Vitalik Buterin at ETHCapeTown

Sure. So I first found out about Bitcoin back in early 2011. The way I got into this space was I found people who were willing to pay in Bitcoin for writing articles about Bitcoin. First there was this blog called Bitcoin Weekly, and then soon after that, this guy from Romania called Mihai Alisie reached out to me saying he was founding a Bitcoin magazine. I became the head writer of that and did that for a couple of years while going through school and starting university. Then about two years after that, I kind of went full-time into the space and started working more heavily on a lot of different Bitcoin projects. Back then, Bitcoin projects were basically all that there was. I started looking into colored coins and some of the more powerful protocols at the time, things like Mastercoin. Eventually I realized that there was an opportunity to make something much more general-purpose than what already existed, and that was where my very earliest proposal for Ethereum came from. Basically, the existing protocols at the time, like Mastercoin, said here are 15 different things that you can do, 15 different types of transactions for them. Instead, I said, well, you just have a programming language with one type of transaction, and inside that programming language you can express the logic to do whatever you want.

It definitely just seemed like obviously the correct way to go. At the time, the number of applications that actually had users aside from Bitcoin itself was basically zero.

Interesting enough, we've kind of come full circle in a way. The very early versions of Ethereum, almost proto-Ethereum, even before the name came along, were trying to generalize Mastercoin's financial derivatives feature. Then after that, the set of applications expanded and people started talking about non-financial applications, like using blockchains to manage decentralized storage systems, domain names like ENS, identity management, and then all this enterprise stuff started happening. In the last couple of months, people have been really excited about decentralized finance. So I thought that was kind of fun, just a lot of topics we're going to talk about.

It's definitely changed a lot, especially just because of how the scale of the project increased. Five years ago, it was basically myself and a couple of other people figuring out everything: the protocol, implementing it, bootstrapping the community, setting up the Ether sale back in 2014, and doing executive security audits. There was this fairly small team that did everything. Over time, the number of people in the ecosystem went up and up. We've had more layers of abstraction added on, so it grew from one team to this big Ethereum Foundation containing multiple teams. Then last year we launched a grant program, and now probably the majority of things the Foundation funds are outside of the Foundation as an organization. So many of the things I focus on have become a bit more high-level. Though there's always some of that. Back in 2014, I wrote a blog post about Schellingcoin, which was this idea for a decentralized oracle, and Augur ended up taking a lot of ideas from that. We talked about a lot of the DAO concepts. Then after that, I focused more on getting proof of stake and sharding figured out. Right now, I basically think that most of the research problems around proof of stake and sharding have been solved. So now it's at the point where the spec for at least the first phase of Ethereum 2.0 is in finalization stages. Phase one is not far behind, but still nothing fundamentally unsolved. Phase two is the thing we're currently focusing on piecing together. I've been heavily involved in some of that, but I also spend my time thinking about more high-level things around DAO design, economics, governance, and all these other topics we'll take a bit deeper into soon.

In the short term, I think the decentralized finance stuff that's been happening is providing things that are useful and have value to people. We've always had this idea in the space that finance is going to come first. The usual reason why I think that's the case is just because if you compare it to existing technology, outside of finance, the sector that's the most terrible is finance. In many countries, even really developed ones like the United States, you'd be surprised just how insanely inefficient it is to move money between accounts. Once you start talking about international payments, it gets even worse. But anywhere outside of payments, people are already used to being able to spin up whatever they want in one click. So the bar that anything decentralized blockchain has to compete with, purely from a usability standpoint, is higher. But I think the technology is starting to get better, to the point where things outside of internal financial applications can really start taking off.

It's definitely a broad category. Most people would say MakerDAO counts as DeFi, Uniswap counts, but these tend to be fairly pure applications that happen inside the crypto space with not too much connection to things outside the blockchain, except maybe over the price feed because you have to connect to the actual value of a dollar. But there are projects starting to poke out and make things that are financial but also more closely connected to real-world things. For example, there have been a couple of projects trying to do blockchain smart contract based insurance. These tend to be parametric insurance applications. They say you put money into a smart contract, and if a flood happens or there's not enough water or the temperature is above some value, you automatically get a payout. There's a project called Hurricane Guard trying to do this for natural disasters, and something in Sri Lanka trying to do it for crop insurance. So more and more things are happening in that area.

You can definitely improve collateral efficiency, but you could argue that the collateral based design of the whole thing is an advantage in some ways. In terms of improving collateral efficiency, one of the interesting things we can think about is that this kind of model exists in Africa and other places too. There is this kind of rotating clubs where people coordinate to put money into a pool. The designs differ, but basically it helps people save. Potentially you could also turn it into an insurance application. If you take these social models where you have groups of people participating together, often they can reduce the amount of capital you need because you can share more.

Identity and reputation is definitely something that's very important. It depends what you mean by identity and reputation because the word can mean five different things. One meaning of identity is just being able to prove that the person trying to spend the money is the same as whoever received the money. This is just user account security, but it's basically an identity problem. The challenge is that the default approach is you just have one key, but if your key gets lost or stolen, you're screwed. I'm a fan of this social recovery approach where you have five friends, each with a share of your key, set up so any three can recover your key if you lose it. You can do a similar thing for key revocation. The nice thing is that if we can create a robust identity management solution, we can migrate it beyond just blockchain to use it as an identity system for off-chain applications. That's one kind of identity. Another second kind of identity is trying to create a token where it's easy to get one but extremely difficult to get many. This is useful for voting schemes and a lot of applications. You could definitely use blockchains as an ingredient, but there's a big design problem of how to actually make these systems. The third kind of identity is reputation, which is not just about proving you are something, but proving you have certain properties. This could be other people attesting to your trustworthiness, or even organizations saying they've verified something about you. Governments saying they verified something about you. Then we start talking about this idea of an ecosystem of identities making claims about each other, which you can definitely leverage for reputation and credit.

There's definitely an opportunity for blockchain based projects to take away in areas where existing infrastructure is either very far behind or just doesn't exist at all. There are parts of the world where you can say, what's the point of this when you have centralized apps you can use to send money to other people in a couple of seconds? But that's not something that exists everywhere. Even going beyond money, things like insurance, giving people the ability to invest in things, and all these other services that go beyond just sending money. Also outside of finance, identity and reputation. I think these socially driven models of identity could potentially leapfrog centralized models of identity in places where centralized models haven't been fully established yet. That's interesting because the information about who is trustworthy and who is willing to vouch for whom is a pretty decentralized thing. The concept that it should be centralized into these formalized institutions and numbers is itself pretty new, so that's also worth looking at.

I've already said a lot of the things I care about. I especially want to try to push things that either move beyond finance into these other areas, or combine elements of finance with elements of these other areas. The other category I haven't mentioned enough is supply chain related stuff. That's an area I'm less familiar with exactly what people have done so far and what the benefits are, but it's definitely worth exploring. In general, innovative approaches to solving these problems in terms of people's ability to connect with each other and trust each other in a verifiable way. Another thing I care about is trying to do all of those things while preserving privacy. Zero-knowledge proof technology has advanced forward leaps and bounds over the last year, and there are tools to start working with it. You definitely want to take privacy seriously in these applications. You don't want the fact that Bob trusted Charlie 60 points, then Bob decided to decrease his rating to 30 points, to be on a global public ledger. For a lot of these applications, there are ways to do zero-knowledge stuff to get the benefits without uncovering everything. One example I like is in cases where you want basically one of the things I dislike about the path many parts of the internet are taking is this trend toward using centralized identity as a crutch for everything. Often the problem you're trying to solve is much smaller. For example, if the problem is just making it difficult for people to create ten thousand fake accounts, you could solve that by capturing everyone's identity, but then you break privacy. There are ways to do it where you still preserve privacy. If you have some existing identity system, on top of that you can put a mixing gadget, and use zero-knowledge proofs to prove you are one of these approved identities without revealing which one. You also reveal a magic number so you can't reuse that identity without revealing which one you've consumed. I'm also pretty interested in that.

Sure. For those who haven't heard this before, Ethereum 2.0 has two major flagship components. One is Casper, our proof of stake algorithm that replaces mining proof of work with something much more efficient. The second is sharding, a massive scalability improvement because you don't need every computer in the network to process every transaction. It splits into three phases. Phase zero does proof of stake. Phase one does sharding for verification of data availability. Phase two does full sharding including verifying transaction execution and smart contracts. At the end of phase two, it's a complete system. But phase zero and phase one are already useful for some things. Currently, the phase zero spec is very close to completion. Phase one is further behind, but the fundamental components are all in place and it's just an editing job. Phase two is still a bit further away, but we're rapidly working on it. In terms of client implementations, there are multiple teams, mostly not part of the Ethereum Foundation itself, though many are supported by grants, building clients that would run as part of the Ethereum 2.0 network. The ones furthest ahead are probably Lighthouse, Nimbus, and maybe Prismatic. Nimbus was the first to publish a testnet, and Lighthouse is going to publish one fairly soon. So it's definitely getting to that stage.

With sharding, the goal is to get about a thousand-fold increase in scalability. We do this by not requiring every computer to process every transaction. Each computer processes a small portion of transactions, which removes the biggest scalability bottleneck. That's one of the things we've done in some of our previous chats. There's a section called Underrated or Overrated where I say a topic and you comment on whether it's underrated or overrated. So let's start with DeFi.

Definitely slightly getting overrated at this point. But the parts that are underrated are the ones that try to push this stuff from theory into practice and figure out what it actually means to provide value to people even outside the crypto space. How do we realize that value? That's something I'd like to see more attention on.

Probably I'd say people talk about them enough at this point.

I think BSV is still overrated for as long as the market cap has multiple digits. Bitcoin Cash, I think, is actually underrated at this point. If you follow the community, they've become considerably more sane since they expunged the Bitcoin SV people. They're getting Schnorr signatures ahead of Bitcoin, and they have real technical talent. It's getting interesting.

Forks in general, I'd say underrated. Especially if you want to launch a new decentralized chain, forking an existing one is even better than doing an ICO or whatever. You get a much better distribution, tap into an existing community, and fewer people hate you. It has a lot of benefits, even if you do a fork without a pre-mine. That space has been underexplored relative to other ways of launching projects.

Depends which circle you're talking in. If the circle is crypto, then definitely overrated. If the circle is Ethereum, honestly overrated as well. Especially in the short term, when talking about governance of Ethereum, most of the things Ethereum needs to survive and prosper in the next couple of years are not really contentious governance issues. We know we need Eth 1.x, we know we need Eth 2.0, Casper, and sharding. I'm not convinced that governance itself can improve things by that much. What can improve things is ecosystem funding.

Definitely still slightly underrated at this point. I'm really happy that the Ethereum community is starting to tackle these issues of public funding for public goods in the ecosystem seriously, and experimenting with different approaches that go beyond the Ethereum Foundation. It's definitely healthy that the Foundation is not the only source of funding, because every organization has weaknesses and blind spots. The more we cover for each other's blind spots, the better chance the ecosystem has.

Depends which aspect. In general, the model has been very successful. The original goals were to push the job of developing these 2.0 clients outside of the Foundation, and the Foundation would develop one client, the Python client, as a reference implementation, but deliberately written in a slow language. The rationale is to prevent the Foundation from becoming too big a point of centralization. In Eth 1.0, we noticed it became this kind of Geth and Parity duopoly. The fact that it was ugly and not a monopoly actually saved us during the denial of service attacks a couple of years ago. But it would be healthier to have even more clients, to have that balance of power in the community, and also to insure against the risk that the Foundation will have internal governance issues and not be able to get its act together. It's one of the incentive strategies: support different approaches and hope that if even one of them succeeds, things can get off the ground. The Foundation has done quite well over the last year and a half in many ways, and the grant program has been one of the biggest successes. However, I think the Eth 2.0 client ecosystem is doing a bit incorrectly in that the level of decentralization has gotten to the point where it's duplicating things too much. We have Nimbus, Prismatic, Lighthouse, Jasper, Pegasus, Trinity, Harmony, and many others. Going from one to four or five is good, but these last five, there's no way we're going to have enough resources to push all ten to the best level of quality. These implementations should try to specialize more in different things. For example, one could specialize in clients, another in developer tools or validator tools. They could even share some components. Instead of having ten things copying the same thing, they should branch out into different areas and philosophies. A light client first approach, mobile first, or light clients for Ethereum inside other blockchains. There are definitely different things you can do, and being more creative and imaginative in that way would be interesting to see more of.

First of all, Asia as a region is starting to do better tech-wise. I was just in Korea last week, and there was a team working on a pretty novel way to build a plasma implementation, and they got pretty far. So it depends on the region. Some regions are handicapped by language. The resources are mostly in English, and if you don't speak English as a first language, that puts a huge barrier to participating. Supporting translations is one thing the community can do. Otherwise, education in general. But if people have the knowledge and skills, a lot of teams are pretty global and would be happy to have contributors from Africa, India, or South America. But there's a long way between the status quo today and everyone who could potentially be contributing having that knowledge. Closing those gaps is definitely important.

It's definitely one of those double-edged swords. A big part depends on the application and what you see as the benefit. But to me, a big part is just empowering people not just as consumers but also as producers. Blockchains do that really well because it's not just that anyone can use an app, it's also that anyone can go ahead and build one. That's something I think is really valuable and we should try to preserve.

The history can only grow, but what we and the client developers are thinking of doing is pruning history, either not storing history older than a year or moving it to a different data structure like IPFS or some kind of sharded data store. That would reduce the size of the state, the amount of information needed to validate the next block. There are proposals to decrease that, like the rent proposal or a stateless client proposal. So there are definitely plans to reduce the storage requirements of an Ethereum node in multiple ways.

As a developer, the environment is going to become more asynchronous. Especially if you have an app that scales to more than one shard, you won't be able to communicate between shards the same way as inside one shard in Eth 1.0 today. Coming up with programming patterns that deal with that reality is going to be one of the challenges. Also, I expect data to be cheaper relative to execution, so approaches that are more data heavy and less execution heavy and state heavy on chain will start making more sense. We're hoping block times get faster and more regular, and that short range reorgs become more rare, so the experience will be closer to what you can expect from a centralized system.

My favorite meme specifically the one you would choose to represent yourself? I don't really have a list of favorite memes. I'm supposed to have a list of memes on hand? I don't keep up with meme culture anymore.

Ultimately, the existing centralized systems and new centralized systems that try to take just enough elements of blockchain to look cool but actually are the same old stuff we've had for decades.

Ethereum is definitely not by itself optimal for every use case. There are plenty of cases where you just want to build centralized things. There are cases where you need decentralization but not a blockchain. There are cases where you want some really high-performance specialized thing, or something that connects into Ethereum. So there are different trade-offs for different applications.

There are definitely things they can do easily that they're not doing already. The simplest one is just banning exchanges. If you make it five times harder for people to get cryptocurrency, that makes it significantly harder for any application, even non-financial ones, to get large scale adoption. Will the space be able to survive if most of the world does that? That's an unproven hypothesis. If they wanted to, there are different ways they could attack it, but then there are ways these platforms could harden themselves. It starts becoming pretty complicated. I'd say it's definitely a cat-and-mouse game.

I honestly think there's not much need for those. The only kind of sidechain that really makes sense is the non-custodial ones like Plasma. Plasma is basically the word for the category of non-custodial sidechains. I'm definitely bullish on those. Otherwise, things that require trusting specific groups of nodes where it's difficult for users far away to determine whether they should be trustworthy, why do that?

Definitely the biggest risk is not being able to build things fast enough. The usual thing people say is if we don't build fast enough, some other blockchain competitor will take over. But the other risk, potentially even larger, is that if the space takes too long, existing centralized systems will just continue improving themselves, and we'll have lost a big opportunity to make enough decentralized applications and data layers that actually become the foundation for a generation of applications built on them.

Decentralization means lots of different things to different people, and it depends on what application you're building. Some of the biggest benefits are trust. If you're building something intended to be infrastructure or a platform that other people build on top of, that should be decentralized because it's the only way to really assure your users that you're not going to pull the rug from under their noses when they've already built a platform. Even the fact that we use the word platform to refer to centralized things maintained by one guy is kind of sad. If we're building things that entire economies end up relying on, it should be common infrastructure. Once it's built, there's not one party that can get hacked or become a monopoly and start extracting rent, or get captured by some political interest that represents only a small subset of users. Having properly decentralized base layers for many things makes sense. I'm also generally interested in this idea of building new kinds of mechanisms and ways to motivate people to collaborate, whether through economic incentives, social incentives, or a combination. The more decentralized approach is how you get better information efficiency. Most of the information about how trustworthy people are, what things are valuable to whom, doesn't exist in central servers. It exists split up among all 8 billion of us. So these more decentralized approaches that allow people to participate in all parts of the ecosystem are really valuable in that sense.