B (3:37)

Yeah. So first to pull back tollgate is this project that allows you to basically sell your Wi Fi for SATs.

A (3:42)

Exactly.

B (3:43)

And then people are able to access it. And then in the traditional Internet infrastructure we have Internet service providers, we have DNS providers, we have all this different, all these middlemen that are involved in networking everyone together. And our IP address is our quote unquote identity. I guess on the Internet and a lot of home Internet, I guess their IP addresses do change a decent amount. Sometimes they change, sometimes they don't. But if you're like hosting a website or something, you basically need a fixed IP address.

A (4:13)

Right? Exactly. Yeah.

B (4:16)

So fibs. Yeah. So continue.

A (4:18)

Yeah. So in a tolgate network, right. The way we envisioned it first is like okay, you have someone that is connected to the legacy ISP if you will, and from there on out you start reselling that Internet and someone can buy it from you and they start reselling it. That assumes a tree structure where someone's on top and someone's at the bottom. But if you have a network that grows and like say you're two, three steps down and there's someone that actually gives a better price for that connection, you're going to switch over. Everyone below you is going to lose their that structure that was there to find them inside that network.

B (4:55)

Right. So in this situation everyone has a nostr public key, private key pair which you can spin up relatively easily. That's your public identity. And then this is why I think it's fascinating because conceptually high level, it makes a lot of sense to me. So then you're able to just connect peer to peer between those public keys without anything else. So I can be connected via Wi Fi or Bluetooth, I can run a long ass ethernet cable to my neighbor and he could run along as ethernet cable to his neighbor and we'd all be connected to each other. And then one of them could then have like some kind of long distance mesh to someone else and then you can continue and all that networking would work natively with noster. That all seems awesome, high level, but how is that possible? That doesn't like what's the nitty gritty that actually makes that reality versus just something that just seems really cool and, and a tweet.

A (5:53)

It is just by decoupling two layers. If you will the physical transport layer and the routing layer. In a lot of cases they're very interconnected. Right. If we have Internet, like IP traffic, it presumes that it uses Ethernet, basically Ethernet or WI Fi. It doesn't run over Bluetooth and every different protocol kind of has its own way of doing that. And what we found is it's better if you have something that works across it and just show. Expose some interfaces to the layer below where. It doesn't matter how it works. All you need to know is just, I want to be able to send some data to appear and I want to be able to get some data to appear. I don't care how you get it there, just get it there. If it's Ethernet, right, You use the Ethernet protocol or WI fi. If it's Bluetooth, you use Bluetooth underneath. But the FIPS protocol, it doesn't have to, doesn't care about that.

B (6:50)

But then how does discovery work?

A (6:55)

That depends on the layer you use or the transport you use. If you use a broadcast medium, say WI fi, you can broadcast the fact that you are a FIPS node and that you want to peer with people. That way people can discover you and send messages back. Oh yeah.

B (7:12)

Work with me here on hypothetical, because I understand hypotheticals better. Let's say you're successful. This project is a smashing success and it's five years in the future. You're in a different country than me. I have your public key. I want to send you information. How. How does me pressing Enter on my side get to you?

A (7:43)

If we are on the other side of the world, that is a problem we hope that we're going to run into with the, with people using it. We're nowhere near there yet. But I, I think the one thing that can help us with that, where if you do mesh without the cryptographic identities and Nostra keys for that matter, I think we can use nodes that are somewhere in between that can help us route that traffic. But this is two step, two, three steps ahead and like routing further away.

B (8:11)

So are these, are these similar to like what a DNS server currently is? How does that look?

A (8:17)

Yeah, in this instance it would be something like that. Like you would maybe publish something to. It's very theoretical. You would publish something to say, a Noster relay that is specialized in this. Right. You can use your, you can use your Noster keys to sign something and like how people can find you, but you can do it in a way where there's not one party involved. It's like one central Authority you look at like, how does that traffic go from A to B? But you can send it to multiple people.

B (8:44)

So the difference would be from the modern day setup is that it'd just be more accessible to run these kind of discovery servers or whatever you're calling them. Like anyone could basically run one and there, there wouldn't be like lock in maybe. Am I reading this correctly or.

A (9:00)

Yeah, if you would use something like this. Yes. But I think we should backtrack a bit first. Like how that.

B (9:05)

Let's backtrack.

A (9:05)

Yeah, whatever you think it works on the local level. So on the local level you discover a peer and you encrypt like you advertise your own pub key. Right. The peer advertises, you can immediately encrypt something to that pub key. So there's a handshake there where two peers use a noise protocol to establish an encrypted connection. And every peer does that amongst each other. And so every hop of the way is individually encrypted. And then on top of that there is the longer routing of like the longer routes. And the way you can find another peer within the mesh, say like a mesh up to 10,000 people, is that every peer holds a bloom filter of the people that they can reach. So if you connect to a peer that can already reach, say like 10, 10,000 other peers, it will give you a bloom filter, which is a very compressed version of a list of npubs that it can reach, which is one kilobyte at this moment, you get that. And then whenever you get a packet that's for any of those 10,000, you know that that peer that's on your left hand side can actually reach it.

B (10:24)

Got it. So in practice, each peer is almost like a DNS server.

A (10:35)

I think it's hard to make that comparison because it's not, it's. It's a different. If it's a different way of doing things.

B (10:44)

Right. It's fundamentally different. But each peer is also my method of discovery.

A (10:50)

Yes. In a smaller local mesh, that is the way you found other.

B (10:54)

Yeah, I mean you could, this is a interoperable way of rolling out like a local community mesh relatively easily.

A (11:02)

Yeah, yeah. And that's. Honestly, that's. If only that works, I would already be very happy. Because you can roll this out in a community. You can host servers, Right. You can, you can host things on an npup that can even physically move around in the network. And if the network gets cut off from the rest of the world, everything just keeps working you can do it if you know, half the network fails, you go over Bluetooth, whatever works, whatever method you have available, that's how you want to route the traffic.

B (11:31)

It's completely agnostic. And some peers can be on Ethernet, some could be on WI Fi, some could be on Bluetooth. Doesn't matter.

A (11:39)

Yeah, or even satellite uplinks.

B (11:42)

Now what if one person have. You guys, is it in scope of the project if like one person has Starlink, for instance, would then the whole community have broader Internet access or no? Is that.

A (11:55)

Yes, you can. So if you do use Starlink, you are using the regular Internet stack IPs, but you can establish peer connections over the Internet. So if I have a mesh here in my house and you have one in your house, we can establish a link over UDP and connect the mesh that way?

B (12:15)

Yeah, I mean, cause like in practice. So let's use a real world example, right. The Internet. The. The government of Iran cut the Internet relatively recently. I don't know if the Internet's on right now or if it's off, but at least during the protests a couple weeks ago, now we have war in Iran. But during the protest a couple weeks ago, government internally cut the Internet because they didn't want communication among the protesters. It's something we see happen all the time. And as a result, bitchat usage spiked. The country now, bitch has its own issues, right, in terms of range and whatnot and capability. But the biggest one that I kept thinking about is those people could communicate internally, which is great. Still an advantage. You know, it's better than the alternative, which is no communication internally. But then they couldn't, for instance, broadcast to the rest of the world and be like, this is what's happening on the ground right now. There was a fog of war that was happening and that was part of the government's desire. That's why when you see large protest movements happen in different places, authoritarian governments often cut the Internet. Now the dream high level in my mind has always been you have like local meshes and then one person smuggles in a Starlink and then all of a sudden the entire community can broadcast globally on Nostr, for instance, right? And say, you know, here's a video of the police abusing us or whatnot, or corruption or whatever they're trying to get out. Presumably this project should be able to offer that kind of capability, no?

A (13:41)

Yes, because you can introduce basically say, quote unquote, a tunnel to the rest of the world from anywhere and that would automatically be discovered by the network. So if yes, you have a starlink that goes up and right. Goes down in another country and you have a mesh there. Well, as long as you knew one of the npubs that was on that side of the mesh before it got cut off, you can actually find that route again.

B (14:05)

That's awesome. I mean, have you talked to the. Have you talked to the bitch at guys at all or.

A (14:14)

No, we just released it actually. So it was two weeks ago. I just met. The idea has been around for a little while and I met Jonathan just in last November and we kept talking about the idea and so that slowly developed over the last few months. And just the last couple of weeks we had a working demo.

B (14:38)

Did you guys. Did you guys meet at Sovereign Engineering?

A (14:42)

No. At a conference in Costa Rica. There was this conference with Nostra conference and I think that was two days. The only thing I did is basically sit in a room with him talking about these ideas on doing addressing with mpubs instead of IP or for DNS as well.

B (15:00)

So what is the Sovereign Engineering connection? Justin Moon introduced us. I believe he met you at Sovereign Engineering.

A (15:06)

Yeah, correct. Yeah.

B (15:08)

What was the Sovereign Engineering experience like? I mean, I love what those guys are doing. I haven't been myself, but I work closely with Gigi at OpenSats and he's heavily involved in Sovereign Engineering. I think it's a fascinating program. Yeah.

A (15:21)

So some of the. The original idea and like addressing with npops is something I worked on at Sovereign Engineering last year. It came from. The idea is like if we can use npubs to address services instead of IP addresses or traditional domain names, then at least we have a static identifier we can use across the web to reach things. Even if the IP address changes. That experiment turned into what's called now no DNS where someone signs cryptographically on in a NOSTR message. All the DNS records publishes that and other people can resolve that as DNS. So that's kind of the predecessor of this idea. We can do it inside the IP network. You still have that structure, but when it changes, you have some flexibility on correcting that, similar to if an IP address changes. Now you use update the DNS record and people can reach you again. You can do it much faster this way. That's nice, but it doesn't resolve the entire issue.

B (16:23)

I didn't realize. So Tollgate is your project as well? I work on it as well.

A (16:29)

Yeah.

B (16:29)

Yeah. So you were in Sovereign Engineering for Tollgate specifically?

A (16:35)

The first one. I was at Sovereign Engineering 3. I didn't know much about Tollgate back then, but I met the guy that started it and from there on we started to. Basically rolled into it and we started to work, work on it together.

B (16:49)

I think it's a super cool project too. So the, so the whole premise of Tollgate is that I'm, I'm paying SATs for Internet access. Are SATs involved in FIPS at all? Like, are people, are people using, is the idea that people will be paying for peering? Or have you thought about that at all?

A (17:10)

Yeah, that, that's my hope for sure. Like the, the whole reason fips exists is because we don't have a way to do this dynamic, this dynamically changing network. So I think it's a, it's a. Payment is a very important part of it because you want to be able to steer traffic based on economic incentives as well. So if, say, one link is very expensive and the other is cheap, well, why not. If you pay for that connection, why not incorporate that in your routing decisions as well? But that's a, that's definitely a next step. But it can be, can be combined.

B (17:44)

I mean, and wouldn't something like that be necessary? Because have you thought at all about DDoS concerns or any, any kind of malicious attacker trying to flood the network or break down the mesh?

A (18:00)

Yeah, there's, there's a couple things there. The mesh, when, when joining the mesh physically, you need to do that with a peer, and that peer, it will be rate limiting how many people can connect at the same time. So that's, that's one defense. Another defense is that to send traffic to an endpoint, you need to know the endpub, because on the routing layer, the nodes don't know for the packets they send, they don't know the npub. They only know a node ID that is derived from the npub. So that gives you some protection. If your NPUB is not known by the whole world, it's hard for someone to DDoS you.

B (18:40)

Interesting, because that would be the attack, Right. If we were going to go down our hypothetical.

A (18:46)

Yeah.

B (18:47)

And you're in, you're in Iran and the government's shutting down the Internet. They would try and jam different types of signals, I guess, and then also they would presumably just spin up, spin up npubs. Right. And just start spamming the network.

A (19:01)

Yeah, I mean, it's a, it's definitely not a, not an easy problem to solve. Right. It's, it's something that's. No, it's probably the hardest current Internet. Yeah.

B (19:10)

Well, the current Internet just Solved it with centralization, right?

A (19:16)

Yes. Yeah. You have Cloudflare to do these things,

B (19:20)

which is a massive middleman honeypot. And then also just Internet infrastructure in general, right. There's just a million central points of failure that can be pressured and shut down. That's why authoritarians can. I mean, it's not as simple as I would basically click a button and cut off the Internet for the entire country. Right?

A (19:35)

Yeah, yeah. So that's why an Internet that is designed to fracture and recombine is very useful.

B (19:45)

Well, what I like about this is it's very pragmatic because one of the trouble, like mesh is something that I've been fascinated with for maybe a decade and there's been many different attempts at doing different things. I think it was exciting that. And obviously partially it was because of Jack Dorsey tweeting it out, but the excitement around bitchat. Yeah, but it was also kind of weird, right? Because if you had actually been following the mesh space for a while, like people were acting like it was the first mesh project that ever existed, but there was a million predecessors that failed beforehand. And always the problem. Well, not always. There's many different. Like you said, it's very hard problem to solve. But one of the core problems I feel like is this bootstrapping issue where you know, you need a critical mass of at least your neighbors to be using it for it to be useful. Right. And people saw with bitch, right? It's like they would download bitchat and then it was just nobody to talk to. Right. Because you don't have neighbors that also download Bitchat. And there's different ways of solving it. You know, it's easier in cities. I like the idea of being in a, like a sports arena or something that already has poor Internet. And maybe in a sports arena everyone wants to talk about the same thing. They're all watching the same game, they all want to talk about the same game. As a high density group of people, you could see it being really useful in that situation. A concert, a music festival, but at scale, to actually like see it work at scale, it's just a really hard problem. But one of the things that I've thought about historically is if you could bootstrap it in a way that it works with modern Internet infrastructure until the modern Internet infrastructure isn't needed anymore, then that makes it way more useful. Because if we're in different parts of the world, but we can use this tech to communicate because of the existing Internet infrastructure, then all of a sudden it's useful today without Actually having the build out of people interconnected all the way between us or whatever. Right. Does that make sense?

A (21:52)

Yeah.

B (21:52)

Yeah.

A (21:53)

And there's a couple of things to unpack there too, is that I really hope we can nerd snipe the people from Bitchat into this. If, you know, if you run a mesh that is cross transports, you can lower the burden of making apps like bitch. Because now Bitchat needs to design its own protocol on like how over Bluetooth the apps are gonna talk together. Whereas if it was just, you can just send TCP IP traffic like through this, like any existing application, like a relay and a Nostra app, then you can do this over Bluetooth without having to specifically design the app to also support that.

B (22:32)

Can we go a little bit deeper there? How does that look in practice?

A (22:37)

So say you would have. You would run a FIPS network and every phone has it running and is peering with each other over both WI FI and Bluetooth. Some people are only Bluetooth. Now on these phones you can host something like a Nostra relay. Now everyone can reach that relay over a NPUB FIPS address from any app on their phone. So now they don't need an app that is specifically designed to be a blue. It can just be like any other app and not have to put in any effort to, to do this. And so you don't just have one app that can do this, but you have pretty much everything under the sun that would work if, if it uses infrastructure like nostril relays that are really good at facilitating so many use cases at once.

B (23:24)

But so how does that look in practice for like an app developer? How would they integrate in that situation? So

A (23:33)

if I would say like white noise. Right, Good example, white noise connects to nostril relays. White noise does what white noise does best. Like do all the encryption, do all the rotation, group management. But they might not want to put so much effort into making a Bluetooth mesh. If they, if someone that uses their app on their phone also connects to a FIPS network, they can just configure a relay like they always would, saying NPUB FIPS in the relay in the relay list, then they can already talk to that relay that's on fips. Now if they then move around inside that mesh and other people join, they go from Bluetooth to WI Fi, they can still reach that relay that is elsewhere in the network. So they don't have to change their app to work over Bluetooth.

B (24:32)

Do they have, do they. They have to add FIP support specifically though, right?

A (24:37)

No, that's the, that's one of the things that's. That I really wanted in here is a. I should have mentioned that earlier is a system where it works for existing applications that expect the IP stack to work. Because I looked at some of the other protocols and what I found is they, they, they often seem to expect everyone to re. Implement all their applications to work with it. There's no way in hell you're going to get any adoption. If everyone needs to rebuild their stack from scratch, there's, there's 40 years of history in libraries, et cetera. Unless it works from scratch or like without any modification in all the applications, it's not going to fly. So the way this works is we add a custom DNS server together with fips, the FIPS daemon. What happens is if any application on your. On, let's say with computers, for now, any application on your computer asks for NPUB fips. What that DNS server does is it gives you back an IPv6 address that is deterministically derived from that npub, which is the same as the node ID of the destination.

B (25:47)

That's wild.

A (25:49)

And then the application will send its traffic like it always would. Right. It's going to, in my browser it's going to be HTTP becomes tcp, becomes IP packets. And then down at the bottom where you know, the system would send it out as an IP packet, we know we just generated that IPv6 adder, so we intercept it and actually send it over FIPS instead.

B (26:10)

And the app just thinks it's using the regular Internet.

A (26:14)

Yes. So that's what we're focusing on most now is like all the applications, right? Ping. When I did the first experiments in sovereign engineering, the Sec 5 I found was like, okay, I make something like this and now I have to rewrite. Fucking Ping app, excuse my French, the Ping application. I'm not going to do that. Like no one in hell is going to rewrite everything from scratch. So it has to work with everything that we already have.

B (26:42)

That's awesome. I mean, I think that's the key, right? That's the key for bootstrapping. Then you don't even have to. The app developers don't have to change a thing.

A (26:57)

Yeah, I think that's a must have. I don't think anything like this can scale if it isn't backwards compatible.

B (27:08)

Okay. I mean, I don't know where to go from here. That's like a mic drop moment for me. I mean if this works, it seems like it's kind of the holy grail that you stumbled on here, sir. I mean, I'm reading through your git repo. What do you want to cover? You know better than me. I, by the way, huge shout out to Justin Moon. He set up this conversation. I told him to join for it. He said he was unnecessary. I think he would have been very helpful.

A (27:36)

We can, we can still try to get him in.

B (27:39)

What do you want to talk about? What is most interesting to you here? You have thousands of people listening to the show. What do you want to talk about?

A (27:49)

Think. I think the opportunity to make the Internet in this general like FIPS doesn't have to replace the Internet if for like long, long range stuff, right? If necessarily, if we're doing a call like this, it's fine if that still exists, like the hierarchy there. We're not, you know, that's not going to change quick. I think it already wins if it works local. And that's exactly, I think the most exciting thing is like if it works local with say half a million people or a million people. You don't even need the traditional Internet. Like the traditional way of doing things is we go out like when we visit a website, right? I go to Amazon.com the way it works now is I send a message to Amazon. It's like, go fetch, go fetch, go fetch. And I think in a world with Nostr where everything, everything in Nostr is a stream, all the data is signed so we can do replication wherever needed. So what we can do with that is we collect information months, we pull it inside the network that we have here. And from there on out, it doesn't matter if that connection to the outside world gets severed because the information is already inside the network on the relays. And then from there we can spread that information. So I think if we have a decent way of hosting that stuff inside this network that can rearrange and is resilient, then it doesn't matter as much when connections outside are unstable. Like you'll just pull in the new stuff whenever you have a connection. And if not, you go about your day and do your things inside your network. You can, you can run your Bitcoin nodes, your cashew mints and Nostra relays and Blossom servers in the, in your town and the main cable gets cut, well, okay, it sucks, but you can still chat whatever you want over all the over white noise. You can still use Primal, right?

B (29:45)

And then, and then the key is when that cable gets uncut or whatever. I don't know what the opposite analogy is when you get reconnected, you can gracefully reconnect to the water network.

A (29:57)

Yeah. And then you can see like what was gone, what is not right. That that boundary Nostr allows us to, to completely remove the boundary between what is offline and online is becomes a spectrum of offline and online. Like offline is. I don't have anything on. I don't have a relay on my computer. I don't have a Blossom server on my computer and I'm not connected to any other node. Then I'm offline where it's like you can be a little bit online where you have one other node.

B (30:30)

Yeah.

A (30:31)

So there's no.

B (30:31)

And then every. Because of Nostr, everything's signed, it's all verifiable. You can gracefully reconcile afterwards.

A (30:38)

Yeah. So the way information flows I think will dramatically change in a Noster world.

B (30:46)

What are the hurdles you face? Like, what are the trade offs here? Like concerns. What are the deal breakers? How are you thinking about that?

A (30:56)

Well, for sure we need a shit ton of testing of this.

B (30:59)

Right.

A (30:59)

People need to. If you have any kind of knowledge related to this, you should test it out, break it. We want feedback as much as possible because of course it's a, it's a, it's a very big aim.

B (31:10)

So we do have this running in the wild right now.

A (31:14)

Well, it's just been two weeks but I, yeah, I have, I have two servers. Public, public servers people can connect to. I could share that.

B (31:21)

Are they connected to any peers? How many people are using this thing?

A (31:25)

Well, right now it's mostly Jonathan, me and maybe one or two other people. So we're hoping for a lot of

B (31:31)

Justin Moon using it yet.

A (31:33)

No, he said he didn't yet. We have to bully him into doing it.

B (31:37)

Fair enough. I'll try and get my feet wet. This is there limitations is like bandwidth users, you know, where are the, where are the points of failure here? As someone who I, I want to straw man it, but I don't have the capability of straw manning it because it's way, way above my pay grade.

A (32:01)

So there are limitations on the, the physical.

B (32:06)

Not strawman use steel man. Sorry. Continue the limitations. I'm sorry, like limitations on like wifi connections or like actual implementations.

A (32:20)

Yeah, WI fi and Ethernet are fine, but it is like the, the lower you get like you have tech like Laura, right there is in the way Internet protocols work. You have something called an mtu which is the minimum. I'll, I'll, I'll find where, what the full meaning is. But it's the amount of bytes you can, you can transfer in one go on a network. So on traditional IP that's like 1500 bytes for every packet. Something like that. You can go up and down in some networks, but if you have transports that are less reliable or lower power, something like Bluetooth uses smaller packets but still works fine. But if you go down to Lora, like Lora Mesh Testic, they might get too small to send. Like the minimum size of a FIPS packet might be too big for those. Although there's some really exciting developments with stuff like Halo, which is a WI FI standard on the Lora frequencies. So I do have hopes that even though it's the same frequencies, the throughput can be increased in the coming years and then maybe that will become feasible. But stuff like satellite uplinks might be already.

B (33:37)

I mean, I don't know how much it matters, but MTU stands for Maximum Transmission Unit. Is that what you were looking for? Yeah, it's basically Ask my bot.

A (33:46)

Yeah, yeah, it's basically how thick the pipe is between the two nodes. Like how much water you can move at once.

B (33:55)

What's happening with the lower projects, are you following it or.

A (33:59)

I'm not following it that much. I think it's cool. But a lot of them have been around for years. Right? I really hope that's my, my. Maybe it's wishful thinking if the throughput can get higher. Which corridor? The guy I work with for, for tollgate, he knows a lot about doing wireless and the things here, like I have to go from secondhand information. But what I know about is that the waves and the frequencies themselves are not necessarily the problem. It is the tech that's used on either end on like how much data you can put through that on that same frequency at once. With newer hardware you can use the same frequencies but actually transmit more data at once. So I, I think if those developments continue, I think those technologies will be help. Will.

B (34:48)

Well, I mean I'm a little bit confused because it doesn't like if I have a lower device. Isn't it using WI fi usually to transmit information?

A (34:59)

I don't know, I don't know exactly what the.

B (35:03)

But it's like whatever their specific protocol is that is actually the limiter or it's.

A (35:10)

Yeah, no, it's a bunch of things. It's the hardware, it's the protocols they use for transmission.

B (35:16)

But the goal is your, I guess your point here, the high level point is that you want low power devices to be able to communicate with this protocol. And there might be some limitations there that you'll have to.

A (35:29)

Yes. Yeah. And I don't want to venture too much into that area. Like, that's not my, my area of expertise. So I don't want to go too much into that.

B (35:36)

And I, I guess the cool part of FIPS is that you have a very, a relatively simple, robust protocol that people that are then more technically sound on, on the, the hardware side could come up with low power ways of handling it. Yeah, right?

A (35:55)

Yes.

B (35:57)

That's awesome. What, are there any other pain points to steel?

A (36:03)

Man?

B (36:07)

I just, I just want to be clear here. Like, I have. I got into the Laura hype. I got into the Gotena hype. Gotena is proprietary as hell. I don't know. I just been through a lot of mesh hype to the point where one bitch act came out. Like, I, I went along with the hype, but like, internally I was fading it. I was like, yeah, okay, so what

A (36:31)

are, what are the things that you hit that this, I guess, disappoint you or.

B (36:36)

I just want it to be a reality, you know? Like the modern Internet sucks, but story of, the story of humanity is us just solving problems by centralizing things. That's the easiest way to do it. And as a result, our lives that are incredibly digital now, the most digital our lives have ever been in existence are like, built on like a shit ton of centralized points of failure. And so I would love to have this dream of robots and people and drones and everything all interconnected in a graceful, robust, peer to peer mesh way. But usually it's just hype and we don't actually see any of it in practice. And maybe this time is different. And I want to ask the hard questions, but I don't have the capability of asking the hard questions. So I'm asking you to ask yourself the hard questions is what I'm trying to do here.

A (37:36)

Yeah, no, I, I think, I think that's a fair ask. It's, it's, it's a very big aim. I think it's a very big aim to say like we can make anything else then work besides ip. I had this, like, I had this idea. This like exactly what I told you. Like this local networking stuff. I think it can, I think it can work and I think it can work because we have the primitives of Noster, Noster and Blossom and Ecash, right? We can transact locally do micropayments. So we have a bunch of new primitives that have never been combined this way. And when I talk To Jonathan, who by the way, is like a very OG open source.

B (38:19)

His reputation precedes himself.

A (38:22)

Yeah, he's been around in bitcoin. Early on he was on the mailing list, the cypherpunk's mailing list. He's done a ton of cryptography and he looked at it and it's like, oh, this might work. And I'm not at his level and. But I think I nerd. Nerd sniped him hard enough to, to come back from his retirement seal of approval. Yeah. So, you know, it's, it's not. You know, you might notice I venture outside of some of my comfort zone as well. I know these, these high over things and how we use these services together, but when it comes to cryptography, I have to have to give it to Jonathan. There's.

B (39:05)

Okay, so let's pull back to something that I'm actually very comfortable with. By the way, when Justin said so, first of all, I saw FIPS because Jonathan posted it. And so I immediately retweeted on Noster because it was Jonathan. So I was like, okay, like it might be hypey, but it's coming from this dude, so I have respect for it automatically. And then Justin was like. Justin was like, you should talk about FIPS on Dispatch. I was like, I would love to. And I was like, I can't believe I'm getting Jonathan on the podcast. And he's like, no, Jonathan doesn't want to speak, but you'll, you'll get Arjun. And Arjun is, is very impressive individual. I met him at Sovereign Engineering. So by the way, that's the background of the show. But Nostr, right? So NOSTR gets a lot of shit, I think, because of people specifically. If I would distill it, it's like, perfect is the enemy of good. Like NOSTR is fascinating to me because it's super simple and robust and doesn't try and solve all the problems. It's like very relatively easy to implement. I mean, things like just the events being just like simple signed JSON, the private keys just being simple, you know, single sig private keys. When you think about NOSTR used in this context, is there a concern that someone's basically digital identity is just,

A (40:29)

you

B (40:29)

know, just a private key that effectively can't really be rotated? Right.

A (40:37)

I'm not too much into the debate of, of rotating keys. I know it's a, I know it's a big topic. One thing I do think is that everyone should like, we use Noster keys in the FIPS context, but it's not meant to be tied to your nostr, like social identity. It is just you should. It should be rotated. We have an open issue for that. Should be rotated by default only when you host something. It should be static.

B (41:05)

Was. So then.

A (41:06)

Sorry, I don't know if that answers your question.

B (41:08)

I mean, it kind of does. I mean, I think like on this, on the social aspect, I don't think it's like that big of a deal. I think you just deal with key security. And so it's a key security problem. I mean, you can't rotate Bitcoin keys. Like, if your Bitcoin keys compromised, you lose your Bitcoin. But you can have many. Right? And like, that's kind of. And also we have all these different mechanisms for accessible cold storage and all this other stuff. So I think you kind of solve it from the key storage piece. And I think especially when you start thinking about agents and stuff using these things, it's more important that you can spin up a ton of keys rather than the individual robustness of a single key. I mean, I've been diving down the AI rabbit hole like pretty hard over the last five weeks or so. Like, my agent at this point has like 25 different noster keys, you know, so. Nice. Who knows? Like, my point is, is the cool part is that you can just spin them up on demand all the time, but in. And so in practice, on your side for this, are you envisioning that peers will constantly be changing their. Or not? Maybe not constantly, but they will be changing their identities often or.

A (42:23)

Yeah, I think they should. If you're, if you're just consuming, like, right. You are like you are right now. Like, you don't. You don't care on your phone when you're out and about what your IP address is. Right. You're not hosting anything.

B (42:33)

No, I have no idea what my IP address is.

A (42:35)

Yeah. And it's actually a good thing that it changes all the time. Similar to Mac addresses when you go hop on different WI FI networks. Like it's. Your phone spoofs it. You should do the same with FIPS keys only.

B (42:47)

I've been using Starlink. I've been using Starlink and I was talking to a graybeard about it and I was like, it's fantastic. Like, blah, blah, blah. And he's like, yeah, but Elon gimped it because the way he did the scaling, your IP address is like shared and rotated all the time and you can't host anything. And I was like, that's A huge advantage. Like if I accidentally leak my IP address, like I'd rather it be rotated and shared all the time.

A (43:11)

Yeah. The IP now is in this, this part of the. Part of the globe and two minutes.

B (43:17)

Like you know, just a region, basically. Yeah, you basically just have the region. So for the actual individual it's actually a benefit. But if you want to host something locally, it's a negative. That's interesting. But then how do. And I guess just peers would just handle that gracefully. They would just have. It doesn't necessarily have to be a static identity for your peers to.

A (43:40)

No, I. And I think if you cycle, it's going to be the same as like you know, walking from, from Wi Fi to 4G. Right. Now if you walk out the door, you get a little hiccup. Right. It would be similar.

B (43:51)

That's interesting. So, yeah, then that doesn't really matter, I guess, unless you're hosting something.

A (43:57)

Yeah.

B (43:57)

But if you're hosting something, how do you think about it in that regard? Like what if. I don't know if these hypotheticals are helpful or not, but what if I'm the person in charge of keeping track of everyone's the total grain that the community has or something? Yeah. And my noster key gets compromised. Is that really that bad of a situation? Couldn't I just broadcast out that like I'm now at this new one? I mean, I don't know how they'd verify that I'm the one saying it.

A (44:34)

Well, that's, that's social, social key rotation then. Right. If, if you lose your. You mean like the social master key?

B (44:41)

Yeah, I don't know what it looks like. A lot of these problems are easy to solve locally because you just like walk to the farmer's market and you're like, yeah, it's just. You can find it at this new address now. Right. I guess an example, a broader example would be. I don't know if you saw but downdetector. Com got sold for a billion dollars. Yeah. Isn't that wild? But anyway, I mean it's fiat games. It's like what is things valued nowadays? But part of the reason is not necessarily the data that you know what down detector is.

A (45:16)

Right.

B (45:17)

Like you go and check if a website's down, but it's not necessarily that they have the data on what things are down. And a lot of it is user reported, some of it is ISP reported, but it's also just a reputation. Right. Like I know if I'm having trouble with something and I'm trying to troubleshoot if it's me or if it's, you know, Cloudflare going down globally if I go to down Detector Duck they have a relatively good reputation that they're telling the truth and have good data. Right. So in this context it would be some npub presumably that is like tracking some information that. And they've been doing it for five years, 10 years. How would they. How do you. Have you even thought about like how they would gracefully move to. Because it's hard to keep a key secure for five years hot.

A (46:06)

Well, I can, I can only talk from the networking perspective.

B (46:10)

Right.

A (46:10)

The social, social rotation. I really don't. Do not know how. How that should be solved. But for the, the routing side, if you say you are hosting an important server, say you are hosting a cashew mint, like that's an important server and the traffic is encrypted with a NOSTR key, which means that the key is probably hot. Right. The key is used actively on the device. In that case it might be dangerous to link that to make that the identity of the mint because currently the mint identity is the domain name. So if that becomes NPUB FIPS and that key leaks because it's always hot, that's can be problematic. So it might still be useful to use, to use something like Node DNS where you have a key maybe in cold storage that then publishes these DNS records on nostr get resolved by name servers that are compatible and that then actually link to the hot key but that can then be rotated because the identity is then tied to self proclaimed. Yes, but at least you can keep one in cold storage and no one should be able to touch it.

B (47:23)

That's a good solution and that's a good example by the way. I'm running my base, basically my community bank. That's like a pretty serious responsibility.

A (47:31)

Yeah.

B (47:34)

And they need to be able to securely connect to me and know it's actually me and not some malicious actor.

A (47:41)

Yeah, I think that's particularly with mints, I think it is a risk that it somewhat uses the domain name as its identity.

B (47:50)

Right.

A (47:50)

It's Mint minibits cache.

B (47:52)

Right.

A (47:52)

It's, it's, it's not an. So if, if that gets rug pulled and they change their domain. Well, it's kind of gone from all the, from all the apps that I use it. So I think those things should probably be npubs to begin with.

B (48:09)

I mean on that note, I mean it would, it would make self hosting these things way easier. Right. And self hosting and then Connecting to them way easier.

A (48:18)

Yeah, that too.

B (48:19)

Specifically privately. Right. Like the powers that be know who. I don't know who runs minibits but the powers that be knows who runs minibits because of the existing networking stack presumably like it's pretty hard to get a private domain. It's pretty hard to have a private IP address that's not linked to your identity. Like he's presumably doxxed to the powers that be. And then existing solutions like things like Tor or whatever have a bunch of their own trade offs and actually solve a bunch of things with centralization themselves and centralized block lists and they have a bunch of. What about latency issues like Tor slow as hell. Would this be slow as hell in

A (49:01)

comparison or it's not, it's not onion routed. So because it's direct peer meant to be, it's meant to be fast. So I, I'm not making any, any big privacy claims here. I, I don't think it's necessarily way better than, than what we have. In some aspects it is like it's encrypted hop to hop and then also end to end. I think that helps. Of course if you are the almighty observer of the entire network. I don't think it's going to be much different from now.

B (49:31)

Fair enough. But right now we're forced to run through the almighty observer. In this case we wouldn't be.

A (49:37)

Yeah, yeah. I think that's actually a big thing with tollgate, right. It changes the financial incentives. So if you have more, more actors, especially on the edges, right. You're always going to have the undersea cables or like major parties, that's fine, it's going to stay that way. But if you have entire communities that self run and don't, you know, don't observe everything by choice, you can at least maybe only deduct that someone in this community looks up this thing which is way better than always being able to point down everything to every individual.

B (50:11)

It's kind of like in that situation it's almost like a hosted VPN, shared VPN kind of privacy model where you have a bunch of people coming out through the same IP address.

A (50:22)

Well, yeah, I think it, I think it helps to introduce more people to convince, to observe rather than just a few. Right now you just go to large ISPs and say like oh, you give me this information. Whereas with tolgate like you have to go to maybe half a million ISPs because everyone has an ISP.

B (50:39)

Right. I mean on the tollgate piece, while I Have you a lot of people's concerns I've heard with the tollgate project is like similar to running a Tor exit node. It's like, let's say I live in Paris and a bunch of random people are just connecting to my WI fi and paying me sats and then looking up maybe questionable stuff.

A (51:00)

Yeah.

B (51:01)

What is your answer to those people on the tollgate side?

A (51:05)

Yeah, I think the people that like have the connections to basically to upstream.

B (51:13)

Right.

A (51:13)

The general broader Internet should probably imply or impose specific block filters to block things that are like clearly illegal. But there's no clear cut, clear cut solution to it. But I think it's also why I want to move to local communities, doing everything local, where you should trust your community enough that everyone's like an honest actor or I want to maybe, maybe it's naive but.

B (51:45)

Well, at least in small communities you definitely could. Yeah, I think it's small towns.

A (51:51)

Yeah, it is a good point. But also it's not really. It's like, are you going to abuse cars for. Who's responsible for producing? Is the person that produces the cars responsible for, you know, anything, anything anyone ever does with a car, it becomes this. I think there's moral debates. That is you take either, either one stance on.

B (52:14)

Yeah, I mean it's not specific to tollgate. I mean you see this already with like coffee shops and stuff. Right. It's particularly in the developing world where you have bad cell connections. It's like every coffee shop becomes an isp. They just don't have tollgate there to charge people for it. Instead it's just, you know, you're, you're buying a coffee so you get the WI fi password.

A (52:35)

Yeah, yeah, but there's no, there's no good answer. My personal opinion is that, you know, the Internet is, should be open, open and free. I think the only, the only counter movement is that you observe everything and start to play police on every, every little thing, which is what we're seemingly heading towards. I don't think that's the right future, making everything like completely open and free. I think that's probably the best way and like solve, solve things in the old fashioned sleuthing way where you, you go after bad people whenever the, the chance occurs.

B (53:12)

I think that's reasonable. I think that's what a lot of us are working towards. I just asked my bot to steal man fips. I think this conversation has been fascinating. I think the project's really cool, to be frank. I obviously need to play around with it. I'm Hoping this conversation leads to more people trying it out, giving feedback.

A (53:36)

Yes, please. We want, we want a lot of people to test it.

B (53:39)

What's the process for, like how does someone go about doing that? How should they think about participating?

A (53:47)

Testing it? Yeah. So currently the repository, there's some, some manuals where you can compile and run it yourself. There's test cases, docker, Docker based meshes you can spin up quite easily. There's like step by step guides. Your LLM can do it. And we're currently working on packaging. Packaging it. Tarball should be ready any, any moment now and I'm working on packaging it for open WRT routers and a bunch more.

B (54:19)

That's awesome.

A (54:19)

So then it should become easier to deploy it as well.

B (54:26)

I. Okay, so I'm gonna play around with it. Hopefully the freaks play around with it, give feedback. Those who can will look over the, the code and specs and maybe provide feedback and advice on that side. And then when I'm a little bit more fluent in it and more people are using it, I would love to have you back on when I can ask more reasonable questions. And I hope I wasn't too much of a struggle here. I think it's still be helpful.

A (54:52)

These questions are, these questions are valid and I have to make like, you know, some of it ventures outside of my, my comfort zone as well. So it's very, there's a lot of things to it and it's really hard to, you know, convey these things in a short manner as well.

B (55:10)

So I asked, I asked Opus 4.6 which is supposed to be the smartest of the models right now to steel man your project. I've never done this before, but I'm just going to read what Opus said. FIPS embodies a principle that the cypherpunks articulated decades ago. Privacy and freedom of communication should be enforced by mathematics not by policy. Laws change, companies get acquired, governments get replaced. But the laws of cryptography don't change. A system where your privacy is guaranteed by encryption rather than by someone's terms of service is fundamentally more robust. FIPS is ambitious, but not delusional. It's technically sound, addresses a real and growing need, is built by someone with relevant deep expertise, integrates with an existing ecosystem, and is designed to work even at small scale. If it gets even 10% of the way to its vision, it can provide genuinely life saving communication infrastructure for people living under authoritarian regimes and a meaningful privacy upgrade for everyone else. The Internet was supposed to be a network appears somewhere along the way it became a network of subjects. Phipps is trying to make it a network appears again.

A (56:18)

You didn't even have to have me on. Could have just.

B (56:20)

Pretty good, right? That was just a one shot prompt to steal man. Phipps and I gave you.

A (56:28)

That's a great summary. That's a great.

B (56:31)

Okay, well I had a great. A great time. I'm excited. I'm excited to see where this project goes. You have any final thoughts for the Freaks before we wrap up?

A (56:40)

Test it, give feedback, break it.

B (56:45)

All that I will link to the GitHub repo. I'm going to link to Sovereign Engineering. I think if you're a developer out there, you should check it out. It's a really great program. Anything else I should link to in the show? Notes for the freaks.

A (56:59)

Tollgate.

B (57:00)

Tollgate. I'll link to Tollgate. You're on Nostr too. Do you have a personal ampub on Nostr?

A (57:06)

Yeah. Yeah.

B (57:08)

Okay. You don't have to say it out loud. You can just send it to me and then I'll put in the show notes as well.

A (57:12)

Awesome.

B (57:13)

I'm not going to make you read the m pub. Okay. Arjun, this was awesome. Thanks for joining.

A (57:19)

Yeah, thanks for having me.

B (57:21)

Freaks. I hope you enjoyed the show. Provide feedback. All links are@ciladispatch.com if you go to primal.net citadel. It's the main Citadel Noster experience. That's the best place to leave feedback. Then the rest of the community can jump in and everyone can comment. I'll tag Arjun there in that post as well. Awesome. Love you all. Stay humble. Stack sats. Peace.

A (57:45)

Cheers.