~sarlev: I’ve been opening these conversations by asking people: what’s the first thing that drew you into the idea that we needed to throw away and rewrite the entire network computing stack?

~master-malwyl: I don’t think I entered through that door. I think my first exposure to Urbit was finding the Twitter account, probably in 2020, and I didn’t really know what it was. It wasn’t immediately clear what Urbit was. It didn’t pattern match anything. I took a look at Wikipedia and there wasn’t much there. I just thought it was some nerds’ passion project. Then, maybe three months later, I found Curtis through his whole other life. I remember listening to a podcast he was on and thinking his interests were quite different from mine.

But I could recognize that he was extremely smart. Whenever I find someone extremely smart on the internet, I tend to go deep and see everything they’ve ever done or said. Shortly thereafter, I noticed the connection. I was, “Oh, that weird project I saw. Oh, he made that.” Then I thought, “Maybe I should take a closer look at this now that I know that this guy’s really smart. Maybe there’s something worth looking into there.” I went wide from there. I searched on YouTube and one of the first things I found was Anthony Arroyo’s three minute explainer of what Urbit was. It was from a challenge to explain Urbit in an elevator pitch style, and he had a kind of charisma. The content of what he was saying I found so fascinating.

I just knew that there were a lot of people here I’d probably feel a kindred, like-mindedness with. Intelligence seemed very high. The subject seemed very interesting. I sensed that there was a rabbit hole I was going to dive into. Then I was reading the Urbit.org docs and the blog. I learned about the identity system and, having a crypto background and being a highly active trader at the time, that was when I found the period of time I’m describing. I was full-on, millions of dollars flying around on FTX and stuff, that’s all I was doing. So naturally I went over to OpenSea and I took a look at this Azimuth thing. I saw these stars and I thought, “Well, one, I have a lot of capital right now. I really like this project. I seem to like the people. Why don’t I buy some of these?”

At the time I think I bought my first star. It was two ETH or one ETH or something. I can’t remember what the price of ETH was at the time. It’s probably similar to what it is now. I was weighing buying a star. I waited a week or two because I was thinking, “I need more time to process and see if I want to actually invest in this thing.” Then I thought, yeah, I’ll buy a star. The moment I bought a star, I was like, “What am I doing? I need to buy a galaxy.” That was, I think, January of 2021.

I remember I bought it and then I thought, “I need to just go get a galaxy.” So I bought a planet on OpenSea, sent Galen a message, and told him I was amazed that it worked. I sent him a peer-to-peer message. I booted this runtime process and then it opened an interface and then I could send a message directly from my computer to his computer and he responded. He was the co-founder of the project and I was like, this is so cool. There’s a wormhole between my computer and his and there’s nobody in the middle. My first message to him was something like, “This is incredible, and how can I buy a galaxy?” That was basically my opening message. Then he said we could talk about it and I was fishing around on the network.

I eventually bought a galaxy secondhand from someone who was given one as an early alpha tester. We negotiated a deal. So I purchased ~del, and my entry point was not—I wouldn’t describe it as having the same experience I think that many engineers or developers have of just being so pessimistic about how broken the whole internet architecture and computing architecture is. I think it’s because I was a computer science dropout. I self-taught programming and then I used that to trade. I never felt the system was broken. Things always just worked. I’d put my mind to something, I’d get it to work, and I didn’t really have much of a reference point of how much better things could be.

I just took it for what it was. This is just how things are built. Maybe if I had a longer career in professional software, but I was always just a hacker. When I was flying high doing all the algo trading stuff, it was literally just a Node.js process running on my home computer. I mean, I was putting up pro numbers, but what the pros were doing was probably so much more sophisticated than what I was doing. So I never felt that sense that we have to start over.

I eventually came to understand that perspective and come to share much of it. But in the beginning it was more of just a fascination, seeing the vision and finding it very compelling. I didn’t share that sense of “oh god, this thing’s fucked—we need to start over.”

~sarlev: You mentioned this story about reaching out to Galen and realizing, “oh my gosh, I’ve got this peer-to-peer wormhole.” That’s one of these really intense feelings of yourseness—“It is mine. I’m not intermediated by anyone.” To extend that, do you have a favorite part of the system that you think will change the way that people interact with their computers or computers interact with their lives? What’s the thing where you’re, “This is going to change the way that people do computing?”

~master-malwyl: I mean, my answer might be Azimuth actually. Your question is about changing computing, but if the question more broadly is how will Urbit change life or the way that people interact with computers and each other, I think my answer is probably Azimuth. Just having a cryptographic identity primitive that is well-designed, I think just basically perfectly designed—or maybe not perfectly but it gets all the important things right—I think that is what I find most potentially disruptive. I love the simplicity of it. It’s such a simple idea, but it can really shake up the foundation of the internet. So maybe I can run with that answer and see where it would go.

But I don’t want to neglect the importance of having a personal server, having a computing environment that is completely under your control that can serve as your emissary to the rest of the world. That’s really crucial for a sane future, and I think it’s becoming more and more obvious to people who are not even in tech. I don’t want to neglect that.

But what I said to ~simfur-ritweb once in a group call was that Azimuth is a wing in the cathedral. I think it’s a way to put it. What’s so important about having an identity primitive is that identity is about as primitive as money. They’re comparable in their elemental nature, and we haven’t had that second sort of revolution with an identity asset that becomes the reserve identity of the world in the same way that some might look at, say, Bitcoin as emblematic of that revolution in the financial or capital sense.

It’s really important that every human has a key pair they can use to sign things and to encrypt things. It is absolutely necessary for a sane future for every human to have that as a basic assumption in how they interact with the rest of the world. You need to be able to sign things and you need to be able to encrypt things and Urbit ID gives you that out of the box. It does so in a way that is cheap. They’re currently like a penny basically or less than a penny to even own one. So the cost is not like a home. In this housing crisis where houses have become a speculative asset, people are forced into serfdom because they’ll never be able to buy a home.

I can’t really imagine being totally locked out of ownership ever being the case with Urbit ID because they’re just so plentiful. The only scenario where that could theoretically become an issue is if there’s just billions of agents and they all need their own or something. But even then, I don’t think it’s realistic.

~sarlev: I would even make the counterpoint to extend that. A lot of people are playing with the Clawdbot (fka Moltbot, fka OpenClaw) these days on Twitter and other agentic things, and they’re putting these agents inside of their own system and they are acting as the person. I think people are going to recognize, “Oh actually, I want you to be acting as an agent subservient to me but with your own identity. I want to know what you’re doing. I want to know what you have access to.”

And of course Azimuth has this lovely feature of the moon-level address space and you’re, “Oh cool, I can have billions of agents personally.” Any world in which I need to have over 4.3 billion active at the same time agents that can’t be breached and rekeyed is a little bit wild.

~master-malwyl: Yeah. I mean, computer processes have names today. When you pull up the list of running background processes on your computer, they have relatively illegible names because they’re not meant to be spoken. But a process needs a name and agents are a process. So yeah, I think the difference between Azimuth and Arvo is that Azimuth is basically done.

There’s not too much that needs to be changed at the kernel level, so to speak. There’s lots that can and should be done with it, which is what I’m currently focused on at the higher-level application level. It’s not a research project. It just needs enterprising individuals to go out and do something with it. That might be my favorite part currently. I see very rich potential in a from-scratch operating system like Arvo, but that’s not where my attention is currently concentrated.

~sarlev: When you think about the potential around Azimuth, obviously there’s lots of exploration in this digital identity, cryptographic identity space from the Nostr “npub” key pair guys where they’re grabbing the thread of identity and communications. You’ve got the Farcaster people doing social with the wallet. Where do you think Urbit ID fits into that broader space?

~master-malwyl: I think that Azimuth’s purity, its design purity, is what makes it the apex identity primitive, in the same way that Bitcoin stripped out all of the unnecessary parts of money and just got to the root of what money as a technology should be in terms of the essential properties. I think Azimuth gets that right for identity. It recognizes and reflects that your identity is just a number fundamentally. It’s just a point in a mathematical space. Those numbers can then be mapped to easier-to-work-with representations.

People have played around with alternative representations. There are basically infinite ways to transform numbers into other things. But fundamentally it gets the insight that what you have is just a number and it needs to be unique and easy to work with. I’m reminded of Zooko’s triangle where you have human meaningfulness. The problem with working with a raw key pair is that while it escapes the personality problem of ENS—where it’s more like a nickname than an identity—key pairs give you this garbled mess that agents and computers can work with just fine, but for humans they’re not easy to work with, you can’t memorize them trivially.

It gets those essential properties right, and what’s missing is just the applications of it. Nostr has a functioning microblogging protocol; I haven’t played around much with it. I’m not sure what else they offer. Farcaster has their IDs and their own microblogging and all that stuff. Azimuth just doesn’t have those—it has Arvo, but Arvo has not been able to achieve any user traction outside of the fanatics up until now for reasons that are understood. I think the design is correct and ready to be used.

~sarlev: What do you want to be able to do with your self-sovereign digital identity? When you’re like, “Hey, I’ve got my Azimuth ID. I’ve got my Urbit ID. It’s on Azimuth. I cryptographically own it. It’s got these key pairs.” What are the things where you’re like this is what I want to do now that I own my own digital identity?

~master-malwyl: Here’s something that I’m not even sure I personally want to do, although I think I kind of do. As an entrepreneur, I recognize this is something that maybe a very large number of people might want to do. I think that being able to attach claims to your identity that are cryptographically verifiable is really interesting. If I have a million dollars in my bank account and I want to prove that to you, I have to go to the bank and get them to write a letter and I’ll sign it, they’ll sign it. It’s very 20th century.

With Urbit ID, something you can do is sign a claim that you own a particular Ethereum account or Bitcoin account or Solana account or whatever and you can link them in a way that is cryptographically verifiable by anyone.

You don’t have to have trust involved. What that can allow you to do is to speak or communicate, interface with the rest of the world, as those assets. The identity can become a kind of cloak around some assets and then those assets can have a voice. They can act agentically. That idea I find very fascinating. Because I don’t think ever before capital has been able to operate in that way where something like a Bitcoin treasury has a voice and now whoever controls it can speak as it. They don’t have to speak as John the fund manager. They can just speak as those bitcoins.

And when people interact with that, it’s almost like a persona in a way. It’s like yes, I’m going to speak with these bitcoins and the bitcoins can speak back to me. It can be bitcoins or it can be equity in a company.

Tokenization of equity is becoming more of a talking point. What happens when the entire investor base of an asset can communicate with itself without having to expose the human identity of the people behind it? What’s stopping a chat room of 20% of all bitcoins getting together and debating whether to fork or whether to make certain changes? It’s the power of coordination and that coordination is made possible because everything can be glued together cryptographically in a way that there’s no trust involved.

~sarlev: I think this is really interesting too. If you were to contend with ENS as one of the leading digital identity primitives that’s out there, there are two big flaws in ENS as an identity system that maybe are actually kind of good for making money for the Ethereum Name Foundation. One is you’re renting them. So you’re like, “Okay, well, you’re renting this identity and you’ve stopped paying. You don’t get to be you anymore.” Good as a business model, bad as a foundational primitive for humanity.

And then two is why is the solution to Zooko’s triangle actually important. You don’t want security leakage that imbues those assets with more identity than is necessary.

~master-malwyl: I’m glad you mentioned the ownership piece there because I forgot to bring it up earlier. That is totally critical, that one be able to own these. Of course one can own just a generic key pair, but then you run into all the other problems.

~sarlev: So how did your relationship with computers change when you started working on Urbit? Has it shifted any way that you do your digital interactions?

~master-malwyl: I think the honest answer is no. I don’t think I’ve changed my habits in any meaningful way as a developer.

One thing that I’m really grateful for is just the education I’ve received in exploring and seeking to understand the system. I learned so much about networking, computer networks, and the substructure of the internet. I barely could have explained what an IP address is. I remember asking some senior engineers at the time: why doesn’t the world just use IP addresses as identities? Why didn’t that happen? Why do we need this extra layer over it? Why don’t we just use IP addresses? They’re basically like Azimuth points. I didn’t understand functional programming very much. I had never used it. There was just so much that I learned.

This experience informs the future that I want to inhabit: one in which every human has something like a personal operating system. It’s a secure environment under their control where they can dictate the border policy of what is allowed to enter into and what is allowed to leave their computing environment, supplemented by all kinds of cryptography to make it all work. That is crucial. It’s like having the right to personhood. It should be like a fundamental human right in the future. It’s that basic.

Where I depart somewhat, at least temporarily, with some people—although I think this is probably a pretty common opinion—is I just think Arvo needs a lot more time in the oven. I say that as someone who’s done Hoon School, I did Core Academy, I’ve written a jet. I’ve done a lot of that stuff. Where I became disillusioned with the timeline I had in mind, how quickly this could take over the world, was shortly after the Uqbar pivot. The original Uqbar pivot. I read a critique of the founder who laid out this critique of why they were going to pivot away from building on Urbit. It was a really strong critique and it was very informed. At that point I’d invested a lot of capital into the address space and I thought, okay, I need to take these criticisms seriously.

I had poked around with Urbit development up until this point. I created a really basic Gall app and got it working end to end, nothing impressive, just to understand the architecture. Then I thought, when these critiques were leveled, “Okay, I’m going to try to build something serious and see if I give up.” If I give up, it’ll be because I finally admitted this thing isn’t ready. If I succeed, I’ll be able to go to that founder and tell him, “No, no, no. You’re just being lazy or whatever. You need to be more determined.” For about six weeks after that critique was published, I went really deep in. I was trying to build a Shopify for Urbit.

I got really far. I mean, it’s all I did day and night for about six weeks. I threw in the towel ultimately because I realized I stopped having fun. It wasn’t fun anymore. To create anything that has commercial potential, it’s probably not going to be all fun. But it started to feel like I was cleaning my entire apartment with a toothbrush. I could do this eventually, but this is insane. Why would I spend my time this way? It doesn’t make sense. I could see how much better it could be in the future, but this is just insane to continue investing my time in. One of the roadblocks I ran into was: to have a Shopify, if I want to have a little web store, I need pictures.

I need to be able to upload a picture and people can view a listing and see the pictures. I thought, well, this is my personal server. Obviously, my photo has to go on my personal server. I’m not just going to put it on AWS and then put a URL in my server. That’s ridiculous. To get native image hosting working at the time—and I think it has improved somewhat since then, maybe some people have created scaffolding to make it better—but there was just no good way to do that. What I did was take the image and convert it to a base64 string and then I would poke the base64 string into my ship as a JSON value. It was a JSON object and one of the values was this monster string.

Then my ship would commit it to the remote scry namespace so that when my client would later request it, or any client would request it, the ship could serve it without hitting the event loop and breaking. It could serve it efficiently. At the time, directed messaging was still very far away, so if you were transmitting it over the network, say a 10 megabyte file, that’s like a thousand 1 kilobyte packets that all need to go through the event loop and get stripped and verified or validated. It was just comically impractical to get something as basic as image hosting. Like, I need to upload a photo to my server and then serve it in a Shopify listing type thing. If I can’t even do that as a developer without feeling like I’m scrubbing the floors with a toothbrush, I’m like, “All right, the scales have fallen from my eyes.”

~sarlev: This was three years ago, maybe roughly two and a half, three years ago?

~master-malwyl: Yeah. Two and a half. Three.

~sarlev: I think that experience around that time was probably not uncommon for people. “Oh, I’ve got a personal server, obviously I should be able to do thing X with it.” The dev tooling isn’t really there and the performance of the underlying system couldn’t handle it. As you know, there’s the vere64 work going to be able to do large files and there’s directed messaging helping the network speed. What other system improvements are you most looking forward to? Maybe not even the deeply technical projects, but what are the other things that you’re looking forward to being easier as a developer or more pleasant as a user? What are the things you really want to see land?

~master-malwyl: I’d be very curious to see how much farther I could get now with LLMs. Even if everything were frozen in time, I wonder whether the sheer power and versatility of LLMs and agents today might solve many problems as a developer.

Even with that, though, we need basic file storage. You need to be able to treat your ship like an S3 bucket, basically. Put your images, your audio files, your movies, whatever. It needs to be a basic kind of “Martian S3.” I would love to be able to back up all kinds of files. I don’t have very good backup habits. I would love to have a node in the sky that was designed to ingest all that kind of stuff and was secure.

The other thing that has happened since then, which I think is the coolest project ever, is Named Data Networking and the whole Directed Messaging undertaking. I’ve probably read that UIP like 40 times. I would visit it and reread it and think about how fundamental it is. I recommend anyone interested in networking or computing give it a read. Named Data Networking and content-centric networking dovetails with what I was saying earlier: every human needs to be able to sign things and encrypt things. That’s really critical because a much better model for the dissemination of information around the world is a model in which data is signed rather than securing channels of communication. You secure the data, and then the manner of dissemination—or how that is routed between those who want to consume it and those who are providing it—just becomes a higher-level concern that doesn’t need to be entangled with the routing. You can just ask “is this the right information?” rather than worrying about who passed it to you.

~sarlev: I think this is one of the points of why Urbit is rewriting so much from the ground up. The project as a whole is actually willing to take on these underlying projects that are foundational to the way that you do computing. I remember being at Reassembly and listening to ~master-morzod’s talk on Named Data Networking, going over the history of networking, how these networks have evolved, and the fact that Urbit is the first production implementation of Named Data Networking that is being used by people in real life is fundamentally fascinating. It’s certainly the case that Urbit is slow right now, but when we’re talking about communicating peer-to-peer on a distributed network, the way Directed Messaging works as a technology is obviously the next layer of how you do networks and data exchange.

You touched a little bit on what you’re working on. Can you tell us more about the project you are working on currently? Why are you drawn to it? What of all we’ve talked about so far has led you in that direction?

~master-malwyl: So I’m building a website. It’s called urbit.xyz. What I’m attempting to do is marshal all of my abilities and knowledge to create something that is actually useful and that capitalizes on all the latent potential I see in Azimuth in particular. That’s really the focus. I’m taking a swing at that to create something that finally digests everything that I’ve learned and takes all the close to 10,000 hours worth of JavaScript that I’ve written in my life—all the algo trading stuff I did, a lot of that was JavaScript—and try to actually produce something that could realistically drive growth for the project.

The way I’ve approached that problem is by putting Azimuth at the center of the picture. One thing that I believe is extremely high priority and that I think I’ve achieved very well is to reduce the friction of onboarding or the friction to trying something to zero, literally zero. I think I’ve done that because my system allows you to use it without having to take action to acquire an Urbit ID. That was a key problem I recognized the importance of solving. So when you land on urbit.xyz, you get an Urbit ID.

Clearing this hurdle of getting someone an Urbit ID ties into the idea of providing a sort of cloak to capital so that it can connect socially. It is personally very interesting to me and I think has the potential to catch fire. I think the best way to approach it is to marry one thing that has clear product-market fit in the crypto universe to Urbit ID. What has product-market fit? Exchanges. And I’m focusing on decentralized exchanges (DEXs) because all activity is totally transparent. One can tag or link every action—every trade, every position, all the liquidity that’s currently committed, the history of that capital, how it all flows—and annotate that with Urbit IDs so that a whole social layer opens up on top of that exchange. I’ve never seen anything quite like this, despite having used literally dozens of exchanges. I think the potential for something like that to catch fire is real.

I’ve been trying to come up with mottos or taglines to briefly convey what I think is the core idea. One term I came up with that I like is “capital unmuted.” The idea is that capital is out there, but it’s on mute, and Urbit ID is a way to unmute the capital. ENS allows you to attach a nickname to some account that might have a billion dollars in it, but there’s no supporting infrastructure for the boring stuff. You need a chat, you need group chat, you need stuff. Someone has to go build that and it’s time-consuming to polish it, as employees of Tlon would understand firsthand. There’s a lot of work that goes into creating a chat app.

~sarlev: It’s not just, “Claude, make me a chat app.”

~master-malwyl: Yeah, because there are a lot of mistakes to make. It’s really, in some ways, as challenging as writing an operating system. Discord is kind of like an operating system. The amount of work that goes into delivering on the desires that people actually have for how they want to interact socially is a complex problem. So yeah, right now I’m doing the boring stuff and focusing on making it good. Then I’m going to do the novel and powerful thing, which is creating that connection between capital and the social network.

Here’s another way I would put it: one of the top two or three things that have product-market fit in crypto is an exchange. Another business that has massive product-market fit for the internet was social networks. There are probably fewer than a dozen of them that are serious, that people use every day, and they have tremendous network effects. They’re very defensible businesses. What I see myself as doing is creating a merger or a marriage of the two. I’m creating a social exchange that is both a crypto exchange and a social network. They’re not slapped together. They’re integrated in a principled way that reflects a vision and recognizes what I think is a latent demand for capital to be able to interact in this different kind of way so that it can coordinate more effectively.

I don’t think it matters unless it actually changes outcomes. If you give 100 bitcoins the ability to chat, cool. Someone could go on there and brag about, “Oh, I have 100 bitcoins.” But unless it actually materially alters outcomes for that project in some way, because they needed to say something that they couldn’t say previously, then I think it’s a toy or a fad. But if the execution is right and I succeed in recognizing exactly what kind of coordination has latent demand, then I think it could change outcomes.

Here’s an example that is a little distant from what I was talking about, but it’s related. I was a creditor of the QuadrigaCX exchange in Canada that blew up. A lot of people lost a lot of money and, in the aftermath, after the exchange shuttered, the creditors had to assemble and organize and there was no good way to do it. We ultimately found each other on Reddit. Somebody created a QuadrigaCX subreddit and we started talking and the lawyers found it and the lawyers swooped in and helped us organize. But there was a lack of a Schelling point. The source of truth was the database of all the trading history and asset balances and which usernames mapped to which deposit addresses and all that stuff. It’s just a database that could be doctored, that could be manipulated.

This might be people’s entire life savings and it’s just a row in a MySQL database or something that some bad actors could go in and modify. Maybe they did. I think there are reasons why it’s kind of inevitable that you’ll be able to use your capital, which is to say the keys that control that capital, to prove that you own it or control it, and then to interact with the world with that attached to you or attached to a throwaway identity or whatever so that you can have a cloak around it. I believe there’s a latent demand for that kind of coordination, that ability to coordinate more effectively.

~sarlev: I remember talking to you and ~radnep-bolled about his time at Facebook and one of the things they found out about people’s social circles is that having different contexts for different things is important. There are a dozen big social networks out there and people will have multiple identities across them. They’ll talk with their football friends on Facebook and their college buddies on iMessage and their internet friends on Reddit or whatever. But as you’re pointing out, there’s this latent gap of where you talk about your financial strategies. You might be totally willing to share them, but it’s kind of gauche in the wrong context.

~master-malwyl: People would maybe use the Wall Street Bets subreddit, but you couldn’t just sign a message that says, “Yeah, I have $5 million in the bank,” next to your username.

Maybe that’s something that people would want to do, not merely to brag, but because if you’re going to play football people want to know that you scored however many touchdowns. It’s like your resume.

~sarlev: Proof that you didn’t just generate this throwaway account to get Reddit karma.

~master-malwyl: Yeah, and we are drowning in noise. The internet is just going to get noisier and noisier and it’s going to get harder to distinguish the signal from that noise. Cryptography gives you a means of establishing a signal. Who said this? Some username said this, but who really said this is $5 million speaking. That’s signal.

~sarlev: Certainly is. I’ll give you my last cookie-cutter question: if you weren’t working on Urbit, what would you be doing instead?

~master-malwyl: That’s a fun question. I think the true answer is probably I would just get back into algo trading. When I was doing algo trading it was all by hand, typing out JavaScript with my fingers. The amount of leverage I have with modern programming tools now, especially the ability to parse natural language—when I was doing algo trading it was all just quantitative. I wasn’t doing any kind of natural language processing. I wasn’t analyzing Reddit or Twitter or any of that kind of information troves. For the people that are deep in it right now it’s probably so intellectually stimulating to be thinking about where’s the alpha and how to systematize it. It’s probably such an interesting problem. It’s probably insanely lucrative. I still think that automated algo trading or whatever you want to call it is the most lucrative thing that any individual can do with their time if they’re good at it and they have enough capital to throw around. I think it’s probably still the best way to convert raw IQ into money. If there’s anything that is just a pure, efficient conversion between IQ and money, it’s probably algo trading.