[Music] And this is going to be a fireside chat with Buckton from Tenderly. We're going to talk about what's happening with the developer ecosystem here and how we can improve it. So, please give him a big round of applause. One, two. Okay.
Welcome. I guess the first thing I want to start off with is just maybe have you walk us through how Tenerly got started? Sure. So, uh, started on a hackathon. This is why we love ETH global and all of the other hackathon events.
So, we were this does anybody actually know where the name comes from? I know some of you from the crowd maybe know. Actually, I don't think even I know this. I thought Patricio Negger know. No.
Okay. Uh we started doing public procurement on chain at first and in Serbian that's also called Oh the tenders. Yeah the tenders and it was the first blockchain hackathon in Serbia and we made like this commit and reveal thing because corruption is a is a thing in Eastern Europe very profitable business and then basically what we did there was trying to fix public procurement on chain. So it was a hyperledger hackathon actually. So we started working on that uh won that particular hackathon and then tried to do that for like four months or something like that and then we I'm going to have to curse but we realize we don't give a about public procurement to be completely honest but before that uh we were working in a developer tooling company and actually the first version of Tenderly was an open source tool to extract the stack trace from a solidity smart contract when something fails.
Uh and then we were like we're going to change the name and yeah bugged them from tenderly. So that's how that worked out. Amazing. It's just easier to keep the same email. It's we've learned that from experience.
Um, walk us back through kind of seven years ago. Actually, maybe more I think it was more than seven years ago now. Um, no, actually about seven years ago. Seven. Yeah.
Um, what the the world was like in terms of actually building EVM tooling and um and kind of how do you think that's changed so far? Um, stuff was so tenderly started as an open source company. I mentioned that particular piece and one of my favorite fireside chats I mentioned this to you last night was actually with eager where we were talking about if you remember eager in on the op fire far side because block scout is fully open source and tendril is fully closed source but yeah so 7 years ago when we started we started as an open source company actually but we were like there's seven companies in the world that came into existence from open source it's four of us engineers who know nothing about business what's the chance that we're going to be company number eight that succeeds with this and then we close source and try to innovate that way to answer your question directly. Uh the difference then there was uh we were trying to basically replicate everything that we used in web 2 because I'm a distributed systems engineer by trade. Uh and none of the stuff uh worked.
So how debugging worked is if your solidity contract failed you would see it on etherscan on like coven if if it were on the bleeding edge but probably some ringy or some other network and then uh or gorly I think actually came out or ringp. Yeah. But basically, you would go there, it would say invalid op code and and a smell silhase that I hated on ether scan. But and then you would have to delete lines of code until your smart contract worked. That was that was debugging.
There was no console log. There was no any of these things. Uh and this is why we did the first stack trace and I honestly think tenderly came into existence. There was this tweet that I saw like uh Greek philosophers had it easy. If nothing existed, I would have invented something as well.
So we basically just started inventing stuff that we needed from web 2. Uh and there was a couple of other companies that were doing it. I mean it was Remix, it was Althia from consensus for for monitoring. Open Zeppelin did like like open Zeppelin is literally the standard library of our of our space and we were trying to do the same thing like tools that we were we were missing. We were trying to replicate that and that's how we were doing product work.
So what are the products now and and kind of what is the new state of the ecosystem that's actually enabled by the teny product suites? I think the interesting uh h so I'm I'm going to try not to make this a tenderly shill. Who here heard about tender or has used tenderly? This warms my heart actually. Okay, cool.
Uh so there's the developer tooling suite which most of you use in the simulation suite. There's the infrastructure like with RPCs. There's the testing infrastructure piece with virtual test nets which a lot of people use and there's a new security vertical actually as well. Uh now when you sign a transaction on a ledger device with a larger screen, we're one of the providers that does the security scan. So we like prevent hacks on the wallet level.
And I actually do agree with Martin, he's not here for me to get Martin points, but I do agree with Martin that that should be done actually on the sequencer level mid to long term. Uh but yeah, the the difference in the system is right now that I think we're stuck in a this very weird thing where Tenderly, for example, is the default developer tool. And I think it's very hard for anyone to make a developer tooling company now because with all of the L2s and everything else, we're all writing to the same database. So in normal web two terms, you create a developer tool and there's like zero cost to running it. Here it's like obscene amounts of cost because of full archival nodes.
We're talking like literal money, not not the time alone. We're talking literal hundreds of thousands of dollars a year at least to to run like the basics of developer developer tooling, not including humans and anything to actually actually build the thing. So, it's very weird. I think that's why like Foundry and some other teams what they're doing is amazing but the whole DeFi money Lego thing and everything else we're highly collaborative space and developer tooling in general like even in web 2 isn't collaborative in nature like nobody is or is not is not I think oh that's interesting because I I I would have uh if there's anything we've seen open source software outside of crypto is way more collaborative no no sorry I misspoke uh no no so the development itself is highly collaborative the usage of the tool is very single player. Understood.
I agree. That's what I meant. That's that's literally enterprise software. Yes. Yes.
And the thing here is because we're all the way I like to explain blockchain to my web two normies friends is uh it's like a MySQL database where I can write directly into your table. It's it's highly weird paradigm. Uh sometimes that actually is. Yes. In PL/SQL one only PL/SQL now is solidity.
But basically uh sorry for the dry humor. you're getting my full personality with cursing and everything else. But um basically it's extremely hard and then our tools need to be highly collaborative by nature. So Foundry for example falls short the moment I need to show you a bug I have locally. I can pick two things.
I can do a Foundry test which most people use and then share it with you. I can go with my laptop to you and if like in office in an office or do a zoom call or I can use something like a tenderly virtual test net to actually share it. But this isn't a tenderly shell. Like because we're all writing to the same database tables, our tools should behave as such. And I think that's why it's a much harder even if we just look at it from a purely cost perspective, running a dev tooling company or any type of infrastructure because it's also up that in our space developer tooling is actually infrastructure.
We only see that now in web 2 with GitHub copilot and similar tools because that objectively is infrastructure. It's running somewhere else. It's not on your computer. uh in in web 3 that was by default if you actually want to get any of these stuff. So so I mean that's like a very very interesting point and I actually do want to spend some time understanding um at least the current state of developer tooling for uh for Ethereum.
But uh maybe a good way to sort of get to that would be understanding like what you think is uh is wrong with the current developer ecosystem in Ethereum. Whether it's uh from the point of view of uh we could do something better on the language to we just don't have the basics that we really need with web two as a comparison to just things that you found people just keep complaining about. Uh but turns out that there are too lot of your control because it's a infrastructure or VM issue. Um loaded question and I could I could probably talk for but let's make some enemies today. Yeah.
Yeah. No, it's not even that. I think even though we have technically let's take Ethereum a minute we technically have two languages right now so solidity and viper that people you can write contracts in Ule you can also etch code on a piece of rock like I understand that you can do stuff like that but uh the thing that's basically hard here is like solidity isn't lacking generics solidity isn't lacking something for us to build apps at all we're lacking a couple of things being content with the PMF being finance I think that's one thing and I think the second thing is then the tooling reflecting that fully and not and all of the languages and everything else. I think what the solidity team has been doing recently is amazing. Uh I really trying trying to use every public possible situation to say that if if uh debugging symbols are shipped in solidity we will lose third of our revenue but this the ecosystem will thrive and we're really actually trying very hard for that thing to happen.
Um, and then I honestly don't think TPS, that's my unpopular opinion, but I don't think TPS is what's uh what's preventing our space from succeeding. Do you have an answer for what is? No, no, no, no. I would I would make an L1 if I had an answer. So, we're just here to support people who are trying to figure it out.
So, yeah. Sorry. Fair enough. Getting a truthful answer. No, this is this is good.
Um, okay. So, we've all I guess I'm answering this as myself. Actually, can I stop you for a second? And I might actually have an answer. I think we just overpromised the timeline, not the value.
I think that we just expected that we're all going to transact with stable coins with Black Rockck and I'm going to send my mom like USDT instead of Serbian dinars or whatever. That is going to happen. We just have to see if my parents will be alive at that point or if or if I'm going to be alive at that point. But if it's it's it's inevitably going to come. We're coming to that situation.
The whole point with a generational leap like this is to either similar to AI for the value to be extremely apparent or for the old system to realize that or for us to fool the powers of old that this is the same thing and they can do the same up stuff they have been doing so far only here. And then it's like aha we actually fooled you. We can now see what you're doing or we can actually prevent you from doing those things and stuff like that. I like that take. Um yeah, perhaps uh we can simplify this by uh by putting crypto on a phone or something.
Yeah. Is the DJ one holder? Yes. Um so I mean I've kind of seen the beginnings of the developer ecosystem here and kind of where we are now. And you've kind of seen waves of all the dev tools that were popular and uh in sort of what they got right and what they didn't get right and we've kind of seen the whole from truffle to uh hard hat to Foundry.
Now, um I guess I'm curious about understanding from your perspective, did those things get better significantly to help actually improve or do you think they were just kind of the the changes because of this is just what's cool now and we'll just kind of take it and and we've just settled on the next best alternative because others are just not fast enough in improving their code. I think hard hit was better than truffle, but I think truffle didn't have to die. I think Foundry is better than hardhead, but hardhead didn't have to die. Uh, I think Waffle is alive because people don't like Python, most people. And that's like people, and by the way, Waffle is amazing.
I just hate Python. So, this is why I wouldn't use No, no, I'm not joking. It's like an amazing amazing framework. But, uh, basically to answer your question, I think we as a space are for the better, but I think we would have been in a much better situation if I don't know. I had this Twitter interaction with Nick actually, who was part of of Truffle.
Um the truffle decoder is amazing like we were actually scared as tenderly that somebody might reuse it to rebuild some of the stuff that that we built. That was completely ignored like by the most of the ecosystem. And then hardhead did some similar stuff and it was completely ignored by the foundry or actually don't know I don't know the in inner workings of these but yeah you don't you don't see it at least from the outside in. Uh I think we constantly forget that somebody did some work as well and that we can learn from it. So that's I I think so I guess how do you think about using those same tools and kind of building these I would say like additional collabor collaboration layers uh for the tently product versus kind of seems like you're kind of creating a parallel version of these tools that are closed source.
Um what's kind of a good way of thinking about actually u being part of the open source community too. I think it's hard. Um, so there's a couple of things. We definitely want to, and this is like one of those don't hold me accountable to this statement type of thing where the marketing team yells at me afterwards, but we do actually want to open source some stuff. Um, it's less of a technological problem.
It's a purely business problem. So what the problem is there that even the joke that they made is that I got approached by so many people in the conference here that they're like, "Ah, Tenderly is so successful. If we were as successful that as people think in this room from a monetary perspective uh I wouldn't be sitting here. I would be sitting somewhere somewhere else. Now what do I actually mean by this statement is that building public goods and developer tooling as public goods is expensive especially in this space if it makes sense.
And the reason why we have to have stuff is closed source is because the system is still growing and we're not a proven business. We are looking like like we have all of the trajectories and signals of a proven business, but are we a generational company? I don't know. I hope I I have to think as the founder. I think I think you're obligated to say yes in this one.
Yes. Marketing team. Yes. But no, basically the thing that um that's hard here is that like why is Foundry successful? It's because Paradigm has $2 billion.
Jorgus is Georges is an amazing leader and the engineers working on that are amazing. But now you see Ithaca and you see uh foundry development actually slowing down a bit. They are hiring so they will continue but by the definition these are hard problems and there's a finite amount of engineering people who can tackle these things. So unless you have a for-profit company like Tangerly which has to use closed source to also is that that's the job of the company and the team. Yes.
Yeah. So it it's a very hard line to thread. Uh I know that I'm not the founder to run an open source company because like all of you that raise hands wouldn't have raised hands if we try to do it in an open source way. Uh and we're going to see how it plays out. I honestly hope and we're really trying hard to work into a future that we don't depend on the developer tooling piece as a company at all.
So we can just open source it because like the Ethereum ecosystem would probably thrive more. But if we do it now or at least if we do it in a dumb way that we would probably do it to be completely honest, we would not have Tenderly. But you would be sitting here with another Eastern European founder saying, "So we had Tenderly, now we had you." Do you think we're for the better? It could be Americans, too.
I don't know. It could be. Yeah. Yeah. Um there's been a lot of experimentation on the VM space um the last couple years.
Uh EVM was kind of the only thing that that was uh was out there for for a few years. Now we have 20 different flavors of it. Um, how's that kind of uh changed your thinking on not just your own product, but like just overall like what do you think is good or bad about these uh people trying different VMs? I think experimentation is good. Uh I'm not one of those that thinks that the Ethereum mainet is moving too slow.
So I think that we're moving for just solidity like the VM or VM solidity. So all of the stuff like something not going down for 10 years is amazing like people are flexing for Erling for example Erikson always flex that Erling had nine nines of availability but then it's like a few aster points Ethereum actually has that particular thing and I think it's amazing and I think we should never sacrifice that for speed. I think it's good that we have L2s that can experiment. My pet peeve is that EV compatible usually doesn't mean EV compatible. So then we had E.
So I guess tell people more on what the difference is and the nuance. So uh for if you're an RPC provider for example, you just stand up nodes. I mean it's not just it's actually an extremely hard business to run operationally and from a DevOps perspective as well. But there's no custom work that you need to do for the virtual machine itself. Now when somebody does even something simple like writing a pre-ompile uh that's not EVM compatible anymore.
It might be even compatible from a perspective of solidity but from a perspective of an any like us specifically. It's not because for tenderly to do a simulation for you where by the way most of the Dows in this room depend on that if they're going to move millions of funds. If we are not 100% compatible with those changes, we're actually going to surface the wrong thing in the governance proposal and then somebody fat fingers like $50 million. I I don't think people hypo hypothetical. Yeah.
No, this actually never happened just to just to be specific. And by the way, this is why the sequencer should protect you. But anyway, um going back to this. So then everybody said we're EVM compatible, which actually wasn't. It it was true, but it stopped being true very quickly.
And then we added the term EVM equivalent, which was true for four and a half months, and then it also stopped being true at a certain point. So what's the difference between equivalent and compatible? Uh I have no idea to be honest. At this point it's both are not EVM basically. I think uh teams like arbitum did a very good job of like doing the EVM plus where the uh stylus so stylus basically gives you way to run web assembly on on on arbitrum orbit chains.
I think that was good because like it's a virtual machine on the side so we could actually ship support very quickly because it's a thing on the side that we can choose to ignore. We didn't but it made prioritizing as a as a infrastructure and tooling provider it made it easier. And then for risk five, for example, I also hope if we do end up shipping it, which I'm not sure if I actually want to happen, but if we do end up shipping it that we're not doing the replacement, but the sim the thing that Tomas said basically of more sidebyside virtual machines and then you pick your pick your flavor. Do you have any uh general thoughts on what specifically can be done better in solidity like as the language spec itself? Or I guess maybe let me let me ask this this way.
What have been the most common complaints that you get from customers? Uh contract verifications. Uh which is very weird. So it's a core part of our work. Uh I don't know egot if your stomach also cramped when I said contract verification.
And there's this huge push by the verifier alliance as well. There was sourcifi as well. Uh and I still don't for the life of me understand how it is such a hard problem because it's moving JSON from one point to another point. But it's still hard like when you abstract it to to its core. Uh I think that's the biggest problem right now when you look at what people care about and it's a very dumb thing to not dumb thing to care about.
It's a dumb thing for it to be a problem but it is a problem. So break that down for me. Why is that important or and also I guess what's particular piece is important? The fact that you just want to be able to say that this contract source code is here and it's verified and just the check mark that matters or do you have additional benefits from having that information accessible? I think there's two benefits.
One is the okay I'm actually interacting with this thing and I know what this thing does which most people don't check but then you have tools like I don't know our transaction scanning and some other providers which actually use that to make that informed choice for you. But so one is the verifiability of the thing like when I'm using unis swap that I'm interacting with unis swap contracts. That's that's one piece but the second piece is let's go back to the I'm writing into your own database table. When I'm integrating with Stripe, I have their documentation and I don't particularly care what happens in the background and then I get an error message in solidity. I integrate with um let's say I'm writing a unis swap hook or whatever and it blows up.
I need to look at other people's code which is also a very strange thing for people when they come into our space and suddenly I need that source and contract verification becomes becomes important. It would be similar that when you integrate every possible thing like chat GPT something doesn't cough up normally here you would like prompt and try to debug it by interacting with it. Here the analogy would be like going into the model and seeing what what's happening. Nope, that makes a lot of sense. Um maybe I guess to uh kind of wrap things up.
Uh I'd be kind of curious to hear what advice you have for uh anywhere from engineers to founders that are kind of starting to build um more complex code on chain. Um what are kind of some of the lessons you've learned over the years that uh you think most people end up sort of realizing too late or realizing that once they know this is a problem it's too much time to correct. Um what can we kind of get ahead of as we're building more complex code? Do most stuff offchain. I honestly think I mean I made that joke last time.
My toxic trait two years ago was uh drinking two beers and going around the conference asking people what they do and then explaining how it can be done with my SQL. Um I mean this is semi semi semi- true but basically uh blockchain should be used for verification much more than for execution. That's why I also don't think the high TPS stuff makes sense and we are moving by the way in that direction with all of the zero knowledge stuff. I mean that's literally literally that particular thing but like I don't need a to-do app on chain. I don't need these things on chain.
In a way I agree but that's not where people are trending. Um so what's a good way to think about balance or just even being more mindful of is there a good framework that helps you say do this or that uh when faced with this particular type of problem or just blanket put data somewhere else only put proofs on chain. Uh I think I'm not the correct person to answer the question. So, this is like while I agree um with gas limit increases, with just more DA kind of being the norm, that's not where people will naturally trend towards because if anything, we're giving them more reasons to not do that. Um and we're making it even simpler.
Um so, how do you kind of still think about Uhhuh. Okay. what to do when everything's going to 10x in capacity. I think it also depends in which reality we're going to live in. uh if we keep the same amount of developers but increase TPS then the incentives are going to be as such for us to do as many things on chain as possible if we grow as an ecosystem which I would love for that reality to exist the thousandx more TPS isn't going to matter like for the sake of like you as a developer now are going to have literally the same experience as a developer like in 5 years one 1000x if we onboard enough developers to eat global and everything else if we don't then yes we can do every button click in a video game as a transaction.
Sure, I really don't want to live in that particular world world unless Tenderly like charges for each of one of those transactions which it unfortunately won't won't happen. It's a good world. Or or the second one is we actually use offchain compute for what it makes sense like T and by the way when I say offchain compute it's like TE it's like VR ZK proofs and like all of these things do it offchain and then post stuff onchain and that's it because we don't need a history of everything that happened. We know to know, we just need to know if something happened. If that makes sense.
Amazing. Well, thank you so much for for this amazing chat and uh please give him a big round of applause. Thank you so much. [Music]
Automatic transcript — names and jargon may be misspelled.