Playback speed
Share post
Share post at current time

#13: Nick Simmons

All about Urbit

Nick Simmons is a founding partner at Octu Ventures, a member-driven venture DAO investing in teams building on Urbit. Urbit is a new computing paradigm that provides complete ownership of your digital world.


0:00 - Intro

1:19 - What actually is Urbit?

5:43 - Urbit ID and Schelling points

9:05 - Why Urbit?

10:23 - Roko Mijic on Urbit vs. TikTok and Crypto

17:32 - Urbit vs. Worldcoin

22:26 - Niche or growth model?

28:50 - Why haven’t Urbit star prices recovered since 2021?

33:13 - Intrinsic value of Urbit address space

36:37 - Urbit as digital land

42:51 - Urbit and DeFi

45:42 - Personal AI on Urbit

51:35 - Urbit-native hardware

55:58 - Urbit design and aesthetics

1:02:15 - Outro

More Episodes



My Twitter:

Subscribe to my Substack:


Theo Jaffee (00:00)

Hi, welcome back to episode 13 of the Theo Jaffee podcast. We're here today with Nick Simmons. Nick is a partner at Octu Ventures, which is a venture firm that invests in companies that build on Urbit.

Nick Simmons (00:13)

Hey, Theo, great to be here.

Theo Jaffee (00:16)

Alright, so first question is, I know the dreaded question for any kind of Urbit guest is, explain for the audience, for people who may not be as technically minded, what actually is Urbit?

Nick Simmons (00:30)

Okay, so the really top level overview here is that you could say that Urbit's a new internet. It's a new way to connect computers and the people that use those computers with applications, platforms, protocols that do all the things that we want computers and the internet to do. So that is sending messages, sharing in social spaces, connecting business, anything like that. Urbit has the potential to build new ways of coordinating

human activity, economic activity, social activity in ways that those of us who work in the project think are probably saner and more stable over the long run. Now, what does that actually mean? To get slightly more technical, Urbit is three key pieces of technology that all unite into this network stack. So there's Urbit OS, which is a completely deterministic functional operating system. So it's a full computational stack and

Right now it runs as a virtual machine on top of any Linux environment, but there's no reason you couldn't run it on a chip. And in fact, shout out to ~mopfel-winrux who is an Octu member and who is working on a project called Nock FPGA to actually design a chip that implements the core Urbit instruction set, which is called Nock. So that's Urbit OS. And the overall goal of Urbit OS is to allow for a computational stack that's

Theo Jaffee (01:56)


Nick Simmons (01:56)

extremely stable. The Urbit Core development team is slowly working towards what they call, you know, Kelvin zero, which is a totally frozen instruction set. So the core of the computational stack just doesn't change and for it to be simple. So right now, I mean, the, the, the Nock instruction set, which can be broadly, broadly analogized to a kernel.

fits on a t -shirt. It's, you know, it's dozens of lines of code. The Linux kernel is millions of lines of code. This has really striking and obvious upstream, you know, social and organizational implications because when you have both extreme complexity and an ever -changing pile of, you know, what programmers call cruft, which are to say overlapping abstractions, patches on patches on patches.

Theo Jaffee (02:51)

Thank you.

Nick Simmons (02:55)

to make a computational system run. You basically require a giant bureaucracy to make that run. It's more technical knowledge and it moves faster than any one person or even small group of people can keep their heads. And so that kind of necessitates a bureaucracy. And computers probably shouldn't require bureaucracy. We're probably past the point of the society where we should be, where that's necessary. And that has all sorts of implications around, okay,

if all of the methods of organizing a bureaucracy start to look totalitarian, or they start to look like they impose a lot high coordination cost, then there's obvious downsides to that. And so the idea of Urbit OS is that you should be able to run a computer and that this gets into why people call Urbit a personal server, a computer that could serve up content to other computers, say other people, and it should be stable.

It should be easy for a layman to run. You should not have to have, say, a certificate in Apache server maintenance in order to run a server, which is basically the case right now in terms of the servers that major internet platforms run. And it should simply be yours. So the old tagline for Urbit was simple, durable, yours. The second part of the Urbit stack is Urbit ID.

So this fun fact, this is actually the first or second implementation, I believe of the ERC-721 technical standard, which basically just means NFTs. So these are Ethereum NFTs and there's three main levels of them, planets, stars and galaxies. Planets are kind of the individual level, Urbit node. Stars are, you could think of them as an ISP or maybe kind of like a community Schelling point node.

Theo Jaffee (04:39)

What do you mean by Schelling point?

Nick Simmons (04:41)

A Schelling point would be just like a coordination point. So, you know, something that a community wants to run in a very technical sense. What stars do is that they are their points for packet routing. So actual, you know, packets of data between any other node in the network and also a peer discovery. So I'll get into this in a minute, but essentially and then we'll go into galaxies. So galaxies, there's so there's about four billion.

planets, about 165,000 stars, and then 256 galaxies. And galaxies are the root nodes of the network. They're the source of truth for major software updates. And they also form kind of a DAO that makes decisions on whether to upgrade the network or change it in any way. The main example of this is that during DeFi summer, the fees on Ethereum obviously went way up. And so the cost and fees to

boot a new Urbit plan of our star became somewhat prohibitive. I think it was about $350 in Ethereum gas fees at one point. And so Tlon which is the company I used to work for, which is the main company that incubated the Urbit network, they actually devised their own layer two roll -up to reduce those fees. And so the Galactic Senate had to vote to decide that this was, you know, these new layer two identities were going to be valid. And so the way this works in practice is that,

Every planet has a star sponsor, every star's galaxy sponsor, and you can leave your sponsoring entity without permission. You just have to find a new one to adopt you. And so this enables a network architecture that roots around damage, which is to say offline nodes. So for example, if your star is not reading packets anymore, it's not online, you simply find a new star.

or if your star decides it doesn't like you anymore, you find a new star. And it doesn't have any say so over letting you go. You have the right of exit. And so, pure discovery in this context is basically that you have to go find someone else's planet for the first time when you wanna talk to them. And so the stars handle that decentralized, initial sniffing out, if you will, of where this other person is based on,

their public key identifier that you have, which is basically their name. So for example, my Urbit planet name is ~simfur-ritwed and my co -founder in Octu is Kenny, AKA ~sicdev-pilnup. And so if for the first time I ever sent Kenny a direct message, a DM on Urbit, my planet sent a packet to my sponsoring star, which then queried a whole bunch of other stars and it found which one was sponsoring

sick death pilnup and it passed on the data but it also let both of us know where in the whole Urbit network topology our planets were sponsored, were located. So that's the big data dump on how Urbit works. I'm sure you have questions or I'm sure there's something like clarification on for the audience. So I'll pause.

Theo Jaffee (08:02)

Yeah, so clarification for the audience would be just like, that was a very good explanation that went into a lot of depth, but just for like a very short, like 20, 30 second explanation of what exactly is Urbit and why should I care? Why should an audience member care? What would you say to them about that?

Nick Simmons (08:20)

Sure, my explanation would be that there's a whole bunch of decisions that have been made over the entire history of the internet that assume certain things to be true and bake in certain qualities of the internet. And they're not all true anymore. We have a whole bunch of conditions that make them irrelevant now. We can build better network forms of communication and collaboration that aren't dependent on that history and those historical choices.

And Urbit is one of the best projects that I found that actually opens up to those possibilities of coordinating people and capital and social connections in new ways. And obviously that needs to be dug into. But fundamentally, this is about allowing people better coordination tools to reflect, to build projects and social graphs that better reflect their actual desires.

Theo Jaffee (09:19)


So, I don't know if you've heard of Roko Mijic, I did not pronounce the last name correctly, but he's somebody who's been involved with Urbit for a while, he used to have his Urbit handle, ~bacsul-lissyl I believe, on his Twitter profile, and he tweeted pretty recently,

Urbit spent 22 years failing to build something there isn't even a market or need for. Think about it. Imagine we wave a magic wand and make Urbit work as intended. People get their own personal serverlets where they can easily host their own content on their own hardware. Meanwhile, in the year of our lord 2024, Gen Z can't even unglue themselves from TikTok for five minutes. You really think they're going to build their own personal website and then also self -host it on their own hardware? And even if they did, would it really make any difference to the world? In 2024, you get away with saying all sorts of naughty things on Twitter thanks to Elon buying it.

And as a bonus, it's all easy to use and easy to read because it's a single app, rather than millions of separate websites. Most of the value of the internet is in, and he bolded this part, most of the value of the internet is in the semantic and social network, the connections, the sorting and filtering.

And these network hubs are easy for rent seeking parasites to capture. Urbit can't really solve that because it doesn't have global state. It's distributed, not decentralized. The fundamental innovation came with Bitcoin and later Ethereum. DApps. DApps are much harder for parasites to capture because they can be made trustless. No middleman. So what do you think about those two big critiques? One is that like, who cares? Like Gen Z is not going to bother to do all this when they can just, you know, download TikTok in a second and then just spend hours on it. And then critique number two is just that, um,

Urbit is distributed, not decentralized, doesn't have global state.

Nick Simmons (10:59)

Yeah, so I actually, I agree that it's very important to have global state. As the first critique, so this is something that the Octu thesis actually gets into. And if anyone's curious, you can go to octu .ventures, you know, mild plug and read our white paper. And we definitely think that Urbit is a technology on the level of Linux or blockchains in general, or TCP IP, where

We're not particularly interested in trying to access, you know, overall, you know, organic adoption into the millions right now merely because the Urbit meme makes sense to people on a mass scale. We're very focused. We can talk a bit more about the, you know, the Octu investment thesis later, but our broad perspective here is that Urbit allows you to build new kinds of businesses, new kinds of products, new kinds of protocols.

that can capture organic interest and solve problems for people, even just starting on a B2B scale. And so I'm really not concerned about the fact that, yeah, people use TikTok. Yeah, people use Twitter and those are, you know, optimized products right now. They give people what they want. What we're interested in is where can't you build an optimized product for a use case, whether it's B2B or, you know, a social,

graph or whatever, although we think that the social graph part will come later, where the actual architecture of the internet as it stands just doesn't allow it. So we're very focused on how do you build tools, how do you open up new landscapes, if you will, rather than how do we scale the Urbit meme as previously defined to millions of people. And the second question around global state, so

First of all, I think that Urbit integrates with blockchains very well. The Urbit Foundation has several partnerships that are very interesting there with Near, and I hear some other stuff as well on the horizon. And obviously, there have been Bitcoin and Ethereum OGs that have been very involved at Urbit for a long time. My Octu co -founder, Kenny, was on the founding team at MakerDAO. Once you get inside the Urbit ecosystem, you meet some very, very kind of like early pressing crypto people.

who all love Urbit and work on building stuff inside the ecosystem. So I think that Urbit absolutely does need to be able to call out to sources of global state, global consensus. And I'm personally a huge, I'm personally very bullish on Urbit ID and reputation as anyone who's met me in the Urbit context can attest, no pun intended. Because I think that we're actually,

very competitive and very far ahead of most of the existing alternative, you know, digital or self -sovereign ID projects or reputation projects. And the fact that Urbit is a deterministic VM and has this network to talk between nodes actually accentuates this quite a bit. And it's also worth mentioning, by the way, that most of the projects that we have either funded at Octu so far or that we see on the horizon are starting to...

all resemble each other in the sense of probably producing a bunch of interaction data between their users and then being able to use that as a pretty high signal way to mutually signal reputation via those interactions in a way that can probably be globally useful. And so I think there are ways to make internal global state and then sure, write it to a blockchain, write it to a global consensus layer to verify it. But...

One interesting thing about blockchains and Urbit is that blockchains for the most part, at least with any application that we've seen so far, have a pretty limited repertoire of things that you do on them and therefore opportunities to create reputation, identity, really granular proof of human, if you will, in the new context.

Certainly there are dApps, but dApp usage is pretty bad for the most part and anything outside of purely financialized transactions. Blockchain social in general is pretty much a flop. Blockchain -based interactions are the only things that have to write the chain. I usually hit this dilemma of either adoption creates fees or you simply never get adoption in the first place because there's too high of a barrier.

Certainly L2s and things like Solana are interesting, although, you know, now Solana has its own uptime issues and obviously there's a very valid critique based on, you know, over centralization there. So I think that like, I would agree that Urbit needs to be a, Urbit needs to be a computing and identity and coordination layer that talks to global consensus layers, AKA blockchains, L1s, what have you. But I,

I see that more as like something where there's a lot of mutual benefit and everyone I talk to pretty much agrees in every ecosystem rather than some failing of Urbit as a complete stack that it won't get.

Theo Jaffee (16:29)

So going back to what you said earlier about Urbit ID and why you're bullish on it and it's ahead of a lot of other kind of digital cryptographic identity projects. What do you think about Worldcoin Sam Altman's Worldcoin?

Nick Simmons (16:42)

Yeah, so it's interesting that you bring up Worldcoin because I did a deep dive on the digital ID space a couple of years ago and I actually I went to the IIW, which is the Internet Identity Workshop, which is the main conference every year for this field. And this was I think 2022 and Worldcoin was there. They did a demo, saw the orb. Worldcoin is interesting because,

they are simultaneously, I think a little bit ahead of the curve on one of the main failure modes, but then encapsulate one of the other giant failure modes that I see here. So to explain this, like,

Worldcoin actually, I think, is a little bit smarter than people give them credit for in the sense that they do have some pretty sophisticated thinking around how do you hash the original biometric data and use that in a way to prove personhood, basically. And I think that some of the critiques there are kind of superficial. However, the real problem that I see with Worldcoin, for the application that I'm talking about, to be clear, I mean, you know,

Sam Altman stated rationale for Worldcoin is to provide this basis for UBI. I can't even really comment on that. I don't have a firm opinion on that because it's so far away from what I'm trying to do with digital ID. But when you look at the history of really most tech paradigms and how they arise and how they get network effect, it's not by...

Theo Jaffee (17:56)

Thank you.

Nick Simmons (18:20)

say, you know, going to the third world and getting a whole bunch of people to scan their eyeballs and then bootstrapping from a whole bunch of people that don't have capital and don't have access to kind of like, you know, elite technological social circles. It's the opposite. Urbit so far, even though a lot of the tooling hasn't been built to make ID and reputation work in a really granular way, already has the opposite.

I mean, if you have any contact with the Urbit community, you understand that this is a very, very special, high value, high trust community of very prescient and capable people with all sorts of interesting connections. And so the way that I think about digital ID usually comes into two failure modes. The first is when you have a good design, but you don't pair it with any kind of use case that actually gets you organic adoption among people whose actions,

constitute high value, high signal in network reputation data. So again, Worldcoin I think is doing a good job of designing some aspects of this. They really have found an interesting way at least to prove personhood, but they're not deploying it among people in San Francisco who are gonna use it as a real use case and we bound that use case that has little takeoff network effects in the same way that say, I don't know, like the Homebrew Computer Club.

started in a garage in Cupertino and a whole bunch of those guys do very interesting stuff. And if you were there, you were part of the scene. It's the truth of music scenes. I actually used to work in music industry. It's really true for anything and blockchains or digital ID is just a way to codify it. The other kind of inverse failure mode is that you have a really good, you know, early network effects. You have a scene.

but you don't legibleize it in a way that actually allows people to, you know, to kind of prove it or attest it. So this could just mean that it's legible and lots of things are legible and that's fine. Or it can mean that you're just not, that you think you're representing reality. You think you're representing these interactions and you're not representing kind of what makes them valuable or you're not proving this is actually someone's, you know, central way that they actually signify what they're up to.

And so for example, I think there have been some digital ID protocols that I think got decent uptake among crypto savvy people, but it didn't give them the opportunity or didn't give them the incentives to actually use this for the majority of their online or even crypto based activity. And so the defection risk there and the fact that it's just not capturing their activity in a way that

really telegraphs commitment to that ID paradigm, that reputation paradigm, means that it's ultimately destined to fail because the defect rate and the flake rate is too high.

Theo Jaffee (21:22)

Alright, so let's talk a little bit about the future of Urbit and different directions it can go. So, do you think that Urbit needs a kind of growth model to survive or could it survive indefinitely as kind of like a niche product?

Nick Simmons (21:37)

Well, that's an interesting question. So, I mean, I pretty firmly believe that first of all, Urbit's growing. If you look at nodes on network, if you look at developer growth, and especially, I think, if you look at capital coming into the ecosystem this year, the growth of new startups and the technical maturity of core parts of the stack, I especially want to call out the aims upgrades. So the networking upgrades that allow for a lot lower latency.

and a lot more simultaneous connections between Urbit nodes. And then the other big thing this year is New Mars, which is an updated runtime. And so much actual, you know, faster computation and larger loom size. So much, much larger memory. The joke inside, you know, Urbit circles is that you want to be able to host the AVI file of Shrek 2 inside your Urbit and play Shrek 2. And now there's also a great,

Theo Jaffee (22:27)

Thank you.

Nick Simmons (22:30)

Twitter thread of the day from Ted Blackman, ~rovnys-ricfer, who's the CTO of the Urbit Foundation around some of the, the loom expansion and runtime improvements that could actually allow some early, you know, machine learning activity inside the Urbit runtime, which is wild. Definitely not something that I think people really saw coming a couple of years ago when it was extremely limited. And so built, you know, building on those technical improvements, I definitely see that there's a lot of growth coming at the same time.

I think that, you know, obviously you would be unfortunate if, you know, if for whatever reason, you know, I don't know, we entered into the mother of all, you know, winters for capital and attention and, you know, usage and so forth. I think that's true of any protocol. However, Urbit exists, like the, you know, the PKI exists, the address space is remarkably well distributed.

You go to the Urbit conferences, you kind of sniff around and I mean, you can also just look on chain, like, you know, how many unique wallet owners for stars and planets and galaxies. And that I think does a particularly meaningful at the galaxy level is that a remarkable array of people are committed to this project and have, you know, and have significant, you know, interest in it, financial and, you know, just attentional and technical. And so,

The protocol is already out there. It has application to reality. It has use cases. It has people building on it. I love the Bitcoin meme of, honey badger don't care. Oh, there's a downturn. Oh, Jim Cramer comes on CNBC and talk shit about you. Who cares? Keep building. So yeah, there's always gonna be upticks and downticks. I really think this is something where I'm fascinated by the idea,

Theo Jaffee (24:04)


Nick Simmons (24:25)

of building just indestructible primitives, indestructible networks. And we can get into this later, but like, I am personally fascinated by the idea of making these networks as unkillable as possible and really kind of crazy extrapolations that that leads to. Can we put an Urbit, you know, node on a, on a CubeSat? Can we inscribe the actual

you know, binary notation of an Urbit, you know, virtual machine on rocks in the desert. Like these are fun thought experiments, but I think they actually lead somewhere productive in the sense of there really are unlike say a lot of legacy systems where the ownership in the social graph is, is illegible and just kind of like dictated by fiat or it's on paper somewhere or, you know, there's some actually,

The degradation of digital information is way, way, way more advanced than people understand. I mean, a huge amount of content that has been produced on the internet, in fact, does not live anywhere. It's actually been lost. It's actually been wiped. So yeah, I think that like there's, with the advent of things like Urbit and blockchains that have distributed ownership, distributed consensus, and also are, you know, writing.

information to a distributed state layer, it really is a different paradigm in terms of what it means for something to be growing, active, or defunct. Like, I think we're already, you know, it's, have you heard of this, have you heard of this kind of, you know, crackpot hypothesis about whether, you know, we could tell if there was an industrial civilization in the fossil record millions of years ago?

Theo Jaffee (26:19)

Um, that sounds vaguely familiar.

Nick Simmons (26:22)

So I mean, it's an interesting thought experiment. I don't think that there's evidence that we had a fossil fuel civilization millions of years ago, but then the question among geologists and so forth is, okay, would it have left enough evidence? And I think there's a loose corollary you could make to something like, or a bit which is how big of a civilizational catastrophe would it take to wipe out the evidence or even kind of like enough of a subset of Urbit IDs or let's say, the Bitcoin blockchain.

anything like that, to the point where if there was interest and if there was a compelling need for it, that you could bootstrap it back up again.

From a game theoretic level, all you would really need is a handful of galaxy IDs that say, okay, this is still the network and we're gonna reconstitute it and we're gonna vote to spawn perhaps more galaxies, more stars if a whole bunch of keys have been lost. And this sounds just kind of like science fictional speculation, but again, I really think that this is an important factor in what's the difference...

in what we're building now for the long term, versus things that don't have these provable interactions on the identity layer and on the kind of like social contract layer.

Theo Jaffee (27:46)

So going back to what you said earlier about Urbit growing, Urbit stars were selling in 2021, they were selling for like $28 ,000 on average on Uniswap and OpenSea. And now they're more like a thousand dollars and they haven't really recovered even as the broader market for cryptographic assets has recovered like significantly. You know, Bitcoin has gone from like 20 ,000 back to like 70. It's hit a new all time high since 2021. So why hasn't Urbit address space recovered too?

Nick Simmons (28:16)

You know, I don't try to speculate too much on price. I think that where, you know, so many things in the crypto ecosystem in general are very meme driven. But I will say this, that I think that there's a general tendency, and Urbit is by far, you know, not the only project where you see this, where tokens or NFTs or something, you know, bespoke like Urbit address spaces and sort of digital land becomes a meme coin for the general idea.

in the absence of other ways to gain exposure. And so purely from a personal perspective here, my suspicion is that the level of interest in Urbit, as far as I can see, and I've been well -placed to observe this, keeps going up. And so it's a very valid question to ask, okay, why hasn't the, you know, why hasn't the star price, you know, kept pace? And I think the answer honestly is that now there are more ways

to essentially invest in the ecosystem rather than simply buying a star in OpenSea. And so for example, Octu does not invest in Urbit address space. We all have a lot of Urbit address space. We're all generally bullish on its value going up. I do think that there's gonna need to be some technical work and some kind of like product and ideation work around how do you make address space specifically? And because the interesting thing is, okay, it's a non -fungible token, right? But,

really, you know, if you just buy a random star on OpenSea that's never been booted, that's never interacted with anybody, sure, maybe you like the sigil. And I mean, the sigils are great, the names are great. But other than that, it is still pretty fungible. And this is even true to the point where for a while there was a project called Wrapped star to make it even more fungible. And so I think that there does need to be some work done on building things that imbue given

chunks of the address space with value, maybe as subnets, maybe as something that, you know, accrues reputation, maybe as web of trust that mutually reinforces reputation on the group level or the subnet level. But as to the price again, I really just think that what happened is that all of a sudden, I mean, when I joined the project, there were two Urbit companies, both of them mostly sold address space. Then there was a little bit of hosting company. Yep.

Theo Jaffee (30:39)


Nick Simmons (30:42)

And now there's a lot of startups. There's a lot of technical projects where you can put your time and energy. There's a lot of grants from the Urbit Foundation to work on stuff. And then now there's actually two projects in the works. There's a Send Chain from ~tiller-tolbus at Chorus One. And then there's another Urbit L1 project from Sunny Agrawal at Cosmos. It's mostly being worked on by Laconic.

that and both of these aim to not only give some some global consensus layer to Urbit as an Urbit native, you know, L1 blockchain, but also has some interesting ideas around how the reputation layer and the economic layer could work in ways that incorporate address space. So I would definitely encourage everyone to keep an eye on those things. I think both of them are kind of like in the, you know, publishing early communicase stage. But I don't ultimately see, you know,

undifferentiated address space, stars basically, as being really, I think, ever again, so much the concentrated, you know, bet or meme coin on the overall health of the ecosystem, at least not for a very long time. And where I'm choosing to put my efforts in Octu is investing in early stage startups that build things on Urbit. And that's where my kind of very opinionated thesis is.

Theo Jaffee (32:09)

Do you think there is any kind of proper intrinsic value of an Urbit star? Like, can you go about valuing Urbit address space or is it more like Bitcoin where there's no real way to value it because there's no cash flow. It's just a purely cryptographic asset.

Nick Simmons (32:26)

Well, I think that both of these things have something in common, which is scarcity and that ultimately you start to get, you know, increasing value within a scarce, you know, namespace or, you know, a scarce currency, you know, store value space in the case of Bitcoin. When you have a whole bunch of people who reach, who reach a social consensus on.

what the value of that scarcity is to them in terms of preventing various kinds of downside risks. In Bitcoin's case, it's, you know, hey, we're worried about money printing. Like there's no actual limit to the amount of fiat money that a government or central bank can print. So we're all going to subscribe to the, you know, social illusion. And I don't say that in a negative way that, hey, it's valuable that there's only ever, you know, 21 million Bitcoins. And so this Bitcoin that I own has

has a value as a percentage of that. Similarly with Urbit, I think that if Urbit got to the point of adoption that Bitcoin had, of course you'd see address space go up. Not financial advice, but that really seems to stand to reason because you have a whole bunch of people agreeing that the scarcity aspect as it represents trust and distribution of the network and so forth also

benefits from that scarcity and they want a piece of that. Now, what happens before that, I think is very interesting. And that's why I say that like subnets and like biting off chunks of this scarcity and maybe either imbuing them with some sort of, you know, economic function, or simply saying, okay, you know, this, this corner of the network, this, you know, thousands or millions of planets are attested to in a given way.

and we're going to carve out this chunk and use it for a given use case. And within that, there's even a greater degree of scarcity. And it's very interesting to think about the network growing with multiple examples of that that then have to have some sort of foreign policy with each other or some sort of equivalency. And it's almost like, I mean, the original metaphor of Urbit as land is very apropos here. When you're on a frontier or you're even, I guess, settling in a different planet,

land is cheap and so people have the ability to go homestead in different parts of it and build their own thing and define it their own way. And then over time, the borders creep to each other and they have to figure out how they're gonna relate to each other. And then the value of land maybe in between goes up. And I think that that's the process that we are gonna have to watch and see with address space.

But it also informs again that I don't see the value in just owning huge swaths of undifferentiated address space and waiting for it to accrue value in the near term. It's much more around, you know, what are the cities you're going to build? How are you going to bring people there? And what are the industries you're going to build there?

Theo Jaffee (35:34)

Can you go into a little more detail about the intention behind the original land metaphor? Because I think that was probably one of Curtis Yarvin's best pieces on Urbit.

Nick Simmons (35:43)

Yeah, I mean, I'm a big fan. Look, I can't speak for Curtis. He's certainly a prolific podcast guest. I encourage you to get in contact. And I know he doesn't often speak about Urbit design stuff. Very good, very good. So I won't presume to speak for Curtis. But as somebody who has watched the growth of the network, I do think that it's

Theo Jaffee (35:57)

It's in the pipeline.

Nick Simmons (36:12)

It was a very good design decision on his part to make it so large, even though it leads to this current price drop that you see while people kind of acclimatize to the homesteading phase.

This really is, Urbit really is a new world. And this is almost a little cliche to say at this point, but I really believe it. And giving people huge swaths of node IDs, whether for humans or for infrastructure points or whatever, is really, really important when you have the ambition to replace the entire, you know, up to the minute,

paradigm of network computing. And by the way, a little bit of a side note here, something that people talk about a lot is that below the planet star galaxy paradigm we've talked about before are something called moons. So moons are derivative identities. They do have their own keys, but they are permanently tied to a planet owner. And the...

Design space for moons has always been a little bit underspecified, but I'm very interested in ways that moons can not only provide IoT device identifiers, which is actually an underexplored direction in IoT in general, in terms of verifying where your data came from, certainly in a deterministic way. ~datnut-pollen, I read a great piece on this back in the year about the Tlon called the lunar IoT in the internet, or something like lunar IoT in the internet, or things like that.

But I'm also very interested in something that I call node dispersal, which is that blockchains and protocols like Urbit, so far, if you look at where nodes are being hosted, it's gonna be a preponderance of, depending on the technical requirements, you're gonna see an awful lot of validator nodes for various blockchains in AWS buckets. And then even if it's actually self -hosted, where's it gonna cluster? It's gonna be San Francisco, Seattle, Berlin, Lisbon, New York.

And that seems like a, I think that we need a scale pass. Like that's an over concentration of, you know, of these supposedly decentralized networks. You know, an interesting way to view the whole proof of work versus proof of stake, you know, debate or argument is that, and this is one lens to look at it from. This is definitely not the only one.

but I find it instructive sometimes, is that proof of work fundamentally is a bet that the distribution of cheap or free energy and the ability to exploit it as it's distributed over the Earth's surface is, you know, and obviously modulo people knowing about something like Bitcoin or something like, I guess, you know, or Dogecoin or, you know,

pre -merge Ethereum, and knowing that you can do this, that this is a thing you can do. But with Bitcoin in particular, the Bitcoin mining has penetrated deep into say, provincial China. It's penetrated deep into the provincial US, deep upstate New York, for example, all over the world. And then it has technical innovations like, oh, flared natural gas on an off,

like an off -grid drill site, you go and you mine Bitcoin with it and it's not economical to export the energy in terms of a gas pipeline. It is economical to export the hashed mathematical results of Bitcoin mining over a packet radio. But then proof of stake is basically a bet much more on, okay, this blockchain is going to scale via

people who have the cultural, you know, imprimatur and the social connection to the people who started it who tend to be, you know, educated Western technical types. I don't think either of these has a monopoly on insight, but I think it's an interesting, you know, dichotomy between how resources are distributed and to the extent that, you know, blockchains are supposed to be this very durable layer of, you know,

distributing information, access, and economic rails across the physical space, how they're distributed. And so in that context, I think that moons are super interesting in terms of how can we achieve physical decentralization via node dispersal across every possible boundary with the Urbit network. Can we put moons as Urbit nodes on devices?

Can we optimize for geographical decentralization? Can we put them in our space? Can we put them across political boundaries? And two follow -up questions here. How can we measure this in the most robust way to actually provide a score, if you will, of how decentralized the network is? And two, what do you do with them? And what's the practical upshot? So.

One super interesting thing about Urbit, right? It's a network of personal servers and you run a client locally and you can dial into APIs, you know, blockchain sources of data, et cetera, et cetera. I assume you've heard about this Wells notice that Uniswap just got.

Theo Jaffee (42:05)

The what that Uniswap just got?

Nick Simmons (42:07)

So Uniswap just got something called a Wells notice from the SEC and we don't know the contents yet, or at least we didn't have as of last night. And a Wells notice basically is just serves as intent that the SEC is gonna bring legal action against you. So basically the story here is that, and this actually goes back to one of Roko's critiques is that, you know, dApps are solving for supposedly things that that Urbit was trying to solve a while ago. Well,

One thing that dApps have not solved and that they are getting in trouble for right now is that the actual front end server layer is the vulnerability. And so this is most obviously true for DeFi right now, but it's also probably going to be true for certain kinds of, you know, AI to consumer apps in terms of your dialing in to use a model, where does that model live, but also where are you and how you access it. So, Urbit excels at this. And actually there's been quite a bit, you know, quite a...

bit of interest from DeFi protocols in, hey, how do you use Urbit to serve up the front end so that someone can use a DAP, can use an AMM, an automated market maker? How do you actually interact with the blockchain rather, in an actually decentralized way for the user? And so one idea around making Urbit node dispersal extremely robust is that,

this allows you to have, you know, umpteenth degree redundancies for mirroring sources of data, mirroring front ends and actually like, you know, keeping, being a backup for any kind of global state, you know, coordination protocol, blockchains or otherwise, there are other options there too, as well as representing reality. So things like,

Po -Apps, proof of tenants protocol, proof of location protocol, things like helium, packet radio, et cetera. Very interested in how we can make Urbit and networks like it actually span the earth and start to instantiate. I mean, the original metaphor is Borges. It's "Tlön, Uqbar, Orbis Tertius". And the story, of course, is about.

a map that is the same size as the territory that it describes. And, you know, Curtis know what he was talking about when he was using that metaphor. And I'm very interested in ways to make that real.

Theo Jaffee (44:38)


So change topics a bit. Let's talk about AI and the future of AI on Urbit. Because I think that like, you know, the perfect application for, you know, having a decentralized personal server would be to run a decentralized personal AI on.

especially because it seems like actually good personal AI is imminent. You know, who knows what Apple is about to announce with Siri in a couple of months. And then OpenAI is coming out with GPT -5 soon after that. Most likely sometime by the end of the year we'll have very good LLMs and people are already building out the infrastructure. So what do you think is the future of AI on Urbit, both implementation and then what people will actually be able to do with it?

Nick Simmons (45:25)

Yeah, so this is super interesting. There's a whole bunch of different threads we could pull. So obviously I think that access to protocolized AI compute and model access and also honestly markets for data sets that you would feed to a model on large and small scales. I think Urbit helps with the protocolization of that quite a bit. And I really don't know anything else that applies nearly as much there.

But there's also very interesting things to say about what you can do with all the interaction data that you create. So each, each Urbit virtual machine that you know, you or I run is an event log. It's just literally a list of computational events that you've performed with your machine and it lives locally and it doesn't have, you know, global state by default, but certainly you can, you can.

create a protocol to write this out to a blockchain whenever you want. And there's actually been some interesting early experiments and kind of like peer to peer local consensus like agreements on that Urbit computational level between orbits. This is something called Agora that Quartus and the Dalten collective experimented with a couple of years back. And so because you have this deterministic event log,

of everything you've done. And that includes, of course, the connections you've made with other Urbits and data you've gotten from other Urbits over the network. It's very interesting to think about training a personal model on your activity as a kind of proof of human and then zero knowledge hashing that and using that as your badge of identity in a very kind of up to the minute way. And again, as we were talking about,

you know, some of the pitfalls of digital ID protocols, this is, I think that proof of human is gonna become an arms race. And that if you don't have the, if you don't have the primacy of someone's everyday, you know, online behavior to draw reputation data from in order to kind of like model them as a human, then you're gonna fall behind pretty quickly. At the same time, you're gonna, you are definitely going to need to, you know,

to not simply leak everyone's data or even just kind of behavior patterns, if naively obscured in ways that leave them vulnerable. And so it's very interesting to think about personal models deployed to prove who you are, to prove belonging, but also to act on your behalf and learn from your behavior. And it does seem like Urbit is the kind of thing that would provide a moat there.

for you to run a personal model in, interact with data set marketplaces, and then command a fleet of virtual assistants on your behalf, which I mean, the moon identity seems to align well with that. But ultimately, I think that the really, my top level thoughts about Urbit AI is that AI is probably going to cause a massive,

sorry, is probably going to cause a massive collapse in the social context and the legacy, you know, reputation, identity, meaning systems that we currently live under. I mean, they were already straining pretty badly. And this is a key part of the Octu thesis is that we think that Urbit has massive potential to build what we call digital civil infrastructure.

which is to say, to encapsulate the values that its users and builders actually watch, digital Jeffersonianism is one way to put this, of allowing people to actually have a defensible moat around their family, their community, their business, their data, and to say, hey, I'm gonna interact in marketplace, I'm gonna interact in society.

but there's a line past which cannot be trespassed. And honestly, I think that AI is gonna make this necessary sooner rather than later. And it's extremely, extremely important that we encode the values that we want into a defensible digital substrate before that happens, because there's really no guarantee that they will simply be uptaken.

whether it's by, you know, mega corps that use AI or not, or whether it's simply by these models in the sense that they're, you know, eating all the data from the internet in the past, you know, 30 or 40 years of written code.

Theo Jaffee (50:31)

And then not just AI, but the rest of the future of Urbit. For example, what could like an Urbit hardware ecosystem look like? Whether it be CPUs, I know you talked about this, Nock FPGA earlier, or like actual full blown.

like computer hardware, like consumer desktop or laptop hardware. And what about Urbit on mobile? Like what could that look like? Would it be more of an app on existing OS's or would it be its own mobile OS or its own mobile device? So yeah, what's, what's the future of the kind of actual like hardware layer of Urbit? How will people interact?

Nick Simmons (51:10)

So yeah, so first of all, this is already very much a thing. So the obvious shout out here is to Native planet, which is a great company in Austin, Texas that builds Urbit hosting hardware. So with a bunch of very, very clever, very cool software integrations to make it easy to run your own planet or star on a custom built hardware box. They have several models, highly recommending.

recommend checking them out, nativeplanet .io. And so yeah, so there's already hardware that's purpose -built to run your, your Urbit nodes locally, to run your Urbit VM locally. I think there's massive potential here. Some of the areas I've already mentioned in terms of node dispersal, I think that Urbit, you know, optimized sensors, you know, space hardware, things like that. And I think there's a, there could be very interesting kind of, you know,

flywheels and mutually complementary economies here where the economic mandate of no dispersal and like, you know, provable, like real geographical decentralization can reduce the costs for the ancillary, you know, business use cases of, hey, we want, you know, sensors here to track X, Y, and Z, or to provide, you know, relays for other networks or, you know, like packet radio, mesh nets, that kind of thing.

I'm very interested in the Urbit sensors, very interested in Urbit hardware that creates a real geographical decentralization. In terms of Urbit apps, I mean, there's some, Tlon has an iPhone app, which works quite well and highly recommend everyone check that out. That's really made it much easier to use Urbit on the go.

If you're asking like, is there going to be an Urbit phone like a salon, like the salon, you know, side of phone, that's certainly an interesting question. Um, I see a lot more hardware interest from urbaners than I did a year ago. Uh, and so I think that people are kind of, you know, trying to wrap their heads around this stuff. One aspect of that, I think maybe accelerating this a little bit is that Apple recently breached what was previously considered to be quite a bright line in taking in.

take a hard line against PWAs, progressive web apps, which were basically a bit of an end run around some of the, you know, frankly pretty absurd strictures on what can and can't get accepted in the app store. The whole question of whether, you know, an entrant to the smartphone business that can get around some of the Apple and Android, you know,

quasi monopolies for apps, especially around crypto integrations is, is certainly super interesting. And I kind of suspect that it's, you know, pressure building behind a dam and that we will get there maybe even sooner than we think at the same time, geopolitics and the geopolitics of chip production and, you know, sophisticated electronic production in general are a big, big question mark, especially with, you know, all the stuff around Taiwan right now. So that could throw a wrench and all that stuff.

I really want to see good thinking and I want to see honestly pitches here around how you unite the Urbit vision with hardware, real use cases, and that geographical and technological kind of like hosting environment, decentralization and dispersal factor.

Theo Jaffee (54:54)

So we've only got a few minutes left, so I'm going to go into a little bit some of the design decisions behind Urbit. So, for example, where did the syllable -based naming conventions come from? I noticed recently a lot of passwords are now very similar to Urbit syllables, where you have a bunch of normal -sounding syllables or English words separated by hyphens. So which came first?

Nick Simmons (55:22)

You mean passwords outside of Urbit

Theo Jaffee (55:25)


Nick Simmons (55:26)

I think that has to be convergent evolution. I think I've noticed a little bit of what you're talking about recently. I mean, the whole topic of password optimization to get around what they used to call a dictionary attack has been going on for a very long time. And of course, now we're going to just use the suggested password managers, browsers and so forth to run. So I couldn't really tell you if there's any, I mean, I'm really highly doubt there's a connection there.

Theo Jaffee (55:34)

us brings us.

Nick Simmons (55:56)

I don't know the full story of exactly why the Urbit, you know, I mean, the syllables are meant to be, you know, human memorable because it's easier than memorizing, you know, a bunch of dotted quads, right? And obviously there is a correspondence between the Urbit syllable names and the, and the sigils. There's a great post by, I believe by Gavin who created the sigils called appropriately creating sigils. If you go,

way back in the urbit .org blog, that's worthy of reading. And yeah, I mean, I'm a big fan. I don't know the full history, but I'm a big fan of how that works because they really are memorable. And I remember at one point, maybe a year or two after I originally joined Tlon, I realized that I probably knew upwards of a hundred people, so easily like over a Dunbar number of people by their Urbit names, their Urbit planet names, rather than,

Theo Jaffee (56:26)

I love that post.

Nick Simmons (56:55)

in most cases, the real names.

Theo Jaffee (56:59)

That's pretty cool. And then what about the influence of Christopher Alexander? I know there's another I can talk about that. I'm actually reading the pattern, a pattern language right now. And I didn't even know, I've known about Urbit for like years. I had no idea that Urbit was inspired by Christopher Alexander.

Nick Simmons (57:04)


Yeah, I'm not a Christopher Alexander expert. So I would recommend talking to Galen or possibly to some of the OG design team at Tlon Ed Fablefaster has some great thoughts on Christopher Alexander. And really, the whole design genesis of Urbit and Tlon's products is fascinating. I will say that one of the things that really attracted me,

to Urbit in the first place was a sense of really like that it was important to know and understand the whole teleological history of technology and product design. And it's actually funny, I was just on another podcast from a hut ventures that will be coming out next week. And one of the at the very end, we talked a little bit about you, hey, you know, what would you be doing if you weren't working in crypto or tech?

And my answer, which has always been true, is that I'd probably be a historian. And I've always been super, super interested in tech history. I have a little bit of a family connection. My grandfather worked in the ENIAC in the 40s. And so I grew up using little punch cards, scratch paper. And I kind of thought that was normal as a little kid. Didn't everyone's grandpa work on computers, right?

Theo Jaffee (58:24)

Oh, I saw it.

Nick Simmons (58:37)

Yeah, it's in the, I'm assuming you went to the Computer History Museum. Yeah, I know, one of my favorite places. I could spend hours in there. I've been there many, many times. I know, right? I know. Yup, yup. I mean, that's when it was started, I think, the museum. And so they're, yeah, they have a backwards looking perspective there maybe. But yeah, so I've always been super interested in,

Theo Jaffee (58:42)

Yeah, yeah, I did. I'll do it.

I just wish it would have been longer. It ended in like 1995, 2000. I missed the last 25 years.

Nick Simmons (59:06)

how this stuff developed. And one of the things that really made clear to me about Urbit is that like, there have been a whole bunch of junctures in the history of computing, the history of network computing and the history of, you know, mass onboarding people into these network computing platforms where particular decisions were made. They were usually contingent on what came before and what the current, you know, ground conditions were at the time. And some of them worked out really, really well for

I'd say the average person and the ability of developers and entrepreneurs to build what they wanted. Like we're definitely living in something that is far from the worst possible world here. Are we living in the best world? No, of course not. Hence why I work in this stuff, hence why we all work in this stuff. But the thing that really impressed upon me was that there were times when I, for example, early on, I think Bill Gates and some others in their early 90s wanted to essentially encircle.

the internet and circle some of these protocols. And instead we got TCP/IP, which is the open protocol. We got SMTP for email, which is the open protocol. Although honestly, the dominance of Gmail as the front end to that is eroding that a bit. There are some pretty scary censorship going on with G -Docs at least.

Theo Jaffee (1:00:19)

Very. Shout out to my last guest.

Nick Simmons (1:00:22)

Oh, really? And so I do think that it's, it's very important to consider, okay, if you're at one of these junctures, and you're in a position to build something or fund something that can, you know, make the right decision, and return agency and return, you know, network effects and coordination ability to the users to the developers to the auditor, nurses, harder than is in the right place, you should do that.

And that's the highest calling you could possibly have as a technologist. And I really think that the degree to which the Urbit ecosystem and everyone I've met in it considers these questions and knows the history and knows, hey, it's important to build this stuff right, is one of the things that grabbed me in the first place and still gets me out of bed in the morning.

Theo Jaffee (1:01:12)

Well, I think that's an excellent place to wrap it up. My answer to where, what would I want to do as a career if I wasn't interested in tech is similar. I would probably want to be an architect. Also Christopher Alexander inspired. So yeah, I'm thrilled to see that he's had an influence on Urbit too. Well, thank you so much, Nick, for coming on the podcast. I really enjoyed this episode.

Nick Simmons (1:01:34)

Absolutely, I enjoyed it too. You asked great questions. And if anyone wants to follow up and continue the conversation, you can drop my Twitter handle in the show notes. And as mentioned, my main project right now is being a founding member and partner at Octu Ventures, which is a member -driven venture DAO that invests in seed stage Urbit projects. And we have some writing on our website, octu .ventures.

And I really try to make that the vessel for most of my energy into Earth these days. And I'd love for people to check that out.

Theo Jaffee (1:02:14)

Alright, well links to everything will be in the description, so thanks again and talk to you later.

Nick Simmons (1:02:18)

Awesome. Thanks to you.

Theo's Substack
Theo Jaffee Podcast
Deep conversations with brilliant people.
Listen on
Substack App
RSS Feed
Appears in episode
Theo Jaffee