[Music] With that, let's get on to our next uh discussion, and that's a fireside chat with Austin from the Ethereum Foundation. Um, and he's also one of the founders of Build Guild. So, please give Austin a big round of applause. [Applause] Hey, Austin. Sir, how are you?
I am uh doing okay. It's going to be a long day. It's hard to sleep when you and I'm still jetlagged. That's the problem. Um we've obviously uh worked together in different capacities for a long time.
But um this is the first time we're actually having a a more of a retrospective chat on how things have been going. And uh the thing I want to start off with is maybe have you tell people what you do and what you've been doing for the last uh five or so years and uh just walk us through how things started and uh and what they look like now. Five years is tough. There were there were some inflection points along the way and I would say that those points happened at ETH global too. If we look at like so pre209 I did things like burner wallet, metatransactions, general devril but then at ETH global 2019 I think ETH New York I think was this moment where I realized like oh like things are finally coming together and the components are there that people can start making products.
So I I would say pre209 it was more like infrastructure and teaching people how to tinker and then after that moment it was more like okay now it's about taking pieces and putting it together and building things. We saw one inch come out of that hackathon. Uh, and then I think there's probably another inflection point when you were interviewing Juan Benet. I don't even know where we were, but you asked him what people should be building and he said something like social networks and all of us were like allergic to people saying put social networks on chain but L2s had come a long ways and so that changed again like how we are teaching things and what kinds of things you can bring on chain. So each of these inflection points mean that we need to teach different things.
And I think so back in the day it was more like here's how to tinker. Here's here's how to build infrastructure. Here's you know here's solidity and how it works. Then it's more like here's all these components and kinds of things that are working and and I think we're getting to a new inflection point that's going to be more like okay now AI can build most of this stuff for you. Oh yeah.
No you know the the ideas. So the things that we teach need to change. So the things that we've been teaching are here's how you build apps and here are here's the curriculum that will take you through as a technical person you can quickly learn how to build on Ethereum and it's actually not that hard to build on Ethereum and we've been building the curriculum and tooling around that and also you know taking a few shots ourselves building a couple apps trying to get people to use them but generally I would say that it's more like curriculum tooling and then still what what Albert called like handtohand. We're still going to universities. We're still going to hackathons.
We're still going to meetups. We're meeting people onetoone and trying to bring them on chain, but trying to do it at scale virtually with things like speedrun Ethereum. This is awesome. Um I want to dig into every single of those uh those uh points, but uh maybe before that, what I'm curious about is do you have like a general idea for like how many uh people have gone through this? It's so being very ethos driven with our curriculum.
We don't collect emails. You know, if if you want to come learn how to build on Ethereum, you can straight up just like go to our GitHub and take our challenges on your own time without us even knowing about you. But we do have a developer pipeline that we are tracking uh through Speedrun Ethereum. And I would say our distribution has been not as good as I wish it was. I think that's the one like a couple things we've failed at is one one of those is like getting better distribution.
Speedrun Ethereum, the way I like to think about it is there's about a 100,000 people that have seen it. There's about 10,000 people that have started it and there's about 3,000 people that have finished it. And I think those are rough numbers, but that's maybe an initial answer to that question. I mean, this is still helpful, right? Because that's 3,000 people that you could not have reached otherwise in terms of getting them to the next step.
Uh, another metric, Scaffold ETH one has more stars than Foundry. Oh, like being at the right place at the right time with the wrong code. There's a lot of that here. Um, maybe for the audience like let's start with like what actually is speedrun Ethereum? Yeah.
Okay. So, uh, starting with Scaffold ETH the tool. Here's how you build an app. We realized that it's pretty easy to figure out how to write solidity and write a small app. It's sort of like what kinds of things can I build?
And that's what speedrun Ethereum was meant to do. It was meant to give I used to do going back to hand to hand. I used to do Zoom calls constantly with developers and just try to just like brain dump them like find a really technical person that's thinking about getting in and just eth pill them and brain dump all the concepts I need. Uh so speedrun Ethereum is kind of like uh a syn a synthesizing of all of those sessions. And so there's all these aha moments that I think that technical people get when they start building on chain that make them realize what kinds of things work on chain.
Uh one one way to kind of explain it is the the web 3 cron job. If you think of a cron job as a technical person, it's like something that needs to run every night to roll your logs or something like that. But if this is a financial app and it's working on chain, then you need to like compound interest every night or something like that and you can't just have a cron job set up on a machine that says compound the interest for our onchain app because that machine can fall over and this the system is much more resilient than that. So what you have to teach the developer is it's not just about writing the code and and the code is very important and the code is law when it's on chain but you also need to include uh you know some some game theory and some economics and some incentives and once you get the developer to think oh it's not just writing this like hard code is law code it's also like providing the right incentives for the right folks and understand that you're moving money around here and if you want that cron job to run no matter matter what, you need to make it so anyone can poke it, right? And only one person can poke it and whoever pokes it gets paid.
And if you have that system there and you have it open source and you have the, you know, poke.exe or whatever ready to go and open source, then anybody can run it and get paid for it. And that's one of those first ahas that a developer needs to have. There's a bunch more and the speedrun kind of takes you through them, but that's an example of it. That's awesome.
Actually, that's a really interesting analogy. Uh but but right now speedrun Ethereum is set up as a a collection of just interesting projects you can build with smart contracts and they kind of go from being easy to hard or more complex use cases and this is people just kind of going through this curriculum on their own trying to learn how to build more complex stuff on chain. Is that is that a good way to think about it or or what else do you add to this? So there's we we do have a pipeline and and we've made sure to kind of create this pipeline where even though we're not tracking you like when you connect your wallet we do know that your address is taking on this challenge and when you turn in a challenge and we run it in our environment and make sure that your smart contract does what it needs to do. We are still kind of tracking you.
So I think we do see Wait, what was the question I missed? No, just like a are we doing more than just kind of having them be self-guided project builders? Yes. Pipeline. The pipeline.
Yeah. Yeah. Okay. So then we have this pipeline of developers and so you're not always on your own as you kind of like So speedrun Ethereum is the top of the funnel for us but then once you get down through the challenges we do have like a batch program and within this batch program we bring you into a telegram channel and we teach you things like are you sure you have good GitHub etiquette and you can make a PR and you can do open source things. we we found the needs of developers and created this batch system and then you know an eventual grant system.
So we do have a pipeline there and that that was like an important part of bringing people in and letting people not have to be completely virtual and giving them a touch point. A lot of the touch points are just like sometimes just me on a Zoom call and that was super powerful for a lot of developers because sometimes they just need to like get their bad idea out and then and then move on to the next thing. So, it's not completely virtual. Uh, it's not completely on your own. Like, we do capture devs, but in a in a very small channel of devs that are like willing to really pop up and come talk to us.
Okay, that's awesome. Um, you've obviously seen I I would say like I think you're probably the top of the list and this is like exceeding ETH Global's cases too of seeing people actually shout at their computer when learning Solidity more than anybody else. shout at their computer like mad at it. Yes. Or just trying to figure out why isn't this working?
Uh because the handtohand part and you were like the the only person who's kind of doing most of this interfacing or historically have been. Um the thing I'm curious about is what do people complain about the most. So I I don't think it's I don't think it's solidity or shouting at it. I think if you're shouting at it, it's a little bit of a skill issue and like you maybe need maybe counterintuitive is is a better phrase. Not shouting.
Yeah. Yeah. I think the hardest thing is knowing what to build that people will actually use. And so that's way farther down the path. But but does that matter when somebody's learning something for the first time because this is more about just fundamentals of just changing how you think about it when you're used to a different programming language or a different norm.
I I think yeah, I think the paradigm is a little bit different in building on chain, but I don't think there's like I don't think it's super hard. I think people who are futing building on Ethereum are futing it for uh views. I think it's actually very easy to build on Ethereum and I I mean nowadays I sit down with a classroom and I oneshot a voting app and it we I use scaffold ETH. It has a cursor rules. I tell it what I want to do and I oneshot the app on chain.
So in five minutes you can build a voting app on chain. And of course like civil resistance and some other things are tricky there. But I don't think the syntax or learning the concepts is really that hard. I I would will say that like we're sort of teaching to where the puck is, not where the puck is going. Like we're teaching things like overcolateralized lending and stable coins.
We don't know exactly where the puck is going. If we did, we'd probably be building those apps, not teaching people how to build those apps. So, we are kind of maybe behind the puck a little bit with the kinds of things we teach, but learning solidity is super easy. I I would even say like, you know, hot take like if you're going through crypto zombies, you may not ever be the guy that or girl that's going to make a protocol or app on chain that a lot of people are going to use because you can just go to solidity by example and scroll through four examples of solidity as a highly technical person and understand how to write solidity. Then the concepts are actually very easy.
Go to speedrun Ethereum and we're going to give you like the the the 10 most important aha moments that you're going to get and they're quite digestible. I think it's farther down the line once you know how to build once you understand things that are working. It's figuring out, you know, the shot to take that people will actually use and how to get revenue. That's the hardest part. That's a I mean that's a very very interesting take because while I agree with everything I I don't think I disagree with anything.
I guess what I'm curious on is at what point is that important for newer devs compared to just having a more thorough curriculum? So you're saying like if there's a hundred devs here already and we have really good curriculum versus like a way to kind of u the thing is like I don't fully agree with what I'm going to say next either. So I'm just trying to see if I can figure out what I'm trying to get out but not everybody should be a founder. Not everybody should start a protocol. Not everybody should do everything right.
Right? So you have a collection of understanding something and most people which is like at least 90% will end up contributing to something instead of creating it if not higher than 90%. And and if you don't agree with that assumption then we we should talk about like what we can do better in this space because then you have a different set of audience and and a prerequisites that you're working for. But if you do agree with that which is kind of the premise entirely that bases E global um you have to work with that as the premise and that means most people are just trying to understand how do we manage and understand and and communicate complexity and understand how do multi- kind of party systems interact with each other and how do I get ready for that world because I am not going to assume that everybody will want to start a protocol or do something onchain and be a solo developer that is in charge of the full stack. Um, so doing more projects helps you obviously understand anywhere from what your limits are to what we know is easy or will end up being complex.
Like think about I remember with scaffold and speed of materium a few years ago like we just said let's deploy our first NFTs when they became popular and then it became let's deploy our first NFTts in an L2 cuz we didn't know really how to fully do that and then we said okay well let's think about how do we make our own lending protocol and I was like wait a second like that's a big jump like what happened like what what do you mean you want to make a lending protocol for fun it's it's awesome that you get to do that but that was a big leap right and this is cool but not everybody will need one and and for some things they end up actually benefiting if one perfect version of that is being used by everybody else. So kind of you have like an AMM like I'd rather use an AMM that's getting popular than uh uh than build one, right? The AMM is clutch, too. When you when you took a jump there from NFT to lending protocol, like the the the DEX challenge on speedrun Ethereum is to me the biggest aha because it really lets you understand what a hypersructure is. And the fact that this this vending machine like right like unis swap only had to provide the hardware of the vending machine.
Anyone can come and provide the liquidity and earn fees and anyone can come swap and the more liquidity there the better the swaps are. And for a technical person to understand that all like you're deploying these vending machines and you make the hardware and everybody else can use them both the LPs and the people I think that's a huge unlock and open that just to interrupt there and I agree right like this is this is cool cuz that is an aha moment for for me and and somebody as a developer saying oh wait a second I can write the same code or just use the same thing that's managing a trillion dollars right now like that's that's incredible as a statement you don't get to do that for I can't just integrate with a bank, but here I can integrate with the biggest bank in the world or on chain. Um, so the thing I guess I was trying to get to was so you're saying that it's not necessarily good curriculum. There's enough things out there. They're up to date whether it's the new tooling or just the the right syntax they need to know.
You don't think that's an issue? And and and that's a very interesting day because I would not have guessed that. Um because when we look at information we look at look when we look at the hacker applications they mostly are complaining about I don't I don't have the best resource they don't know about scaffold eth so so it's now a discovery problem uh and less about the quality of the content which is a really interesting take and by now it's not now the last two years there was every problem now it's about how do we teach the LLMs and how do we make it more accessible for AI interesting no that's great right so so now you have okay we we have figured figured out the the basic things that we need to teach everybody and we have a good good answer for like 90% of the people will find this very useful. Let's just get this in the hands of more and more people and AI is definitely a great way to do that. We are already seeing people uh leverage a lot of these toolings um for just general code.
Um how do we make that more accessible for crypto development from your perspective just better products making AI or making you just exactly like just helping them leverage AI to do more things on chain. I think so, uh, the cursor rules was a big unlock for us. And like LLM's.ext there, there are ways that you speak the language of the LLM differently than you speak the language of a new developer coming into the space. And I think now we're at that inflection point where we need to kind of change our documentation to be more read by AI.
And you go to Chad GBT and you say, Chad GBT, teach me how to build on chain. and it's going to basically give you the speedrun Ethereum rundown in a nice format. I think that's a great world, right? But how do you think about the situation where I think it's unique to our space which is I can't just trust the output of an LLM for a smart contract and and and that's not a one easy problem but two like this is exactly where we get criticized in our industry for money being lost and and when you're dealing with money it's it's a harder problem to solve. Um, and and if we're leveraging AI to let even newcomers just one shot for for the lack of kind of vocabulary here, do these things, I I think we create more problems than we solve.
That's my worry for now. I think for the next six months, maybe. Fair enough. Like like I I I'm on the side of in six months my laptop's going to be smarter than me, which is a little scary. My nail already is in the world.
Yeah. Yes. Like what's the test? Yeah. So I I think that like yeah there's there's a changing environment and I uh I do think that it is a little scary to think about a brand new dev oneshotting an app but I think we are at that point and maybe it's a hot take but I think that a lot of the hard problems in the last few years were you've got to be very technical you've got to be very safe you've got to have an audit or you've at least got to put it out there and let it get lendy.
with a little bit of money. I mean, you should still do all of those things probably, but I think the accessibility to building these things is opening up very broadly. And any jabron can come in and, you know, maybe not one shot, but they could vibe code, get very close, yes, they could vibe code an app that people can use. They can put it on an L2. It costs them a fraction of a cent.
Uh, that thing can, you know, balloon up in TVL. You should probably build some locks into your system. You shouldn't be using a test net. You should be yoloing everything to L2s with locks in it. Uh but I think it's far more accessible and uh the the highly technical deep down developers out there I think can still play this role of auditor.
But I think the product creators and even the you know guys and gals that are going to work at a and all the other places in the in the ecosystem I don't think they need to be hugely technical anymore. I think you need to have a systems mind and you need to think algorithmically but I don't think you need to know how to deploy uh create two with a specific salt. I think you can just kind of vibe coat that and I I know now that a little scary but I don't think it's going to be scary in six months. I think maybe six months to a year we will start seeing like when you see something sloppy you know it was a human and and the AI is much high much more higher and tighter. Yeah, we just look at the m dashes in the comments on the code and I think right.
Yeah. Um so I I do want to kind of um come back to one point and and also kind of try to wrap this up but um one thing that's sort of interesting that something you said was like people are trying to understand what are some interesting projects they can build and that's kind of what ends up being the the determining factor for them staying longterm um or just being given the opportunity. So, so it feels like it's not just a distribution and and discovery. It's also giving people an idea for what they can do beyond just make something for fun. Um, do you have any thoughts on like what we should be doing more of?
What do others ecosystems get right? Um, things we can just steal that we know are working and and why are we not doing those? I So, even though they're not all founders, I think the word for this is founder support. And I've been searching for this word and I heard someone at the EF say it the other day and I was like that that's it founder support. We need to be doing better founder support.
And I think that even if you are going to go work at Uniswap or A or somewhere else you kind of think as a founder at first and you think about you know folks are programming money I can get rich too. I want to program money and get rich and you kind of come in for that but then you stay for the ethos. But I think what we need to do better is of course distribution, get more people flowing through that pipeline, understand how to use the tools, understand the curriculum, understand the aha moments, but then at that moment when you are powerful enough to build the decks, but you don't know what kinds of things to build. I think there needs to be like almost like a human touch point. And I know that doesn't scale very well, but but that that makes sense because then you're optimizing for like higher probability of success and and it's okay to put extra human effort on that those cases.
And Ethereum, I think Mike from Rainbow says this, Ethereum is a shephering kind of ecosystem. It it is complex because there's so much stuff, but also there's all these people like competing to farm grants around you and just like, you know, toxicity around you. having a shepherd be like this wallet, go get this wallet and then this curriculum, go do this curriculum. You kind of need a shepherd to bring you in. And I think that if we can channel people through and then just have a lot of shepherds waiting for them and and have that touch point and then provide just like hella mentor and hella uh founder support where it's like giving them the next steps, the connections.
Here's all these companies. here's all these GitHub repos. Go start to contribute. Go apply for these companies. And then, you know, here's all the things that have PMF right now.
Kind of skating to where the puck is at first. Go build a few of those things and and cut your teeth on that stuff. And I think then we just need to be way better at founder support. We, you know, let people pitch their dumb ideas, realize they're dumb, and iterate on them. And I think there's there's so much to that where when I came in and I felt like I was a founder going to I don't know ETH Denver in 2018 or something like that and I felt so ignored.
I had this game on mobile on chain. It ran out of shame. It was so much fun and I showed a few people and they were blown away by it. But really I was alone and I know that like every founder feels this. They come in and they think they have a good idea and they show it to a few people just thinking that it's going to explode and like money's just going to start hitting them in the face and it's doesn't work that way.
It's so much harder. And I think getting people from I can build a DEX to I can build a protocol or I can go work for a protocol is the next hardest thing that we need to do. And I think that's where we need to be uh the most focused. I think that's a perfect message for uh for everybody in our space and and a note for us to end on. Um, Austin, thank you so much for this amazing chat and uh we'll be here.
Yeah, thank you. [Music]
Automatic transcript — names and jargon may be misspelled.
