a16z Podcast - a16z Podcast: Crypto and the Evolution of Open Source

Episode Date: August 20, 2018

with Devon Zuegel (@devonzuegel), Denis Nazarov (@iiterature), and Jesse Walden (@jessewldn) The open source movement enabled so much in computing, including the collaborative building of libraries -- that is, building blocks of code that developers could combine together to build applications. But as these applications grew to massive scale, those libraries ended up being somewhat asymmetrical for "nights-and-weekend" developers (compared to say, the disproportionate resources of a large company with billions of users and big data). Blockchains, however -- enabled by cryptotokens that align incentives among stakeholders -- shift open source development from libraries, to the creation of shared, open, permissionless services. Instead of being siloed and repetitively produced as if from the industrial factory era, any smart contract developed on Ethereum becomes a shared service that can interact with any other service... incentivizing developers to improve on existing services, build on top of them, and enable combinatorial innovation at greater scale than ever before. But if decentralized networks are to win the third era of the internet, how will we resolve challenges such as single-purpose services (another form of consolidation), community conflicts, and other issues? In this video, freelance software engineer (and blockchain app developer) and writer (and urban watcher) Devon Zuegel guest-interviews a16z crypto partners Denis Nazarov and Jesse Walden, the co-founders of Mediachain Labs (which was acquired by Spotify in 2017). They draw on their past experiences leading open source development of a decentralized media attribution protocol for connecting creators to their audience, and what the implications of "services vs. libraries" could be for creatives now. And what about identity, stablecoins and crypto finance, and more? Finally, they extend their previous analogy of cities and network effects and how it fits the idea of libraries vs. services in crypto. Please note that the a16z crypto fund is a separate legal entity managed by CNK Capital Management, L.L.C. (“CNK”), a registered investor advisor with the Securities and Exchange Commission. a16z crypto is legally independent and operationally separate from the Andreessen Horowitz family of fund and AH Capital Management, L.L.C. (“AHCM”).  In any case, the content provided here is for informational purposes only, and does NOT constitute an offer or solicitation to purchase any investment solution or a recommendation to buy or sell a security; nor it is to be taken as legal, business, investment, or tax advice. In fact, none of the information in this or other content on a16zcrypto.com should be relied on in any manner as advice. You should consult your own advisers as to legal, business, tax and other related matters concerning any investment. Furthermore, the content is not directed to any investor or potential investor, and may not be used or relied upon in evaluating the merits of any investment and must not be taken as a basis for any investment decision. No investment in any fund advised by CNK or AHCM may be made prior to receipt of definitive offering documentation and due diligence materials. Finally, views expressed are those of the individual a16z crypto personnel quoted therein and are not the views of CNK, AHCM, or their respective affiliates.  Please see https://a16zcrypto.com/disclosures/ and https://a16zcrypto.com/disclaimers for further information.

Hi, everyone. Welcome to the A6 and Z podcast. I'm Sonal. Today's episode is on crypto and the evolution of open source. And in particular, the idea of libraries, as in traditional open source versus services as with smart contracts and blockchain, which are open, permissionless, and more importantly, may address asymmetries between nights and weekends developers and entities with disproportionate resources. What are the implications for developers, for creatives, for identity, for crypto finance, and more. Devin Zougal guest hosts this conversation with Dennis, Nazarov and Jesse Walden, the co-founders in Media Chain Labs, which is acquired by Spotify in 2017. They draw on this past experience leading open source development of a decentralized media attribution protocol for connecting creators to their audiences and consider what could be different as crypto evolves. Both Dennis and Jesse are now with A6 and Z crypto. Speaking of, please note that A6 and Z crypto is an independent fund managed by CNK Capital Management, a registered investment advisor with the SEC. It's separate from the Andreessen Horowitz family of funds. The content here is for information only and should not be taken as legal business tax or investment advice. It does not

constitute an offer or a solicitation to purchase an investment or a recommendation to buy or sell a security. In fact, the content is not directed to any investor or potential investor and may not be used to evaluate or make any investment. CNK is not seeking investors and investment in any fund advised by CNK may not be made without first getting final offering docs and diligence materials anyway. For more details, please also see A6NCrypto.com slash disclosures and A6NCripto.com slash disclaimers. Finally, you can find this episode as a video as well with links and more on A6NC's YouTube channel. Hi, my name is Devin and I'm a software engineer and writer based out of San Francisco. Today, I'm talking to Dennis and Jesse about the evolution of open source

and the evolution of crypto networks as well. Dennis, you've written quite a bit about the dichotomy between libraries and services. Can you expand a little bit on that distinction for us? Well, in crypto, you know, crypto has been kind of described as the natural evolution of the open source movement that open source started as kind of an ideology and, you know, a community with certain values that believed in software that should be openly licensed, free to use by anyone, and kind of the idea that anyone should be able to collaborate on it. People would have identify some problems with that, where mostly around financial incentives and tragedy of the commons, this common resource ends up being used by companies in commercial settings, but there's

sort of no incentive structures to maintain it long term. And crypto, with the introduction of these tokenized networks, kind of changes the power dynamics. And I think it's important to kind of dig in to that and talk about the nuance. The GitHub model is a really good way of thinking about it is a GitHub repo is basically a collection of text. You know, it's a collection of code, and GitHub allows multiple parties to collaborate on it and to improve it, submit pull requests, open issues. So it's sort of beautiful when it works.

You know, a code base can evolve with the help of the community. But there's some systemic asymmetries where, while code is, you know, it's a useful collection of, a program is a useful collection of different pieces of code and modules, etc. Code becomes powerful and useful when it's instantiated.

So something, you know, like MySQL is an open source database. It is used by big companies, but it's also used by weekend developers when they're doing a tutorial to build a web application over the weekend. And kind of the asymmetry comes in, you know, the deployment of that library.

So when Facebook or a large entity like that, that uses a database, it has millions or billions of users, it has tons of data, that deployment becomes much more valuable. And in the case of a weekend developer, they're starting from scratch. They need a computer, a virtual machine, or some AWS service where they run some server on it and they deploy the database, but they have no users, they start from scratch. It's also just like raw engineering power.

Facebook has hundreds of engineers to deploy something like MySQL. And then the weekend hacker has to do it themselves. And it's just like there's an asymmetry in terms of like how you leverage the open source libraries. Yeah. So I think a great point is that a developer has to reinvent kind of all the base functionality of an application. I think a useful analogy is that, you know, like an industrial production that, you know, in order to make any product you need a factory. Like all factories require, you know, a brick building. They require some power source.

tons of employees and assembly line, sort of like there's 90% of that stack is redundant between all production and everyone is trying to do this unique 10% thing. And the costs of spinning up production is very expensive. So I think, you know, all open source is this kind of great community iteration process.

There's still some systemic problems with the power distribution. So if you're established a company, open source is way more valuable to, it does way more to you. And I think what's interesting in blockchain chains. You know, Ethereum is described as the world computer.

There's this, the idea that we share a computational substrate, we share a data substrate and like a file system substrate. So any service that is deployed to this network becomes this instantiation. It becomes a service versus a library. So I think that's the big analogy we want to talk about today is that sort of open source code is libraries. have to copy the code and re-instantiate it, kind of deploy it from scratch, put in tons of resources.

But in blockchains, kind of the metaphor is these services. A service provides some utility, it shares state. It's like the singleton deployment. It's like one logical thing. So like Ethereum is a single database. So any smart contract deployed on Ethereum lives in this kind of shared universe where if a smart contract provides some utility,

like identity, for example, and there's a good identity deployment, any new developer building on Ethereum, they're not thinking about redeploying identity from scratch. So in a web application, you need login. Everyone has to re-reploy login. But in Ethereum, it's a totally different metaphor. Anything that becomes useful can be reused by anyone else, and the participants are incentivized to improve an already existing service

versus trying to compete. One thing you talked about was how, in conventional open source, There are these libraries that everyone can reuse building blocks, but they still have to re-implement them. They have to redeploy them themselves. Can you talk a little bit about the difference between using the crypto building blocks and using these library building blocks? The way that I think about it is on blockchains, when you deploy an application or a service, you're deploying one logical thing. There's like a canonical instance of it, and it has an address on the chain that anyone can discover. And it's actually, you know, if this thing that's been deployed is useful,

it's easier for everyone to converge on it. Like, we could call it a showing point. People discover this thing, and they start to use it, because it's easier to do that than it is to redeploy it, which would require you to copy the thing, you know, redeploy it, and create a second instantiation of it. So, you know, on blockchains, you could say it's easier to sort of like follow the rules that already exist, and those rules are open for anyone to see

to ensure that you can trust them and accordingly you can you find the thing and you just start building on it and that's the easiest thing to do it's a default setting and that diverges from open source because again with open source database like MongoDB is something I can download

but I then have to like spin it up and redeploy it in my instance is completely different from Dennis's and completely different from the one you're using And so, you know, this is just the fundamental divergence. On blockchain, we get one canonical thing that multiple parties converge on. And in traditional open source, you know, everything is fractured into its own silo. I think maybe connecting it to our previous discussion about cities as analogies to blockchain networks. I think you can think of, you know, libraries or code as blueprints, you know, that describe how to do something.

So blueprints are obviously very helpful, you know. If I have a blueprint that tells me how to build a road system in a city that's very useful, a blueprint that describes how to build a power grid is very useful. But, you know, to run with the open source analogy, open source requires me to take the blueprint for roads and, you know, build my own city from scratch. I have to implement the road system. I have to implement the power grid. I have to implement shipping services.

I have to get actual people in there and for kind of emergent properties of, like, you know, a vibrant city emerging. But I think blockchains are more like the road system already exists. You don't have to worry about it. So a road system is a service that's been deployed by someone else. Now there's cars flowing through it. There's gas stations are starting to be built. There's storefronts.

So anyone that wants to participate in the economy and kind of the infrastructure of the city just gets to build on top of the roads, build on top of the power grid. And I think with the same open and collaborative properties of open source, You know, the fact that we don't have to rebuild cities from scratch to do some new thing. We can just, you know, join an already vibrant city. It's, again, these compounding effects where it just becomes much easier to deploy services. And I also think the kinds of, you know, applications you think of building becomes fundamentally different because it becomes, you think about how is what I'm contributing, you know, going to impact the ecology,

the symbiotic system. One of the infamous examples of people building on other folks' data caches is the LinkedIn API. I know a lot of companies built services on top of the recruiting API and retrieving people's profiles and so on. And so when LinkedIn shut down that service, just lots of businesses literally went out of business. They could not function anymore because they were based on that. Can you talk a little bit about why someone would want to add data and add, maintenance support to a network like this if they don't have that proprietary control, like LinkedIn currently does?

I think here it's fundamentally different. Like the value of the service, the reason the service sees third-party applications as its customers, like its survival is dependent on being symbiotic and kind of enabling of services on top of it. And I think another argument is that if a token is used to mediate access and use of service, that there's huge incentives to build a very great service, because if the more and more ecosystem participants rely on you, the more valuable you will be. There's huge incentives for entrepreneurs and ecosystem members to support that service, but also it's predicated on

openness and participation of other parties in ways that it has not been in Web 2 and previous generations. Yeah, I would, I mean, I keep coming back to this example of cities, but maybe like One thing that's driving the incentives, or aligning the incentives, I should say, for people to collaborate on blockchain network is maybe akin to, like, city governments. You know, like, there's, the city provides this infrastructure, and then hopefully, you know, people use the infrastructure to start deploying businesses that make the city more valuable, and, you know, the city's able to collect taxes on that. And the analogy would be in blockchains that the more people are using the network, like the more valuable. the currency, because the currency is what powers all the commerce, and that's sort of like the tax, and that goes to the value of all the city's inhabitants. So there's this alignment for everyone to grow the whole ecosystem together. It's like a tie to raise all boats. Because, you know,

people can converge on these canonical instances, and there's attribution built into the system. You can look at the smart contract and figure out who deployed it. And so the result is that as this thing that I've built spreads and becomes more and more popular, there's sort of a direct correlation between its popularity and its use and the value it's able to generate if the incentives are set up properly. So one interesting way to think about this is in media there's this concept of the arts, there's this concept of derivative works

where I'll make a song for example and then let's say it becomes popular there's lots of people out there who are maybe inspired by this song and want to remix it and today it's very hard to coordinate this process of remixing the song how do I share royalties with the original creator with blockchains you can have the song

be this canonical thing that's deployed to the shared network, and you can have the rules for how this thing can be reused encoded in the thing itself. And again, it's easier to follow the rules than to break them because the rules are legible to software.

To break them, I have to literally copy the thing, change the rules, and redeploy it. So that's harder to do. The more reuse there is, the more remixes that are made of this song, the more value can drive back to the original creator. And that's like a fundamentally new thing

where, like, in the open source, the old, you know, old open source metaphor, the song gets copied a million times, like, and, you know, like, it goes viral on the internet, but none of that value drives back to the creator. And so that's, like, you know, another metaphor, I think, and this applies to anything that can be deployed on a blockchain. So whether it's a song or a piece of software, you know, it enables reuse, it enables sort of new innovations on top in a way that drives value back to the original creator. And so I think the real point here is that there's a huge incentive. to create original works, and then there's, you know, it's easier than ever to be inspired and iterate on that in a way that, you know, benefits all the participants in the ecosystem. Yeah, a lot of the conversation around copyright over the last few decades has been around trying to prohibit uses that, because the value doesn't currently accrue back to the creator,

usually. This is something I run into on Twitter where I first, when I first started tweeting, I wanted to always credit any images that I included, which, you know, it's a nice thought. But over time, that becomes really hard to do. And so, you know, I'll admit, I just sort of stopped crediting all the images that I used, which maybe I shouldn't do that. But it shows that my desire to credit the artist is there. It's just really hard to track down. If it was built directly into the image, it would happen naturally.

Right. And so this is actually something that we tried to build ourselves back in 2014 with the, that we started called Media Chain. It was literally trying to solve that exact problem. And the way that we always talked about Media Chain was a canonical content registry. That was like the internal codename for it. And so the idea was, yeah, you'd have a canonical ID for every creative work. And as this work propagated the Internet, you'd be able to append all the information it picked up along the way. Like, you know, people's comments and, you know, retweets and whatnot. And it would lead to this really rich history. And it would also serve

as a channel through which value can flow back to the creator. And so I think we were too early on this idea. Like there was no sort of, we didn't have the incentives figured out. And that was because, you know, at the time, smart contract platforms like Ethereum didn't exist. Like the idea of creating your own programmable incentives was like not a widely distributed idea. But now that we have that now that we're in the sort of smart contract era, we're starting to see a lot of these ideas surface. And I think so like NFTs, non-fungible tokens, are an embodiment of the same idea where you have a canonical identifier for a thing. It could be an image, a song, you know, anything, literally anything. And hooked to it is metadata and the metadata is

arbitrary. And it can be application specific. It can be who made it. It could be you know, it could be all the retweets, like you can start appending and expanding the universe of what metadata is captured by an NFT and by this canonical identifier, and that enables sort of all kinds of interesting, like reuse, all kinds of way to surface this information and new contexts. And so that's, you know, a direction that's generally very exciting. It's like, you know, we'll have just way more creativity and sort of reuse of works. And again, like, I keep coming back to the media examples, but I think this applies

more generally to software and composability. This is something that was explored a little bit in the pre-Internet days. I don't know if you guys are familiar with Zanadu and Ted Nelson's work. Do you guys have an idea of why that didn't take off then and sort of why you see it might be able to work now? Yeah, I think we looked at Zanadu a lot and it was a huge inspiration to kind of the early thinking and kind of block-bockchains being the reincarnation of that architecture in many ways. I mean, I think, you know, the reason it didn't happen is because, you know, HTML and kind of the architecture was these unidirectional links that you would just embed, you know, I think it was just like an architectural problem. The easiest,

you know, this like stateless kind of documents linking to each other, it just works one way. You would have a link that would go in one direction. and kind of this unidirectional backpinging propagation that Zanaddy talked about where every time, you know, an image, like the vision was that if you quote a piece of text, you are literally embedding a snippet of that document, like the canonical document, and you should be able to expand it and see, read more, and kind of just, I think it's an incredible vision, but like the architecture just was not there at the time. And then blockchains with, you know, with Jesse was talking about with these canonical identifiers,

that once something is incarnated on a blockchain, you have this one ID, and whenever anything else references to it, you reference to the original, that kind of this, just a byproduct of like this ledger architecture is you have these very, very strong links that are immutable that don't die. Well, it's shared state that was missing from Zanadu. There was no shared state, and so it was harder. The default setting was not to point back. the default setting was to copy

and to reproduce, and then the pointing back thing became like an excessive thing that you had to do. It was annoying. With blockchains, the default setting is to reference the canonical because it's harder not to. It's harder to break the rules than to follow them. I think that's the... Yeah, even like

embedding an image, you know, if I want to post an image on... I found an image on Twitter, I want to post it on Instagram, I have to like re-upload it just like architecturally. You take the bits, put them on your phone, and then you upload the bits to Instagram, just because, like, there's one CDN, which is silo from the other CDN, so you have to re-upload things, and then, but like the vision of, like, IPFS is that you, it's content address, right? You, the idea of the thing gets an address in and of itself. So I think, yeah, I think the content analogies are super interesting, and I think maybe just pulling back to just the Web 2 metaphors, I think, I think, I think what we're really describing is very similar to APIs.

I think you mentioned before, you know, so I think like even going back further, we take, something we take for granted is like credit cards, that this universal payment system Visa MasterCard that becomes this digital payment rails that could be abstracted to interact with basically any merchant. If you have a credit card, you can use, you can, the internet becomes a global storefront, you know, For a human, they can find products and everything, and the service of payment makes them all interoperable. I think we take that for granted, but even though we have these competing services,

they're all united by a payment layer. And I think blockchain services will look similar to that. And I think, you know, there's services like Twilio. You can think of like some machine learning service where you submit, you upload an image, and it tells you, you know, it's 30% cat, 40% dog, whatever, some machine learning API. But the problem there is, in the Web 2 model, APIs are black boxes.

And, you know, you have an interface that you can integrate into your application as a developer, but you don't know what happens behind the scenes. And I think what's different about blockchain is all of these services, the interface is open, but also the implementation is open. So, you know, it's not only can you, like, benefit from its utility, you can, like, pull back the covers and see how it's working in real time. And it's not even that you know what the blueprint looks like. You know what I mean?

It's like one thing if I knew all of Twilio's code was open source, I know I can inspect the actual running instance of Twilio. If it was like a machine learning model, I could know like what exactly it's learned and like tweak the parameters in real time. I think that's also fundamentally different and kind of this openness in terms of both data and the code itself and the ability for anyone to continue.

contribute improvements to it and also just reuse it, it's again, you know, it will be compounding innovation that is unprecedented, you know, in the open source movement for that. Can you give some examples of prime candidates of types of services that could be provided in this way? Well, yeah, so I think in, on the, there's, there's two examples I'll give, like there's, I mentioned some media stuff earlier, but I think what's happening right now, that's really fascinating is more on the software side. So one area that is today a risky area is what we call crypto finance.

And so you have all these emerging protocols like decentralized exchange, for example, is a set of smart contracts on a blockchain that enable people to trade tokens. This is a great, like an early primitive, and it fits with all the speculation going on in the market. And it's got some early adoption. But what's really cool is now that you have these exchange contracts, very quickly we've seen developers start to build on top of them. So they're referencing the canonical thing, and they're already building on top new things, like, for example, lending protocols or derivative protocols. And these lending and derivative protocols use the exchange functionality in order to build this new financial application or this new financial product. And again, this is all decentralized because at the root,

the base protocols are decentralized. Stable coins are another example. So the goal is to maintain a currency that's pegged to something like the U.S. dollar. In order to do that, you need to, in some cases, manage collateral that underpins the value of the dollar. And to do that, sometimes you have to sell off collateral when it goes up in price, and when it goes down, you need to buy more. And so, again, you can have a stable coin because we have decentralized exchange. So there's this whole suite of new financial products, very risky new financial products that are being deployed and super fast. All of this has happened in the last year just by virtue of starting with the primitive of decentralized exchange.

So we're starting to see this compounding effect take place. And it's finance, that's the most natural application. But I think it will extend to other stuff. So to bring it back to the media side of things, we were talking earlier about songs and having a canonical. ID for a song. And then making it possible to, you know, much more easy to remix that song. And, you know, what you end up with is something like, you know, like a universal library or universal, you know, you could end up with every song ever. And the benefit to consumers are, you know, is that now developers can compete to build the best listening experience. You know, instead of having to

redeploy something like Spotify or iTunes, you know, where everyone has to build their own library up from scratch, you just have the library. And you now, like, compete for better recommendations, like, better playlist, you know, better, like, visual experiences of the music. And so, like, it will just lead to sort of more, you know, more creativity on the part of developers, more creativity on the part of the creators who are making the content, and then better experiences for them users. What's the incentive for someone to put a song into this chain? Well, I think you know, the theory is that the more the song is listened to, the more the song is remixed and reused and sort of propagating in the network, the more value drives back to

the original or the canonical. Exactly. Like you could have a royalty and, you know, like if you set the royalty too high, like chances are your song's not going to get reused. So I think there'll be market forces that can make the price converge on the right one. But this is, I think, fundamentally better for creators in the way things work today, where it's like this very clunky manual process where you have to coordinate all of these relationships in contracts and there's tons of middlemen and lawyer fees and basically the creator ends up with nothing. Here it's like this free flow of ideas with market pricing that goes directly back to the

creator. And so I think, yeah, it's better for consumers, better for creators, better for developers that want to build new experiences. And now the creators would be aligned to try to encourage further uses of it as well, as opposed to trying to block off access. Exactly. Like, you know, that strategy never worked. Like the strategy of trying to lock down content on the Internet, like forget about it because bits are infinitely reproducible just by virtue of copy and paste. And so, you know, the key thing is it needs to be convenient. and Spotify and streaming companies like Spotify have realized that

and they sort of aggregated the library and just made it super convenient to listen to the songs. In a certain sense, they created the canonical song by just aggregating it and making it easily accessible. Here, the opportunity is to do that, that same thing, but in an open way, where there's no sort of middleman brokering access. Right, because right now,

if you wanted to build a Spotify competitor, Good luck, really. You're building from a library of zero and you have to build up this whole. Right. And so, yeah, so that's a good point, which is to say that I'm not, I wouldn't encourage, you know, blockchain projects that want to go and compete with Spotify directly, but there's this new territory that's opened up to figure out new sort of modes of creative expression that benefit from these properties and hopefully we'll see some

emergent new types of media develop as a result. And I think the key difference will be that this media will be collaborative and potentially a derivative in nature. So on the threat of crypto finance, I know something you've been thinking about is identity in the context of credit scoring in decentralized networks. Can you talk a little bit about that? Yeah, for sure. So in the world right now, there's a lot of building blocks that people have to completely recreate, redeploy. Something that the conventional Web 2.0 has made some progress on is Oath as a tool that people can use as you're building a little side project. You don't want to have to build a whole complicated login system.

You plug in Oath. This is a huge step forward, but it is not really sufficient for everything. It's centralized. It's just kind of janky. And it doesn't give you all the information that you could put. possibly want, because it's dependent on Google and Facebook and Twitter and so on, to give you the information that they're willing to give, which is pretty limited. And as we discussed with LinkedIn earlier, they could theoretically shut that off at any

point. They probably won't because that would cause an uproar, but it's theoretically possible. So something I've been thinking a lot about is what if we had a canonical source of identity information that allowed people to do something sort of similar to OAuth, but they would know that that data would always be there, that they would always have access to it, and that it wasn't controlled by some proprietary source that may or may not open it up in the future. And better yet, the developer using the system would have an incentive to improve it for their own sake if they were building something and there was something about this new OAuth system

or whatever we'd want to call it, that could be implemented better for their uses, they would have a reason to go in and add in that new feature, create a new branch, merge it, and now the whole system is better for everybody. Right, and the first application, or one application, could be creditworthiness, and then that, in turn, could enable lots of new crypto finance products. Right, so there's, again, back to this compounding factor. That's so exciting. I think another lens to think about the benefits and the risks is through trade, through transportation, innovation,

through information, innovation. We've seen like a rise of globalization that global markets make trade way more efficient, make communication more efficient. You know, electronic payment systems make it much easier for countries to trade with each other to transact with each other. And, you know, there's obviously benefits through that, that, you know, we, we, we, we, can scale up production of iPhones, for example, and, you know, ship them to anyone in the world that can afford them, you know, everyone can potentially afford a smartphone, but there's

also problems with globalization that, you know, certain countries that are kind of in a better economic position can kind of dominate the marketplace, for example. So I think a way to think about these blockchain services is that, you know, if this theory we're talking about is correct, that blockchain services, you know, maybe like information wants to be free, like blockchain services want to be like singular monopolies that dominate the entire network, that there should be one service doing one thing, and it'll be most efficient if it's just the only one that everyone uses. But there's obviously problems with that, that, you know, maybe it's not the best implementation. Maybe changing it and evolving, it is difficult and expensive, and

it exerts a new kind of dominant control on the participants in the ecosystem. So I think that's something that remains to be seen. Yeah, to pop a few things off the stack. Earlier, when you brought up the city's analogy, it made me think of exactly this issue where there's a strong network effect in San Francisco, which is where I live. But also, you know, when you walk around the streets of San Francisco, this place is falling apart. Our infrastructure is not very well designed. our governance has a lot to be desired.

But I can't even really imagine leaving San Francisco. This is where all the action is. This is where the cool opportunities are. People are building awesome stuff. So I as an individual or really any developer is caught between this choice of, well, do you want to be where all the people are? But do you also want to deal with all of these problems

that are inherent in the system? And so how do you guys see crypto networks dealing with this problem where they may agglomerate all of these benefits but then be headed in sort of a messed up direction. I think an argument the community is always using is the argument of a fork. If you don't like it, you can fork the network and start your own network. I think this goes back to the earlier argument I made about the asymmetry of open source to

establish players and big companies with tons of users to like developers that start from scratch, just because I can run my SQL, doesn't mean I'm going to get a billion users and all the data Facebook has. In the same way that, like you were saying, if you want to leave San Francisco because you don't like it, there's all this economic opportunity here, there's, you know, nice restaurants, culture, et cetera. If you start your own city, it's going to be like a pretty sad place to live. So I think, you know, something to give you mind while a fork is technically possible,

there's kind of all these kind of social hurdles to get over to actually make it functional and to get a critical mess. So something to think about. Yeah, and then aside from forking, the other proposal is sort of formal governance or some, you know, on-chain governance. And that today is, you know, is like a big aspiration for the space. It would signal that the participants using the service could, you know, change it and evolve it. But it remains to be seen whether, you know, that people can coer and avoid sort of the many attacks that you can have in voting systems. So in practice, the way that a lot of these services operate today is more like,

I would say, traditional companies with a decentralized open backend. There's still like an organization responsible for the smart contract, but their backend is open and transparent. And that, for now, seems to be like a sufficient level of trust, level of trust where you trust that the code won't change because you can audit it, but you also have some degree of trust placed in the company that sort of deployed it in the first place. And I hope that, you know, they will continue to evolve and rally the community towards a new shelling point, towards the new canonical thing. And that becomes a really

interesting sort of signaling challenge that these companies will need to undertake. So it is forking, but it's not forking to a completely new universe. It's just redirecting attention from what is today the canonical thing to what should be the canonical thing tomorrow. Right. And even if it doesn't totally solve this problem of recreating a whole new community, it's certainly better than where we're at right now. With open source, all you have is that blueprint.

You don't have the grid to go with it. Some people are leaving San Francisco because of this critical mass. So there's a point where the concentration of value and the extreme conditions that it creates actually are forcing function for people to fork off and create their own network because there's enough value in doing so.

So like leaving San Francisco for Austin or something like that, there's enough value in Austin to make it worthwhile and sort of offset the cost of leaving the mothership. Yeah, it's interesting you come back to that analogy, actually. I live in a group house of 10, soon to be 16 people. And we've jokingly tossed around the idea of, you know, we have enough income and individual people and cool people that we could hang out with. Why don't we just buy a house in like Sonoma or something or even a little farther out where it can be cheaper? And we could build our own little community that's totally

separate, which is not great because I think we want people to continue contributing to San Francisco, to continue contributing to this whole network that we've been building, and to try to fix some of the problems that are there, as opposed to escaping from them. It may be the case that it is this classic pendulum where, you know, the pendulum swings in one direction, and we had all of this forward progress and, you know, communal effort, but then if the incentives aren't designed properly, it swings back in the other direction towards fractured services. And so the important thing is designing a service where all of those incentives of the participants are truly aligned. And that is the really difficult challenge

in the space. Cool. Awesome. Thanks, guys. Thank you. Thank you.