Unchained - The Chopping Block: Anatoly Yakovenko on Why Solana Is Building the SAGA Phone - Ep. 373

Episode Date: July 14, 2022

Welcome to The Chopping Block! Crypto insiders Haseeb Qureshi, Tom Schmidt, and Tarun Chitra chop it up about the latest news in the digital asset industry. In this episode, Anatoly Yakovenko, Tsar of Solana, joins the show to discuss the use cases of Solana, whether it’s actually a blockchain, how Solana scales, and much more!    Show topics: What Anatoly thinks of Andre Cronje's opinion that Solana is not a blockchain Why Solana has had so many instances of downtime Whether Solana is the NASDAQ of blockchains Why Solana changed its approach to fee markets and how it is different from other blockchains Why Anatoly is bearish on generic Layer2s How to prevent a mempool to blow up Whether Anatoly is pro app-chains Why Anatoly would have had Solana build a fee model from the start What proof of history is and what its problems are  How vertical scaling and horizontal scaling differ What the use cases of Solana are  Why Solana decided to build the SAGA phone How a web3 phone enables new ways of innovation Why Solana doesn’t see Avalanche or BSC or other EVM chains as competitors but Mysten and Aptos Why the execution environment is where all innovation happens Why Tarun thinks rollups on Solana is a good idea Whether EVM could work on Solana and why Anatoly cares less about EVM on Solana than Solidity What Anatoly’s biggest lessons from building Solana are Why Haseeb admires Anatoly and what the similarities are between him and Vitalik What Anatoly advises people who are building in crypto       Hosts Haseeb Qureshi, managing partner at Dragonfly Capital https://twitter.com/hosseeb Tom Schmidt, general partner at Dragonfly Capital https://twitter.com/tomhschmidt Tarun Chitra, managing partner at Robot Ventures https://twitter.com/tarunchitra  Robert Leshner, founder of Compound  https://twitter.com/rleshner    Guest: Anatoly Yakovenko Twitter: https://twitter.com/aeyakovenko   Episode Links   Solana Serum: https://www.projectserum.com/ Some of the Solana outages and consequences:  September 2021: https://www.coindesk.com/markets/2021/09/14/solana-validators-ready-potential-restart-amid-blockchain-outage/ January 2022: https://www.theblock.co/post/131278/traders-are-complaining-about-solanas-performance-raising-questions-about-its-status-as-a-wall-street-darling  May 2022: https://www.theblock.co/linked/144639/solana-restarted-after-seven-hour-outage-caused-by-surge-of-transactions  Solana’s blockchain clock 30 minutes behind: https://www.theblock.co/post/149112/solanas-blockchain-clock-loses-track-of-time-now-running-30-minutes-behind  June 2022: https://www.coindesk.com/tech/2022/06/02/solana-halted-by-bug-linked-to-certain-cold-storage-transactions/ Popularity of NFTs on Solana: https://www.theblock.co/post/153214/magic-eden-raises-series-b-funding-solana-nft-unicorn Solana phone https://solana.com/news/saga-reveal https://dune.com/queries/952048/1656567 https://www.coindesk.com/business/2022/06/23/solana-labs-is-building-a-web3-mobile-phone/  Fee market: https://www.reddit.com/r/solana/comments/u3p8xc/eli3_solana_fee_market/ Aptos/Mysten:  https://www.coindesk.com/business/2022/07/11/web-3-startup-mysten-labs-aiming-for-2b-valuation-in-latest-funding-report/ https://www.coindesk.com/business/2022/05/31/paypal-ventures-invested-in-team-reviving-diem-blockchain/ Paper on Proof of History: https://www.shoup.net/papers/poh.pdf Learn more about your ad choices. Visit megaphone.fm/adchoices

Not a dividend. It's a tale of two-clan. Now, your losses are on someone else's balance. Generally speaking, air drops are kind of pointless anyways. Unnamed trading firms who are very involved. D5.Eat is the ultimate pump. DFI protocols are the antidote to this problem. Hello, everybody. Welcome to the chopping block.

Every couple weeks, the four of us get together and give the industry insider's perspective on the topics of the day. So, quick intros. First out, we've got Tom, the DFI Maven, and Master of Memes. Next, we have Robert, Crypto Connoisseur, and Captain of Compound. Then we have Turun, the Gigabrain, and Grand Pubod, Contlet. And then we have myself, A Sieb, I'm Head Hype Man at Dragonfly. And today we have joining us a special guest, long in demand, finally was able to come on

the show, Anatoly Yaakovenko, Tsar at Solana. So, quick caveat. The four of us are early stage investors in crypto, but I want to caveat that nothing we say here is investment advice or legal advice or even life advice. Anatoly, welcome. First off, all of your things have alliteration except ZR. Do we spell Zahar with an S-B now? I was originally going to intro him as CTO of Solana, but Anatoly specifically requested

Zah or senior staff engineer. And so we went with Zarr because it had a little more punch to it. But don't blame me. They blame Anatoly. Anyway, but Anatoly, it's great to have you on. I have to tell you, we have been talking about Salana almost every other episode. It's pretty hard to run a show in crypto without talking about Solana constantly.

That's been our goal. That's really, that's it. Just like, how do we get people to talk about it? That's why you're constantly having to deal with crap like this. And you have a perfect track record of 50% positive and 50% negative talk. That's very true. That's very true. We have it many times gotten pretty heated in our discussions about it.

Truly an Ethereum killer then because I think that was Ethereum's place when Bitcoin was the only game in town. And it really split. That really split the conversation into people that were either pro or against it. Solana is officially the most contentious blockchain today. That is for sure. So you might remember we had a guest a little while back, Andrei Kronier, who's, I guess, now kind of semi-retired but doing his own thing.

We used to be kind of chief guy at Phantom among a bunch of other projects. And the original impetus for us bring you on was that there was a quote tweet of a snippet from the show where he claimed that, I think it was like, Solana is not even a blockchain. And he was talking all the shit about Solana. And we were like, okay, well, we should get Anatolian for the counterpoint. Do you know, Andre, by any chance? Yeah, he's awesome. Yeah, I've talked to him a bunch.

I would have loved for him to work on Solana in those early days when we're trying to bootstrap serum. But, you know, he was doing his own thing when we were doing our own thing. Learning Rust for like a solidity dev is really tough. because they went through a lot of like shating in those early Ethereum days. And when you do that as a developer, you kind of like become a master of your domain, right?

You know the nons and crannies. And throwing all that away and moving to a new system, it hurts. Like you have to kind of give up a part of your brain. It takes years for you to get to that point. So like after like 10 years, you'll be like, why is like half my brain occupied with this useless details?

I'm just gonna like remove that entire plan. part and throw it out. But like it'll take some time for that to happen. It happened with me in C++ for like for a decade. I knew every new cranny in C++ and then at one point was like, why is half of my brain occupied with useless details and just like snip, amputated, it was all gone. I have to ask you directly like, you know, I know that I remember you commented on the on the tweet thread when we, the snippet of him claiming. that Solana is not a real blockchain. And I think there are kind of two things that I think have been the crux of what people argue about on Twitter about Solana. The one thing is how decentralized

the Solana. The second thing is how stable or how live or how much uptime does Solana have. So right off the bat, I think those are probably the two things that we should jump into. What is your response to the critics, the naysayers about those subjects? Both of those are kind of the same thing. If you look at it, look at like fundamentally what is a what do these systems provide it's extreme security and the extreme part comes from the fact that in a catastrophic failure yellowstone blows up something like that that there's at least one copy of the state that survives and it's provably that that's a copy because there's cryptographic signatures from the quorum that we can validate and from that

single copy we can recover the whole state machine and it continues that's really a what the system's supposed to do. And it provides some ways for people to update those state and continue running. The fact that there's uptime issues is kind of irrelevant because they're all temporary. It doesn't really matter if there's a four-hour period between blocks. So it doesn't really impact the fundamental thing that this thing provides, which is catastrophic failure, like security against catastrophic failure. But it does impact the applications built on top.

That is like, that is us eating it and like dealing with those problems and it sucks. And it's 100% bad. And like obviously all the engineering effort is on that. But when you think about it from like, what is the purpose of these systems is this like massive amount of security that is so massive that you trusted more than Google. That's like the really cool thing. And when you kind of dig into it like, do I trust that Google will lose my. state or even a combination of Google, Microsoft, and Amazon, is it possible for all of them to fuck up and lose my state?

That's more likely than a network like Solana with thousands of participants. They're all running different servers, different configurations all over the world, right? That's the cool thing about it. The uptime thing, yeah, the uptime thing definitely sucks. But that's like day-to-day business like for applications. and a, you know, we for it, not just on Twitter, right? Like, users, folks like Magic Eden, companies with real revenues, step in, right? Like, had like $100 million third quarter, right?

Or whatever, second quarter. They depend on the network running to generate those dollars, right? Like, so. That is all true when we contrast, like, centralized systems with decentralized systems, being able to verify backup state is all true. I think the criticisms, or maybe the comparisons are more across blockchains, right? Because any blockchain has a lot of those principles, right? Even like, I don't know, something random, like Ethereum Classic is, you know, it's true that you could like, you know,

replay, recreate the state and anybody can sort of, you know, audit it. The question is, like, for day-to-day use cases of, you know, this product, it's like, where in that spectrum is, like, acceptable to lie, right? Like, is it acceptable for this to have, you know, one transaction per day and it's sort of guaranteed up time and but unusable for like pretty much anybody or on there's a salon end of the spectrum? I'm curious how you sort of think about that component of the tradeoff. So there's a lot of people already working on the other quadrant, which is they built a system like, you know, for a work-based one that has really nice properties that allow

to have basically an infinite uptime because there's no definition of Bitcoin or a POW system being down. Two hours between blocks, right, like last year, nobody said Bitcoin was down. They're just like, oh, people are moving miners and now it's going to take a little longer for blocks sometimes. Two hours between blocks and Solana, it's down. Right? Like, so there is a definition of being down in Salana because we're now in a different ballgame altogether. It's providing a block every four to milliseconds. There's literally no gaps between blocks by design. So when there are gaps that are even like 10 minutes, everyone's page your duties off and we're

trying to figure out what the fuck is happening and why. So I think the same, the same thing. The This is where I think things are just different in these new class of chains that need this constant uptime and constant throughput. There's no way we can be service the world if there's downtime, but also if the solution is let's increase fees for everyone. And like the network is effectively only up for a very small percentage of users at that point. So that brings us to fee markets, but before we get there, I kind of want to drill in on this point a little bit deeper.

So when we had Kyle on a previous show, we sort of asked him like, hey, so what's, this was, this is right after, I think, the last, you know, kind of notable period of Salana downtime. And Kyle's answer to like, okay, what's going, what's the deal with all the downtime that we've had this year? His answer was like, look, Solana made some very explicit tradeoffs about trying to eke out very high performance at the expense of, of stability. And they went to market faster, they got things out the door quicker, and now they're paying the cost. And that cost has to be paid down

one way or another. And look, it is unique among blockchains, right? It's pretty clear that there are no other major blockchains that have the stability issues that Solana has, but there also no other blockchers that have the adoption that Solana has or the throughput that Solana has. But his framing of it was that this is basically

a choice that Salana made that it's now paying down. Do you agree with that characterization, or do you think that that's not the way you described? We didn't do it intentionally. We made the wrong decisions at the right time. But it wasn't like... What do you mean by that? When we were designing this thing,

literally the use case we were designing for was basically serum. It was a central limit order book. Our like benchmark that we had internally was basically a very dumb central limit order book. And the idea was that all the transactions are going to be very small because limit orders, buy bids, asks, cancels. They don't take any compute. The biggest part of that whole computationally pipeline is the signature verification. So a fee model is very, very dumb.

Most transactions are going to be extremely fast to process, and the system should charge per signature. And if the number of signatures per second that is processing exceeds some threshold, we double the fees for everyone. And then that forces like spam to back off. That's a very, very rudimentary design for a system. that mostly runs something like serum. We actually got serum was like the first major use case, right?

That we were like, right, surprisingly. Never happens in startups. But what we were wrong about was all the complexity around like composable defy didn't just call serum. They wanted to call the lending protocol plus the AMM plus serum at the same time and do these complicated swaps. and like radium, like build an AMM and top of serum that like really took the limits of computational what the system can do. And now you ended up with really, really heavy transactions that had hotspots on particular parts of the state that were just really, really expensive to handle.

And all the bots and everybody in the world that wanted to take that like $5 or whatever ARB would just slam them like immediately. And this is something that we didn't expect that use case and what we built were CRSERM and just totally didn't work for everything else. So that technical debt was unintentionally there. The mistake that we made or the not mistake that we make is we let the design be open for these other use cases to spring up with the idea is like, okay, people are going to try random shit. We don't know what actual product market fit is in crypto. So we didn't lock down the system to the point that it only works for serum because that would have been a disaster. Right. Like none of the stuff that you see on Solana right now would have been possible.

That would have been, like, I think a bigger mistake than allowing the stuff to flourish and then obviously blow up in our face a couple times that we had to like hustle to fix. That kind of brings us to one of the things that we talked about on the show as well is about fee markets on Solana. and I know that in the past, you were relatively opposed to the idea of imposing fee markets because you've described it as wanting Solana to be the NASDAQ of blockchain. Is that the, I think that's the quote that I've seen before. That all comes from that idea that, like, what we're going to run is mostly a central limiter book transactions by volume, and those don't need fee markets, really. Right.

And the fee markets mean that you need something approaching a mempool, you need to be ordering transactions, you'd be doing all the stuff that if it's just a giant stream of, things that are happening, you don't need to do that amount of work. But it seems like the fee market has been the approach that you guys have landed on, although a version of a fee market that's quite different than most of the blockchains. Talk us through what made you change your mind and or the community changed their mind about the fee market approach. And what is the approach that you're taking? How does it differ from other blockchains? So this is again, had we not made those early mistakes, we wouldn't have discovered this, right? Like, it wouldn't

have been obvious to me at that time because I'm an engineer that is very learned practically. I have to see the data, like see how the system works, how it's broken. And that was something that we just couldn't foresee. And basically what we discovered, and this is obvious to me now in retrospect, is that these systems have very traditional database hotspot problems. You have a whole bunch of state. In Solana, the state is organized as a key value store. each key represents a public key effectively and each value is an account with data and those accounts

represent both user state and program state. There's really no distinction. So you have like an AMM, there's a program that doesn't have any state that processes these key values and the value would be the AMM market and the key would point to that AMM market. So when you have a hotspot, it means everybody's calling the exact same AMM key value. They're constantly hitting that one market because they want to swap those arms or the same serum market. So it's not like a smart contract specific. It's specific to a single state.

And that hotspot, it doesn't matter how fast we make it. Like imagine it's a switch and everybody wants to flip the same switch in the room. It doesn't matter how fast you flip that switch. Like if we made it slip in one nanosecond, it wouldn't really matter. matter. If it could flip a million switches in parallel, it wouldn't really matter. What matters is if you have a hundred people that all want to flip that switch first, you have a hotspot and everyone throws a whole bunch of transactions, only one of which can actually go through. And that problem is solved by a mempool. But it's solved in a way that creates one giant heap of requests to all

access a single global state machine. And that means when you have an NFT launch, you know, or or airdrop to all the ape holders and all of a sudden fees go up, you still have DeFi protocols that are like, I need to land this liquidation. I have a channeling Oracle that needs to line this liquidation. Now I have to outbid all these stupid NFT people to land to do my like day-to-day process. And now you have a bidding war between hotspots that create these like really, really bad scenarios, right, on Ethereum.

And the only way to fix that is to isolate them. And this is my bearish argument against generic L2s. L2s segregate state logically, right? Different L2s have different states that they own. But they don't, they're not, they're all generic. So you can't like, they're not going to solve fees because you're still going to have an L2 that has NFTs and DFI at the same time. And if you have an NFT drop and liquidations that all need to occur at the same time,

the memple for that L2 is going to blow up. The only way to do this is to actually isolate these transactions by normal database transaction isolation, make sure that stuff that's processing hotspot A does not impact hotspot B. So there's no way I could have predicted that this is what we needed to build before we actually saw this problem. Right? Like, I don't know. Maybe smarter people have seen it, but like... I would say one thing. Your description of this problem does seem to be very...

very pro-app chain, because in the sense of Cosmos, Cosmos actually has mainly been built around trying to isolate these types of state congestion issues. So I guess, you know, from knowing what you know now, what design decisions would you have changed and say? Yeah, we would have built this, we would have built this fee model like, I think from the start. And the model is pretty simple. Because we already know, which, what, like, we already did all the hard work, which means we force steps to specify in their transaction, which accounts are going to need to read, which accounts are going to need to write.

So that's actually the hard work is forcing that design onto all the applications. Once you have that information, all you're doing is sorting the transactions into buckets and then making sure that priority, like, is based on some kind of fee. So both of those things we added, it's just that once the network is mature, releases, and moving shit out into an ecosystem takes longer than us yoloing a change in, like, you know, March 2020 when it was pretty easy to make changes like that. So there was this paper in the last six months by this sort of famous cryptographer, Victor Schoop, about, you know, proof of history, kind of like, not quite having a lot of the desired properties. And also, you know, I think in some of the downtime scenarios, the proof of history module was sort of one of the things

that seemed to have a lot more trouble. If you went back in time, would you still have proof of history? So, Turin, can you very briefly explain what proof of history is and how it figures in Islam? So proof of history is effectively a way of doing sort of a minimal proof of work like system of like doing a repeated hash that someone can verify, that allows you to sort of have some notion of like sequencing of slots that in a more linear fashion. The problem is it sort of doesn't have certain guarantees and also can cause a lot of extra traffic on the network

at the worst times, at the times when when things are kind of have the most general traffic. So what was the extra traffic? Like what's the extra traffic? The proof sizes? Yeah, basically people spamming the proofs effectively when they can't get any other slots in, any other transactions that they're gossiping. Like, at least that was sort of the argument. We haven't seen that being like a major bottleneck. So if the chain gets into a state where blocks are getting dropped by propagation is just not happening. The currently, like, the nose just retry.

Like, it doesn't really matter. Like, they just retry their previous block. and the chain and proof history kind of stalls in effect. What happens during an outage is you have a complicated system. It's a database, right? It's trying to run in constant memory because that's a requirement. And you have a scenario where there's like, these people have built bots that spam it with like 100 gigabits worth of traffic.

BitFenex called that a denial DDoS, right? But that's not really DDoS level at this point. DDoSs are much, much higher now. It's just normal bot traffic. People really, really motivated to get early in an NFT drop. This is not even for liquidations or complex finance or HFT. This is purely to get an NFT drop. And the 4 million or 5 million packets per second that are being processed are moving

through these cues on stages. And if one of those cues has unbounded allocation, the cue blows up, memory grows really, really quickly and the validator just runs out of memory and crashes. And then it tries to restart. And to restart, it takes time to process the current state, hash a bunch of stuff, verify hashes, and catch up with the network. If a third of the nodes run out of memory in that condition, you basically are in a situation where it becomes really, really hard for them to catch up with the block producers that are still trying to produce blocks. And the network then is in a state that's Like, it's easier for the validators to decide that, okay, we can let it run and see if it catches up.

Eventually, it could. Or we just get everyone together and we, like, kill this particular shred version, which is, like, a network ID for the current, how we currently propagate blocks, increment it, and then continue. That means all the old block, all the old proposed blocks get dropped automatically, and then people continue from the last known slot. And that's just a decision that they make wherever they, whenever they feel like. But it's not a, we haven't seen like profus or anything like that fail, like cause a catastrophic failure or cascading effects yet. Well, I assume everybody followed that. That's, that, well, what you described sounded super obvious. I don't even know why we had to discuss it.

So like the bugs that we have are like dumb out of memory programming bugs. It's a complicated database. somebody somewhere, probably me, if you could blame it, wrote that code and didn't correctly manage the state. I want to go back, I want to go back to the point you're making about isolation. I remember back, before Solana was live, before all these LFEL ones was live, I remember you did, it was you and Alex Skidanov from near, you guys were sitting on a couch, I think it was like East Denver or something, and you were arguing about basically vertical scaling versus horizontal scaling, right? And the meme at that time was that Salana's vertical scaling. Solana is you take a big,

beefy server and you run it all in the same machine and near, and E2.2.0 at some point, you know, in that moment in time, we thought E2.0 was going to be actually have shards. The sharded approach is you have, you know, a lot of small machines and you horizontally scale. As you're talking about isolation, isolation is almost, you know, kind of to the dot description of what horizontal scaling gives you versus vertical scaling. How of your thoughts about sharding versus the Solana, you know, just big, beefy machine approach. How is that evolved over time? Or are you basically like, look, I think we made the right choice.

So I think we made the right choice. I think the problem with the other choice and kind of described it as unless you have app-specific chains, you're not really isolating anything. And with app-specific chains, you're breaking composability. And that becomes really, really hard on like all the really cool, weird defy use cases. That's like the trade-off, right? Like you stick NFTs in a separate thing. You stick your lending protocol in another thing.

So it's liquidations don't touch any other lending protocol. Like you'd have to have like compound and Ava in separate app chains. Because like what if both need to be liquidated at the same time, right? Like that becomes that that's the issue there. And that sucks. I think fundamentally that that sucks from the magic of crypto, the magic of smart contracts. Now that makes sense. At the same time, it also seems surprising that, you know, Salana, it's very clear, especially from your description, that when you guys were designing Salana, you were designing it around, the use case that you were imagining was serum. The NASDAQ of blockchains, right? You guys see very clearly this is going to be the defy chain. And it's not DFI's NFTs and pictures of monkeys that has ended up being the predominant thing that keeps taking down Solana. How do you think about that? Or does that strike you of anything?

I think it's awesome. Honestly, I think the internet is, is like, typically the best versions of the internet is just sharing pictures of cats, right, like doing funny shit. And I think it's awesome to see that NFTs are basically like people having fun with this stuff and not really taking it too seriously. That's been, I think, more interesting to me than trying to like convince, I don't know, the Goldman's hacks of the world that lending their, their. borrow and lending books should be on chain. Robert can do that. Like that. I'm happy to do that. I love that conversation.

I don't get to enjoy the. You enjoy plenty of pictures. That's true. I do enjoy the pictures. Like the NASDAQ use case was like interesting to me because I'm a, I like high performance system. So that needs a lot of throughput. Defi that doesn't need a lot of throughput was just not something that I was super

interested in because it can run somewhere else, right? So there was no, there's no need for Solana if the only defy use cases are borrow lending. But if Central Limit Order books exist, then I think it's obviously that you need something like Solana. And now that NFTs are a thing, I think when you're talking about minting, like we're, we see 100 to 200,000 NFTs minted per day on the network. And that's with like, you know, few million humans. out there messing around with the stuff. If we go to hundreds and millions of users and now we're talking five, 10 million NFTs minted per day, you need a network like Solana right to support that. And that's interesting. So speaking of different use cases for Solana, one of the big news stories

from the last, it was a couple weeks ago, but we're having you on. We should we should talk about it when you're here is the saga phone. So the saga phone is a Solana branded phone that is going to have native salana support from the get-go. And there's been a lot of conversation about the saga phone, how likely it to succeed, whether how long ago this was conceived of, whether it was intentionally announced into the bottom of a bear. If you look on chain, it looks like there are a couple thousand pre-orders for this thing already.

So I'm first of all really curious of just, how do these things happen? Is this like a Solana foundation is like, hey, guys, we should do a phone? is a third party that's like, hey, guys, we're thinking about doing a salon of phone, and you guys are like, cool, here's some branding and you guys are, like, did this come, was this your idea? Was this someone else's idea? Like, how did this happen? It was my idea, and it happened because I met through, do you know race capital guys, Edith, Chris McCann? Through like them, like Alfred on their team, he's like, I was just, I think what the space needs is like Apple and Google to really open up to

crypto. That would really like, I think, accelerate everything. And the only way that's going to happen is if somebody builds an alternative model and shows that it works. And like, this is what I would do. And they're like, you should go talk to Jason X. He was like the first employee essential basically, first or second. Architects at iPad Pro, awesome entrepreneur, awesome builder, like super hardware nerd. And they built an awesome phone. And when I talked to him, I was like, we should go do this. And it was because of him that this happened, right? I just met the right person to do it.

There's no way I think I could push this idea through a big corporation. Like you can, right? You can make deals and do all this other stuff. But you really need like a driven CEO that kind of gets division and is willing to push through this. Because it's not just like, we don't want this to be like the previous crypto phones or it's just a wallet or something like that. There's a fundamental difference with sticking the signing infrastructure inside the trusted element. So the high level operating system, Android and the wallets cannot steal your keys.

You create an environment where Phantom doesn't know your seed frights. And you have a, that means that we can have things like tap to pay, like you go to store and you hit the button to pay, load whatever wallet you want the developers can innovate with. And you know that that wallet can never steal your funds and that wallet's UI can now be used to pay for whatever purchases you want. And that opens up a whole like new channel channel of innovation, right? Devs can build cool shit that can integrate with like merchants. It could be bi-directional channels. Every time you buy something, you get NFTs or whatever. Like all that stuff that people dream about like with crypto and programmable money is now possible because the software that you're using and interacting with never has access.

to the seed phrase. Right? And that level of security, I think, is necessary. And this is something that when I was at Qualcomm, this is like what this point? Six years ago, all this existed, all this existed, all the application stuff you can build in trust zones. All that stuff is there. They just don't enable it because there's no provable product market fit for it.

So we want to drive that forward. I don't know when this is going to break out, like when it's going to be relevant. but I do believe like somewhere in the next three to five years, we're going to see a moment where every mobile phone out there has signing inside the trusted element, right? That has like Web 3 enabled natively. Like it's just obvious to me, right? Well, that is one thing that you guys have been amazingly effective at

is breaking out of the Cold Star problem. And I mean, that's the struggle that every layer one goes through and there's also a struggle that any new line of hardware goes through. is, you know, okay, it's one thing to have, okay, I have my beautiful Solano phone, I can look at it and pet it, but if nobody actually, if there are no merchants,

if there are no applications, if there's no, you know, blah, blah, blah, that if there aren't enough people who own this phone, then you're not going to have enough developers who are creating experiences that are native for this phone, like everyone's going to assume

you're on mobile web, not enough people are going to integrate whatever SDK you need in order to interact with the trusted element. So how do you think about that? How does this phone break out of that, you know, no man's land? So I worked on like every phone that failed, the WebOS phones.

You guys remember those? The Amazon Fire, the Metro, the Windows Metro one, a couple other ones that were just code names. Basically, it's really, really hard. And the thing that I think how we can possibly succeed and it's not guaranteed is that I think the big store marketplaces right now are so bad to Web 3 that 50,000 Web 3 users that have this device, but no restrictions in the store with how Web3 apps can talk to them as a better distribution channel for crypto-native developers in the big app stores.

Like Magic Eden can't really ship an awesome marketplace on mobile because Google wants 20% of fees on user-generated content that Magic Eaton doesn't control a $10,000 NFT. Magic Eden cannot charge the user 12,000 in their mobile experience when it's listed for 10,000 in the web. It just doesn't make sense. Nor can they eat that cost, right? Because they don't own that property, right? Like, all of, all of the big app store business models are around the idea that the content, the mobile app is owned by the content provider. The content provider is the creator and they can charge wherever they want, right? And then Google can put a tax on that. And that doesn't work in a digital owned economy. That's just not like, you have to treat digital objects the same way that you're

to physical ones, like eBay, right? You're not going to charge 20% markup on the eBay mobile app. And that model is just, you know, they have a lot of VPs, a lot of layers of decision making. I don't know what's it going to take for that to like percolate to the top and for them to change your policies. I hope is that like maybe we make enough noise that that happens. And that's a good outcome, right?

Like I think for the space is that even if we fail commercially, but we change both Google and Apple treat rep 3, that's a win. But, like, how we succeed, I think, is we get it to hardcore crypto devs. It's open source. So, like, Robert, if you still code, go submit the Ethereum integration. It'll, it'll, okay. You know, I've never could. Hey, hey, hey, hey, a secret is Robert used to make Ti-83 games.

Okay, yeah. That's true. And that was programming in basic, basically. If there's still Ethereum founders at code. I think they're mostly retired now after prices have come down. If there's anybody, if there's still Ethereum founders at code, it is honestly open source. We'll literally take more changes than MetaMask would take a lot of changes.

Fair enough. Fair enough. Okay, I have another Salana-based question. kind of changing gears a little bit. By the way, we got a lot of questions about the Saga phone. So I think people, it's a good, it's a good grounding for that story. One of the questions we also got

from a bunch of people is comparisons with Aptos and Miston, which are the new Facebook slash move-based chains. I think you responded in some of the, the Twitter comments that you weren't super familiar with what they're doing, but I think I see them as the most common comparison to you guys.

And I spoke with someone from Salana Labs who was telling me that internally at Solana, you guys don't see Avalanche or BSC or any of these guys as your primary competitor. You see Aptos and Mistin as your primary competitor. Is that true? And please elaborate on that if it is. I think so I think I have this theory that it's a little controversial that and obviously the reason behind Salana's design comes from that is that execution is like the really hard part in all these systems. Consensus is relatively simple, like compared to the runtime.

And this is why you've seen a lot of innovation and consensus because you can build all these consensus systems. The security failure and consensus is extremely unlikely one. It requires collusion from a large set of participants, right? And then partition the network and then do a double spent. Everybody, this is not going to happen, right now because the number of participants in crypto that are don't want to collude that rather make money, right? Like the traditional way is much, much larger than the number of like

attackers that are willing to stake and like it's just extremely unlikely. And in the execution land, building a VM, any bug could be a catastrophic failure. You're running untrusted user code with untrusted inputs and any exploit in that virtual machine can blow the whole thing up. And they can blow up it from a total loss of funds, catastrophic, or application-specific one, or like a reliability failure, like you saw with the Shanghai attacks. But building EVM was, I think, the hardest problem in launching Ethereum. And that's why I think you've seen a lot of stickiness for EVM. Because, like, when you're building a new consensus engine, you can go to market much, much faster by bolting EVM on top of it. And like, look, you've got

an environment that's secure, right? You're using Gath, right? Like, everyone trusts it. You don't have to go, like, re-implement this extremely complex wheel, right? And that's actually where all the innovation and the cool, like, like, this is where you can actually innovate and build cool shit and capture a lot of values in the execution later. Consensus systems, nobody devs don't care about what they do. They just care that they work.

They hate us if it's something fair. But what they wanted to do is just work. The execution environment is where devs play, they innovate. And developers need new tools like I think every two years because they're curious much. So when somebody actually like the team that built Move is extremely awesome. They're like really smart folks. The design is really good. I think Move itself is a is a really good attempt at building a more secure role.

to build smart contracts. So I think there's an opportunity for them to actually build their own ecosystem because it's new. And it's going to attract a bunch of curious devs. And if it's good, that could be enough for them to get stickiness. I'm not worried about yet another EVM chain.

It's just like, okay, they're all splitting themselves up, right? Like, they're fighting over the same group of developers. They're not creating anything new. And like, I was going to ask, like, going kind of back to the app chain, you know, bit for a moment. You know, why do we think, why do you think we haven't seen other chains with the SolanoVM

using different consensus mechanisms yet the same way we have with EVM? Like, you know, if we think, hey, it was a great developer experience, people love Rust. Like, why is Salon only one? Especially when, you know, if I'm, of an app chain, I don't necessarily need the co-location. I might want to run my own decks even, like step in. Like, you know, why am I paying a premium for this co-location block space versus just, you know, running myself? transactions in Salon are so cheap that it's going to be more expensive to run your own app chain

that that part is true i think for step in like stephen even though it being like having i don't know seven times more transaction volume and user volume than uniswap it'd be more expensive for them to host oval servers it's also like a pain in the app i think to to do that so i'm not sure i don't know there's like one or two I've heard, but they're like pretty low like credibility, I think, or haven't really gotten traction. I mean, it took a long time for there to be other really upstanding EVM chains. Obviously, Ethereum Classic was the first. Ethereum was quite a bit, you know, simpler and earlier back then in its life.

And Ethereum Classic, of course, it had tons of stumbles. It took two years to go from like, right, to literally people have to build a whole blockchain. and both like it took avalanche, finance smart chain, all those guys the same amount of time to build whatever they built like their EVM based chain. We had to build consensus plus a runtime in that same amount of time. Yeah, definitely a heavier lift. I'm thinking of something even almost like X-Dye, right? It's like, I mean, maybe like Audius is a good example where it's like started out an X-di, migrated to Solana, better performance, cheaper block space. But like, you know, I feel like that's kind of the trend that we've seen, right? Like even DYDX moving from like Ethereum L1,

Starkware, now to Cosmos, you know, people just sort of shopping around for like block space. What would you get by launching your own Solana chain? Well, I mean, I think sort of the argument would be, hey, maybe there's a premium to having your own token do validation or being able to capture MEV or I think especially, you know, when fees going up, you know, maybe there's an argument that, hey, we can have, you know, cheaper fees by having in our own chain. So the way that Solano fees are designed is that there is no advantage to moving to an L2. Right? Like the fee problem isn't a block space constraint problem. It's a hotspot problem. So when you have 10 people that want to flip the same switch, it doesn't matter if that switch is flipped in the zero knowledge environment, in an L2, in an app chain.

You still have 10 people that want to flip the same switch. So they will bid up to flip it. If you isolate them, that then they don't impact anyone else's in any. other switches that are also doing their own thing. And if we can achieve that at Solana at the L1, there's no advantage to launching an app chain. The only reason you might want to do that is like... That's true. That's true to an extent, right? It's not true. Let's say there are five hotspots simultaneously. Or let's assume that, you know, you've got one of these crazy NFT drops and, you know, people are just absolutely bombarding all the RPC endpoints. They're bombarding all of the leaders. and the leaders are basically being dedosed

with all the crazy requests that are going through. Certainly one can imagine that, look, if we just spin up a new Solana, now look, it's hard because Salana is a moving target. Salana is not like Ethereum where basically it changes once every six months. So that I think is another element of like, look, you just have to constantly know how to do the DevOps for Solana without being tied to Solana.

That's just really hard. I think those reliability issues would come with you, right? So whatever you spin up, it's the same code base, same RFC. Yeah, it's at worst because you don't know how to run it. So, yeah, I think it'll be, I think there will be a time, almost certainly, if Solana continues to be successful, that Solana's codebase will solidify. And there will be people who run Salana Classic or whatever, you know, there will be other versions of Solana and some of them will gain legitimacy.

some of them might even be run by exchanges and some, you know, I don't know, gate.io or something will run their chain and it'll be based on Solana. And they'll make the same arguments and you'll have the same combination of flattery and annoyingness that will come from having another chain based on a codebase.

So if somebody's listening to this, there's a pull request that removed the move loader from the network before Maynett launch. The reason why we removed it is because running move as a native loader would have exposed that runtime meant that if there's a bug in that runtime it's a bug in all of the state machine and that's a catastrophic failure but you can literally take run your own version of salana with move and launch before aptos and suzzi

i actually i actually think and sorry again my internet's i actually think the idea of salana roll-ups is a very well uh an actually good idea in that You could imagine a roll-up where the fee structure is different based on the use case. So, like, basically, it's still running the Solana VM. So any bridging is at least, like, guaranteed to have certain execution properties to be bitwise identical, at least in the actual execution, maybe not in the fees. But the fees can be adjusted to, like, which parts of the runtime are more used in certain applications or seller applications. And in general, I think there's a lot of the modularity in the networking stack in the Salana codebase.

actually lends itself quite well to basically having roll-ups that are running the Salon Runtime. I actually kind of imagine that, I know Haseeb was thinking more on the, of the adversarial salana runtime, like Fobie does B&B, but like does it with Solana or something. That'd be fine. Anatoli's vision. Anatoly's vision. Oh, my goodness.

We need to create a bounty for Anatoly's vision. It only runs a serum and that's it. I think, for instance, the Solana runtime, that's special purpose to do certain ZK things a lot better than the current runtime. Or, like, you know, these things that at least guaranteed that, hey, bitwise compatibility is easier. And you don't have to have all of the problems we've seen with basically every bridge to Solana. And really, like, the translation layer, where the translation layer is the part that, like, between, multiple virtual machines that aren't bit like identical leads to these like kind of nuanced bugs right so there i don't know i could see a world where salana rollups are there is like such a

thing because like if there's already that much developer interest people will start trying to specialize it in some way shape or form i mean how do you feel about that if like say people started sure that'd be cool if there's like stuff if there's system calls that they need or like hooks into the VM like i don't think we have an e-val where you can tell the runtime to go evaluate itself. But that's relatively easy to add. If people need that, we'll definitely add it. Or terrifying.

You heard it here first. Salana roll-ups coming at some point. I mean, if people want to build it, I don't want to stop them. Sure. Speaking of roll-ups, though, so I think last year there was a lot of hullabaloo around neon labs, which is supposed to be this roll-up, unclear if it's actually a roll-up, or more just like a virtualized instance of the EVM on top of Solana. And I've heard through the grapevine, this is total speculative, I don't know if this is true,

but I've heard through the grapevine that people are writing off whether they're ever going to come to market or whether even that was the right approach. And people are like, ah, you know, whatever. People made a big deal out of it, but that's not the direction that you want to be focusing on. What is your, what, what say Anatoli about the prospects for EVM on Solana? Does it make sense? Does it not make sense? Do you even care about it? I think there's two projects.

One is neon, and I think it's live on Tesla, so you can go to mess around with it. I don't know what's blocking them from Mainet, but it should be pretty close. But like the hard challenge is like you're taking Ethereum, a version of Ethereum that was built in Rust, and you compiled it as a Solana program, and then you're hooking up all the APIs that MetaMask expects Gath to implement against that thing. Right. And the problem is that like you take a Ethereum transaction and you submit it to Solana and it takes more time than is allowed for Salana transaction to take. So you have to create a continuation where you cash like you cash the partial computation of this EVM as it's churning and you tell it to like, okay, stop and then I'm going to go send another continuation for you to continue and then crank in until it's over. And you have to make that system to be. compatible with whatever MetaMask expects. So how that like fits, I think is pretty close, but I think there's like a little,

I don't know if they got everything right, but I think they're getting closer and closer to it working. Yeah, if I recall, they had like a super tight gas limit for transactions that could execute in reasonable time. That's a hard, hard approach. And I think if it's not live yet,

it's pretty close. Like go, go to mess around with it and see what's broken. Like, I'm sure they, they need people to, to like, just play with it and tell them what,

what sucks. There's another one that is one that I like more personally because this is just part of the stack that I like. And that's taking solidity and implementing it as an LVM front end. And then going through LVM, just generating a native salon of program. But that approach means that the result is not going to be compatible with EVM, right? It's a different state model. You should re-audit that code. A failure of the EWASM project, perhaps you could argue it was a project management failure on the behalf of the Ethereum Foundation. On the other hand, you could also argue that it's the LLVM runtime and the EVM just have too many thorns that don't quite, you know, fit together. And that was really the problem.

And I think, yeah, it's that one, like, but again, the reason why they needed that to work, byte for byte can be exactly compatible. That's the same challenge that Nienna's working on. That's a really hard problem. With Solang, we're kind of like saying it's not EVM compatible, so didn't even worry about that. But it compiles solidity and runs as you expect, right? It's just not EVM compatible. And that, that I think is a pretty cool approach that's pretty flexible.

And like if you're solidity dev, you can go use that right now and you can go get started and build native Solana program. in solidity. And that's actually something that I think I would like to see Move implement it as a native Solana like through Lovium pass. Like I think this is something that like whatever,

whenever there's a new runtime or a new language, that's the approach that I want to use to like keep adding the support for these systems. There's a grant, I think from the Salana Foundation to the Zig Lang team. Zig is like a, it's, talk about like C++ and then Rust and then Zick. Zig is like this. It's a very small nerdy

community on building high performance like C like next version of C basically. So like stuff like this I think is awesome to enable for devs because these are new tools and devs are curious. They need new stuff to go play around and like innovate with. So do I care do I care about EVM and Solana? I care about solidity on Solana more so than EVEV. because I want devs to have tools to do stuff with. That's the long story. But not specifically because I care about, like, trying to get Ethereum Demps to move over.

I just want to, like, have the tools there for them to go do whatever they want. So I will say one thing. You do actually care about EVM and not Solana because the EVM semantics, and we've already seen this with all the ethelowlops, where the fact that you can't just redeploy on some of them and it, like, actually gives you, like, bit-wise compatible liquidations. It turns out there's some condition under which, like, ZKSink doesn't do the correct liquidation in compound.

Like, that type of stuff, you really want to prevent people from doing, and you do want the, like, by-code-level semantics to have the same. Or you're just honest about it, right? And it's something like Soling, you recompile it. It's a new program, and you re-audit it. You actually rethink about it and see what, in this new environment, like, what, what actually happens. And that may mean that you may need to maintain two

versions of compound. And that's fine, right? But like you should just understand that behavior. I do think it will be, it does get hairy from keeping track of like what changes correspond to what audit and like

the kind of like overload for developers will mean that mistakes will be made no matter what in such a scenario. And it's of course always true in crypto that developers are surprisingly lazy, or that some developers, I should say, are surprisingly lazy and will... The best developers are lazy. Well, the words developers are lazy in different ways and the best developers are lazy. I will say that. Okay, so we, this whole episode, we've kind of done a really

deep dive on Solana and kind of your thinking and how Salana has evolved as a blockchain. I want to shift gears a little bit, talk more about kind of your journey. So, you know, I remember back in the 2018, 2019, bare market, when you were just first fundraising for Solana. I remember, actually, one of the very first talks I went to as an investor. I just started my career as an investor. And I went to, I think it was like 500 startups talk.

And you were up there and you were like, I used to be at Qualcomm. I did all this low-level stuff with whatever it was you were working on super low-level stuff related to phones and whatnot. And I was just like, wow, this guy is like super legit. This is like... That works?

Yeah, exactly, exactly. And, you know, I remember back in the day, Solana, you guys really struggled to get, to get allies and to build a coalition behind you guys. Struggle to get funding even. Absolutely. Absolutely. You went from basically being the, you know, the ugly duckling that was struggling to really get, you know, funding and attention and energy and resources to, you know, having, you know, having. you talked about serum, kind of putting the wind at your back and creating this tremendous momentum behind Solana

to now make you, you know, when I look on the block, it's funny enough, they have four tickers at the top, and Solana is one of them. Salana is not the number four of cryptocurrency, but in most people's minds, it is. And what has that, what have you learned from that transformation that you would not have expected when you first started being like,

I'm going to build a blockchain? If you could go back and tell your previous self, like, hey, guess what? Anatoli, here's what happens when you build one of the biggest blockions in the world. What would you have been surprised to hear back then? The more success you have, the more work you have to do, I guess. I don't know. Like, I think what I would have put in really surprised that, like, the biggest use case is going to be pictures of, like,

it would just not have made any sense to me, right? Like, if I went back in time, like, four years ago and I was like, look. Does it make sense to you now? I have, I've become a PFB collector and it does make sense to me and it puts me back in a state of like playing Ultima online in those super early days of massive MMOs where for the first time you kind of had community on the internet that was interactive in like on the internet itself and you could customize stuff and you were doing things. These were like the first guilds that were being formed and things like that. and I remember geeking out about that and feeling like I was part of something. And that, I kind of feel the same way about PFPs. So I don't know if that's like, most people feel that way, but like I kind of do.

So that's like my reason for enjoying them. I have to say, you know, so we talk a lot of good about Solana. We also admire Salana because, I mean, that's part of the reason why we talk about it so much. But most of all, I think I admire you for, one, not only having gone through that hero's journey of building something that with nobody on your side, not nobody, but a ton of people on your side, to building one of the giants of the industry, but also the spirit with which you've done it. We were talking before about how most L1 founders, as they become more and more successful, they become more rambunctious, they kind of, they turn into assholes, they start picking fights.

and you've somehow managed to carry yourself in Solana, despite the company in which you, you know, Solana obviously has got now, it's fair share of shills and crazy people and whatever just because it's so big that that's what happens. But you carry yourself with a Vitalik-esque level of grace and thoughtfulness, which has been amazing to see you. That's a huge compliment.

Yeah, I mean it. Thank you. Yeah, thank you. That's a huge compliment. I think I got lucky that I was older when I started, of this. So past my rambunctious age. And I should ask that.

If you, if Anatoly had started Salana in his 20s, how do you think Salana would have gone differently? I don't know if it would have made it. Actually, actually, a related question. You know, right now there's actually this kind of narrative that I won't say from who, but there's many people who are doing this of like, hey, we need to like have like 18-year-olds building everything. like 17-year-olds or 12-year-olds.

And how do you feel about this- narrative of like, how do you feel about this current, like, narrative of like, oh, like, yeah, we should only have kids, like, building any crypto project. Because, like, I kind of feel like it ignores the fact that, like, these things are hard to build. You can't just, like, only have, like, people with a lot of excitement.

You also have to have, like, a lot of different types of people. I would trust dads over 18-year-olds. Like, especially somebody, like, like that has been through like, you know, with a two-year-olds that's been through that, and there's still like coding. Or moms. Yeah, or moms. Moms.

Yeah. Or moms. Like anyone would like it that has raised kids has gone through some shit. But they're able to power through a startup. You know, I want to reaffirm that statement because I actually also believe it. I think one of the things that most people forget is that when we're building an industry like crypto, where in a lot of cases, people, entire net worth is at stake, it requires a level of seriousness in the approach to engineering

systems that are stable enough that you're not going to vaporize someone's life savings. And like there's many ways to go about this. But I think in general, not always the case, in general, the older you are, I think the more seriously you take that sort of like sacred, you know, requirement of like, you know, in duty towards building. not that there's not 18-year-olds who feel the same way, but I think in general, the older you are, the more seriously you take that.

And it's one of the reasons why I think that, you know, in many cases, you know, we'll see, you know, older founders and teams building things, you know. Tom, as the resident 19-year-old, what's your take on this question? I think it's, you know, different specialties. It's like Ender's Gabe. You can take the children and you can morph them to build amazing MEV bots.

But if you need to do something responsible, then you need to be a little older. So you got to just sort of segment. It's true. It's true. It takes all types, you know. And Vitalik was a obviously very, very young man when he first came up with the idea for Ethereum. And it seems very clear that he needed to be young and bright-eyed and to not be able to understand the objections that more seasoned people were telling him about why this wasn't possible. But it also seems true that it was important that Satoshi was probably not a young man.

and that Satoshi, whoever he or she or they were, had a steadiness to them that allowed them to shepherd this thing into existence that required, to your point, Robert, a lot of seriousness and a lot of poise in order to build something of that nature. So I think, yeah, I like that view, Tom, that it takes different types of energy and different types of people in different stages of their lives

to usher in different kinds of innovations. Yeah, and again, I think my main point here is just more like, there is sort of this youth fetishization aspect of this that seems to be occurring a lot. And like, I just generally, anytime I see that kind of feel a little weird. And so because Anatolai was sort of like, you know, he brought up the fact that because he was older, it was helpful. I think like that's like a good message in some ways.

You don't have to be some like 19 year old. I think you have survivor biased, right? You have a lot of young kids try it. And the ones that make it are the outliers, the Vitalics. And he would have made it in anything, I think, given the same. size of his brain. But most startups, I think, tend to succeed when they're founded and driven and led by more seasoned folks with like 10, with 10 year plus experience and whatever industry they're in. Because it kind of, that you need more than just right at enthusiasm, right? You need

a network of people to hire from. You need like expertise. You need gut instincts that are home by that experience and that. The only way to do that is, you know, by like spending your, your butt in the seat for that much time. So I want to close out on one last thing. So I think a lot of people in crypto look up to you, Anatoly, as being one of the preeminent founders in the space and a real success story for obvious reasons. What would be your advice to, I mean, look, right now it's obviously a trying time. It's a bare market. A lot of people have lost confidence about their prospects to be able to build something in crypto. What would be your advice to the entrepreneur might be listening to I think right now it's really hard to actually, I think, launch a company in the middle of a bull run because there's so much noise.

And it's really hard to get to product market fit in a bull run because of that noise. And this is why post bull run, you see a lot of those companies fail. Right. Like despite them raising money, despite everything else, maybe seemingly going their way, they actually miss the most important part, which is product market fit. But in a bear market, there's a lot less noise. And if you can iterate and build and create, like get those 100 users, that's actually a much stronger signal than anything else. So right now is the best time to build.

And you're building for the next Bull Run with the hope that like, you know, like how many, it was like four of you, four DFI project, Kampa, Navi, Uniswap, that caused the defy summer and like the real, the bull run, right? It only took four, right? That's it. So four entrepreneurs that join now and build something really, really good are going to cause the next cycle. And that's all you need. Well, that is a beautiful note to end on.

And I totally thank you for coming and sharing your wisdom with us and bearing with our brady questions about Solana. Until next time, thank you, sir. Please keep Solana up because we've got a lot of need for it. And we look forward to seeing the next stage of evolution. Yeah. Thank you so much. Awesome. All right. Thanks, everybody.

That's it.