not individual. So yeah, this way some question about be it the CDR data, the mobile phone metadata, or any type of location data. Who would do it? You need to set up a kind of probably a legal entity with this contractual agreement which would need to be most likely compatible with the GDPR, the European legislation. So it's really like, yeah, who would do it? What type of contractual agreement? And then the question of data portability. How do you make sure when you are transferring your data from say your Google history location data to this new ledger that these data are not altered? So thanks a lot for the, go into it. Good morning, everyone. I've been asked to stay nice and close to the podium so everybody can hear me online which is probably pretty good because I tend to wander and now this will keep me from falling off the stage. It looks like part of my slide didn't come up there but my name is Matthew Pritchard. I am a Housing, Land and Property Rights Specialist based at McGill University in Montreal but I'm here today as one of the founding members of a relatively new Housing, Land and Property Rights Lab known as digitally designed Housing, Land and Property Rights. We specialize in HLP programs in conflict affected environments. So I'd like to start today by asking the question that brought us all here. Does your phone know where you live? I think that Xavier just shared, well, a huge amount of information on the types of data that our phones collect on us. And now for me, does my phone know where I live? The answer is a definite yes, even though I seem to be continually trying to restrict what my different phones based in different countries know. In fact, I carry usually like most of you two phones. I have one that has a Canadian number based at home and then recently I spent a lot of time in South Sudan so I have my South Sudanese phone. And now what's interesting is that both of those phones, regardless of where I am, really tell me where I work, where I live, where my children go to school, where I travel for work. And even when I am way off the grid, right? Even when I go for a week or two without having access to cell service, my phones can give a pretty good idea of where I am. But I'd like to push this question a little bit further. If your phones know where you live now, do they know where you used to live? And Xavier gave us a fantastic example of this from when he moved flats. And for me, I sort of poked through my own phones yesterday on the flight down. And I found out that looking at pictures, social media, audio, video, receipts that I have on my different devices, I have detailed information of about 10 previous homes that I have lived in, in five countries, going back to about 2,000. Now that is a huge amount of information. And beyond telling me that I seem to have moved a little too much over the last 20 years, it's not super useful in my everyday life. But what if I was a refugee or an internally displaced person, right? How useful would those data points be? And this is where I and our team at digitally designed HLP start to get really interested and excited. Specifically as a lab specializing in the housing line and property rights for displaced populations, we spend a lot of time thinking about how to improve return and restitution programs. And the main question we focus on at DDHLP is how can we use different types of information that people already have on their devices to facilitate return and restitution of displaced populations whose homes have been destroyed, occupied, damaged, or transacted? And by transacted I mean sold on. So in other words, could the information that people have on their mobile devices be used to improve how refugees and IDPs reclaim or seek restitution for their housing line and their property rights after a conflict is ended? And now I really dislike doing this but asking questions and then answering them right away. But in typical academic fashion that's generally how I go through life or try to. And so I would say the answer to this question is resounding yes, right? That these data can absolutely be used to improve the ways that claims are made but they currently are not being used, right? So why is this so important? Why do I get so excited talking about data and housing line and property rights? Well, for a lot of reasons, I'm a pretty excitable guy but what I will focus on today is that we have the opportunity to use technology and information that displaced populations already have to do three main things. And the first is to dramatically streamline the return and restitution process which can get incredibly complicated and take quite a long time. And the second is that addressing housing line and property rights issues after conflict not only has significant benefits for the livelihood security of displaced populations but also hosts communities and reduces the likelihood of future conflict. And then finally for this brief introduction using technology and data that displaced populations already have and in many cases are using themselves can lower the costs and speed up the implementation of an incredibly challenging but important component of transitional justice. So why a digital upgrade? Well, it's been clear for some time that the ways in which displaced populations monitor conflict have changed. Yet the programs designed to meet their immediate and longer-term needs are relatively slow to respond. And this is especially true for how countries design, sorry, countries, donors and implementing partners address the destruction, occupation and secondary transaction of housing line and property rights. And it makes sense, right? It makes sense that protection actors are relatively conservative with how they approach new programming. But what's interesting is that the way we approach return and restitution has changed very little since the Marshall Plan, right? So we're going back over 70 years. Yet as programs struggle to evolve, refugees and IDPs are actively using technology to monitor and share information on their housing line and property rights in real time. And what we're seeing here are the intersection of two key trends. And it looks like we've had some conversion issues between Mac and PC. So we'll just pretend like I meant to do this to highlight the different points that I'm making. But the two key trends are the first trend relates to the prevalence of smartphones, right? So Ulyss said before that I think it was Pew that the penetration of smartphones was about 45% in developing countries. Well in some areas it's much higher than that, right? So 86% of Syrian households in Lebanon own a mobile phone. From DDHLP's own work in Erbil in Iraq, we know that there is an average of three mobile phones per IDP household, okay? So there's a huge prevalence in many of these environments. The second trend relates to what IDPs and refugees are doing with those devices, right? Specifically, IDPs and refugees are actively using mobile devices to not only monitor their housing land and property rights in home areas, and by home area I mean pre-displacement, but they're also sharing information with their networks, right? With family, with friends, and with protection actors. For example, 36% of IDPs in Erbil are currently using mobile phones and social media to not only actively monitor their housing land and property in home areas, but they're sharing that information, right? They're sharing up-to-date information, pictures, videos, and documents about their HLP. And even in remote areas of South Sudan where I've been spending most of my time the last few years, where people in some camps won't have access to a network for weeks or months at a time, displaced populations are using mobile phones to store information, right? And in many cases, we'll transfer that information physically via Bluetooth or cable connection. So we know that mobile devices are used to monitor HLP in home areas, but why does this matter? Well, it matters because forced displacement, the destruction, occupation, and transaction of housing land and property rights in conflict-affected contexts really present a primary development challenge to displace populations, to host communities, and the establishment of durable peace. Grievances linked to the loss of housing land and property rights, they do not abate, right? If we ignore them, they do not go away. Instead, they grow and can lead to subsequent problems and sadly, but often, the resurgence of armed conflict. And this shouldn't be surprising, right? Because in some areas, large amounts of housing land and property rights are degraded, they're destroyed, they're looted, they're occupied, or, or, and they're transacted, they're sold on. Yet, at the same time, oh, sorry, they're destroyed or made otherwise unusable. But at the same time, we have the opposite, right? We have the HLP that's destroyed, and then we have the HLP where homes or entire neighborhoods are reoccupied, right? Sometimes they're occupied by people who have been displaced from other areas of the country for a relatively temporary period of time. In other cases, we have economic opportunists, right? Who will use displacement as an opportunity to not only come in and to grab other property, but in some cases, completely change its use, right? To build it up, to destroy a home and build a business, to destroy a business and build a hotel. But regardless of who is occupying that HLP, they often make investments, right? Short-term occupants will fix leaking pipes or perhaps a hole in the roof. Longer-term occupants may change absolutely everything about a home or a former business. And then over the years, right? These, this land, this property, it can be transacted. It can be sold on often numerous times so that the people who in the future, thank you, have a document, they may not know that that land or that property actually belonged originally to a displaced population. And so the troubling reality is that number of claims that proliferate after a conflict has ended becomes so fraught and with enmity that they can and often do destabilize a peace process. So given that the failure to address the destruction, occupation, and transaction of HLP can undermine peace processes, what is the conventional approach to dealing with these issues? Well, most large-scale restitution programs are based on a set of mass claims techniques designed to address hundreds of thousands of claims as quickly and as cheaply as possible. And this is done by first raising awareness about the requirements for returns, gathering, assembling, and organizing evidence, putting that into a claims database, and then engaging in what are called mass claims processing techniques. Now, the overall goal is to screen, categorize, and corroborate what are often hundreds of thousands, if not millions of property claims so that decisions can be grouped and applied as quickly as possible. And speed is of utmost importance here because the goal is to encourage people to opt in, right, the longer it takes, the less incentive displaced populations have to opt in, they decide to stay where they are or perhaps to take matters into their own hands in potentially destabilizing ways. So, boy, that is difficult to read. Although mass claims programs need to be accessible, right, it should be no surprise that they are generally very time-consuming, costly, and cumbersome. Returning populations are often ill-equipped to prepare evidence and organize claims, and the time taken to develop these large mass claims programs really discourages returns and restitution. And one of the biggest problems is that mass claims programs overwhelmingly focus on that third point there, the adequacy of documentary evidence. But we know that in most conflict-affected environments, displaced populations no longer have or never had the types of evidence needed to participate in mass claims programs, right? And this is because the property was never registered, right? Or if it was registered, that document has been lost and any backup that ever existed was purposefully targeted, right, so we know that the locations that store land property documents are often among the first to be targeted in conflicts, right? Clean the slate and then we can repopulate an urban area or a rural area. Now the widespread lack of documentary evidence is especially problematic or complicated by the fact that new arrivals often do have documentary evidence. And that's because they work very hard to come in and to get legal title, sometimes through corruption, sometimes through confusion, to in an attempt to solidify their claims to the property of displaced populations. Now the key takeaway here is that the majority of displaced households with legitimate housing land and property rights claims often and generally do not have the documentary evidence required to successfully make a claim. And this is where the data from mobile devices can come in. Can these data be used to define relevant versus non-relevant documentary evidence? So absolutely they can. And again, I apologize, I'd sent along a PDF version so we could avoid this, but we'll just, you just have to listen to me because I don't think you can read my slides. The most obvious step is to look beyond formal titles to the multiple types of evidence that can be used to triangulate individual community land and housing property rights. So what do we mean by alternative forms of evidence? Well, the easiest and most straightforward are detailed information or intimate knowledge about housing land and property rights. About your home, about your fence, about something that was just never right. I know I have a lot of that at my house. But also previous investments. Additionally, we rely on detailed histories of acquisition and occupation. And then we get to the multimedia. Then we get to photos, videos, audio, most of which are geotagged. So they can be linked to specific locations and dates and times. So videos of birthday parties, of retirement parties, photos of children. These can all be used to triangulate housing land and property rights claims. Then let's add in what Xavier spoke about, right? We have huge amounts of data from call detail records, from geolocation history. And then finally, what about bills, right? Bills for electricity, for water delivery, for sewage pumping, for generator repair. All of these can be used as alternative forms of evidence. Now, mass claims programs can take all of this information, right? Not just documentary evidence, but supplement it with what Julia said, you know, all of these everyday sources of information and start to look for patterns. And then patterns that emerge from a certain group can be used to triangulate evidence, right? And here we're moving beyond documentary evidence to make judgments for an entire group of people, which is what mass claims aims to do. So bringing in alternative forms of evidence allows transitional justice efforts to move quicker and at significantly lower costs than the current approach to mass claims. And the goal is not to replace the current approach to mass claims, but to use existing information. And again, this is what I get excited about, that people already have on hand or can get access to, to increase the accessibility and buy-in of traditional mass claims programs. Okay, so how can we make this happen? Well, we have a significant opportunity to upgrade how we approach mass claims, simply again by using that information and technology that people have already. Now the main issue is that displaced populations are generally unaware that they have or can get information that can be used to solidify HLP claims. And so this means that the first step in any upgrade is to increase awareness about alternative forms of evidence. And the second step is then to prioritize the collection, storage, and corroboration of a wide variety of evidence for claims beyond government-issued land documentation. Now neither of these steps need to wait until after a conflict is over. And in fact, the longer we wait, right, the greater the risk of losing key forms of evidence as people are displaced, as people die in camps, right? So we don't have to wait until after a war is over to start documenting, to start increasing awareness about these alternative forms of evidence. And now at DDHLP, we are combining our knowledge and experience with mass claims programs with three things. And that's the prevalence of technology amongst displaced populations, advances in database construction and fit for purpose cadastres, as well as techniques for securely handling big data to engage in evidence collection while displacement is occurring, while war and conflict are ongoing. Okay, so what does this look like in practice? Well, the first part is relatively well known. And that is that armed conflict leads to large-scale displacements and displaced populations settle in camps or seek out formal or informal alternatives. But regardless of where they are, refugee camps, IDP camps, squatting or with family in cities, IDPs and refugees are actively monitoring what is happening at home, right? What is happening with their home? What is happening with their land? And then start to share this information through their networks. Whether or not they know it, a lot of this information can be used as evidence when it comes to reclaiming housing land and property rights. And this is where DDHLP comes in, right? Through tailored and web-based platforms, displaced populations can access information about relevant laws and organizations and then learn about alternative forms of evidence that can be used to make claims in the future. At its most basic level, as I said, or this most basic function, this tool dramatically increases awareness about what constitutes evidence and increases the opportunity for displaced populations to organize and collect this information. But we also need to go a step further. And in addition to raising awareness, our web and app-based platforms will provide displaced populations with the opportunity to securely protect existing evidence by backing it up. So multiple forms of evidence can then be used in any number of ways, right? They can be incorporated into mass claims techniques or alternatively, they don't have to be used at all. But they are there, they are protected. And to finish, the end goal of a digital upgrade rooted in alternative forms of evidence is not to radically change what is being done. Rather, the goal is the same as the opportunity. And that is to use mobile technology and information that people already have to improve the standard and outdated approach to the return and restitution of housing, land and property. And this is where a digital upgrade comes in, right? It does not require huge investments. Rather, we need to simply extend and operationalize what people are actually doing. And all of my information was supposed to be here, but given that there were so many problems with the presentation, I think that New America would be happy to provide a copy of my presentation, perhaps in PDF form. And then you can have a copy of all the slides as well as email and all that good stuff. So thank you. Hi, oh, good morning. I'll start off with introducing myself. I'm Heather Dahl. I'm CEO and executive director of Sovereign Foundation. So you're probably wondering how does someone like me end up dedicating her life to self-sovereign identity. And this is what I live and breathe 24-7 these days. It's because over a decade ago, I started working on threat intelligence solutions, which to break that down is we monitored and played the underground, which was very interesting. Setting up honey pots, going into the dark web, buying things off markets, and truly seeing what is out there that not only cyber criminals take from you, other companies take from you that then cyber criminals take from them, and also what people communicate with each other, which has been stolen without them knowing it. Why didn't we do that? Because when you talk to those companies, they're gonna tell you, we're secure. Some crazy companies will say, we're 110% secure, which right then and there we know we got issues. But why do they say that? It's because for a lot of security teams, that's their job, they're telling their CEO we're secure. But we developed threat intelligence solutions to figure out, are you truly secure? Because if we can find your 2020 strategic plan being sold in the underground for about $30,000, and I can show it to you, I'm gonna see it probably not secure, and I'm gonna show you evidence of that compromise and breach from the outside to prove that no, you have a leak and we need to go find that. But here is why I ended up in digital identity, because when you look at the extremely disturbing things that are being sold and bought about all of us and our businesses and our organizations, here's what it comes down to. One, a problem with how data was stored. Centralized data repositories. So old. So if you were thinking, I need to be ready to centralize data repository, just pull out your bell bottom pants, wear your Britney Spears belly shirt, and you're in the same fashion as when you look at emerging technology today. And the other one is digital identity. There were problems with identity. So let's unpack this statement. Does your phone know where you live? How do you know it's your phone? How does your phone know it's you? How do you really know that's where you live and how does it know that's truly a property? Because on the internet today, nobody knows it's you. And that's the problem. If you can't solve the problem of how the data is stored and how it's identified, then all the other solutions that follow after is basically recreating the same problem that we have today. And that is why I'm leading the Sovereign Foundation and that's why we have a team of hundreds of top technologists around the world working to solve the problem of digital identity. So when you go to develop the next generation of solutions, you're not doing it with the old generation of identity. So we all know this cartoon. It's the most published cartoon the New Yorker has put out. On the internet, no one knows your dog. Well, the fact is on the internet, no one knows you're a dog. And so you're getting advertised cat toys. And then your trail that you may or may not be a dog is being monetized by others to sell you hamster toys. No one knows you're a dog on the internet because no one knows who you were because the fact is the internet was not created with identity. Let's look at how the internet was created. It was created from machine to machine communication and a trusted network. Whether that was ARPANET, whether that was the private internet that emerged out of Southern California, which actually one of the original node founders, Rodney Joppe, lives here in Washington DC and is fantastic to talk with about why we got into the problems we have today is because it was not developed with identity in mind. And how do you know that? Because we were affiliated first with universities usually, right? And it was the university that had the ID to another university, which was a trusted relationship. They never imagined people would wanna be using this thing. And so then the university would give us an identity. We can see that in our email addresses, right? At one point I was Heather at Willamette.edu. Well then life went on and we realized more people needed to use that internet than people affiliated with employers and universities. And so then one day in 1998, I got HCDoll at yahoo.com. What did that affiliate me to? Yahoo, not me, yahoo, my identity's yahoo. They trust I filled out the form I got. And that's why we're using this identity today. So on the internet we've evolved into a system that we run an entire network where no one truly has their own identity. We have identities that we attest to that others own and control. So the fact is today the old model of identity fails. It fails for everything that we're developing forward if we don't change the way that we incorporate identity. The fact is individuals are reliant on other providers for their identity. Those organizations control own, provide access to identity. It creates a fragmented system of username and passwords. I won't even go into the problems with that. Third party data trails continually breach our privacy. And in the talk from the tech community I'm in it, we always talk about the amazing things our data trails can do for us. Look at the flip side. And we are starting to understand the amazing things that data trails do to hinder our lives, to hurt our lives, to exploit our lives. And so how do we balance both sides of data trails? We know the cost of business with the current identity model from a security perspective is just not sustainable. We can see that with the lawsuits and the breaches. And here's what's interesting, IoT. It multiplies all these problems exponentially. We haven't even started to go down the road of what IoT is gonna require from security and privacy on the old identity models. So here, let me break down the identity models as you go through and you think about what you're gonna look at with property rights. We have a centralized model here. These are your driver's licenses, your educational degree, and in fact, they're usually paper-based and not digitally portable. My day this morning is the perfect example. A little crazy gain out of the house. Guess who left her wallet at home, right? So if I wanna go have a beer after this talk, I wanna go across the street to Old Abbot, have a few beers, and someone feels like flattering me by asking me for my identity. This is all I have. And I'm not gonna get my beer because they're not gonna take this. Nothing in here is bartender gonna say, yeah, you're over 21. That's the problem we have with centralized identity today. You have to carry it around with you. It's not digitally portable. It can be changed. Then we go to hear the federated IDs. These are social IDs, Facebook login, Google login, government IDs that they give you through some citizenship cards. First of all, they lack interoperability. My bank will never take Facebook login to lock into my bank account. And it also creates the man in the middle. You've got to go through a third party that can shut you down. It is prone to censorship. And when we're dealing with populations of displaced people, populations of people who may have been moved out of their country, populations of people who are not welcomed by others, they're prone to censorship. And then they have to rely on one single government or one single organization to provide proof of their identity. That's what we have today. And then you've got the fragmented. Oftentimes these are talked about the saviors of identity, the fragmented. These are your data graphs. These are your data trails. These are your proprietary analytics. The fact is they're really trustless in their marketing. Here's how to centralize. This is you and the organization they give you your identity. Why is that important? Because the state of Maryland can attest that I am over 21 and then I can operate a car. I can't drive a semi truck, but I can operate a car. And I can show that and I can get a beer and I can buy a Mazda, okay? The federated. Someone sits between you. Well, this is actually a real problem because, well, I refused to use Facebook login for anything. So I'm always having to register my phony email address to get to anything. But you have some standards here that are emerging and so those are adopted. But the federated identity model is prone to censorship and control. You always have to go through someone else to get your identity. And then we've got the fragmented where you have a bunch of devices and organizations collecting a bunch of things about you. Then you got some third party data. It all comes together and miraculously, it proves your identity. It proves that if you're a dog, you're a cat. It proves maybe a portion of what you are, but it doesn't prove who you are. It may prove one version of you, Heather Dahl the individual, but does it prove Heather Dahl the mother? Heather Dahl the CEO. Heather Dahl the daughter. It does improve the context in which I'm trying to have a relationship with you. So the fact is we do need a new model of identity to support the growing digital economy. You know, the line between our physical world and our digital world is becoming indistinguishable. Hence, I leave without my phone. But everything these days, even if I wasn't able to own a phone, a good part of my life is stored in the digital world. So what do we need from our identity to make this digital world one that can empower those that technology has left behind? Everything needs to be trusted, verifiable, importable, that you trust the issuer of the identity. Whether that's a government passport, business record, law enforcement document, that it's use case centric, that you're only sharing information that you need to share to receive the service or transaction. It's crucial, especially when you're relying on extreme amounts of data, that it's decentralized, that the data storage isn't one large data silo that's susceptible to hacks and breaches, or that it's susceptible to being shut down by governments or organizations, or that a company goes bankrupt. And then it's based on identity for all, that this identity can be implemented in cases with low resources, limited budget, slow and tech adoption, low literacy rates, and with limited connectivity. It's not just me that is talking about the case of this emerging decentralized identity or self-sovereign identity, forced to research when it looked global industry tech research firms, they tend to advise Fortune 1000. Published earlier this year that there's an emerging model of decentralized identity, and it will ultimately replace the model today. So when you think today about the systems that you're going to develop with property and land registration, think about what's about to happen in 10 years. Don't think about the identity that we were offered 10 years ago. And that's why I prutes show this, because when you have companies in the Fortune 1000 starting to think this way, it's important that you also think this way too. They talk about how digital businesses need to start looking at ways to implement customer identities where they can bring their identity with them. So what does this look like? Microsoft actually announced earlier this year that they're getting a way of username and passwords and you may have taken that with finally I don't have to update my password. But no, don't read it that way. Read it that way, the identity is changing and this is the first glimpse. Decentralized identity is moving into commercial adoption with products being released later this year. SSI's self-sovereign identity is gonna be recognized in over 55% of the identity access managements by 2021. And so as you look today and think what is the decentralized identity and how does it work? I put some extra slides in here because in two weeks you will remember nothing I said. But you will have my slides. You're gonna say, let me unpack what she talked about. Self-sovereign identity, and this is the global network that sovereign relies on, is what we call an identity meta system. You don't call sovereign and say I wanna buy a package. Now what we provide is an open source meaning that anyone, any engineer, can go in and use this code without paying a fee. It's accessible. And basically at the sovereign foundation we run the public blockchain in which decentralized identities are anchored to. Decentralized identities then are held in a digital agent app or a wallet so you can keep your identity in a wallet on your phone so I could go and just show my Maryland ID to a bartender and you could scan it and give me my beer. And then I'll go show you the next slide here to show you how it works. The most important thing that the sovereign foundation reduces is what we call the phone home problem. That when I go buy my beer after I'm done talking to you the state of Maryland doesn't know it. Because the bartender, when he verifies my credential he verifies it to a private blockchain that is anchored to a public blockchain. He doesn't ping the state of Maryland database to prove that I'm over 21. And so this is how it works. Actually I'm gonna jump through. This is how it works. And this is actually a use case that's going on today. The government of British Columbia actually issues all business registrations using sovereign identity. So when you think about property rights registrations I think it's a very nice inspiration. When you register a business in the state of actually, there we go, in British Columbia. British Columbia goes through the usual KYC that share a business, et cetera and they issue you an incorporation credential, a self sovereign incorporation credential which then goes to the CEO. It's held in her digital wallet. Now she decides she needs to buy a new tour bus for her eco tours. So she goes to her local bank, ATB Financial which is a very large bank in Canada. And she is able to show that her company is registered in British Columbia. And she's able to do that using zero knowledge proof. Maybe there's more information in that registration than she cares to share with ATB Financial and all they want to do is verify that she's actually registered. What we call zero knowledge proofs only provides the information to verify our needs. It doesn't give them everything that you have on an identity. So they're able to accept that credential from her digital wallet and then they read and verify the credential notice on the public ledger. They're not going to British Columbia and they're not letting the government know that she's applied for a loan. So why do they accept that existing trust relationship? Because as the bank I trust the government of British Columbia just like the bartender's gonna trust the state of Maryland to have gone through and prove that I'm over 21. Same type of relationship we have in the real world. So they can basically read the data, read the sovereign ledger and say yes, loan done. But I think this is a great example of what can be done with property rights as far as being able to give people and organizations proof of what they hold and do so in a self sovereign way meaning they can control the access to the information and the ability to share it. See you ledger is another case, very similar. They're issuing all credit union members a self sovereign identity so they can show proof of a credit. I respond as another case and with that, I'll wrap it up and we can take questions later. Thanks Heather. Is this thing on? So Emmanuel is going to join us remotely from New York so we're gonna beam him in in just a moment here. Hi Emmanuel, can you hear me? Yes. Hi, how are you? Very well thanks, thank you for joining us remotely. Is that audio good in the room, can you hear him? Okay, and I'll be operating with Flickr so just let me know. Okay, so yeah, so should I just start and share my screen? Yes? Yeah, I think you can go ahead and share your screen. Okay, so you can start screen sharing while the other parts of me is sharing, so okay. I've got a copy of your slides up. Okay. So if you can just tell me when you want me to advance them and we can go through in the same order. Okay, all right, great. All right, so let's start. So good morning everybody. So I hope the sound is okay. So I'm sorry I could travel to DC from New York today for I mean positive personal reasons, but yeah, nevertheless I had prevented me from traveling so I'm sorry about that. So but I'm very happy to present so for about like 15 minutes and to share some thoughts on using personal data for probably good. So I looked at the slides that were just presented in the two previous presentations and I think so what I'm going to talk about is like a good compliment. I mean it's definitely, you will see some common ideas and some common principles. I'm most familiar with the work of a flowminder for reasons I will explain. And so hopefully it sort of provides maybe a more general framework and also some actionable tools for doing the kinds of things that have been presented in the preceding presentations. So you see my some of my affiliations here. So I'm not going to spend too much time on that. So I'm mostly the director and co-founder of an NGO called Data Pop Alliance that is affiliated with the MIT and also the Harvard Humanity Initiative and flowminder is one of the core members of Data Pop. And I'm also the director of a project that I will talk about which is called OPAL which stands for Open Algorithms. So yeah, I will also explain. So all right, so let's go to the next slide please. So yeah, so we, I don't think there is, we shouldn't spend too much time on it but so far the past decade has been the decade of the data revolution or the digital revolution, the big data revolution with all the comparisons with data being the new oil and so on and so forth. So there's of course, there's been a great deal of excitement, also a great deal of skepticism, increasingly concerns and fears about this like all-wellion, dystopian future, the fears of a techno, centric techno-led revolution like increases in power, imbalances. But at the same time, the argument is that it's both here to stay and there is probably a lot of good that can come out of it. It's not that like the world is in a great shape and nothing needs to be addressed or fixed. So I think one of the biggest questions that we all are asking ourselves is basically how can we leverage data and especially the personal data which are the core of the data revolution so our behavioral data, data about our behaviors for the greater good and how can this be done. So next slide. And so to sort of highlight or stress some of what I think are sort of like structuring let's say parameters of this discussion. So this is a cartoon that I did several years ago now about six, seven years ago. So political cartooning is also something that I do sort of like on the side but also I use it to share thoughts on the topic. So this cartoon hopefully is sort of self-explanatory. So trust me, I'm a data scientist. And so, and it zooms on the microscope. So what it tries to capture or try to capture is both well simply the fact that indeed today we can see like human societies at these very fine levels of temporal geographic and also like complex levels of granularity. So all the intricacies and complexities of our lives and our interactions as social animals. And so through especially the use of cell phone data as like sensors and the risks that it can pose for privacy to start with and also the power imbalance. So you see this gentleman who's basically observing us and would possibly make decisions good or bad decision but on our behalf without these people being watched having any say in how these are there or our data are being used. So I think that these are the crux questions of that we are talking about today. And so next slide. And so maybe to the question, I mean to the, well let me just comment on this other cartoon first. So which I did six months ago at a meeting in Brussels of Euro stats. So the European, like say organization of official statistical agencies in Europe. And so on the policy side, of course there's policy making side. There are lots of expectations also that these kinds of data would yield. So indicators and better statistics and as you get into SEG monitoring and that will be able to do use cases and better planning and better all sorts of good things. So the data reality, data revolution is here. But the reality is that for all the work and the very good work that a lot of people have done including FlowMinder and others. But FlowMinder is of course one of the pioneering organization in this space. There's still little evidence or translation into policy. When I'm asked like you know what show me or tell me like one policy that was changed or designed on the basis of so-called data analytics it's very hard to find still. And so the joke here in this cartoon is that yes we can monitor poverty and inequality and lots of things at very fine levels of generics. But the bad news that we still can't do anything about it because what gets in the way? It's politics, it's business, it's inertia, it's bureaucracy. It's a whole range of things that make the link very difficult. And so next slide. And so maybe to the question of does your phone know where you are or where you live? I think that's the title. I mean the answer is of course, yes but. So yes, we are being monitored at night, the phone, all the apps know where you are but they know more where the SIM card is. And that's something also from a scientific and policy standpoint that is important to keep in mind that there are still, in many countries, SIM cards are shared and so it doesn't mean that one phone is one person. And so that has both scientific and political implications. But the general starting point of all these discussions is that we as a community of scientists and practitioners have realized a couple of years ago, 10, 15, 20, but especially over the past 10 years that yes, our phones, our data, our data crumbs said a lot about us personally and at the aggregate level. So I'm sure you're familiar with those like machine learning models, why and how it's possible to infer things like poverty, inequality, but also of course, population density, age structure, and so on and so forth using cell phone metadata. And so some of the use cases that Xavier presented or use that is just that the way a poor community looks like in cell phone data is very different from the way a rich community looks like. They call it different times, they call different distances. So you have those digital signatures and the real world outcomes or features that are correlated sort of like systematically. And so you find those correlations and that's what you use. So this little cartoon to try to explain like the gist of these like machine learning predictive models, how can you predict or infer some real world variable from the way we use our cell phones at the aggregate but also at the individual level but especially at the aggregate. So now next slide. But so of course, so that has a lot of, there's a lot of questions and concerns and obstacles in this statement making saying, well, there is evidence that we can do that that it could help them do their urban planning but then how does that exactly work? Can it be done systematically or do we have to do this one by one through non-disclosure agreements, through contracts because then it's sort of like always the same organizations that tend to have access to those data. So as much as you want to do good and be a do-gooder you sort of also use your privilege. You reinforce power and balances and there are lots of organizations and academic teams that do not have access to the right people. So that was sort of like five, six years ago where we sort of realized that yeah, there's this power, there's political economy in there that can actually prevent the diffusion of these techniques and keep them in the hands of a few. But I think the big picture, like the big question that we're asking is sort of like, yeah, how can we have it both ways? How can we leverage the wealth of information and also these tools as levers of change? I will try to explain what I mean by that for the greater good. And so the sort of starting point or endpoint as of today but starting point of this longer term, larger discussion is this idea of trying to create sort of like human artificial intelligence. So human AI. So the term is not from me, it's from Sandy Pentland at MIT. But so we've worked on it since and we wrote this paper here that you see. So you can see the title page here, like towards a human artificial intelligence for human development, leveraging those kinds of personal data. So what does that mean and what will you take? So it means, it really means two things when I think we're talking, when we talk about human AI. It's both using those tools, so artificial intelligence, so machine learning, so using those techniques and tools but also being inspired by those techniques and tools. And so what that means is is as giant, like artificial intelligence, so there is of course collective intelligence, there is culture, there is evolution. And if you think about it, what we as human societies learn from their mistakes and people learn from their mistakes and learn when they do something that works well, but we don't learn very well. So it's taking a lot of time for things like vaccination which we think there is science to prove that it works and should be done to actually be accepted widely as being something you should do. So if we were an algorithm and well we would probably say, yes, like you have to do this and Amazon is very good at that, Kayak is very good at doing that because they learn very fast from like these trial and errors based on data. So the big picture kind of question in framing is, okay, as human societies could we do that? So both be inspired by artificial intelligence and use artificial intelligence to our ends. And when I say inspired, it's not saying that everything is great about artificial intelligence. I mean, there are things that are very risky. Artificial AI gets things wrong. If the sample is biased, if the data is not good, then it leads to very bad decisions and outcomes of the AI. So we also have to learn from the weaknesses and the limitations of the AI but apply it to human societies and led by humans. So it's a sort of like human machine kind of relationship that we're envisioning, but for the benefit of human societies and also other like natural species and the environment. So next slide to become a bit more concrete. So if we think that it would be a good idea to base policies and programs and our own decisions more on data and evidence, the way an AI would do it. What, why is it not really happening? Why are we still talking about whether or not climate change is caused to any extent by human activity when it, well, seems to be the case? Why are we, so all the, there is increasingly this notion that what we, our policies and governments and are not really fact or evidence-based, so why is that? So I would say that first, so yeah, first some impediments that we have to work on. So there is this lack of appropriate data, culture, capacity systems and connection. So that's sort of like a structural big picture kind of impediment. We don't have the systems in place to do that. So it may be actually risky to do that and most people or most organizations don't have the capacities. We work a lot as a, for many others in, so let's call them like developing countries for like a better term. Senegal, for instance, I mean, the National Historical Office is trying to do the census every 10 years. Now we're telling them you should use big data so there's a bit of a stretch. The second reason is that, of course, there are bad incentives from powerful agents. They are not all bad, obviously, but still, so some of them use, try to use these techniques, very powerful techniques to their advantage or others just say, well, I'm fine with the sort of like alternative fact narrative so I don't need your data, I don't need your insights and your facts. I'll keep doing what I've been doing for a while and it seems to be working out well for me. There is also the fact that in this human AI, if we don't agree on facts, then we can't learn. If there is no sense of also culture of compromise of consensus, then we can't learn. If you wanna learn what reduces crime, well, the first thing is to be able to agree that something is a crime, for instance, otherwise you can't adjust, you can't learn and adjust. And also last, so that's more let's say the political or political economy argument or reason, is that there is, as I mentioned earlier, no real killer case, something we can point to and say like this has saved lives in this country, in this specific instance, contrary to what you can tell about vaccines, for example, so given the risks for privacy, et cetera, which I'm getting to now, it makes it hard to make the case convincingly that this is something that is worth doing. So next slide. So what do we do with, yeah, what can we do with all of that? So hopefully, yes, I think this last section of my presentation echoes a lot of, I think what has been said before. So there is clearly this question of how can we share, but also use, and I will come to that in the end. So first of all, like share, or actually, yeah, it's more share, not the data, but the insights. So extract, let's say like the value from these very personal sensitive data without turning this into an Orwellian to use the term that seems to be like mandatory, an Orwellian kind of a society. So we wrote a paper that was published in Scientific Data Nature last year with a fairly large group of people, including people from Flamender and UN Global Pulse and others, about two different models. And one of those models is this question and answer model, which is, which has inspired and is reflected and promoted by the OPAL approach and the OPAL project, which I talk about. So next slide. So OPAL, and really I've looked at this time, it echoes especially the previous presentation. So it's based on what I think we all think is sound, which is don't share the data, like don't share the personal data, don't pool them in a large lake or something. Otherwise, it's a recipe for if there is a bridge, then everything is great. So these are like federated data systems. The data are stored, collected, stay in the servers where they are first collected. And then there are open algorithms that can run on this data. So let's take as the case of OPAL, the case of telecom operators data. So we work with Orange and Sonatelle in, sorry, Sonatelle in Senegal and Telefonica in Colombia. And so we will send an open, so we send an open algorithms. We can discuss what open means, but let's use open algorithms for now. And then you get things like a population density estimations or poverty estimations. So it's also quite similar to what FloMinder is doing with FloKit and what others are doing, but we really try to do this like in a box so that you can get, arrive somewhere and install that. So it's a question and answer model and so there's auditing and logging and so we try to increase accountability, et cetera. So next slide. So what is, yeah, the map didn't, so this is just to explain exactly what we're doing. So I mentioned some of the, yeah, you can go to the next slide. So we're doing in Senegal and Colombia to actually show that and how it works. So this ability to mine, send questions to sensitive data and only get the answer. So something that I think is key is not to think that it's only like a supply problem, that it's only that the reason why these data are not used or data or even statistics are not used as much and as well as they should is not only because there is no data available, such that if you fix that, it doesn't mean that all of a sudden governments and decision makers around the world are gonna say, oh, if only I had known that this was bad for the environment. In most cases, we know or they know. So it's not just that. So here, what I'm getting at is that there is, of course, a technology component. We want to be able to do it safely and at scale in a privacy preserving manner and you have like blockchain stuff and so all sorts of like technological ways of ensuring that you cannot re-identify people, et cetera. But there is also a governance side and the governance side is about ensuring that people can have a say in how there are data are used that people have the capacity. So forums to express how they want their data to be used or not to be used, but also the technical capacity. So we try and do some capacity building and trainings and improving data literacy, et cetera, such that they can also use themselves and weigh in, et cetera. And of course, what we do is a drop in the ocean, clearly in there, the needs are huge. But still we're trying. So we've done OPAL in two countries, as I mentioned. The goal is to expand to more countries and more industries in the next couple of years by building this OPAL in a box, so like system where you would have a technological, a set of technological features, parameters, systems and also a set of governance, rules and standards and sort of like a toolkit that you could install to do that. So next slide, and I'm almost done. So, yeah, so OPAL is operational now in Senegal, so this is a screenshot of someone from Berlin. So asking, can you tell me the population density estimation in Senegal of all communes in Senegal? So of course it's very small, you can see, but you get a sense, so it's like API based. This algorithm is calibrated. And so the question of calibration, I think is going to be defining criterion as to whether this is gonna work in the long run or not because there are all sorts of biases when using large datasets. And also like cultural and anthropological reasons why you cannot just say, oh, this is what the cell phone data is telling me. So therefore, since there's a 100% penetration rate, I'm gonna assume that this is representative. It's not representative. So next slide. So this is the vision for like OPAL and a lot of those approaches in the future, which is it's not one industry. So here we start with telco, but we also want to be able to mine or query national statistical office, so NSO national statistical office data. A lot of those data are also very sensitive. So micro data collected by NSOs are very sensitive. So we would like researchers, other parts of governments, civil society organizations, businesses to also be able to query those data, but in a way that does not lead to or increase the risk of breaches and leaks. Energy is also very promising. And in the future, we'd like this to be integrative such that if you have a question in the form of an open algorithm, you can run on those different datasets and then you have the results. Next slide. Yeah, next slide. I will finish in two minutes. I'm sorry to interrupt, but I realize you can't see the time cards in the room. So we're nearing the time for the Q and A, so if you could wrap it up in the next minute or two, that'd be great. Thank you. Yeah, I'll be done in less than two minutes. So, and as I said, it's not just tech. So there needs to be also this more like this connecting tissue that people understand what's happening can have a same in what's happening. So that's also one of the big priorities of both OPAL and the work that Datapop does, which it's not just about a supply side problem, it's also about a more like a systemic problem of power imbalance and lack of capacities. So next slide, and I think there are two slides left. So yeah, I think this notion of a, so the goal is really, it's quite transformative. I mean, the vision is that where people would be able to pool their data, and so into like, for instance of data cooperatives, and then everybody would have a saying, like, do I want my data to be there? I have a right to have a copy of my data, but you can pool them the same way you pooled your money in banks, and that would allow societies, people as data emitters as a group collectively to ask questions, and then they can check if the government is doing a good job if, as a Lyft driver, are you paid the same thing as the rest of the aggregates? I mean, you can ask a lot of questions, and so the point here is that the technology is there, or almost there, and what is missing is now the governance of this, and GDPR is a first step, and then there are lots of other governance steps that need to be put in place for this to work. So next, and I think the last slide. Yeah, we can just keep that. I mean, this is just an example of the kind of work that Datapop does, and so we take a regional approach to trying to implement those kinds of projects and systems at scale, concretely, regionally, with regional partners. And so last, I believe this is the last slide, this is the second to last, but yeah, so the main messages here, so the previous slide, the main messages now is the time to do this systematically and at scale, so to really not, yes, stop piloting testing but really do this systematically and at scale, and it's possible, and the last messages are, yeah, so what needs to be taken into account are the politics of it, there are lots of powerful interests, so lots of politics going on, make sure that everything we do is based on the latest tech and science, a couple of years ago, we thought that just anonymizing data was okay, but then we realized that actually we can make a lot of inferences just based on even pseudonymized data. Ethics and rights, and I would just stress that privacy is not just about safety, but it's also privacy as agency, so it is the right and ability to say what you want to reveal about yourself, so it's not just about protecting the data, it's also about allowing people to actually have a say in how their data are used or not used, and then there is of course the economics. The economics is there will need to be a business model, regulations, subsidies that make it sustainable from an economic standpoint, because right now, at the moment, it's a bit like the far west of the very early years of how are we going to, as companies and society, how are we going to, how do we value that, and how do we take a freight such that it can actually run systematically, thank you. Thank you very much, Emmanuel. So we're gonna move on to the Q and A, it's gonna be a little bit abbreviated as we're running on here, but if the other presenters would come and join me on the stage. Okay, so given that we're a little pressed for time, I'm not gonna hog all of their attention and ask my own questions, but I think we'll start by throwing it out to some of the questions from the remote viewers. Tim, do you have a microphone? Yeah, if you'd like. Not sure if I'm right. Yeah, you're right. Let's pick on some of that, perhaps we can turn to just like, one minute, when I first hear that question, I think the important part is looking at the privacy policy involved without data, GDPR compliance of the data, the governance around the data, and then how do you know that's the accurate data, so the identity around the data set. That's how I start unpacking the question from a technical perspective. I'm not entrenched in the data analytics world, but what I would say is you have service, digital services in which you create data about yourself. For instance, I host an Airbnb condo and I'm a super host and I've had five star ratings for 100 visits. What does that say about me as a potential employee? Would you want to hire someone that has that kind of track record? Now, how do I prove that in a credential to someone to show I could just get a screen capture, but how can I take that data set that I've generated about me and my performance and use it elsewhere in my life separate from Airbnb? And that's where the self-sovereign identity comes in, is taking those data sets, putting them into a credential that an individual can use to prove something about themselves to another party in a privacy-preserving and secure way. We'll put it another way. I mean, there's an issue of data provenance. Collectively, you produce all of this data by your interactions with Google, with Airbnb, with Uber, with whoever, but bring it back together to create a picture of yourself improving that it belongs to you and it was about you is really challenging and digital identity, obviously, is a core part of addressing that challenge. And we'll also have to chime in, or next question. So, if I understood the question from the web viewer, the idea was, okay, can you reverse engineer and aggregate that back to the individual level? For every one day, at least not from under. So, yeah, so maybe just, yeah, so on that point that you just mentioned, so yeah, so the latest research about reverse engineering or like re-identity, it's re-identity, re-identity is somehow someone is going to be able to look through the window or that you find your number so what happens is that, so you anonymize data and then so everybody looks like, looks like, yeah, you don't know who's who, but then it's possible you were so unique, so this is the work of E. Alexander Mongeau and others, were so unique that then you can say this data trail belongs to one individual and one individual only and then what happens or the risk is that if there is another data set that is connected to that individual that you carry around with you if you have a phone, for instance, then you may find another data set that has your name and number, so it's by cross-referencing different data sets that you can actually re-identify people, so this is the risk that, yeah, some of the attack risks that people are working on and just one quick, first Airbnb, so with respect to the technology, et cetera, but I'm really sort of concerned that of course, we all know that there can be value in being able to say, well, yeah, trust me, I'm trustworthy and here is some example, I'm a good Airbnb host or I have those, I mean, the credit score, the American credit score is about proving your trustworthiness at least financially, but I would say my concern or my fear is that only a few people are able to host like Airbnb, for instance, and that so it could actually, this culture of digital credibility can be abused and it can become sort of like an obsession and you have a lot of people who are sort of obsessed about their digital persona and how they appear online, what they say on Facebook and then if you say something wrong on Facebook and Twitter, you can actually ruin your life. So that's more a cultural thing, but I think we have to keep in mind that, yeah, people can make mistakes, that it's not because you're a bad Airbnb host, that, yeah, that you're a bad employer, sorry, a bad worker, and that it might not be you also hosting the Airbnb. Okay, that's... Is it really you? Thank you, Emmanuel. I think we're gonna move on to the next question. So you say, why don't people believe facts? And so you comment, you say we use this technology and on and on, they say, well, that's interesting, but I still don't really believe you. And if you say, well, this displaced widow, listen to how she knows what her fence looks like, she knows how far it was from the town, and they say, well, maybe she does, she's a woman, so maybe she doesn't, or they say, okay, she does, but maybe her husband's family should take it instead. And then also another example for this question about, Xavier, you're posed at the end, I guess a question about a data trust. You said, we don't have the answers, but what happens to all this data? So in the end, when you collect it, what, if a country says, thanks so much, this is all ours now, and you can't use it, and you can't use your development work, and isn't it so interesting that everybody's going to these big privacy trials in the European Union courts, and maybe we'll try it too. So, and maybe you have some insight into that as well. I would say 80% of the time that I spend today is talking with government and policy agencies, officials, and a lot of attorneys. Governance is very important, trust frameworks, governance frameworks around how these systems are used, and I spend a lot of work working on a governance framework, and I think the Sovereign Foundation is one of the most mature in the world. But what's more important in my situation with Sovereign is that our network is actually being used by governments. It's not only Canada and British Columbia, it's Ontario, it's actually, I think we are up to six in Canada, it's New Zealand, it's being implemented in Afghanistan, it's actually, I was down talking with the Federal Reserve Bank of South Africa two weeks ago, because they're moving towards adopting Sovereign and one of their largest financial institutions will be giving Sovereign IDs to the unbanked or those who can have accounts of $100 or less. So when I go talk to policy makers, I actually have use cases that are at commercial or being used by millions of people, and I think that helps explain that this isn't just a concept, this isn't just about managing data, this is about looking at a new way of using identity in our digital lives. So that starts to address the gap between policy and the technology. Thank you, so just a quick comment on, since we're talking about mobile phone use and specifically our gender gap, particularly in low and middle income countries. So that I think has your implications for whatever applications are going to be mobile phone based. Even if we're to put that aside, what the question has to be obtained consent, especially when we don't know what these kinds of things could be used for. Yeah, sure, I mean that's a good question. Depending on the regulation you are under, research might be kind of enough of an excuse for not having to ask expressively the consent to use the data from the data subject. Now, I mean typically the use case we're talking about today, which is like using this type of data to strengthen property rights. To me could be really a good case of motivating individuals to share this type of data. And you could have really like, to me could make sense to offer to developing a kind of wallet whereby they're going to know that okay, sharing this data there are going to make me a better position to claim a social benefit, to access some services and perhaps to claim property rights. And then you would have like a direct opt-in close for research or for humanitarian and development organizations. So I think that would be really ideal is this type of bottom up approach to large scale data collection when we're thinking about using at scale these data to support an orient policy intervention. When we're talking about property rights, there is no question here we're not going to use the data of someone. It's this someone who is going to use his own data to claim this property. Understand of your question, you had the question about gender and then the question about agency, which are obviously interdependent. I think the question about gender, I mean obviously there is a disproportionate use amongst men but that also can change as we get into camp scenarios, especially in locations where camps are overwhelmingly female. And actually in those cases, use of mobile devices spikes significantly amongst women when compared to the sort of overall population. So in some refugee camps and some IDP camps you see significantly higher mobile device penetration amongst women than you would in the general population for a number of reasons, right? One, I mean camps are full of women in a lot of places because men of fighting age are out or are dying. But then also you're seeing phones being given so that people can keep in touch. Not necessarily an answer to your question of how to address the problem but it does change the penetration a little bit. And then at least when it comes to the way digitally designed HLP approaches agency, I think it's a little different because instead of coming at it from working with governments and saying this is the tool, it's down to the individual user, right? And I think that that's where Emmanuel's point about data literacy and data agency are incredibly important, right? Is to focus on this is what you have or this is what you may be able to get. And then these are the options that you can use them for, right? And then you get to make the decision what level of information, what type of information, how much you share or is it just about increasing awareness so that you know, right? Coming down the line when you're making HLP claims what can count as evidence. You may not gather any of that or you may keep it on your phone instead of backing it up to the cloud but you make those decisions yourself. Your question is near and dear to my heart because before sovereign I spent six years working on solutions that prevented the exploitation of children. First of all that I respond to use case I mentioned the problem that they were solving and are solving is in Thailand you have fishing boats that go out with a certain number of workers in order to save costs, they come back with fewer. And so the problem is that they're solving how do you make sure all workers placed on boats return from sea? And you have workers, you have no devices, no digital footprints, they're using biometrics to a decentralized identity but also doing it a privacy preserving way to remove the biometric at certain points in the transaction. It's being conducted in very remote locations and why is it being able to do that? You don't need connectivity to use sovereign because it's something that we call a state proof where it captures a mirror of the network at the time of capture and you decide how updated you need the information. It's using very low cost devices, raspberry pies which can be developed in 50 bucks or less. And the individuals don't have to carry any paper or anything on them in order to prove that they're on the boat and off the boat. So I think that's an important use case to start addressing those. When we talk about property rights, oftentimes we think about the individual owner. We have to also think about guardianship of those individuals. And I'm about to publish paper on guardianship in about two to three weeks. I think we'll be inspirational on thinking through those issues. Yeah, I apologize for coming here but we are running a little bit over time. And also to everyone else in the room who may have questions that they weren't able to ask in the afternoon sessions after lunch which is about to begin, we'll have a chance to speak directly and at length with all of the speakers here to address those questions. So thank you all very much. Thank you.