# Self-Hosted, Mesh and Decentralized Infra in Practice | A. Burns, B.C. Royes, L. Steininger, Calle - Channel: [Ethereum Cypherpunk Congress](https://streameth.org/ethereum-cypherpunk-congress) - Date: 2026-07-09 - Duration: 32:37 - Topics: web3, privacy, now, crypto, cryptography, blockchain, data, security, human right, rights, tech, technology, internet, open source, free, freedom, ethereum, hackers, ethics, cypherpunk, dev, developer, dapp, decentralization, bitcoin, computer, surveillance, cyber, peer2peer, p2p, love, solidity, zk, zero knowledge, education, academy, w3pn, neo, summit, berlin, neocypherpunk - Watch: https://streameth.org/watch/yt-lYNUCj5iQ6s - YouTube: https://www.youtube.com/watch?v=lYNUCj5iQ6s ## Description “Building the Stack: Self-Hosted, Mesh and Decentralized Infrastructure in Practice” Adam Burns, Bradley Clark Royes, Liz Steininger, Calle Moderated by: Beth McCarthy Uniting technical and cultural perspectives on stewarding p2p, self-hosted, and decentralized technologies, from a diverse panel of builders with expertise spanning "the stack". From the networked level (Adam's work with wireless networks at Freifunk) to infrastructure (Bradley's AI compute node and Liz's cryptographically secure storage) to protocol (Calle's OS bitcoin protocols Cashu and Bitchat), "the stack" designs itself toward secure self-sovereignty. This panel explores the unique challenges of each use case, and the tensions that can arise from contradictions in intent and implementation across levels of the stack – for example, accessibility and openness vs. privacy. Identifying and understanding these points of tension and interdependence through a systemic lens empowers us to build more resilient infrastructure, from tooling to hardware to the community of contributors to maintain it. Speakers Adam Burns — free2air, Dyne X: https://x.com/free2air Bradley Clark Royes — Foresight Institute https://foresight.org/ LinkedIn: https://www.linkedin.com/in/bradleyclarkroyes/ Liz Steininger — Least Authority, Private Storage https://leastauthority.com/ X: https://x.com/lizstininger Calle — Bitchat, Cashu X: https://x.com/callebtc Cashu: https://cashu.space/ Moderated by Beth McCarthy — Web3Privacy Now X: https://x.com/ontologymachine Follow Web3Privacy Now: 🌐 https://new.web3privacy.info 𝕏 https://x.com/web3privacy 🦋 https://bsky.app/profile/web3privacy.info 📸 https://www.instagram.com/web3privacy_now/ 🎤 Neocypherpunk Summit: https://s26ber.web3privacy.info/ Subscribe for more talks on privacy, cryptography, digital rights, decentralized infrastructure, and the future of the open web. ## Transcript Cool. So, uh I'm excited to follow the panel that we just had with uh this panel that like goes deeper into what challenges we have with building a decentralized stack, what that means. Um and uh addressing some of the things that came up in the last one around how can we be more resilient through thinking about uh decentralized infrastructures in a systemic way from the protocol level to the hardware level to um building the communities that are using uh and maintaining and contributing to these tools. So if each of you could share um your name and what you're working on and what part of the decentralized stack you would consider that to be and then we'll dig in a bit to what that means. So, thank you. Uh, I'm Kelly. Um, first and foremost, I'm a Bitcoin developer. Um, as a Bitcoiner, it's interesting to listen to these conversations because in the Bitcoin space, we've been hammering down on decentralization since 16 years as our primary objective. And, uh, have done many painful decisions to keep Bitcoin as decentralized as possible. I work on Cashew, which is a Charmian e-cash system, basically a u like incredibly private way of exchanging value uh, which can be hosted by anyone. So we're targeting communities, websites, services. So there are many many different cachements that people run. And I also work on Bitchatad which is a um offline decentralized meshbased messenger that works when the internet goes down. Has basic zero infrastructure that means your phone is the infrastructure and uh it's been downloaded by millions of people in the world. U many political events have been um have used bit uh in the past year. So um I'm happy to join the panel. Thank you. I'm Liz from Least Authority. Um we work in different areas within decentraliz within the decentralized space. Uh when it comes to the stack, we've worked on distributed storage um secure distributed storage and maintaining privacy within that within that type of storage. And then also working on uh file transfer, looking at it also from the security perspective and making sure it's private from the sense of not using identifying information in order to allow file transfer and also file storage. And so from the angle of privacy of not even requiring certain personal information in order to perform these tasks. Um and so that's what we've been focusing on. We also help out on the security side with other projects by doing audits and stuff like that. Um but yeah, in terms of this space with a stack, it's been storage and file transfer. >> Thanks. Hi. Uh I'm Bradley. Uh my academic background is in cognitive science, but I am currently working as the AI node manager here in Berlin for the Foresight Institute. Uh Foresight is a 40-year-old uh SFbased nonprofit that provides resources. Originally that was grants. Um, now it's grants, co-working, residency space, fellowships, and compute. Uh, and we just opened up a new node physical location here in Berlin in April. Um, I think that's enough for now, actually. And, and I'm the node manager. I think I mentioned that. >> Hi, my name's Adam. I have a past involving uh, building the infrastructure of the internet from the '9s onwards. And I have an interest in the the liinal uh boundary between the human and the machine and and and technology and media itself and its impact. Um lately, uh I found well I founded the Australian chapter of the Internet Society in the uh '90s and have continued the political uh and um social side of my interest through um establishing a community network uh in Europe, one of the first and joining the Internet Society's community network infrastructure uh special interest group. I'm interested in the way that the infrastructure that we build and have around us has an impact on our minds and cultures. >> Awesome. So, I'm super inspired by uh the different angle that all of you are bringing to solving this problem. So, uh we all know that it can be challenging and a labor of love and an expression of values to use open source software. So, anyone that was interacting with our speaker management system and having some challenges there, for example, you know, you um the I guess it's a sacrifice that we make in a certain way to have things be more um challenging or tricky uh because it's also very important to um you know to support uh these open source efforts and uh to have privacy embedded by design in ways that we communicate with each other in this space. So I'm curious to hear from each of you guys what um like what does it mean to have the kind of the decentralized version of the area that you are working in. So like why self- sovereign and decentralized compute you know why peer-to-p peer mesh networking you know etc like why um you know for storage for bitcoin protocols like what is the reason that it is um yeah so important to uh to have these infrastructural designs and to adhere to these values in um in your work and in general in uh in comparison to this centralized options >> maybe I want to continue the thought that Kyle just offered um in the previous panel, the guy from Brave who said that decentralization in itself is not a goal per se and I tend to agree with that because decentralization itself is has no ethical uh component to it. The underlying reasons why decentralizations matter, decentralization matters is and this is something that we can learn from the original ideas of Bitcoin is that if something is decentralized then it means one it cannot be turned off. Oh, that's >> someone is hacking the light system. Decentralization means it cannot be turned off by any individual. That means it's resilient. We see this from the tour network. the world like the establishment doesn't like tour but it's still open the estab or torance the establishment doesn't like torrent they're fighting against it for 20 years and it's still running great everyone should use it and the same is true for Bitcoin and this this this touches on the economic side of things Bitcoin is a is a um system that that shakes the power of the economic um empire and uh the same is true for communication and I think communication is almost the hardest nut to crack here. Um because we've seen many attempts at trying to build something that is resilient and and uh cannot fail. It's surprisingly hard because the other examples that I just gave work really well. So uh for communication and like I'm now giving the example of bit. It's a uh the biggest issue for communication is the hardware infrastructure because we barely own like a tiny part of the infrastructure in your pocket which is your phone. That's the thing that you own and everything above it like the meme with the you know stacking everything on top is is not in your control. So that means surveillance is part of it and it can be rocked under your ass and you cannot use it for political reform or or uh uprisings or demonstration. So there's many risks associated with that and I think those are kind of the goals that we're working towards. For specifically, it means because you're limited to the things that you own that the range is limited, right? So there's like the lema is the range is is limited because you don't own the infrastructure of the internet above it. So we're trying to kind of build that from the ground up. But I know that there's also people trying to build it from the top down and hopefully we'll kind of reach uh each other and limit, you know, and and and be able to communicate through the entire stack. But I think we're still far away from that. >> There are other alternatives of course as well. You could build from the middle out. Just suspend physics. >> I can give one example. So uh foresight is not trying to build a hyperscaler of what we're building here, but we're trying to build a research grade cluster. Uh and I think this is one example where it's very different from what you have in your hand. We're not a big nonprofit. We our portfolio is like five million a year that we give out in grants about 3.4 of that is to secure AI but nonetheless we're setting up something here in Berlin and in SF as well that could run simultaneously for multiple research labs without needing queuing. So and we're hoping to see that and and we're hoping that we're not just the only let's say a small nonprofit or similar organization that starts to build out clusters like this. We're hoping that many realize okay maybe I can't compete but if we come together we can still provide something at an like a hyper local level that is much more than someone could ever do at home. So it's just one version of this. I don't think it's the answer, but this is the way that we're thinking about it. >> Yeah. And to the decentralization question of this idea, um for us it's been for security by design and looking at how to make things secure. But I mean the word security when you think about it, it it's needs to be framed within what are your goals for the security um and who is trying to control what from who else. And in the case of like decentralization, you really want to empower these end users. So talking about what you have in control in your pocket versus the other things that you don't have control of, how can you enable through protocol development, through use of cryptography, the users to be able to control what's happening with their data even when they don't control the other parts of the system. So um and this is the kind of security angle that we come from. It's because it's it's relative to user control which is necessary for freedom and you need to have this also to maintain your privacy because no one can really care for your privacy better than yourself and you cannot have true privacy without that autonomy of control over your own data. And so this all is related. So the decentralization part comes too if every endpoint every end user can control themsel control their own data. You need to have it decentralized otherwise you know somebody else is in control of some point and even when you work on infrastructure that is c has points of centralization how can you bring it back down to that enduser control so that you yeah like I said for privacy and security purposes. Yeah, and I'd like to add to that, uh, if I might, Liz. Um, I think it's very important for us all to have a healthy interest in the infrastructure around us. And, um, one of my pet interests is the the the way that, uh, the infrastructure watches and uses us. And I'm interested also in central decentralized services and the internet as particularly in my case I have a hobby with the DNS system and how it can be used abused break broken apart and democratized. Um so I think the interplay between someone mentioned that they couldn't get a good definition sometimes of decentralized. I think it's it's a spectrum of um existence and there's no real purity in this thing. But the uh important thing is to keep our eyes open on the impact that it has on us. >> Definitely. So to that point I think you know in this space we love to talk about the trilmma and um I think here we have an interesting trillemma between you know security accessibility and um my brain just blanked on what the other one was. Um yeah, decentralization. Exactly. So yeah, within this trilmma then um I feel like you know it so often has this feeling of like you have to pick two of those like and like what you know what would you view as either you know a counterpoint against that or ways that that expresses in your area of work? >> I can take that. So um I think you know I believe in the trillemma and there's theoretical underpinnings for it. Uh but we have the freedom to choose more than one system. That is the good part uh the good message here because you don't have to rely on only one thing uh for the entire trillemma. Uh you can build in layers. So I'll make an example. Bitcoin is extremely decentralized but as I said in the beginning we made some painful decisions to keep it decentralized. meaning it's extremely slow. It only carries like one megabytes of data or few megabytes of data per block per 10 minutes. And that is a deliberate choice to ensure decentralization because that means that everyone in the world with a Raspberry Pi can run a Bitcoin node. And that is kind of something to optimize for and you suffer in terms of for example UX and uh but the nice thing is that we can build on top of Bitcoin. The Cashew is a protocol that builds on top of Bitcoin. Lightning is a protocol that does it where you um sometimes give away another dimension of that trlemma but you get insane speed, transaction speed, privacy and so on on top of it. So I think we can be smart about the stack that we built and we don't have to have all uh properties of everything in every single layer. >> Yeah. And to add that the answer might be that they're all mult multiple stack choices. you know, you can choose the whole stack. You can choose different things at the at the different um at the different layers because this idea that everything is 100% decentralized is not a good way to think of it either. That you have to make these trade-offs and you have to determine for yourself where you want to make certain trade-offs. So, at that moment in time, is your threat model one that you know you can trade off maybe some uh security or privacy elements for some, you know, convenience or something? But perhaps in other things that you do, you would rather do it slower and harder because you want the security asurances and stuff. So, and I think that this is where again the choice comes in that if we have multiple stacks, multiple options at the layers and we communicate well to users as to what their choices are then you know they can determine for themselves at what points do they really require decentralization or privacy or you know the answers to this trillemma. And yeah, I really like the point that you made about communicating with users because I think to that point of like the accessibility part really is like a red zone of the trilmma triangle when there is not that um that communication but then that also requires you know a contributor infrastructure of people that have the time to make this documentation to uh communicate with community and to be really clear. So I think you know in terms of what are we continuing to design for this is one that's very dear to me and I'm really happy that you mentioned it. >> Yeah. And on the on the counter side also how do we hold accountable those who are maybe telling users things that aren't true or you know like for marketing purposes for whatever. I mean that's where the incentives sometimes are misaligned in the space but importantly we have to be able to like have those like even the the talk before the difficult discussions to like have the conversations openly and hold people accountable about like what are we telling users and what are they actually getting to Adam in just a minute but that reminded me of a conversation I had with someone recently where they were saying you know coming from the comm's team you know we want to reach out to community in a way that is simple and enough to make sense and get people galvanized um you know to be using this tooling and to engage. But then at what point is distilling that like starting to become not as accurate and on what layer of the stack does that you know lack of accuracy even show is it someone that's really digging into the codebase or to someone that's using it as an end user and it's not the um you know the outcome or uh downstream effect that they expected. So yeah, definitely a challenge. And Adam, >> yeah, I think my my takeaway or pickup from this is uh also I think it's to do uh with simplicity. Uh and you know, I think there is a maximum layer of cards that we can stack in our edifice. Um that is healthy for human humanity. I think um regardless of the rest of the the the structure and we should we should be aware of that and uh look at the impacts these these um structures have on us. Yeah, >> there's a term in cognitive science called affordances, which is basically what is the easiest action for a user to do. And I think that >> frictionless sword, >> a frictionless sword. >> Frictionless. >> Yes, frictionless. But it exactly. And I think one thing that I'm seeing a lot is we're building a protocol and then we're saying, "Oh, yeah. Yeah, we're going to build some really nice UX on top of it." But I would actually say we should start the other way around. And again, I'm not a not a technical person, but I do believe that with the tools that we have nowadays, especially with generative AI, we have a lot more ability to say, you know what, I'm not just going to push something out there that works well for me. I actually can just run this through Opus 4.8 or Fable if it comes back online and design something on top of it, a harness on top of it that people can actually use. I also think I also think that we're going to get very close to in the future. This is this is just speculation. We're all going to have our own kind of UX that is living on whatever devices we use. And then the the infra the the the protocols, the the apps, whatever it is, is going to kind of like fit in as an API or whatever it is underneath that. And so I think we should start really spending a lot more time not just saying, "Oh yeah, the comm's team will do it." From the ground up, I would love to see more people. What I'm saying because I'm I'm every day I'm seeing researchers work on stuff and build stuff and I'm seeing software and I'm looking at it over their shoulders and they're trying to show me and they're saying this is so exciting, Bradley. This is going to change the world. It's an open source palunteer. Everyone's going to use it. This this just happened to me and I'm look and I'm looking at the Yes, exactly. And I'm looking at the UX and I'm like, no, >> there's a pink and fluffy palunteer. >> I would love to see it. Yeah. So, that's just I'd love to see more of that. >> That's a good point. Yeah. We can't expect adoption of tools. We can't expect users to actually have choices without being fully like knowledgeable of what their choices are and if we're not building that from the ground up. Yeah. >> And I want to suggest there's one really really simple way to make sure that happens. Whatever you're building, get it in front of people that are non-technical. Get it in front of people that don't know that your industry and see what happens. I think so many builders do not do that until years in and thousands of coding hours into whatever they're building, they haven't actually got it in front of people. >> Yeah. And be insecure about it. be insecure about your own assumptions because that is a danger once you start going down that rabbit hole. >> I I want to contribute maybe a slightly contrarian view on that. Uh although I largely agree with what you said is I'm a developer. I've been writing code for almost 30 years at this point and I barely have written a line of code in the last year or so and everyone knows why. It's because of AI. So AI has completely changed the way we develop software. We can do it faster. We can do it with more precision. we can do things that we weren't able to do as developers. Um, it also has a secondary effect for users uh themselves, which is it's it has a potential to solve one of the big like the biggest UX struggles that we had so far because things that were complicated before are now a simple prompt away. And this is not only does not only apply for writing software, but also for using software. So you can tell your agent to set up a GPG key for you, set up uh your mail client and upload your key and share it with everyone you need with a single sentence. And that is something that would have required maybe you to study the whole thing for a week before which you don't need to do anymore. So from the users perspective, things are changing rapidly as well. I think we don't feel it as as much right now. And I like to to top it off even more contrarian, I think even UI itself is disappearing. Uh so the the two-dimensional surface that we interacted with for many years is probably going away and will be replaced with something much more verbal, more human and that might surprise us in the next coming years where things that we thought were impossible from the users perspective will become suddenly a lot more tangible. >> Yeah, I don't know. I'm sort of over this um singularity theme of blending with our environment and the body dissolving. It just doesn't sound that appealing to me. Taken too extreme. >> I'm referring to simple chat interfaces. We have them today already, so you don't need to wait. >> One thought I have, and this is more of a question because I I haven't looked into this, but with the AIE EU Act, I'm sure some of you know more. Isn't it true that when it comes to wearables, which OpenAI is pushing really strong towards and a couple of the companies as well, that that every single user has the right to have access to the stream of their data precisely as OpenAI or any one of the wearable providers would be would be getting it? I think this was something that I read like a year ago about this act that's coming into effect in September this year, like in a month or two. I'm hoping someone here tells me more about this. We can tell the audience. >> Yeah, I'm I'm not sure about that particular bit of legislation. I just think that trying to legislate way out of a scale that's possible or practical um to understand let alone enforce it just sounds out of out of belief really. So I guess one uh question that arises from that or um that was on my mind with uh yeah what um I guess how does AI inter interfere with each or interfere interact with or um get impacted by your different areas of work. So I know obviously for you doing compute for AI it's uh you know directly related but I guess you know in terms of uh privacy for example I know that there's many um threats that did not you know exists before the types of attacks that people are now able to orchestrate like if you want to speak to that uh from any of your perspectives. I mean excfiltration of data using AI agent is real. Uh everyone should be very wary about that what they run on their computer. Obviously I think most people here will agree or know this but um to come to your point of what it enables I just want to you know double down on what I just said before. Please ask your agent to set a PGP key you know uh that's it. I think you will feel you know those who haven't tried it yet. It's it's natural that you're um you know a bit skeptical about it but the models have in the last year this has just happened in the last year have become so incredibly good um that you can basically ask them almost anything that you can do on a computer they can do for you. I'll make an example for infrastructure uh for example setting up a lightning node has always been a pain in the ass. Uh ask your agent to set up a lightning node. It it h it's done in two two minutes. So there are many things that you thought you cannot do and you can do them right now. Uh like that's the thing that you should maybe take home with you if you haven't tried it yet. Please do. >> I mean most people are still using day-to-day most systems that they're using are not fully respecting their privacy as much as they should be. even though we've had, you know, over a decade of GDPR and all these things that are supposed to be helping us as users have our privacy be respected. And on one hand, AI is going to make this a lot worse, I think. And it already is. Like that the data is there. It's more accessible. I mean, there's so many different ways you can slice it with AI, whether it's agents and whether it's using um uh AI to help, you know, uh exploit vulnerabilities to access things. uh social um like taking advantage of people, fishing, social engineering, all of it. AI is just making everything faster um at different speeds in different areas, but it's it's coming. And then so that's scary. That's not good. We didn't really get privacy by design embedded in all of our core systems prior to this, but at the same time, uh all those problems will hopefully cause more action. So maybe people will start to realize that oh okay we didn't really feel the pain before because oh they just took all my data but nobody took all my money or something and but now now that if if that advances faster then maybe we'll all care more maybe the people who can impact this where the different challenges need to be addressed maybe they will so I'm going to stay optimistic but it sucks that things have to get worse before they get better. That is yeah very optimistic and comforting way of framing that challenge. So definitely uh do you do you guys have any more thoughts on that? >> Um I think when it comes to providing research grade hardware or small microclusters at a local level I think it's quite simple that there's no one in between a researcher and the hardware they're using to run their their their experiments. And I just think that that means there's no like corporate interests that are in play and like the model theoretically in the model weights of these running an open source model there's no one watching what happens and we're not even watching what's happening there. So um in terms of AI privacy as practical steps that you can do today is run local models. They're getting pretty good. U you still need some expensive hardware for it but it's very expected that this will change within the year or two. So um local models are getting surprisingly good. There's also end to-end encrypted models out there uh providers that can prove to you in a secure enclave with an attestation that they cannot read the data that you send to them. So for sensitive applications those might be a better option. Uh but just the simplest thing that you can do today is to not use a credit card to pay for AI. Otherwise your identity is directly linked to that account. So uh you can use chatgpt or clot or something like that. uh if you make sure I mean you should always be wary of what you send to them but at least don't use your credit card that is linked to your account if you can somehow manage that >> important to keep in mind. Um are you going to add something or >> Yeah. Okay. Um >> yeah one thought comes to mind in fact um my neurosurgeon once said to me he's seen hundreds of brains but never seen a thought. Um and in the same way beware when the furniture starts talking back to you as to who uh they they are owned by and what in the agency is there for. Definitely. Well, um looking at the time, if there's any uh anything in particular that you guys want to add and we can also take a question or so um if there's uh an important point that has not yet been highlighted from any of you um yes, >> maybe to loop back on the beginning of our conversation. So, I work on Bit Chat, which is a um Bluetooth messenger. You can download it right now. It's perfect for a conference like this. can talk to each other on chit talk about the panel for example um without internet uh the um to loop back to the layer concept that we brought one of the biggest challenges that we had have is range but um and this is also you know addressing everyone here is there are other protocols that can do mesh communication on a longer range such as Laura mesh core meshtastic uh other technologies so um we think that the way to go is to marry them similar to the you know how we build layers in crypto basically is have something that works on the ground on the street in a city block that is for example bit have something that works between the city blocks between the cities that might be mesh core have something that maybe works between uh the countries and that could be I don't know maybe some future satellite open source satellite project that that doesn't exist yet so uh I hope we'll be able to for communication do something similar on a physical level that we have been doing on a digital level for many internet applications and protocols. >> No, go ahead. >> Okay. I just kind of wanted to jump in on this idea of like the physical level or kind of what I call what I do like the the the social infrastructure level really bringing people together and I wanted to just share a little piece of lore that might be fun to the people here. uh foresight co-founder Christine coined the term open source at an event not well dissimilar to this one but back in 1997. It used to be called free software before that. And they were having a really hard time having it gain traction. People didn't like the name free software cuz they thought oh it means it's going to be free. I don't want to develop a protocol. It's just going to be free for everybody. And uh she originally coined this term. It made a huge difference in how open software was supported and developed. And then here we are today. So many of us are building open open source software and caring about it. And so I just wanted to kind of like there's one thing I wanted to say is I think these kind of meetings and these kind of spaces are also really really important for all of this because we get the interiplinary communication because we actually meet somebody eye to eye that we wouldn't because we can talk about these things. So it's that's close to my heart but I just kind of wanted to shout that out there is that we also should continue to prioritize events like this and Beth thanks for having us. Yeah, I think when it comes to um building the stack just I would want to add that uh the there's been amazing advancements with cryptography and cryptographic protocol like uh different approaches with cryptography and cryptographic protocols that were theoretical in the past and you know very much in academia and known that this could solve stuff but now they're actually being implemented and applied to scenarios that can solve our problems. So um we can look at things uh you know within the stack and see things from a completely different perspective now and solve things um solve some of these problems that we saw as ch that we had as challenges in the past there's now new novel solutions for and to me that's incredibly exciting and I look forward to seeing you know the these conversations at these events every year as we start to see like the the dynamic shift and the paradigms that we have challenges of like they move because we have new tools at our disposal. >> And if Adam can close us out. >> Oh, thank you. Yeah, I would also just uh want to um support what the panelists have said so far, but I'd also support um and want to suggest that people examine the existing infrastructure around them when deciding what to build. So um you know my particular interest involves the domain name system but there are other projects are well uh as well to look at. There's a project that that attempts to duplicate the in the existing internet in all its structures and forms. This is fascinating. It's a no uh allegiance to the project but it's called dn um42.org. um and it it replicates all the governance structures as well as the technical one. And I think seeing models at that detail of both the human and the the mechanical or the digital um is very valuable when you're starting to think about designing new alternatives um involving yeah meshtastic and other alternate forms of of uh communication. So yeah, take a look and check these things out. treat it with the seriosity that it deserves. >> Awesome. >> And play. >> Yes, that's the most important part. So, yeah, thank you guys. And since we didn't have time for Q&A, if you want to join us in the um little workshop area that is speaker office hours, then some of us can head over there. If you don't know how to get there, it's like you turn out of this door and then there's the workshop room area which hopefully you visited to see department of decentralization zone and stuff and we'll be in the room one if you want to continue this conversation and uh yeah chat with these awesome speakers. So thank you so much. Thanks for the panel and thanks guys.