All right, cool. Thanks for making it, even though it's like the last talk of the day. So I'm excited to talk a little bit about how we use Ed Proto for funding impact and resource allocation. And luckily I'm actually not here by myself, but there's also Shafi in the back and Adam here. So if you have additional questions later, especially on the technical side, both of them can definitely help you more than I can because I'm actually a political economist. By now I also have acquired some technical knowledge, but it's not my expertise, but it is theirs.
Cool. And also I will in this talk sometimes say we are building this, we are building that. It is not just kind of the hypersets foundation that I'm stewarding, but there are other organizations like Gainforest and MaEarth that are building with us these impact funding mechanisms. And hypersarts is one of the primitives that we are stewarding. So that's why the hypersets collective is the word that we use for like the organizations and the people, the contributors who help build this. I deeply believe that right now in the current conditions, what we choose to fund is the world that we are creating.
And the exciting thing is that Ed Proto has a lot to say about this. And this is in that way, I think one of the most important use cases where we can apply Ed Proto 2. Why this is important is because there is something that we collectively value that doesn't have a normal uh business model. And this is actually a lot of things that we care about. Digital public goods, open source software, regional plant projects, science, journalism, community organizing. Those are actually the things that make a good life. And we are actually not really funding them.
They are all underfunded. If we would put more resources into this, we would actually create a better world. How I am thinking about this is that when we make decisions together what we want to fund, we have at least these three layers. We have funding mechanisms. This can be grants, milestone bounties, price competitions, crowdfunding, etc. etc. So they are different funding mechanisms. Then there are different decision-making mechanisms. That can be a jury in a foundation. That can be also in crowdfunding, kind of just like the crowd deciding. It can be any democratic mechanism, deliberate deliberative mechanism.
It can be just a funder saying it's my money, I'm the dictator about this money, that's also fine. But these are different decision-making mechanisms, which I think now is an exciting part that we will see a lot of innovation in the future, because with like kind of human AI collaboration, how we can make decisions in the future will really change. And this also interacts with the context layer, because now the AIs can read for us much more context than what we can. So from my perspective, sometimes I say it's like the dream of a political scientist come true, because you can come up with funding mechanism and decision making mechanisms.
In the past, the voters just didn't behave the way you wanted them to behave. They were not informed, they didn't actually do what would be necessary to actually reach good decisions. So the context layer that's where atproto comes in. Because context, like the context for an AI, is what do we put into the decision making mechanism. So in the case of regionative land project, this is, for example, satellite imagery, this is bioacoustics data, but also community input and expert input. So in the case when there's a regional land project, I grew up in a small town of 300 people in Germany.
If there's a is if there's a project, the people in the town they know if it's good. Or if it's just some bullshit carbon project that is not actually helping stewarding nature. So why do we not just ask the people that already know? And then, yes, let's add some technical data from satellite imagery that support that that um that opinion, or if they actually disagree with each other, then we can also have a conflict resolution mechanism where, like, okay, we see there is some kind of a conflict. Um, but this is exciting because atproto makes it easy to collect this context and it makes it possible that we are not just collecting the data, but we are also collecting who said it.
Because that is important when we analyze it, because it depends on your your reputation, if this is if this really should be considered in the decision making process or not. So this is all the kind of a loop because the project actually create the work, the impact, then there will be third party measurements and evaluations, community input, all of that. And so if we do that loop over and over, we can also over time improve that decision making mechanism because this is not a static system, this is ideally a self-improving system, similar to how we now train AIs to become better over time with this context and the tools that we have here now.
We can also over time make better decisions that what we have today. Trust signals don't accumulate. That's exactly what we see from social media, but it's the same in funding data. You have like a crowdfunding platform, you have the foundation that raises like that gives us out grants with data. So these are all data silos, and the projects have to put in their data over and over again. So we think we can change that. And we are collaborating with a bunch of projects here that are working on this. We are trying to kind of help build one context that different platforms can tap into.
And if you don't recognize all these logos, we will get into that. This is where I normally spend a lot of time on, which is nice because here I don't. But all of these reasons why you want decentralized social media also apply to why do you want to have decentralized information for decision making mechanisms that concern the public good. Another one is in the ideal case, we will actually create a lot of network effects. So network effects, if there are more funders, obviously that's good for projects in this network, because when you have your data, more funders can see that data.
But also for projects, it's good if you have a lot of evaluators because you will get better information to the funders. And I mean it's only good for the good projects, not for the other ones. But there's everybody benefits from more actors contributing to this network. And what we of course want is that the value compounds across the network and not inside of a platform. Another reason is that when we think about the projects and funders and evaluators today, this is how we might think about it, but already today you don't write your grant application yourself anymore.
You actually let it let somebody write it in AI, and the funders they get so many applications they have to use AI to sort them and find out, like at least pre-select some of them, and then they make decisions on what to fund. So this is the world that we probably have today. And I think atproto makes it possible that we put like shared observable records in the middle, and I think this is really important to make the system accountable to what we as humans actually still want to fund. This is kind of an overview of like the shift that we want to get to what we believe is really a modern funding stack, and it's all that that I already introduced.
It is users control their data, we have many evaluators, we have many interoperable funding mechanisms. It is about narratives, verified data, and trust-based funding, not just about I have really nice pictures. Um, it is about human error, collective intelligence, and we are building self-reinforcing systems here. All right, um, switching gears a little bit. This was like the the high level where do we want to go? This is a little bit of how. So a hypersert is kind of a living record of work. So it starts with just a record, um, who did what, when and where.
This can be for the past or for the future, so it can be a proposal as well. Um, this is kind of the record here, and in the ad proto world, of course, this is just like one record, and in our case, there's an activity claim. Then the evaluate is all the data that can be added to that to really make sense of it. And this can be self-evaluations or additional data from third parties, but that's why we say it's a living record, because it over time, more and more information about this work will accumulate. And that makes it possible to really make good funding decisions, funding decisions either as grants, because you also observe the the the other work that they have done in the past, you have the reputation, or you have outcome based funding where you see somebody has actually proves that their work was meaningful, and that's why they get funding afterwards.
All right, let's get into use cases. MarEarth is a um a project that um basically wants to fund the regeneration, and that means funding a lot of projects. And when I say a lot of projects, we actually need to fund hundreds of thousands of projects in the world that are really connected to the land that they are stewarding. And this is not just we like we cannot scale the projects up because then the projects actually lose the connection with the land. So this is really a challenge of a kind of decentralized sense-making system. How do we actually know where all the projects are?
Um they are taking on this challenge. They're using hyper certs to record the data and add kind of the other data and stay interoperable with other platforms that also work in um in this space. They will have a uh crowdfunding um campaign in June, I believe. One second. So in June, um crowdfunding campaign for 100 uh land projects with a matching fund of 500,000. So this really will get adoption um latest then. Maybe some other platforms um do already some prototypes before, but that's our really the first big use case. 100 projects will be onboarded, um, and then all the communities will join to um to fund these projects.
Um another project, Gainforest, and Shafi the is the co-founder of that in the back, um, also works in the same space, and that's exciting because Gainforest also works on how do we help communities deploy technology like bioacoustics to give them tools to prove what they have done. And so when communities that are part of the crowdfunding campaign for my Earth go to GainForce and work with them and create data, then this data is recorded on NetProto and shows up on the crowdfunding platform for my Earth. Um the time is not working anymore. Oh, I have a time here.
Okay, cool. Yeah, let me know if I when I have like five minutes. Cool. Um so here we really see already why the usefulness, like the usefulness of the interoperability between platforms. Um GainForce also is building um a platform to help the communities fundraise directly. This is called BoomiSerts. So we have here Boomi certs where you can see um kind of the the different communities and you can fund them directly here. So a Boomi Cert is a hyper cert, and you have basically like a very familiar interface, but it's based on the atproto shared shared records.
Um another use case, uh Protocol Apps. So I actually started the project about hyper certs AT Protocol Apps, then spun it out to the HyperSerts Foundation. Um Protocol Apps will actually use hyper certs in their platform or like use it right now already if you apply to a grant for PL research, you actually create a hypersert. And actually I because we have like how much time time? 17. 17, okay. Then actually, because uh the Boomi CERTs platform here can also show it to you directly. There's for example bees and trees. Um, and then you can see the site boundaries where it actually is, you can donate to it.
Um so this is still in beta, but this is like one of the platforms. Then when we have uh protocol labs, can actually see that I can sign in to the grants platform here with my proto account, which is cool, and then start an applic application, and here you see it was saved as a hypercert on their personal data server. Um then another um application is Simocracy. Smocracy is the idea that the decision making mechanisms uh is like painfully painful and costs a lot of time. So, what if we would have just a digital twin for ourselves who acts on our behalf to make to have all the discussions and make the decisions for us.
So we ran a experiment at funding the commons in San Francisco two weeks ago, where there it was, I don't know who knows Frontier Towers, but it's basically a co-working place. There are multiple floor leads that organize the co-working space and we distributed 5,000 US dollars to things that happened in the co-working space. And the way it was done is all of them had like a 30-minute structured interview that created their AI agent, and then the and anybody could propose things. Every proposal was a hyper thread. Then the agents were actually discussing among themselves what should be funded with a complicated mechanisms that's called the S process, which is a really cool process where you aggregate different marginal utility curves, which most people cannot do.
But it is actually really like in theory, like an amazing process that's used by the survival and flourishing fund. And then the result was like we then compared the result to what the humans actually would have decided themselves. And there was not a clear this is clearly better, but it was a reasonable thing that the AI AI agents actually decided. And the exciting part of this is this is like a really bare bare model of this, and it will just improve over time. And that's the cool thing. That's like we we have a model that kind of works already, but it only has like possibilities to improve.
So that's Symracy and so we had the Frontier Towers, and you can actually see all the the AI agents. Um you can also chat with them to get to know them. They have like a constitution, values, etc. Um, another application, and this actually comes out of the idea of um if you have a conference like this, we have the beautiful signs of like who funded this this conference, but actually in the background there are lots more contributors that are not on there, like the volunteers. And so can we actually, when we have work in a hypercert, can we show not only um the financial contributors but also the in-kind contributors, the volunteers, and um this is in that way just a different view on um kind of the uh the work and the contributors.
So when we take, for example, a GitHub repo here, and I paste it in here. We can fetch all the contributors, and I guess this is actually the repo for the website. So we can see, I mean, this is like I don't say this is the actual contribution because this is just based on the GitHub commits, but you can see who actually contributed to it. And if you now say, actually, we all know that uh Boris actually didn't do it, uh, then we can decrease Boris uh that's right. So that and and now you can think of okay, what is the community mechanism to actually do this right?
And you can also say now, okay, actually the financial contributors are also meaningful, let's add them as well. And um then, I mean, there are some settings to make it beautiful, you can have like a background, etc. etc. You can share it anywhere. Um, but you can also think of like every open source project could have that on their website with a donate button, and if you donate, you also come to the um to the hyperboard. So I think uh these are some of the use cases. Um of the use cases and um switch gears again.
Except I maybe pause here if somebody has a question in between. Um whether you have looked at value flows. Value flows. Yeah, I mean, if there's a specific solution, I think there are like quite a few projects that have thought about how how like value flows and um also like source cred and others. There have been like projects for um attribution. Um I think this is something where we want to really focus on the context layer, and then lots of different attribution mechanisms can be built on top of it because we are not the ones who can say you are this community should use this.
It needs to come out of the community. Can you explain more how project data collection and evaluation on the project network? How the the lexicon records work in that way? Or like how does an organization that wants to be funded, right? Like why kind of what is that we need a function online? What they needed to do. Yeah. So this this would be, for example, like this would be a specific platform that is built on top of this. So then you would talk about my Earth, Gain Forest Democracy, and how they would implement it, because we are just implementing kind of the lexicons that can be used by them.
And in MaEarth case, for example, in the first round, so it's basically a hundred projects will be selected by like more classical selection panel or like actually like together with partner organizations that there is a minimum quality that we already know about the 100 projects, and then on the um on the side, the projects can add more of their own data. Um we will work with like trusted evaluators that we know that can also add data. So this this is in that way the choice of Mar Earth, how they want to use the the hyper search standard that we are proposing.
So it's not that we like it's kind of the platform's choice. Yeah, but happy to chat more about it afterwards. Cool. So changing gears a little bit into what is what is the underlying technology that we have because we had some challenges from the user experience where how do we actually build for land stewards? Um how do we build not only for land stewards in the US and Europe, but actually for land stewards in the global majority. And um how do we explain to them what Ed Proto is, or can we just avoid that? Um so what you see here is basically the um we have of course here defined the lexicons for the projects, etc.
We have the personal data servers, but we also have also something called the certified group service. Um I will go into that in a second what that is. Um this is something that has not been built, but the project was originally very Web3 token centric. Um and there might be still a use case where you want to um add kind of an on-chain component to it when you for example, similar to carbon credits, when you want to make sure that somebody really can prove that they funded exactly how much of an impact. I think we can also find solutions that are more atproto native, um, but potentially there's a part where um on-chain data might be useful.
Um then yeah, the platforms I I have talked about already. Um something else that we um built is kind of a login experience that is uh just password less. Um because the user experience for my earth, when you go to my earth, you log in, and it tells you something about AdProto, it tells you something about a different service that you don't know. You actually already use uh loose users. And so what we did is you basically sign up, you put in your email, you get a one-time password, you put it in, you you have signed up, you have signed up, and you have a user account, and then later on you can set your your AdProto handle, you can set your password afterwards.
Um so this is kind of uh certified. Um the login, it's not from the UI, it's not very settments. Um from the login, it's not very exciting because from the UI it's just like very very basic. Um, but it's not something that is that easy actually to do um in the Ad Proto ecosystem. And this is uh something where if you want to know more about the technical details, uh talk to Adam. Um it really is about reducing the friction during onboarding and what you saw actually on this slide. If somebody comes to any of these platforms, they still can just sign in with AdProto or Bluesky with their normal um normal way, but they can also use our password less uh login.
Um cool. Certified group service. Um and I will switch briefly. Because I think I can also show it to you, actually. I will show you the exciting way to log in with OTP. So I think I don't know if there has been an easier log into the Atmosphere before. But here you can see there are different groups. There are some groups that have been and some groups that I have publicly acknowledged, and I have just created a lot of test groups. Some others that I have not publicly acknowledged, so publicly it's not known that I'm part of it.
Here the hyperserts foundation and kind of the very common UI as well. I can just like switch to the hyperserts foundation. I'm logged into the profile. I can see the members.
And that's kind of the uh certified group service. And this is in that way, there's an overlap with what we heard today. I don't know who was there from Brittany, Brittany. So and we will check how we have built it, how they have built it. But there's definitely something where I'm very excited about how we can allow groups to have accounts because in our use case, when we have land stewards, they have projects, they might have volunteers, the volunteers take pictures of tree planting, and they should publish that for the organization, and they should potentially not be able to interact with all the other data entries.
So this is kind of the interesting role-based access control that we can implement here.
Two minutes, all right. This is how we implemented the group service. So when like the group has a vanilla PDS, that also means they can transfer the data whenever they want to anywhere else. And the user when they want to write into the group repo, just ask the PLC registry because the group is also registered as a normal DID. And in the PLC registry, there's a service that the user then knows, okay. This is the group service that I have to ask. The group service checks can this user act on behalf of this group. If so, then the group service has the app password to actually write into the group service, and then it's also recorded who did what for uh on behalf of the group.
Um and this is interesting for role-based access control, but also for so many more use cases. If we have um private spaces, then who is the owner of the private space, maybe it should be a group. And maybe the governance mechanism should be an addition to the group service. Um cool. If you want to build with hyperserts, um we have built a couple of things, a scaffold app that's easy to use as an example. There's a CLI. Um you can also just point your agent to hyperscan.dev slash agents. Um there's an indexer, Windex is our data.
Um I introduced certified as an easy login experience, but you can also use any other. And um hyperscan.dev um gets you to an overview of everything that happens in the hyperserts ecosystem. And additional tools, also like badges and um identity link, which um potentially links like a web 3 wallet to your identity that I haven't really talked about, but all of that you can also find on hyperscan. Um yeah, and we are here to talk. Um if you want to build with us, um please switch out to see on the bottom or just talk to us uh right after this or um while we are here at the conference.
Um where are you based there? Uh everywhere. Um I'm based in general in Berlin, but traveling a bunch. And then we have uh London and Zurich, Paris, we have a developer in Bhutan, um, and then the also the My Earth crew that we are building very closely with. Um they're also kind of global. Um yeah. Thank you.