ZKPodcast: Urbit with Galen Wolf-Pauly from Tlon

Transcript

[Music] welcome to zero knowledge a podcast where we talk about the latest in zero knowledge research and the decentralized web the show is hosted by me Anna and me Fredrik Hey so before we start in I want to speak to you briefly so this is a bit unusual but to be honest we live in unusual times so this show is released weekly and we sometimes record some of our episodes in

advance to have a small backlog of shows we then air those over the following weeks this by the way I'm just giving you a little bit of a look under the hood on how we do this over here the thing is with things changing as quickly as they are kind of on the world stage at the moment our episodes that we're airing may not always reflect the latest goings on or maybe people's sentiments and generally this has been intentional in a way we've wanted these episodes to act as a resource and not necessarily be tied to any specific time and so many of them are focused on a topic the thing is the past few weeks have been really wild and so I just wanted to kind of check in

with you and say hello and thanks for continuing to listen we plan on continuing to make these over the next few weeks months maybe years but if it sounds like an episode is coming from a different era like it's because we recorded them in advance when things were maybe a bit more normal and hopefully they'll be normal again soon anyway just a heads up given the speed of change at the moment I imagine there could be some twists in the coming months for example the DK summit the event put on by this podcast of which I've done four can no longer go on as planned for the fifth edition so it was planned for

March 31st but like many events happening this month and next and maybe after that we weren't able to host it as planned so the new plan is to host it online which will be a bit of an experiment for us and so we really need you to bear with us as we put it together but this change could also mean a really neat opportunity for more of you to be able to participate so I'll be sharing more about this over the next few days and weeks if you already have a ticket you can get refunded as there will be a like a new cheaper ticket for the online version available if you want to keep your existing ticket that would actually help a lot in recouping some of the funds that we've lost on the in-person event for this I will definitely create some

special bonus content like thinking maybe we make an extra meetup or something like that so yeah I just keep that in mind if you have never heard of the zk summit and you're interested well please send in an application I will share the link in the show notes once you've applied I can then send you a link to the online versions ticket yeah I think it's gonna be really interesting I can't wait to see sort of the regulars who come to the summit every time and I'm also excited to see if we can you know have some new participants join us this time yeah so hope to see there back to this episode this week we chat with Galen from the project Herbet we recorded this a few weeks ago it's a

somewhat controversial project as well as being a really interesting one we spent most of this episode figuring out what Herbet is and then at the end we chat a little bit about what makes it a bit of a lightning rod it's not necessarily a blockchain project and it's not necessarily zero knowledge but we were curious and we hope you are too okay so to wrap up my kind of unusual long monologue here I hope you're feeling really safe and good wherever you are I hope you continue to make time to stay curious about math tech and the humans that make up the decentralized / internet world I think it helps to focus us to continue to think about a

potential future that we want and hey if all else fails just watch Star Trek that's what I do when I'm feeling worried about the world anyway thanks a lot and now here's our interview with Galen from orbit so today we're sitting with Galen who's the CEO of Tuan plon is the company that works on the project Herbet so welcome to the show thanks for having me and we have Frederick over in Berlin hello hello in this episode we wanted to explore this project orbit and Herbet as far as I understand is not a blockchain project but it's very similar to a

blockchain project or lives in the same world as a lot of the blockchain stuff that we've talked about and I think both Fredrik and I are really curious to find out a little bit more about what this project is what it plans on being and if there are connections to maybe some of the other topics that we've already covered on the show yeah so maybe to fill in how we overlap with the blockchain world first and then we can get into like what Herbet is concretely and why it matters or why we think it matters so in the simplest terms orbit is an identity system and an operating system or what we call an operating system is really an overlay OS so it's a piece of software that runs on any UNIX machine with an internet connection to

log in or authenticate with that OS which also runs so network you need an IDE that ID is registered through a series of or like a suite of aetherium contracts those are live and deployed all of those IDs are ERC 7:21 so they're like non fungible tokens that are you know represent your identity and basically like hold a keychain that you used to authenticate yourself and like you know as encryption authentication keys on the network for setting packets back and forth there's kind of like a hierarchy of keys for how you secure that identity we can get into what that is so there's definitely overlap at the identity level and then the operating system at the very lowest level is a

determines is like a single function and that function is deterministic so your Herbet is like a pure function of its state which of course is something that anyone is familiar with blockchains is like that looks like a little bit like a blockchain burbot doesn't do consensus like every node runs independently but because they all share the same computing model it's really easy to share programs like send data back and forth then you get all these kind of like higher-order benefits which are also a little bit like people are familiar with from these sort of more like global computers where they all share the same state so urbis don't share the same state would be really easy to build a blockchain on top of Herbet we often talk about it like if you're gonna you

know today in order for me to interact with to spend money or to interact with general-purpose blockchains I need a computer and on that computer I have a mnemonic or in my head or whatever I have a hardware wallet so you need a key and you need a system to call out to a chain herb it's super compact super simple so we think of it like herb it's a good complement to a blockchain like it's fast it's super simple easy to reason about its security and so using it alongside of blockchain makes a lot of sense so we've always kind of like coexisted with this world but we're not directly in it so I think I mean that the largest way that it's similar to the locking world I think is in this sort of

vision statement in what you're trying to achieve you talk like if you go to Herbet org you talk about decentralized applications you talk about a lot of the same goals even like trying to remove the power from mega-corporations and put it into the hands of the people what all blockchains talk about this these same things so like the goals are there the vision is there but decentralized app has to mean something else in your context because block chains talk about decentralized apps or the pillars upon which you build decentralized apps our consensus and Byzantine fault tolerance and those are not things that you have as you just said so we're like how do

you see decentralized apps differently from what a blockchain would see them as yeah yeah so Herbet is very much aimed at personal computing in the most mundane of senses like the operating system side right I mean really like the to use together so you know one way to think about the OS is almost like think of it like it's a meta protocol looks a little bit like a database and inside that database you can basically define something between like a stored procedure and the program that also makes it extremely easy to have that program share data across the network compute with other nodes and so on so when it comes to things like chat

sharing documents sharing links sharing like other kinds of type data when I look at the like using basically sort of like a global blockchain based computer for doing things like that I mean it's just technically really really hard right like you have to get that state to update among a whole bunch of nodes all at once and all I want to do is like send a message to my friend so Herbet sort of like takes just like another approach just because our goal really is to like solve the like personal computing problem like one urban node one person so this thing needs to store your data do your computation I think you can think about building an urban app on the OS side is kind of like you define a protocol and then build an

interface for that protocol it comes with a number of protocols kind of out of the box and you could like build new interfaces for the existing ones so it's a little bit different than I think when I think of like really canonical you know blockchain based decentralized applications ones that like really really need a blockchain something like augur enosis things that require global consent they need to be secured by global consensus a lot of sense to use a blockchain but when you're looking at something that needs to have really high throughput needs to be fast needs to be responsive I mean just from pure technical standpoint I think like Herbet makes like a bunch of trade-offs such that you can you know accomplish those things and I think then it's also the thinking is sort of like well if you needed to send money as a part of a

message or you know register a vote or whatever it might be where you need trusted computing like where you need a blockchain then you should just call up to the chain like have Herbet handle the you know signing and sending of a transaction does that make sense thing is though like Herbet as a project and the concepts around it it started like kind of before thank you right how old is the project so the project started is like a personal kind of like independent PhD in 2002 is those sort of reported number I think like I always like I'm like you know no one really knows you could think of it like urban walked out of the desert in 2013 and no one knew like where it come from

see it Kurt Kurtis worked on orbit as an as independently for quite a long time he built the colonel for the first mobile web browser Joe I think is kind of like important context it's like someone who worked on basically getting a bunch of you know devices with a lot with under terrible latency conditions right - speaker out how to compute together and I think he looked at that prom initially was like this is a nightmare this is really really hard with the existing stack I think I could rethink this and then disappear for 10 years to think about that problem and so I found urban in 2013 once it was sort of like a basic prototype I see so it's more like the origin story puts it yeah a little bit I get to think of you know

Bitcoin there was already you know e gold or whatever right there are like early digital cache projects have some of the the cipher it doesn't predate like the cypherpunks or anything crazy like that but yeah orbits I think actually there's a blog post I should know this but the truth is I actually did or I don't know the exact dates but I'm pretty sure the blog posts describing the PKI model so we talk about urban address space as digital land because it's sort of like bundled into blocks of increasing size there's no mining it's you can think of it almost like it's pre mined as sort of a existent distribution scheme we talk about that model I'm pretty sure was documented and published either I think before Bitcoin the Bitcoin white paper

so I don't know I always think of them as kind of like coming out of the same like primordial cyberpunk soup or something like that but with just like slightly different goals yeah it's it's funny like my first interaction north Herbet was probably 2014-15 something like that and it was like me and a friend this is probably at like whew there something yeah exactly those me and a friend sort of found this project and we started digging and it's all this crazy all these compilers and like odd languages and like lots of stuff to geek out on and you download this thing you run it and it's basically IRC but you have some unique ID assigned to and

it was like okay like it I don't really understand what's going on with like behind its IRC that's cool but it's like I I know there's a lot of here but I I don't exactly know it is but it was it was still a cool thing and it sort of but even then like the vision was their tutor lore and like the website back then talked you know about sort of this OS and a global computer that replaces everything and like everyone will just live in their orbit and like you won't interact outside of it and like I don't know yeah stuff like that and then you went in and its chance yeah it was a weird contrast but it was it was so cool

yeah anyway that was just a random story my question kind of like is it's actually like an import all most like an important it's good context I think it for people to know even like what the state of the project was at that point cuz I do it's like if all the way that orbit has evolved is important but yeah yeah go for it well if your orbit org today there's this background image of video of an ocean is this reference to people saying that orbit is taking a boil the ocean approach that I should just say yes the thinking although I'm you're I'm gonna take the bait because I

like what we have been thinking about in this regard I think that Herbet evolved out of this basic intuition that like something else was possible and and like I think definitely Curtis and many of us really in the early there's like not totally sure exactly what that looked like but everyone who worked on orbit came from like had a very technical or had a lot of experience with the existing technology and a lot of like frustration with it basically like it was like this just doesn't sort of match up with the values of kind of like basically like the early computing movement right like computers were supposed to be these really wonderful tools why am i stuck talking to people on facebook like that I can't really extend or change or do anything with it just felt like extremely restrictive and if

you follow that rabbit hole all the way down without dragging everyone through that that's five guests like I think that you know that sort of terminal thinking is like okay actually this is a technical problem like let's not get into what exactly you want the user experience to be like but at the end of the day like if someone is running the server for me they're like deciding how I'm gonna compute like how I'm gonna interact with people how long my data is gonna live so on and so forth and so that technical problem is a big problem so you're basically thinking like everything that we put on top of Unix to make the modern Internet work is not gonna work like we're basically throw that away and start over which is a totally insane thing to say but maybe

also we would I would tend to think actually probably true and it's so it's a huge there's a huge amount of technical work there and you're like all right I got it I don't have stuff to fix you know we'll just we'll figure out exactly what people do with this later but we know that it has something to do with like it's going to you know ink it's gonna sort of absorb a lot of your daily computing sort of experience so to draw it back to what you're noticing on the website I think as we've gotten closer to being able to deliver something that people can use in a way that is still richer than chat although chad has definitely something people use her before probably more than anything else the thinking was sort of like what you really want is an extremely calm straightforward a simple way to bring

together a bunch of different modes of communicating and collaborating so for any for you guys to get this podcast done for us to get work done for parody to get work done for anyone who uses the modern internet I'm like switching between a bunch of different things and there's something just like anxiety inducing about that experience and what you want is for your computers to not get in your way you want them to actually kind of be able to do the multitude of things that you need in order to collaborate with people you want to talk to them you want to share documents you want to share links you want to share all kinds of different types of content you want to be able to edit them and you want that to last forever and you don't want to have to worry about anyone else

like any company getting in your way and so sort of the image of that just purely from an aesthetic standpoint was like it should just be like you alone in the landscape there's like imagine the sort of like the modern world disappears and you're left alone was kind of like an e-ink device that runs a computer that like will you know as mental last a thousand years kind of thing so the thing we're trying to evoke there is this sort of like sense of calm and like kind of lack of dependency that I think the system is built to embody like actually be um yeah no no it would if if you had in 2015 it's like definitely not clear to us the funny thing about doing any kind of any project like this right

is like there's all this stuff you just don't know yet and but you can't really fully concede that fact you have to just be like no I think we know where we're going and then eventually you refine and kind of adjust but anyway I don't know maybe that get some background to the applicant feel I think there's an interesting thing in what you just said of owning your own data an artifact of someone else owning our data is a proliferation of apps and yeah and it is I to me it is stress inducing as well like this guy over here wants his spreadsheets shown in some way this guy over here wants it in a third app this guy over here wants it on a fourth so in the company we end up using five different tools to do the same thing but

really it's the same source data you should just have four different ways of viewing it because it the way of viewing it tailors to the person but that's something that we haven't really started to address in any decentralized app space like people in the blockchain world and etherion world and otherwise they talk about this as a potential but they all still just go and build their DAP and have their UI on it and like don't really encourage or to some degree even enable other people to build their UI so nothing yeah that's interesting like I mean in a way the kind of great irony

or like it's just a funny thing about the way that we've gotten up to this project is that it is at once built on kind of like intuition about things that are fundamentally flawed in existing technology and a total focus on the user experience being kind of the only thing that matters and so when you say like yeah and 2015 I use this thing it's basically IRC like we even still like the most frustrating thing about Herbet is basically that well like we believe completely that the user experience is the thing that drives the adoption of new technology it's the only thing that will validate our own contrarian approach to building technology but it's not quite there yet and so when I look

at this landscape of yeah I mean I landscape basically people replicating what I think real anti-patterns is like people replicating centralized services in a decentralized way I mean I support these efforts like I think it's in spirit the right thing but you're still from a user experience standpoint you're not delivering something that is good enough basically to attract just in the most practical storms like not that you're not going to attract enough people to this thing because it's not exciting enough it's not compelling enough it's not new enough like that's just like the single hardest problem like if you can actually deliver a new experience of computing then all of these kind of like protocol issues these these things will go away because people

actually want to use something that you know works in a particular way delivers to them and you user experience and yeah I'm sort of stubbornly feel like the only thing that matters is unification like I need I just if you add another app to my home screen like you're not making my life better basically like you need to remove I want this stuff to go away I want to actually get simpler and if you can't make it simpler then you're not you have it actually solved a problem necessarily and yet when it has been made simpler for example like Google kind of taking over a lot of the you know taking over the email taking over the docs taking over whatever else certain about me search was first but taking over what they've taken over we're still sort of assuming that that company I mean we know that that company

is then gonna everything right and I guess that's like they're giving you the simplicity part they're saying like oh you know everyone will have an account everyone can log in it's a little bit easier than like doing these other kind of apps right where you'd have to share anything and yet yeah they're not really gonna offer you a chance to own it it sounds like you I'm here the focus is really on the ownership I think it's like that we think of like the sort of guiding design principles are you know it should be simple to use for an ordinary person and as to understand as a developer it should be durable it should last it should outlive you like it should last forever and it should be yours it should belong to you and in every way so you

wanna you want those things to work together I mean Facebook's always like a simple example right because they're all these other social networks Facebook one because like purely on interface basically right as like it was simpler it was more straightforward it was cleaner and I remember originally you thought it was private introduced this private you got to choose who saw what whereas MySpace you didn't and Friendster you didn't yeah that was one of the biggest things that it offered and I remember I remember when it came out that's funny it was all kind of like about cuz it was I feel like Facebook yeah they there's so many strange and things about it but it was it was also like you're this small commute it was all about this small community which I like think about what we think about a lot because that

feels like the thing that has been somewhat erased from the internet but I think that in the web 2.0 era the big winners were ones that won on user experience and we traded you know privacy and durability or control for really good user experience which I just done like that's a totally sensible trade actually from a kind of economic standpoint and I think that's kind of just that that's the challenge to anyone who cares about ownership or decentralization you have to like see it as a feature you have to be able to deliver it as a feature like I said and you someone has to be able to sense materially that that is doing something for them I think in the case of it and so when I look at like certainly with

Bitcoin like Bitcoin does me in very certain cases it does make your life easier right like it's it's delivering like a very straightforward feature of maybe remittance or if for many people obviously store a value it's on that on you know on those two verticals it's like it winds unilaterally and that's why it works and it doesn't matter interface in terms of like what do I use to login or authenticate whatever it doesn't really matter it's just like technically it wins but I think outside of many of those things or like outside of those things for Bitcoin there might be a few for a theory of the effort has to be at like you know at the kind of like at the level of an ordinary user like people

have to start to think about it in that way I think sort of like people in this space have to think about it that way I think that's how we try and think about it in order to like just like be broadly successful and I fully concede it we're not there yet but I think that's like yeah we try and think about it getting closer I think I I want to start jumping into the tech and talking about yeah what it actually is but before then I just wanted to throw another anecdote at you because I think it's also potentially interesting conversation all right I was at this was in conference and you make some references again in your website and your FAQ FAQ is pretty comprehensive by the way Thanks and you make some references to was a match sort

of a core core foundational technology and there was this guy at this conference who who said you know laid out this vision of we're gonna build everything on Wazza Wazza MS gonna be like the the foundational VM that that every machine runs and then we just build one unified OS on top of wasum and then apps on top of that and everything will be great and I went you know raised my hand and said what you're describing sounds very much like orbit and his reply was yeah but I live in the real world yeah I imagine that when you started this and started building everything from scratch II like what I'm

picking out from what you're talking about you did this because you live in the real world or that's when you see right where yeah no no the real world has a bunch of and we need to like start fresh and this guy is like no no I live in the real world because I kwasind can interact with what what exists in whatever what's what what's your thoughts on that yeah like we'll see basically I think that's that's kind of a totally fair I think it's worth maybe understood there's there they come from different the motivation is very different given I can see that I'm not my familiar as wasum is probably not you know could could be deeper maybe you can you can

correct me potentially but the primary motivating idea behind the Erb of em knock is basically that you never ever ever ever ever ever change it because you never change it you can ship upgrades to the system over the network forever so if I have an orbit today and actually hopefully later this year we're gonna try to effectively guarantee this and I put it in a drawer for five years and I plug it in again in yet like five years later it'll just upgrade itself until it works again and the reason that it's it's kind of a I mean

the idea is to just be as dumb as possible right you're like you want your VM to never change because then you can change everything else and my understanding I mean x86 is not like this which is why is a big part of why you end up with this complicated software stack on top of your hardware and wasum while it's a nice layer it's not still not simple enough to actually accomplish this ultimately user-facing goal right like and it's I would say while it's a goal for them to change less frequently like they don't want to change much yeah but it's still under control of w3c yeah and they're still make our corporations basically in

control of it if they want something in it it'll happen yeah I think there's a the culture of urban infrastructure development and and I think most of the work that we do is much more from the standpoint of like crafts or crafts people whereas I think that something like wasum to me is like this super professionalized world of software development and I think that professionalized software development is kind of concedes that you're going to need this industrial scale armature of people and companies basically to maintain the world software and our approach is like you know if you make it

really really simple if you make it really really well then maybe you actually don't need that and that's much more in the interest of the consume of like or people because then they can actually potentially understand the tools that they use and improve them which I would argue much better reflects the physical world that has far outlived any of our technical tools and I think these are just totally different attitudes and the thing that bothers me is ultimately that when you build a really complicated tool it's like who is gonna maintain that thing and if it means that yeah giant companies have to maintain it I mean it's not out of ill-will that I don't want that it's

like I just I just how do I know whether how that's going to last or how they're what's gonna happen to their you know their profit motives like what's what where is this thing gonna go it's hard for me to reason about so I think it's then reasonable to be skeptical of like how those things might evolve and so yeah our attitude is like no make it as simple as possible make it as understandable as possible for everyone who uses it they're just totally different things and we may be you know maybe impossible or maybe we may be fully can see that we might be wrong about that but seems like a good idea see so I think next up we wanted to move a little bit more into the technology of orbit and what it's made of you I think

in the introduction you did mention a few pieces yeah there's a clear like focus on identity technologies there's a focus on like you kind of called it an overlay OS yeah I when you said that I actually made a little note I wanted to dig into what that what you think of when you say that yeah and then yeah there's a few other pieces that I think we should explore but why don't we start there sure yeah what when I just give kind of like an overview of what these pieces are technically how they fit together and try and just be as technically as specific as possible so starting with the identity system the identity system was actually originally a part of urban

itself orbit was just the software package you installed on top of of Unix we found that in order for us to actually deploy this thing and let people own it orbit itself would have to be totally secure which is a really tall order given just the scale of what we're trying to accomplish and this thing called the theorem already existed which would accomplish exactly what we needed to do and by factoring out the identity system we would end up with something that's totally general-purpose and reusable so an orbit ID is basically just a number think of it as like any number between 0 and 4 billion so what we want for an ordinary user is effectively to have like a username and password like experience you have this

name that sounds like a name in a foreign language like Rafa Rob dial or Suriname diver these strange synthetic names they're pretty easy to remember each one of them is a number under the hood right we just take the number and turn it into something you can remember the lower numbers issue the higher ones so at the very top there are 256 what we call galaxies galaxies each issues 256 stars a star each issues 65,000 planets simpler to think of in base 2 so it's just 2 to the 8th to the 16th to the 32nd there are also moons underneath planets or to a 64th save gigantic number moons you can think of it sort of

like galaxies are kind of like the root DNS nodes they're the the core infrastructure of the network they actually can vote to upgrade the rules of the identity system like that's sort of built into the etherion contracts stars are kind of like your local ISP they help do peer discovery they probably provide services to into planets and then planets are for individuals and moons are for their devices so the way these this distribution works is that a effective like a sponsor node signs you're like sets the initial keys and then you perform every subsequent key update so that's how I can kind of like give you

an ID and then of course from then on your I can't take it back right if I'm a star and I give you a planet you need a sponsor to be on the network and we can talk about how the OS uses this scheme but it could be used by anything if I sponsor you mean like an existing user needs to yeah and use something an existing super node needs to just acknowledge that you are sponsored by it okay the design of this system is motivated by two primary things for any PKI or identity system to function in a decentralized way you need some way some fundamental way a preventing civil attack right preventing identities from being reused over and over effectively

to combat spam and abuse so by making a finite we say well they by default should have some value and that value should be above what you would make spamming or abusing the network and you also want the network to be somehow accountable you want there to be you want it to be like the foundation for reputation systems it's not a reputation system on its own but you want there to be some way of knowing when I meet a stranger on this network like the default should be I should trust them at least a little bit right versus like a random IP address if anyone for anyone who's ever run like failed to ban or something like that like a random IP address you definitely do not trust right a random orbit ID you should trust why would you trust that though like why

couldn't it be a malicious IP hopefully it cost him 20 bucks right so a gmail address I think costs if you bought go you know on the dark web or I always I always hate when people say X is so big you go on tour and find a market place sorry I wish that's what journalists would say when they talk about the dark web and like it's so it's well-known like the dark wetness something serious anyway you go buy a bunch of email addresses they cost you like I don't know 25 cents or something like that there's a price and so the idea being that if someone has a little bit of skin in the game they don't want to abuse that address because then everyone would blacklist them so I'm not saying that all of a sudden everyone on the network is friends but they're

probably not malicious is it is the thinking how do you keep that costs above the value made though because like that mean that the problem with ipv4 and and I assume the problem will be even bigger with ipv6 and Gmail addresses and everything is that it's too cheap right it is too cheap you get an email address and so I can buy one and spend the out of it and yank it and yeah and and then and I move on to the next one so how do you ensure that the value of a planet actually stays at 20 bucks or whatever it may be yeah we don't we we can't the thinking is simply that you you let the market do that and you let the sort of scarcity

trésor shape of the address space do that and that you effectively put trust into all of the people who also run the network so like the price of a planet right now is about 20 bucks why I mean there are plenty of planets like a wall give you a planet but I think the reason that hovers around that is partially because everyone who's selling them anyone who owns really cares about the network itself so they want to sort of protect it from you know anyone buying and mass to try and sort it to try and you know I've use it basically but ultimately I think as herb it grows you're absolutely right that like it's not perfect it's it makes trade-offs that I think are there they make these

trade-offs mostly in the in the in service of simplicity because I think if you think about trying to design it's very difficult to design another mechanism at least whenever I've reasoned about it I think I mean the total addressable space should be this isn't it the same as ipv4 yeah it is ipv4 yeah yeah so I mean in theory you should end up with a structure like ipv4 but I think the problem with ipv4 is we've come to the point where we just assume that identity is not tied to IP anymore yeah and like an IP can be owned by me today and by some other random person that you know has a contract with his

ISP tomorrow and so they'll be randomly sort of shuffled around and therefore we can't build a reputation system on it yeah that's have an identity that you explicitly from day one say there's going to be a reputation system around it then it's also like a good reputation IP will cost way more than a bad one yeah the way I often explain this is like you can think about Herbet as digital land I think right it's this sort of vast territory that unlike land which of course has natural resources which make parts of it you know more desirable than others existing neighborhoods and so on herb it is more like this vast grid and as it gets used it becomes it sort of takes shape and

neighborhoods develop so certain stars may provide specific services in the way of like one of them runs as your extra layer and one of them runs machine learning services and some are faster than others and so on so definitely I think price differentiation and actually you see some of this already will be vast like they will there will be you know a sort of unused uninitialized piece of outer space is totally different than one that has been operated that's effectively almost like a business these are also likely infrastructure providers so you need to the ideas that you host inner 'but somewhere so very likely and we work on this a little bit there's actually a separate company now orbit live there

are two or three other ones kind of on getting started up that where the thinking is that they would provide hosting and prevent do exactly this so it's sort of like you go somewhere you buy a name you buy hosting and you get kind of like the complete package from a provider who is also running infrastructure nodes right now we're kind of out of the scope of what is orbitty technically it's just this set of numbers of special numbers that have a distribution pattern at the beginning one like thing to mention I think is like at the beginning all of this was effectively owned by Curtis who wrote it he sold half of it to the company for almost nothing when we started the company or when I mean I was actually not quite

there yet and then the company and Curtis both gave almost half of their ownership to a future foundation which is still sort of cared for by Toulon and then since then you know in 20 unfortunately missed the boat honest but in like 2014-2015 you get a galaxy just by like writing a little bit of code and it was just easy to get large chunks of address space lots of developers some of it was sold to investors and then as its you know urban has kind of gotten off the ground for the most part yeah we we sell address space other people's any now it sort of trades on the open market but it's all non fungible anyway it's sometimes people are always curious about the history of you know how did

this stuff end up and which which parts of what you just described ours like etherium smart contracts are built out of a theorem customer contracts because you started sort of at the galaxy being oh yeah just down two planets so we don't register moons on chain because they would be they're just too many of them so galaxies stars planets are all registered on chain there's an interface for interacting with those contracts that we built and deployed so if you buy one somewhere now you can log in at Bridge daughter but organ-like manager identity so and then yet we give away these little cards that have a key on them when you register this key it'll will derive keys for you using our key pattern which is more user name and

password like trying to kind of obscure the difficulty of tip 39 or whatever but yeah all that stuff is is all done on a theorem and doesn't depend on the OS at all got it so each one of the things is a unique and ft yep yeah they're all enough team is connected to each other they're they somewhat depend on each other like the sponsorship tree is registered on chain basically and then Herbet the urban OS like listens to aetherium and propagates key updates through aetherium there's kind of a key chain for each n of T so I guess yeah maybe dive slightly deeper into the technology the idea is that each one has an ownership key a spawn key a transfer key and a management what's

called the management key ownership is obviously kind of the thinking is like put ownership in a safe probably with transfer spawn you could keep on you know in a locked drawer on your desk and management is like I don't know like in an envelope on your desk so management you use to boot a node spawn you use to spawn children or sponsor other nodes and then ownership is the thing that you're transferring ownership are effectively the same I'm just wondering cuz like we actually did an episode a while ago on NF TS or I think we actually they were on ERC the ERC process yeah and we looked at ER c 721 and there were like these I don't know what actually I haven't stayed up to date on this but there was some new and

our er C proposals where you'd have like like if you had an NF T that was in a game and you wanted that game character to own a hat mm-hmm you could actually use these there was like new ERC's protocols for how they would own one another yeah and I'm just wondering if if you guys use that at all we may not even know about that maybe that's embarrassing I'm not sure I think we probably finished the contracts in early 2017 and spent almost the entire year auditing them because we wanted to be completely certain that they were secure we deployed them in well very quietly in late 2017 and then announced them in

early 2019 okay and late 2018 sorry that's right like like it's like we did that and then I I think we kind of like it works the idea like obviously I was talking about before seconds did not change it so yeah we don't we don't we don't think about or we don't keep up to date I feel like really okay yeah go ahead well I was just wondering like what happens if theorem changes yeah that's a good question and it's one that we get a lot especially from our community who is many of whom are super skeptical of the theorem the thinking is basically that we have you know this ownership registry the pretty consistently updating as ownership changes hands and if there was

anything that you know if the theorem was going to go through a really crazy fork or who knows you know any no or theorem actually is whatever being 51% attacked the quorum of galaxies could basically just call a block number and just say okay look at this number we gotta move and so hang on to your keys we're going somewhere else ideally we would actually what we'd like to see is that once there's you know when we went on chain there were about five thousand owners total maybe like three somewhere between three and five thousand stars galaxies planets that's obviously they continue to it's kind of

like on a solid linear rise it's a lot of people to coordinate but the thinking being that once the urban OS nodes are more more of them are online it's still a probably pretty good number of them online today but once it felt like you could get consensus from inside orbit itself we just move to orbit itself yeah we just have like the galaxies run their own consensus and it'd be sort of self totally self run the theorem is great we've always thought it I was like bootstrapping and could go on for a while it if the room works like we have no real look like complaints about developer experience is good but yeah there's like this sense where we're like well it's also kind of a liability you just don't know what's gonna happen

there I mean especially with the original philosophy of no dependencies and like actually yeah trying to build sort of a new thing but it but it also feels like it's an inevitability like it's not like oh yeah may be able to live on a theorem forever it feels like no not once you know we can run consensus and whatever have proof of stake or whatever within the galaxy sets within orbit itself then that's that's the obvious move right so yeah exactly yeah it's more like when to get there rather than it I do think that it's really great that in

the time frame that we've done it we've been able to give people like moving to aetherium has a great compte centralizing effect because now that people actually hold their own keys and they can independently sell these things move them around start them from a community standpoint sauce it's fantastic like so I think is definitely the right practical move but then yeah there's this kind of when we think about the super long term of Herbet which we obviously do or think of this as a super long term project yeah there's this kind of like alright there might also be some liabilities here that we like don't actually want but we'll see I think we just try and treat it as practically as

possible let's dive into the other side which is a bit OS and knock hoon all of sharise crazy fun things yeah something that you don't talk a lot about though and where I would want to start out like starting out here with most of these kinds of projects is not working yeah how does networking work what's the protocol I assume you don't use lib p2p yeah yeah we've thought about it actually kind of but we don't use let p2p well live p2p should use our identity system I think the reason we don't use live to be p2p actually because it's unauthenticated from an identity standpoint okay let's get to networking is it kind of at the top of the stack so urban OS you were asking

earlier why do we call this an overlay OS and the thinking is sort of like okay you can get a UNIX box with an internet connection anywhere in the world there's probably one in your pocket like UNIX plus tcp/ip and UDP is like not going away everything on top of that I don't know so what we do is we basically build a formalized interface into UNIX which is using Libby V where we get file system events network events HTTP so it's I mean file system UDP HTTP there used to be TCP we don't really use it anymore and terminal events and then we convert those into that event stream is

what gets consumed at the very lowest level our virtual machine and Wayne nock has pretty sure 13 op coats and then the event it fits on a t-shirt it's extremely concise just the spec and it's a you know basically super super compact touring complete virtual machine we on top of that have a programming language or our own language called hoon hoon compiles itself to knock which is kind of cool as its self because we update it over the network it has to be self hosted from a technical standpoint I find I always find this incredibly refreshing because if I you know if my stack trace like terminates

inside of the language I can actually just go read the compiler or go read the standard library I understand what's going on in the language on top of the language we have a basically an event dispatch which acts almost as like a kernel between a bunch of different kernel modules that provide a similar set of affordances to what you would think of from the web stack so file system Network protocol application sandbox secret storage and a build system also probably forgetting something there timers few odds and ends utilities from there then and so that whole suite of things we call arvo so sort of like arvo who knock is the sort of inside of orbit stack that whole

thing clocks in at about 50,000 lines of code which is super compact I mean I don't know anyone any technical person listening it's like that's like something that most of our infrastructure people like are comfortable traversing the entire operating system code base which is really great right like it's it's 50,000 lines of hune right yes yeah which to be fair is like yeah yeah several hundreds of thousands of lines of Eva Haskell or something I wonder if it would be that bad but yeah it would be longer it's so yeah important to note so sorry is like hoon is purely functional but it uses no reserved words so it uses what we call runes which are effectively the same as rivers reserved

words they're just like the language operators and so it's it looks really weird and it's super compact at first people who love right you know people learn to love writing hoon and do really really like it the original thinking was basically that and this is true although it makes we're really steep learning curve is like when you look at hoon you don't really read the code so much as you just sort of see it because you aren't doing this other operation of separating reserved words from from like variable names I mean people are super critical of this I think for good reason but one of the things I always find funny about is it

every time someone wants to criticize this they act as if they're like the first person to discover this we're like yeah it's weird it's it's a you know it's a very particular flavor but it's also not like completely unprecedented there's ap oh there's K well what's this K exactly the the theory exists and there's other languages that do these kinds of things yeah it's not for everyone but it's also way easier than it looks we run a periodic sort of MOOC called hoon school which anyone can come and sign up for and we find that in so for for a new infrastructure hire for a new person who's here you know really working full time learning hoon who's like an experienced programmer a week

two weeks it's not that hard it's like if the first day you're like this is insane what the hell and then by day three or something you know usually getting your figure it out for even for people who've never we find actually for people never program before sometimes it's even easier to teach them hoon cuz for people who have a lot of professional programming experience you're kind of unlearning some stuff to learn hoon people are like well what you know why do we even have this language ultimate least because hoon compiles to knock like hoon is basically like see to knock as assembly language is a very thin wrapper over knock it kind of feels like knock you could write other languages or you can like compile other language to knock potentially we've done a little

bit of research work in this area but hoon is not we don't expect hoon to be like the Python of Irbid or something like that like there will be other probably more expressive language is currently actually one way we think about this is sort of like urban should almost be like personal fire base plus identity so like urban should have much better HTTP API is such that you communicate with Herbet primarily using just like JavaScript it could be your Raspberry Pi that's a sensor that's just calling out to urban in an authenticated way so Herbet has like defined protocols that then you don't have to learn for most people you don't learn whom you just like push data into orbit pull data out of orbit and orbit is kind of like this

dumb event transceiver that just lives forever anyway no talk about networking we've walked from like the thin OS shim getting some events yeah through knock through hoon and now we're talking to other computers somehow yes yeah yeah so import I guess you really made me realized one thing we forgot is we get events in and of course we push event we push effects out so an event comes in something happens and then an effect comes back out so in the case of networking UDP packet comes in that may trigger any number of things inside of the event dispatch sequence of the kernel and then an effect comes back out

so how do we I mean to be honest like there's only so deeply that I can dive into the networking stack but the basic idea is that we because of the way because knock is deterministic and because we use sequence numbers Eames which is what we call our networking protocol it can provide exactly once messaging between urban OS nodes in a way where from the standpoint of an application programmer if I want to operate on remote data I just act effectively like it's local data and the networking protocol you know sort of abstract away

any of what you would need to know about yeah just you don't you certainly you only think about messages and from an at the application level you only think about like OS level events and it may be an event that is actually addressed to another node in the network so the whole point of this identity system right is such that I can just say hey tell surah Anam Tov what's up you know and the networking layer knows how to go and find that / like actually can use the ID hierarchy to go find the pier then can you know figure out the underlying IP address and move that over to you know get that into effectively their OS at the OS level and where you

know the programmer doesn't ever have to really know anything about the about the networking stack it sounds a little bit like Erlang networking where you have a process ID and like the programmer doesn't necessarily need to care of that process ID is local or if it's a remote node it's kind of resolves actually I'm not that familiar but people keep telling me that we have a lot of overlap with some with Erlang so how does this differ than from live p2p like how is the actually I don't know if we've ever covered this but how does live P 2p send a message from one so I mean the PHP is like it's it's an agnostic networking protocol and it's meant to be generic so loopy tree is really just a little

snippet of header data that goes with like a connection handshake saying I'm speaking Lib p2p with these protocols what are you speaking and and you they can agree on which sub protocol they talk they do Lib p2p does a lot of peer discovery though right like isn't that a big part of it you still need a hat in the address but it'll find the other peer not so I mean the thing with lipids B is you one should really just think about it as the this little header message of like so that's because that's then like both forwards and backwards compatible but discovery is like a sub protocol kind of this so you can go I I speak Lib B to B

here's a discovery request and then if the other person speaks Lib pates be they know how to interpret that yeah but I'm so when I talk about networking I'm one of the things I'm always curious about is discovery like how do you do discovery is it a CAD ammonia kind of thing or have you found your own way yeah so we use the we use the sponsorship tree so like if we're two planets and we want to talk there's our two cases right we've never spoken or we didn't you know never been connected and maybe we've been connected but you're not responding and in both case we effectively do the same thing I asked my star who s their galaxy who asks your galaxy who s your star and we

send if we've never spoken before respectfully spend a first message through where you always are you know your your black broadcasting to your parent to your sponsor just like hey here's where here's my IP we don't presently do anything to hide your IP which is I think part of the thinking of that you know orbit is not a dark net but actually Ames is transport agnostic and could be run over tour we've people have thought about working on that which would you know change some of this but in any event just from a pure discovery standpoint goes up the hierarchy down to you and then I attach to my message you know here's my IP address right so my first message gets to you by going through our sponsors and then you talk

directly back to me and we continue to talk until you know whatever maybe you're taught your herb it's running on your phone and you switch ip's and then we go back through the hierarchy if you're if the connection has been dropped pretty simple it's similar to DNS basically and so you can see how maybe then the sponsors you know quality of service matters you know you want to be under a sponsor that's online and responsive and so on yeah it's very simple the DHD networking is cool but all right seems hard from from that perspective of aims right it's like simple it's it's definitely harder I mean it sounds like you know like you

said with the service quality of galaxies it really like I keep having this in the back of my head that there should be a blockchain within within orbit that is a proof of stake chain because then you can slash the galaxies if they aren't online or are you serving DNS requests correctly all this stuff what we cool I agree that yeah that's oh you already have all of the basics in place yeah all of this yeah yeah yeah no I this is like gets it's just never quite made it to the top of our properties but like people have talked about this quite a lot there's some tons of interesting you because you also kind of get stake in some ways for free just given someone's ability to authenticate

as a certain ID right so then their ability to run a chain is or like to you know to participate in a chain particular way is really easy to reason about you already have civil resistance is to some degree although it kind of becomes problematic when that's how folk well whatever the other thing that I want to dig into and and ask if like how you're doing it and and what you plan on doing is with knock because you you have this super minimal vm yeah 13 are codes or whatever they obviously need to be translated to x86 in some way yeah is that currently being interpreted is there a compiler how does that work and yes I wasn't sure what do you imagine

that it how it will work in the future one of our first batch of interns was completely and this was in like 2014 was obsessed with the idea of building a NOC ASIC or even an octave which people still love to think about and I would love to see happen although it's Hardware scary basically you know yeah I mean it would be incredible right to have a machine that actually run like it's like physically real or realize in that way anyway for the time being yes NOC is interpreted in see NOC has one very important opcode which is basically the jet and jetting basically says hey instead of running this the computation

in here call a registered function in the case of certainly crypto algorithms right don't want to have to re-implement that I mean you just like yeah we we like to wear our seatbelts like we use these the existing implementations as much as possible so in NOC you know when you're gonna run 2:56 or whatever that's running in a reference implementation and see that's not running in Hoon code I see sort of like host functions and was them yeah yeah exactly yeah I think that is exactly right and then there's also like in simplicity has the same concept which is similar I'm just saying like I mean it makes a lot of so make sense for

crypto it also makes sense for things like if he had a tensor flood chip right you're like yeah definitely not gonna run this machine learning algorithm it would take the life of the universe please run it on this piece of metal that is currently written in C yeah I think I mentioned that so we have an interpreter called ver it's written in C it's pretty it works reasonably well we are sort of always want it to be faster and wanted to be easier to reason about it has its own memory model it's it's pretty intense to piece of C code so we've been working on actually using Gras truffle which is the sort of like the JIT to like Java JIT to write a knock interpreter and so that's kind of

like one leg of research work around here that seems pretty promising so there may be a job a bit java-based knock interpreter reasonably soon which also would be from what we can tell be significantly faster and would also let us do 64-bit so herb is also limited to 32-bit at the moment or just always driving us a little bit crazy but of course we would fix it yeah I mean it feels like if you can get to a stage where you're getting it it's you're like that much closer to the Machine level as well yeah yeah yeah another interesting like intermediate between running an interpreter on a UNIX machine and like having an ASIC is

having a unique kernel that just runs yeah interpreter right and like you're you're still just booting into an orbit yeah so we've done we'd like to provide hosting ourselves at the very least for our own urban nodes so to make it really easy for someone just go from like an email link right to like a running orbit OS node and so I've been working on you know kubernetes based automation for this and part of that of course is like packaging or bit up such that yeah there's nothing else on there you're not like you don't have to like go run your VPS you basically just say spawn me an image that only runs out of it that far yeah like we made it that far which also of course splits out like

persistence right in in a kubernetes like way I think that yeah that the JIT project is definitely the like immediate future though just because I think it's like if we're doing something as unusual as knock and trying to make it performant that's just like the most it seems like that's that's our that's our best bet but you know we're still if like Chris Lattner all the sudden is like looking for jobs and happily talk to him there's lots of fun hard interpretive problems so you just mentioned before real briefly that you don't roll your own crypto you tend to use like existing libraries but I'm wondering if do you are there any like I don't know like privacy technologies or systems cryptographic things that you're

exploring like what what kind of cryptography exists throughout the system a lot but it's mostly unremarkable I see so yeah so basically like there's sort of like the standard suite of algorithms you would need to interact with a number of known ball chains and for us to do networking mentioning sending assigning transactions to to these networks does that mean you directly connect to those networks like do you connect p2p to those networks or do you send to like a local node that then forwards to their network yeah really good question so for aetherium basically it's the latter we've been working on sort of bolting on a Bitcoin SPV node to orbit itself so

yeah soon you'll be able to just like effectively set a flag or some not sure exactly set it up but like you know you can just say please also build an SBB note on here and then your bit can you know talk directly to Bitcoin which is cool and I'd like to see that expand it'd be nice to be able to do that with aetherium also the Bitcoin project and pretty sure actually has pretty much completed as in the last week or two and I think there might even be a someone was working a lightning oh - and there might even be like the option to do have full Bitcoin node at some point this has been like one area of work we have a lot of like sort of urban attracts Bitcoin maximalist and I like Bitcoin I mean it's make sense to

me so we've been trying to has her better support bitcoins from natively within within orbit cool that actually like we talked we did talk a little bit about your community what is lon and like how big is it like what is this thing that's building this larger project and are there a ton of outside like contributors that's yeah it's pretty varied so for a long time till on was five six you know until basically like early 2018 we funded ourselves so early on tuan was funded by and recent founders fund Tim Draper sent in some sort of old-school Bitcoin people and then we started to sustain ourselves by doing small address

space sales where we in 2016 and 2017 sold little batches of of stars and so the community I think probably from that period maybe 2014 to 2017 started to grow like we had you know increasing outside contribution and a small community of people that started to spring up we started hosting meetups which I feel like was like a good things that have always been like monthly or bimonthly things that we hosted our office and then yeah so in 2018 sold a slightly larger chunk of the network privately and we grew Twan to about 23 and that's about the size that it is now during that time I think later in the

year another company got started called herb adult live it's totally separate I mean we know them obviously but they're more concerned with basically like onboarding and making sure that the experience of like buying an address getting on the network is really good and then only in yeah probably really just like over the course of 2018-2019 like the community grew pretty significantly I think like putting things on chain and just the overall well while Herbet is still slightly painful to use like you know you need to install this thing yourself the user experience I think has gotten better to the point that people just hang out on orbit and there are sort of small communities that form on the network itself so yeah we're

kind of like the primary kernel developer but we don't even hold that you sort of like the economic incentive has in fact acted is like a decentralizing force basically right it's like other people are interested in participating in this thing and so they've gotten involved either purely because they want to own a piece of it or because they want to build a company around it or whatever we already talked about Kurtis as the guy who started as he had a lot of the original vision of NOC and who and then like a lot of this is coming from him obviously and it's sort of whenever I talk to people about Herbet he is one of the people you know that starts getting talked about both

for good and bad like I have both sides where you know it crazy dude who like invented a bunch of and like it's cool to be able to do that but then also like this whole political thing that I personally don't really know what it's about but you mentioned it in your own faq as well so I assume this is something that gets discussed a lot he was a part of the company there was a lot of controversy he left the company what does that all mean for for both the project and for the company so I got involved in orbit entirely because of the technology like I mean I was sort of amazed that someone had gotten this thing to even work at all and of course later found out that yeah Kurtis has a

reputation as a writer and as like a sort of as a blogger and I guess a political thinker but I always knew Kurtis just as like you know I'd like a basically a personal professional relationship working with this person and I liked working with him he's super super polarizing for I think good reason I was trying to put this in terms of kind of like the question I asked myself because I think when people are skeptical of Kurtis I'm well yeah I was really skeptical Kurtis like I completely understand why people would that way the question for me was like what is the goal like is there anything in this system you know technically in terms of its ambitions and so on like let's assume he's Satoshi like like what should we just be suspicious about about

this thing and I think I mean at this point like I'm probably the most I'm the person with the most kind of like sway in terms of what we work on or whatever you know it's like like effectively kind of inherited leadership of the project I mean I took over a CEO of the company in 2015 I guess I've been in charge of this thing for some time and yeah I feel like I sort of like take complete and total responsibility for sort of like what the goals of Orbitz are and certainly like in terms of like if I thought curses political thinking actually had a huge impact on on burbot in a way that was suspicious or was not in fact in the interest of people using Irbid and I probably like wouldn't be working on it but it's a yeah the curse is always like

a huge PR liability and in some ways actually Curtis was the person to start this project but he was never the person to finish this project and it was the the nice thing has been seeing other people take ownership of it and take it on people who don't like her at all you know get in get involved basically because they like believe and what Herbet has the potential to do so yeah so I mean the practical thing worth mentioning right is yeah in early 2019 Curtis left he has no voting interest in the company no voting interest in the network but he does own some of the networks have like he owns no galaxies but does own some

number of stars which I like really should know it's it's not a it's nowhere near a impactful number in terms of control especially because they're not galaxies and that was his own choice I think he understands his own position as like a super controversial person and was just like you guys should take this and do what you can with it I do think the right way to think about Curtis is this kind of like mad genius type you know who then kind of just wanted to go and like and sort of disappear like the main thing that Curtis had was like the serve hubris to even attempt this project right like most people does not work on something like this is so crazy and so you can eat someone like that to

get it off the ground and that is definitely yeah it's like that's not what you need to actually make it work and make it make sense and make it understandable and function as a community and so on interesting yeah yeah urban is much bigger than Curtis and I think that's evidenced in the group of people who work on it which is very is like widely varied I think we also think about urban as sort of urban aspires to be neutral infrastructure if there's any reason that that is not true you know or we are not able to achieve that goal like we're also I think pretty we welcome critique and criticism of the sort of culture of working on this thing

is one that is spirited and open and straightforward and so I think that that's like the main thing I'm always like okay if you're worried about something you should just come talk to us about it and get you know get involved in some way that this thing if there's something you think we missed like we want to talk about it that's it so pretty much always been that way but I think that it's sort of like it's reasonable to be skeptical and we too are skeptical basically cool I think that's really helpful to help us understand how you think about it and it sounds it's so funny because like that idea of the person starting the project may not be the person to lead it onwards I feel like we maybe see replicas of that in some of the projects that we've

encountered in the blockchain space as well I think that's the case for all startups like it is extremely extremely rare that a founder should ever run a company forever yes at some point the founder should leave and just be the founder I mean this is like they can do something else in the company were probably not leading yeah definitely yeah this also comes through and a lot of think of some more specific examples but the other thing that's funny about this is that there are many like Curtis was able to assert you know to be able to produce a virtual machine self-hosted language and

operating environment that actually runs its own network pretty much as a solo project as insane like it's totally nuts it's amazing and pretty much everything that if you looked at orbit in 20th between 2013 and 2016 even that seems technically weird or annoying about Herbet probably actually is weird and annoying and should be fixed and so the thing that has almost the entire system at this point has been rewritten by the people who are here and even now I mean we talk about changing and upgrading every single piece of this thing like there are things parts of it that need to be more intelligible there are parts of it that are actually just completely

wrong and should be fixed those for the most part actually they're less and less of those but I think that's an important thing to understand about like the role of this person is it's like he tapped into something I think a lot of people felt was important and needed to be done but there was nothing about this beyond that it's like every it's it's owned by a bunch of other people it's become it's become a collective project so I want to say thank you so much for coming on the show yeah thanks so much for having me and helping us understand this mystery project that is perfect thank you very much and to our listeners thanks for listening thanks for listening

[Music]