Solana's MEV Inflection Point: Users vs Protocol | Deep Dive

Eugene

So it's really a question of, like, sovereignty over the block space. Does the network own the block space? Does the network own these, like, account hotspots, or is that something that belongs to the application that created the hotspot? So, like, who sort of deserves that economic value? Kind of a philosophical debate at the end of the day of who owns the block space and deserves the economic rights to it.

Lucas

Hey, everyone. This episode is brought to you by, say, the blazing fast parallelized blockchain which is unlocking Solana like performance for the vast ocean of ETH devs out there. Now, you're going to be hearing all about se and their new v two upgrade, but if you take away one thing, the EVM is here to stay. There are some problems with it which we're going to get into later in the episode. But say, and especially their v two upgrade is helping solve that.

So thank you very much, say, for making this episode possible. Hey, guys, big shout out to uniswap. Uniswap is delivering the best on chain trading experience, period, bar none. They've got a bunch of cool new features coming out, like uniswap x limit orders, real time data insights. We're going to be hearing all about those features later in the program.

But for now, Uniswap, thanks for making this episode possible. We'll get right back to it. All right, everyone, welcome back to another. Episode of Bell Curve. Before we jump in, quick disclaimer, the.

Views expressed by my co host today. Are their personal views, and they do. Not represent the views of any organization with which the co hosts are associated with. Nothing in the episode is construed or. Relied upon as financial, technical, tax, legal, or other advice.

You know the deal. Now let's jump into the episode.

Lucas

All right, everyone, welcome back to another episode of Bell Curve. Today I'm joined by two big guests, Eugene from Ellipsis and Lucas from Geeto. Fellas, welcome. Hey, thanks for having us on again. Jinx.

Guys, I'm really pumped to have the two of you on here. We're just mentioning before we got on, I think this is almost the one year anniversary of the first Solana Mev episode that we did with Hasu as part of that Mev season, which was a ton of fun and one of the strongest episodes of the season, I thought. And it felt like a really appropriate time to revisit the topic of MeV on Solana because it feels like it's at a little bit of an inflection point in terms of the overall approach to dealing with and mitigating MeV and is going in a very different direction than Ethereum is going. And I want to maybe before we get into so the interesting philosophical divergence that's happening right now for folks that might be a little bit less familiar, I think many listeners will have at least a high level mental model for how MEV works and sort of the transaction supply chain on Ethereum. Why don't you guys just give just a little bit of an overview about what is sort of the, some of the critical differences in between the Ethereum transaction supply chain versus how things work over in Solana land?

Eugene

I can take a stab here and then Lucas can correct me on anything I'm wrong about. I think there's a few big differences between, let's call it Solana block building and Ethereum block building. The first big one is where transactions get sent to by default. So in the Ethereum world previously, the vast majority of transactions sent directly to the mempool. These days they're sent more to private mempools for mev protection.

This is never the case on Solana, where the default behavior has been when a user sends a transaction to an RPC, the RPC will forward the transaction directly to the leader, and then the next future leaders as well. So there's no like big public gossip mempool, which means there's less opportunity for direct front running or direct back running without something out of protocol building up the mempool. And then the other really big difference in my mind is continuous block building versus discrete block building, where under MeV Boost today. And even without Mev Boost, the default behavior for block building on Ethereum is effectively to let a bunch of transactions pile up in the mempool for 11.9 seconds of a twelve second slot, and then build the most valuable block possible in that last bit of time. Whereas on Solana, the default behavior is for the leader to be building the block incrementally, continuously as transactions are coming in.

And so this trades off building the theoretically most optimal block, with pre conformations for users by default, partial blocks getting streamed out to other validators. And one big engineering difference here is the cost to build and propagate and replay the block is actually somewhat significant relative to the block time. So if you were to build the blocks completely in it in a discrete way, you'll run into some computational issues and you run into some network issues trying to propagate the block and make sure it gets included. Lucas, anything to add there? No, I think covered it pretty well.

I think the block streaming piece that he was just talking about is definitely the much harder piece when you start to look at MEV. But I think that was pretty good overview of how that works. Can you say more about that block streaming piece? Actually wasn't something I was super familiar with. I'm sort of loosely aware of how the attestations work in Ethereum, but what's up with the additional.

Lucas

Why is that where so much of the computational load and work is in Solana? Yeah, so I guess Eugene said that Ethereum, you can basically wait 11.9 seconds and then build your block and emit it, and the entire thing is emitted on this kind of like interval, the slot time interval on Solana, the validator is basically, every validator kind of has their own like internal. I don't even know if I want to call it a mempool, but they just have kind of their own internal state of what the pending transactions are. And they are processed as soon as the slot starts. So like a time equals zero, they start processing those transactions and will start propagating them out to the rest of the network immediately.

And it's basically doing that until it reaches 400 milliseconds or the block is full. While that's happening, it's processing, it's sending them out over turbine as shreds. While that's happening, every other RPC and validator on the network is receiving those shreds, reassembling it to the transactions and actually replaying it. It's cool. It's like a video stream, as opposed to Ethereum.

It's choppy. Like on Ethereum, I guess you get the whole picture at once, but on Solana you're kind of just building the picture. Like the lines and the pixels are being streamed out to other validators and they're reassembling it live. Super interesting, and maybe to Eugene underlined this point that you brought out about. I think the discrete block building versus continuous block building is super critical because the approach that Ethereum has largely taken around MeV is running these sorts of auctions.

Lucas

There's sort of this principle where you want to extend out the length of time that you have to run this auction because you get more data right, you have more view of transactions that are going on. You can build a more profitable block. I think one of the other big differences in between Ethereum versus Solana is this idea of deterministic ordering. The way that the fee market works on Ethereum is you can essentially pay pretty granularly to slot where you want one or multiple different transactions to fall within the block. Whereas my understanding of the way that it works mostly in Solana today, is that you don't quite have that option as much.

Right. There's. You do have. You do have a priority fee that you can sort of pay for, but because of something called jitter, it's just you don't nearly as much have that much granularity in terms of where you can place your transaction. And to avoid myself, the embarrassment of explaining jitter, maybe, Eugene, you could give an explanation of why that is, and correct me if I just said anything wrong there.

Eugene

Yeah, that all sounds pretty accurate to me. I think one way to think about it is on Ethereum, if I send a transaction with a slightly higher priority fee than Lucas transaction, and we're both in the same block, I am very, very likely to have my transaction land in front of Lucas's. Or if exactly one of our transactions lands, it's very likely to be mine. Whereas on Solana, with continuous block building, and given the way the default block builder, or what we call the scheduler on the Solana side, the way that's implemented, you don't really have any such guarantee. So there's some combination of latency and priority fee, and what we call jitter, which is sort of this randomness which affects who lands and how early you land in the block.

And so this creates an incentive to spam, because if I send ten transactions instead of one, I'm just going to be a little bit more likely to land ahead because of this randomness factor. And so we call that jitter. And as we understand it today, there are a couple main sources of jitter on Solana today, being the first, the network layer. As transactions are sent to the validator, in what order do they actually, are they actually received by the validator? And the other is within the scheduler itself.

So the block builder, because it's receiving these transactions and trying to include them and execute them continuously, there is some jitter that. That gets created there as well. Yeah, I think a lot of, and I guess to comment on that, I think a lot of the focus, up until maybe a few weeks ago, public, I guess the public focus on this problem, on Solano, is with respect to the scheduler. But now we're kind of seeing a lot of this move to the networking layer. And, you know, I guess back in the last run, and maybe up until, like, mid 2022, Solana was using UDP as the networking protocol to send transactions.

And basically, people are just blasting leader with transactions like 50 gigabits per second of like NFT mints and arbitrage and stuff like that. So they added quic, which kind of lets you like back pressure senders, and you can do quality service based on the amount of stake you have or other. You know, kind of just make up whatever rules you want, but it kind of lets you back pressure people like TCP does. And now we're kind of seeing people start to, you know, map searchers are pretty clever and we'll find ways to take advantage of things. And I don't want to go too much in detail, but you're kind of seeing people start to abuse or take advantage of the quic implementation.

Lucas

I actually do want to get into this a little bit because just for folks who might not even be aware, when you say networking, can you just describe what that ultimately is? Then? Can you also just define what quic is? Because it feels like that's a really core component of the networking approach on Solana. Yeah, so earlier when Eugene was talking about the transaction supply chain on Solana, basically you send your transaction to an RPC server.

It can be like Helios, Triton, maybe you run your own, whatever, it's not super important, but that will look up the current leader and then the next one or two leaders, and it will forward that transaction to the validator over the Internet. And the protocol that it uses for that is quic, it's cuic, it's a protocol. I think it was invented at Google. And it's basically UDP is connectionless. And then there's something called TCP, which is connection oriented.

So like, I have to establish a connection with you and there's like some handshake process and QWik is kind of like a combination of those two things where it's, it's using UDP, but it's, there's some connection to it. And I guess the one benefit is that it doesn't have head of line blocking like TCP does. And I don't know how important it is to get into head of line blocking on this call, but basically, yeah, the validator uses quic to RPC clients and validators use quic to communicate. There's a quick server running inside the validator, which I think it's 500 unstaked connections and 2000 stake connections. I think the theory is that staked senders should get more priority or better quality of service than the unstaked connections.

And yeah, I think that's, that's kind of a high level of how the quick stuff works. Awesome. So I want to get into maybe to just summarize the high level construction of each one of these systems, just one put up against one another. I want to start to get into sort of the pros and cons and just kind of formally define what the different approaches are that Ethereum versus Solana has taken in terms of, frankly, just dealing with MeV writ large. So on the Ethereum system, this is changing.

Lucas

But basically you could very simplistically think about it, that all of these transactions that are happening in DeFi or whatever gets sent to this public mempool that everyone can see, various specialized actors called searchers, put those things into bundles. They submit them to a builder, which goes to a relay, and then ultimately that gets signed by proposer and then propagated to the network. And the defining characteristics of this system, at least from my perspective, is this sort of idea that MeV is going to happen no matter what. The trade off of everything being very democratic and out in the open, is that users end up getting some, like front run, basically front run and back run, which is called sandwiched. So even though we like this idea of transactions being public, although I will say more and more transactions are moving towards sources of private order flow, the negative externality of that is that users ultimately end up getting taken advantage of.

And the way that Ethereum has dealt with this is to return that MeV to initially proposers, to validators. So it's like, okay, well, not great for users of the network, but ultimately it's going to go to proposers more recently, through things like order flow auctions, it's being returned actually to users. Solana, it looks like, is taking a, well, there's very different construction of the protocol. There isn't a public mem. Cool.

Everything just gets forwarded to the leader. But one of the disadvantages of the approach that Solana is taking is some of the negative externalities that the Ethereum approaches up to avoid, which is spam. And then also maybe we can talk about this a little bit later, because spamming is very important to being able to land arbitrages and things like that on chain. Then there's a huge push and incentive towards minimizing latency, which can lead to co locating services, things like that, which we decided that probably wouldn't be good for the overall health of the network or the fairness of the network at large. So it looks like, I mean, you guys tell me, like my, from my vantage point, it looks like there were a lot of reasons to be not super happy with the approach that Ethereum has taken towards MeV.

I think it's just a really tricky, thorny problem. There's no good one way to do it. The biggest criticism of the ETH approach to MeV is that the user seems to come last, I think, in many of these situations, or they have for a long time. Solana, it seems like it's taking another approach here, putting the user first. So, Lucas, maybe I can segue this right after you to talk about why Jitto is discontinuing its mempool.

But my follow up question to you guys is going to be then how do we mitigate some of these negative externalities that Ethereum is trying to avoid, namely the spam and the colocation, maybe, Lucas, you talked a little bit about how am I viewing this sort of ideological divergence correctly, and then maybe we can get into the whole mempool part of Jito. I think so. I think, like, Ethereum is starting to move to this like, thought process, or I don't know what you want to call it, but, like, they don't really, like, no one's really going to be executing anything on the l one, or they, you know, they're kind of pushing everything to these l two s. And I feel like if you look at the Ethereum blog posts and the way that people talk, they don't really ever talk about the users. It's kind of just like ultrasound money and whatever narrative they're trying to push, whatever the flavor of the week is.

I guess the common theme is that there's not really any focus on the user.

Yeah, I think, like Solana, there's definitely different network dynamics where I think this can work. Like, the decision that we made, which I'll get into in a bit, is like, I think it will work. I think the network is like, small enough and has seen enough shit that people have a lot of stake in seeing the network succeed. And so I think that the validator set's much smaller. There's a lot of branded validators that have a reputation, and I think that the dynamics are just different.

And also Solana has something to prove, and that is that it's the best place to build, it's the best place to do execution, and has a very bright future ahead of it. And I think ignoring users is not going to be a successful strategy. We haven't really seen any PBS or frontrunning on l two s. Why do you think that is? It's because they actually care about the users.

And they want to see people stick around and no one's going to interact on l two if they're getting screwed over judo. I had this product called the Mempool. Is like one of the first things we built in. Like, started building in like beginning of 2022 and basically let you see transactions before they were executed. We released it from the very beginning when we had like 3% of stake started growing to 2030.

Defi was kind of dead for a while. And then all of a sudden in December, there was a massive uptick in meme coin trading and just DeFi volumes in general. I think it was the first week of December that we started seeing a lot of this, and we just started seeing behavior that we didn't really like.

We tried a few things to stop it. We tried to prevent sandwiching through code, but kind of ended up being this cat and mouse game where people were going around it. And we realized that it would have been like a really hard and tricky game to play from an engineering standpoint, but also operationally. So we decided to shelve that. And then we tried a few other things, but basically decided that we should probably be spending our time focusing on more positive things.

And so we ended up just disabling that product. And, yeah, I mean, I think the. It wasn't like an overnight decision. We talked to a lot of people about it and, you know, we've been talking about it and trying to do something since December or January. But overall, I mean, I think seems like the feedback from the community is super positive.

And the Mev tips, like last week, I think we were like a few hundred Sol from breaking our record for the amount of MeV tips going through the system, even when the mempool is enabled. So there's still a massive appetite for Mev and getting into the block fest on Solana. I think we're actually doing more validator tips than Ethereum now doing around a million dollars a day. So it's pretty cool to see that too. Like, you know, we ate a lot of shit for 2022 in the first part of the year, but, yeah, pretty cool to see that.

And then we're also seeing massive adoption in the network, even since we disabled the mempool. So I think we're at like 64, 65% of stake today. We're right around 71% of stake.

Lucas

That's awesome, man. That is phenomenal. I've got a lot more questions for you, actually, to follow up on that. But Eugene, I wanted to get your take on everything that Lucas said. About just putting the user first, and maybe your comment on the ideology and different approach and direction that Solana is taking to Ethereum here.

Eugene

For sure. I think the problem space is pretty different. There's no apples to apples comparison here, but at a high level, I view Ethereum as, from a cultural standpoint, really emphasizing incentive compatibility. You see a lot of the narrative around PBS, and the reason for PBS to exist is around centralization of stake, and stake flowing to whoever is going to be able to offer the highest apys. So with something like PBS, you're going to be able to roughly equalize the staking yields across different validators, unsophisticated and sophisticated validators.

So really helping to equalize the solo staking opportunity. And Solana has a very different approach, which is to prioritize the user experience over all else. I think we've seen that a couple of different times. The Jito decision is one that's quite obvious, where users who are playing some of these shitcoin games on chain just really getting screwed by sandwiching, where when the user sets 50% slippage, what they're really trying to say is, not that I'm really happy to get filled 50% worse than the current price, what they're trying to say is, I want a market order. And once you insert front running into the equation, your market order just turns out to give you the worst possible execution.

And we also see that today with, with Solana really focusing on these issues with the network layer and the, and the scheduler, where the user experience on Solana has been degraded over the last few months because it's been operating at such high capacity, such high throughput. And so I think that's the biggest cultural difference. But again, the problems are quite different, where if you go back to 2020, 2019, the sort of origin story of flashbots, just when DeFi was taking off on Ethereum, because the mempool was public, there were all these games that were getting played between these defi traders, some combination of arbitrage and sandwiching. And this is just happening in public, this is happening in the mempool. Besides externalities to users, there's also these externalities to the network.

And the focus was on mitigating these as quickly as possible. And at that time, remember, this is in the proof of work world. There are only a handful of miners with significant stake, and they were already starting to one spin up their own searching operations, which would create better economics for some miners over other miners. And some were already making under the table arrangements with searchers and so I think avoiding accelerationism was a lot harder for Ethereum in 2020 than it potentially is for Solana today. And it's not to say that, like everything that was accelerationist on Ethereum happened, the most high profile example of this was there was an attempt to build reorg as a service on Ethereum, and that effectively got blocked by the social layer, even though it would have been locally good for greedy validators to participate in.

Lucas

Yeah, I think that's such an interesting bit of lore, actually. You can go back on an old episode of Zero Knowledge with one of the, I believe one of the project Blanc guys talking about this. It's an X flashbots guy that was doing that. Um, and it was the social layer. I think the way at least he tells it right publicly on this podcast episode, is that, uh, it was a DM from Vitalik that sort of got him to stop, which was the ultimate, uh, social layer.

But maybe actually to, uh. Maybe just because we're getting only the Solana perspective here. So, to give you the ETH perspective, what I think some folks might say in response to what you guys just mentioned, is that one of the other big differences in Ethereum versus Solana, even though Lucas, those are super impressive numbers. Like, the amount of mev overall in Ethereum for a long time was much, much larger than Solana, and I believe is still larger today. And so I think I'd say maybe it's a little bit less about prioritizing ultrasound money, but it's more about.

I know that some of the folks over at Ethereum land were extremely concerned about the impact of latency and colocation, and very frankly, worried about the impact of spam as well. So the way that I've always understood it is like, I don't know if there's a way to ultimately arrive at some sort of Shangri la type solution, but, you know, if you don't charge, you know, I think the. Maybe the counterargument to what you guys are describing is that, yes, well, Solana is a very small community today. Ultimately, if the network is going to be successful over time, it's not going to be like that. And just saying, hey, like, we shouldn't, you know, sandwich users, that's not going to work as the network ultimately ends up scaling, and you need some sort of structural, something structural in place to get people to stop doing that.

So that's what I guess I'm not really still understanding in terms of the vision here for Solana. And actually, we were talking about beforehand it looks like this is one of the things that people mentioned, lucas, when you guys discontinued your mempool, it's like, well, this was open source, right? So what's to stop another team from just coming in and spinning this up? And I think it looks like another team has actually done that. Although maybe they're exaggerating some of their, their metrics a little bit.

Hey, everyone wanted to give a big shout out to today's title sponsor. Say. Now I want to talk to you guys a little bit about why I. Think say is cool from a design. Standpoint, a big problem that it solves for ETH devs out there and then some cool stuff that say has coming up.

The reason I like say from an architecture perspective is, again, it's a very fast blockchain, parallelization, all of that stuff. But SE has essentially been custom building block space, which is for consumer apps, indexes. Now, they have some very cool features which enable that. So twin turbo consensus, optimistic parallelization, SADB, all of this stuff allows you to reduce the time to finality, make for very, very fast transactions. So if you're building a consumer app.

Lucas

Or a Dex, this is basically the blockchain for you. If you've been building the EVM, you love the EVM, but there are some restrictions about it that don't support your app. So maybe you can't do fast enough transactions or it's not parallelized. Whatever it is, you can now take all that stuff that you built. You don't have to start from scratch, and you can build it on, say, now, recently they've launched v two, but also public Devnet.

So the way that you can follow that and keep up to date is go and follow, say, network on Twitter. All right, thanks, guys. Hey, everyone. This episode is brought to you by Uniswap, delivering the best on chain trading experience, period, bar none. Now, if you know the history of Bell Curve, you know, I did an entire season on Dexs with Dan Robinson, a paradigm.

That one was a blast. We ended up talking about Uniswap quite a bit. So I'm a huge fan of the product and team. There are three really cool new features that I want to talk to you about today. One, limit orders.

Two, real time data and insights. And three, Uniswap X. So, limit orders, you guys know what those are. But now you can buy and sell 700 tokens at your price on your terms. Also, we're going to get into this later, but it's powered by Uniswap X.

You don't have to pay any gas fees to real time data and insights. Now, when you log on to this Uniswap interface, you can see real time charts, transaction logs, pool data, and project information. Looks really neat. All that information is in one place. You're basically good to go just from Uniswap.

And then finally there's uniswap x. You get zero gas fees, no fees on failed transactions, and you get to source off chain and on chain liquidity. And if you want to nerd out about the construction, I think it's really cool. There's an RFQ based system. They farm it out to different fillers that you're getting the best execution price.

So finally, if you want to go check out all this stuff, which I highly recommend that you do, click the link at the bottom of the show notes so they know that I sent you. Cheers, guys. I still am curious about like, what's the ultimate solution to like spam here and some of the negative externalities of things like co location, like what's the end game? Yeah, it's tough. I think no one really knows exactly what the end game is going to look like here, especially if you do have significant amounts of real state on the l one.

Eugene

The problem looks very different for l two s, and no one, I think, is representing this Jito decision, removing the mempool as this thing that is going to fix the network forever. But it does mitigate a lot of the issues that we were seeing in the first couple months of this year. I would guess on the order of high six figures, low seven figures per day of user welfare being lost to the sandwich bots. And I think that's already a pretty big win. It's definitely not a long term solution if you play everything out to the equilibrium in the long run.

Yeah, you probably have some construction that looks like PBS ish. You probably have something that looks like either public mempools or private mempools, because that's like the economically most efficient thing to do, or at least more efficient than what we have on Solana today. But the barrier to getting there, I think, is a lot higher on Solana because of the throughput, because of the pace at which the network runs, compared to Ethereum, where, yeah, kind of like time has stopped for 12 seconds between each block and that's kind of an eternity in compute time.

But at the very least it spies us some time. I think everyone realizes you can't rely on the social layer forever. No one expects to rely on the social layer forever, but if you can buy another year or two of better user welfare and then give some of the big brains a little bit more time to work through what some of the longer term solutions might look like, that's still a huge win for the users and for the network as a whole.

Lucas

Yeah, I think just that is one thing. I think that it just feels like a really stark difference, and we'll just have to see how it ultimately ends up playing out. I think what Ethereum would say is we are thinking about the users. We're thinking on a very long time horizon that we don't want a network which is co opted by a couple big entities that co locate with different validators, and they just get to land all of their transactions. And then in the long run, even though it's not great for users, today, at least we're building a more equitable network.

The issue with that, from where I see it, is that actually they have wound up with like three builders, right, that build all of the blocks, and there are two large staking operators. It's Lido and Coinbase. And actually, interestingly, I'm not sure if you guys have heard anything about this and not sure how much I heard from one of these builders that actually, that whole assumption that users would go to where yield is highest is actually not necessarily true. And the convenience of something like Coinbase is much stickier than everyone thought it was ultimately going to be. So maybe some of the core assumptions that went into the current PBS implementation over on Ethereum weren't necessarily sound either.

But those are sort of the pros and cons. The way that I see it is, ethereum does have this very long term oriented view, but some of the negative externalities I think we see playing out already. Whereas on Solana today, we've chosen to prioritize the users today, and to not create infrastructure that's ultimately going to sandwich them. But we don't really have a long term solution about how to avoid that holistically in the long run. Yeah, and on the Ethereum side today.

Eugene

So, first of all, I totally agree with that representation that a lot of stake is not particularly sensitive to the staking yield. If you're getting 4% or 4.5%, most people, or a lot of capital, just doesn't care. I think one issue with a lot of these economic, a lot of these economic models, they don't actually take into account all of the preferences of the user. Like, just kind of nice to have staking within Coinbase, even if I'm paying 25% of the yield back to the Coinbase entity.

A lot of capital may also have some sort of compliance requirements that prevents them from running their own validators or whatnot. And so a lot of the economic models sort of break down at the edges here. And what you really care about is how strong each of these incentives are. Where if the staking difference was instead of 4% versus 4.5%, like 4% versus 20%, then, yeah, maybe you would see, or almost certainly you would see a lot more gravitating towards the 20%. And so in that sense, you could call PBS a success, where the staking rewards are close enough between all of these different types of staking opportunities.

But the convenience factor is pretty big. Yeah, the convenience factor, I feel like, is just one of these things. Frankly, it's what makes base pretty interesting. Over on the Ethereum roll upside. It's just like they have so many users.

Lucas

I'm sure that user flow is going to be really easy. Yeah, it's just maybe an interesting thing to think about. I've got a question for you guys, because I have a lot of trouble reasoning through this, which is, I think, zooming out of both of these different systems and the trade offs that they've chosen to make. We are trying to avoid some of what we view the problem of tradfi being. Some of the problems in tradfi is, one, there's a lot of regulatory capture and lock in.

Two, there aren't very many entities which are responsible for doing things like internalizing order flow, matching buyers and sellers, et cetera. And theoretically, over time, that could lead to pretty unfair, unbalanced practices. I think everyone's goal, whether it's Ethereum or Solana, is to set up a system that mitigates some of these trade offs. But it does feel like we're sort of starting to gravitate like the laws of, you know, economics and physics are starting to pull on our networks here and over in Ethereum land. Like, we have wound up with three builders that build all the blocks, albeit we have a really decentralized proposer set.

It feels like over in Solana, maybe what we're a little bit more okay with is we don't have this PBS separation yet, so we do have more validators, but it feels like there are more companies and maybe searchers that are kind of doing like, mini co located, like they're running their own validator so that they can place transactions more quickly, I guess, like from, from your guys perspective, when you, when you really zoom out and you're like, what are the. What is the end state that we're trying to avoid and what are we doing better here? Like, what are some of the trade offs that you feel like are red lines that we just should not, should not cross versus, I don't know, just, like, being totally realistic and pragmatic. Like, what are some of the maybe less ideal sort of trade offs that we're going to have to accept but aren't the end of the world, if you know what I mean.

Yeah, I think trying to. I think one of Solana's strongest, like, cells is low latency. But that kind of turns into this game where, you know, you just over a long enough time horizon, maybe a bunch of stake ends up next to, like, the major sexes, like, in Tokyo or something. So I think trying something I think a lot about is, like, trying to avoid that while also not slowing Solana down or impacting the performance of the network so much that it loses one of the main selling points. I guess if you think about MeV acceleration on Solana, I think the most optimal thing would probably just be every leader is four consecutive slots.

You could just hold all the transactions and just pack your last block. With all these transactions that have been delayed a lot, obviously not ideal. And I think you'd probably piss some market makers and searchers and other things like that off, but you would make a lot of money. So I think that's something that I think a lot about is how do we kind of balance those two things? How do we get low latency, but also not make it low enough to where you just have people, you just have these large forces of gravity around economic activity on, like, the geographic scale.

So you know something, the way that might. That might look is like, I guess we kind of try to avoid this right now with our block engine where we're running 200 millisecond auctions, but that's still, like, kind of long. So maybe we can, like, bump those down to, like, 50 or 20 milliseconds or something like that on a longer time horizon.

But, yeah, I think that's, like, the thing I think a lot about. Kind of curious to hear maybe a different angle from Eugene.

Eugene

Yeah, at a high level, I agree. I mean, at the end of the day, no matter how much research you do, the speed of light still exists. Economies of scale still exist.

You know, when I think about democratizing finance, I think of that on, like, a few different angles. One of them is accessibility, and one of them is decreasing the barrier to entry. The really nice thing you get by lowering the barrier to entry and lowering the barrier to innovation is you increase competition, which not in all cases, but generally improves the end user outcomes. And so what we do see in tradfi today is there's a lot of capture. You know, you have these entities that have spent tens of billions of dollars in R and D, and they want to protect those investments, and they're in these pretty entrenched positions, and it's very difficult to compete with them.

And that's the type of thing we'd like to avoid in crypto, to the extent that it's possible.

And I think, like, again, going back to the ethereum side of things, the goal of PBS was to democratize access to MeV, to all proposers. So the less sophisticated as well as the more sophisticated proposers, because we know there's centralizing forces there. And the goal explicitly was to move that force from the proposer layer, the validator layer, the consensus layer, over to this new blockbuilder layer. And I don't think anyone really thought that you're going to have 20 competitive block builders, each with like 5% market share. There's just a bunch of centralizing forces here, but at least they're not on the validator layer.

And even today, we're already seeing these forms of entrenchment by these blockbuilders, mostly in the form of private order flow. So there's like two types of private order flow. One of them is order flow that a builder generates themselves. So this is a builder that's doing their own searching, or perhaps, you know, someday, like coinbase or binance wanted to start up a builder, they send a lot of transactions on the network. So that's sort of their own order flow that they can choose to keep for themselves.

And then there's this other type of private order flow, which is coming from the external world. I believe the biggest sources today are like some combination of these private auction like things, like Mevboost, Mev Blocker, as well as from the Telegram trading apps. And that is a little bit more worrying to me, because that really does increase the barrier to entry on the blockbuilding side, where if the top three blockbuilders have access to this flow, and that's organized through these, I don't want to call them like, backroom deals, but very non public deals, a new block builder is not really going to be able to compete with them. And in fact, like, the more market share you have, the easier it is for you to give better guarantees to some telegram trading app that you're going to be able to include their transactions in a timely manner. It's going to be less costly for you to do that.

So there are some, let's call them economies of scale there. And then of course, there's this verticalization, which again relates to the speed of light, where, because the speed of light exists and it is not infinite, you get this verticalization and co location between the searching layer, the building layer and the relay layer. And if we really care about lowering the barriers to entry and increasing geographic decentralization, that's just something you're always going to have to fight against. And at the end of the day, the network values do have to come into play here because the most economically efficient system is going to be one, even if it is on a blockchain. It's like everything is in the same data center to lower the latency as much as possible.

And there just has to be this value of geographic decentralization. And at some point the network has to decide in some way that this is a strong priority.

Lucas

I agree with you on that. I think that eventually, like the sort of, there's this early wave of early, there is the early wave of adopters in crypto, very ideologically driven for good reason, right? Like, and that's what has created so much staying powders, staying power, such a strong community, et cetera. But ultimately there needs to be some sort of market value to the decentralization that we're creating here. Otherwise, like most of the world isn't ideologically motivated.

Right? Like we need people to actually use this because they have a selfish reason to do that. I do think that exists, though, actually maybe a, maybe a dumb or contrarian take that I don't really wish, but like, it would take one government crackdown to show you the value of this, and it could be this Overton window shift and everyone just wakes up like, oh yeah, this isn't a nice to have, it's a need to have. So I agree with you on that, Eugene. I think that's, yeah, I think censorship resistance is very important to the network, but the users are not going to care about it until they're getting censored.

Eugene

And then from the network layer, at that point it might be too late. You might have to do things that are a little bit unpopular or the people who care the most about the users today would disagree with from a prioritization perspective, but hedge against some of these tail risks down the road. Can we talk a little bit about some of the just what's going on in the UX of Solana right now. So I feel like the amount of activity that's happening on Solana has exploded recently. You guys would know that better than anyone, certainly than me.

Lucas

And I don't know, maybe you have some statistics or something off the top of your head, but it feels like Solana is actually winning in terms of things like Dex volumes, even transaction fees that are landing on chain. I saw surpassed bitcoin, or got very close to bitcoin, which is pretty nuts. But one of the side effects of that, which is maybe slightly less good, is the UX has been impacted here. So there have been more failed transactions. It's taking longer for transactions to succeed.

The fees have started to go up, and the network has become a little bit less usable in recent days. So I'm curious, like you were, I think, Lucas, you were mentioning this a little bit. There was a debate between, I think there's the margin, five folks in labs about, is this a scheduler issue versus a networking issue? It seems like we've aligned on that. It's a networking issue, but, yeah.

What's, I guess, in your guys mind, what's driving some of the degradation in UX on the Solana side of things? And do we have a fix? Yeah. So, yeah, there's been insane increase, like record volumes. There's a lot of meme coin activity, which at the surface seems like pretty insane.

But if that's what people are doing on your network, then you need to make sure you're taking care of them. And so I think last time I looked, validators were getting around 50,000 packets per second. 50,000 transactions per second. And, you know, blocks are like roughly, I think they're like a few hundred or like 1000 non vote transactions. So there's, there's not enough block space.

There's a few things that are going on right now. One is kind of looking at this networking layer a little closer and figuring out how to make it perform better. There's some scheduler changes that are coming up in 118, which is the next software release, which should be live in like the next one to one and a half months, which basically the schedulers change to have less jitter and respect the priority fees more. So I was actually like, if you look at blocks now and you like, order, if you look at the priority fee throughout the block, it's basically random. I looked at some of the blocks where there's some canaries that are with a very small amount of stake running 118 with this new scheduler.

If you look at those blocks, the priority fees are nice and ordered. So I think fixing the networking layer, fixing scheduler is huge. There's a lot of discussion around fees. I think I've been a part of, like 20 different conversations, very heated conversations around fees. You know, does Solana need dynamic base fee?

What does that look like? Things like that. And we're trying to help out with some of our infrastructure as well at judo, trying to, I think we have capacity to help out, kind of speed these transactions up. So providing people that infrastructure and things like that. But, yeah, I mean, these are all like, obviously, in a perfect world, these things would be working perfectly and everyone would be happy.

But Solana is like hitting some massive growing pains. And so I think it's good problems to have and shows that people want to use Solana and people are excited about trading on it. I also, I want to, just because I've been pushing a little bit on the ethereum side of things, I do want to push back on some of these people, some of these takes that I've seen where it's like, oh, you know, Slan is a meme coin chain. Sort of looking down at it as being illegitimate. Like, a lot of meme coinery that used to happen on Ethereum, and I didn't see anyone pushing back saying that was a legitimate activity back then, you know, so I feel like that's not a very fair criticism that I've seen come out.

Lucas

So I don't know. Yeah. Just wonder. It's. I mean, yeah, it's pretty crazy.

If you go to, I think it's analytics. Soulscan IO. If you scroll down, they have a. There's a chart showing the number of new tokens per day. And let's see, I mean, back in, like, let's see, back in like, December, there's maybe 500 or 1000.

Today there's been 7000 so far. We peaked a few days ago, March 15, there was almost 10,000 new tokens launching a day. Man, that is nuts. And you know what? I just feel like every freaking discussion that I have these days devolves into meme coins.

Lucas

I don't really want to give a take on meme coins, but I do think it's indicative of where new activities happening. Like, I just don't think you could look at this any other way, that this is extremely positive that this is happening on salon and probably a pretty good indicator of, like, when new users come where they're ultimately going to land, right? Like, it feels like this bodes. This is a good forward looking indicator for who's going to end up acquiring more users this cycle. So.

For sure, yeah, I mean, this is like a pretty crazy metric. And then one more I'll share is this is defi llama Solana volume. And, yeah, you can see, like, you know, let's go 200, roughly $200 million a day in November. And then, you know, a few days ago, it was hitting, like, above 3 billion. Looks like we're right around one and a half billion today.

So there's just been this crazy uptick and, you know, good problems to have. But it's also starting to stress some of the engineering and potentially economics of Solana. We call that a champagne problem, but, yeah, also a real problem. What do you guys think? I would be curious to get your take in.

Lucas

Eugene, you're at our office as a couple months ago, we were talking about Solano fee markets in general, but it does feel like that there's kind of this debate that was brewing now is maybe like moving a little bit more into the fore and talking about doing some sort of fee overhaul on the Solana network. Like, you know, ultimately, now everyone talks about 1559 within the context of the burn and tokenomics and stuff like that. But originally it was really intended as a Ux upgrade for Ethereum, right? Like, before that, Ethereum has obviously much higher fees than Solana. There was essentially a first price auction.

I don't know if you guys remember interacting on Ethereum back then, but it was impossible. It was like you had to place. You had to place an order, some amount of gas. You just guessed that most of the time it didn't land. You found out ten minutes later.

You're like, God damn it. So originally 1559, even though now it's always gets touted about the tokenomics, there's really just a Ux upgrade. I think the question is, like, is Solana going to get one of those eventually? And maybe dynamic base fees, just like 1559 would be a part of that. Eugene, what's your take on that?

Maybe you could summarize some of the debate that's happening internally in the Solana camp for folks who aren't as in the weeds as you are. Yeah. So the fundamental thing that's happening is the demand for block space exceeds the supply of block space. And so you need some mechanism for allocating the block space among that demand. And so, like, one very natural thing to do is increase the price until you've equalized supply and demand.

Eugene

Obviously, you would also like to increase the amount of block space, but you can't do that forever.

And so, yeah, there's this, like, big ux effect on Solana today. I think basically all the arguments you see on Twitter about Solana's failed transactions are wrong. There's a lot of nuance here. The first piece is that there's multiple types of failed transactions. So when you send a transaction to the network, let's say it's like six months ago, you're transferring from, just transferring funds from one address to another.

You sign the transaction sent off to the network, and less than a second later, you hear back from the network, okay, you're good to go. You paid like one 100th of a penny. And now what happens is a lot of times, your transaction just never makes it on chain, because, again, the demand exceeds the supply, which means plenty of transactions never show up. So there's that type of failed transaction, which is the user sends it off, they wait for a minute, and they just, like, never hear back from the network. Essentially, like, you send your transaction off, and then it just, like, disappeared into the ether.

And then there's this other type of failed transaction, which is your transaction landed on chain, but the state change that you wanted did not occur. This would usually be some type of trade where you're buying some meme coin, you said 10% slippage, and then by the time your transaction hit the scheduler and was scheduled for execution and made it into the block, the price had moved enough such that your trade doesn't actually go through. And it is really difficult to observe the former, because you can't. Like, if all you're looking at is public chain data, you will never be aware of all these transactions that have been sent out that just never made it to the ledger itself. And so only, like, RPC providers and validators have access to this data.

And then, of course, a lot of those transactions that are coming in, if there's 50,000 of them, I assume a bunch of them are these like, spammed arbitrages and NFT mints. When those are happening, and, you know, at Ellipsis Labs, we're application developers and we have some amount of users. And when it's really difficult for users to land the transactions, it means our application UX is a lot worse. It means the users are complaining all the time. And the only lever that app developers today have to pool is increasing the priority fee.

And because this mechanism doesn't work particularly well with the current scheduler, it's not very effective. And what ends up happening is the application developers just like continue to increase the priority fee, so massive ux degradation. And at the end of the day, in the same way that the meme coin traders just want to place a market order for some meme coin, the vast majority of users just want to know, hey Solana, just tell me how much it's going to fucking cost to land my transaction and then I'm going to decide if I'm going to pay it or not. And if I decide to pay it, you're going to include my transaction really quickly. And for something like this, you are going to need some type of fee model, whether it's enforced by consensus or not, where you're really giving the choice back to the user in terms of are the transactions going to make it on chain or not?

And so I view the scheduler issues or the fee issues as separate than the networking layer issues. I think there's pretty significant room for improvement on both sides and there is progress being made on both sides. But even if you had a perfect networking layer, at the end of the day when the demand for block space exceeds the supply, you're going to need some sort of mechanism here. And one very natural thing to do looks sort of like 1559, where, you know, we can toss out anything about the burn. The real important piece there is this natural fee escalator where if you're seeing a bunch of blocks that are really full, that probably means that the market price or the market clearing price for the fixed supply is actually higher.

And so you need to increase that price. And if you can do that in a predictable way, then you can.

Deliver. The UX that we want to have. And this is the basis behind one of the proposals that is out there to implement an in protocol fee model. It's a little bit different from 1559 because there's not just a global block space limit of 48 million compute units or gas units per block, there's also a per state, per account compute limit. And so this proposal that's out there, which is called SIMD 110, implements this sort of local fee market where you have to pay for each account that you touch in a transaction and the amount that you pay is based on prior utilization of those accounts.

Eugene

And I think in ideal world you would implement it both at the per account layer as well as at the global layer. Because again, the demand is just like really, really high relative to the supply today. And you are going to run into some UX issues here no matter what. Because from the end user perspective, I think the vast majority of user demand is really insensitive to whether the price to land on chain is, you know, one 100th of a cent versus $0.01 or maybe even versus ten cents. And the folks you're going to price out here are going to be the spam bots, which I think is still a pretty good outcome.

But I think if activity stays at is basically no matter what sort of fee mechanism you put in. Yeah, the fees are going to be higher than they were six months ago, and that's just a function of the demand increasing. So how similar, Eugene? I think back in, whenever we did this research hanging in the office, people were talking about something called pros, which were program rebate able account write fees, which is sort of similar. Right.

Lucas

It addressed this idea that you can basically write lock any number of accounts now for free or up to like 64 or something crazy like that. And this would be an automatic payment for doing that, which then the owner of the account could choose to rebate. I mean, is this sort of a similar type idea? And do we need to fix something with. Because basically what it sounds like you're saying is that for the way the base fee gets set in 1559 over on Ethereum, it's sort of a.

There's a global. You're not. You're not doing pricing based on discrete pieces of state or anything like that. It's just this very global sort of unit of account which has sort of pros and it has cons as well. Like, you can make the argument that just because NFT mints are popping off, that shouldn't really impact defi activity and vice versa.

So. Yeah, sorry, that was like five questions in one. But is this basically just a more. It's almost like, should listeners be thinking about this as sort of the 1559 base fee? But you can be much more specific based on the type of state that's being impacted.

Is that the long and short of it? Yeah, that's exactly the way to think about it. Where the gas model in Ethereum is such that you're basically trying to price all the resources, all the limited resources of the network into a single dimension that we call gas. Now, technically, there are multiple dimensions with the blob space markets as well. Yeah.

Eugene

Which operate on a different or a very similar model, but like separate, completely separate from the gas market. And on Solana, there's many millions of accounts, many millions of pieces of state. And this proposal would be to implement this type of market on every single one of those pieces of state. And that's done, at least in this proposal at the network layer or like a Solana wide enforced by consensus type of deal. And the pro proposal that you're referring to, which came from mango markets, I believe, in December of 2022.

So quite a while ago is more to put that fee decision in the hands of the application developer. So the application developer will say sort of like, hey, I own this piece of state and I'm going to charge anyone who touches the state some fee. And then the idea behind that was, if you're some sort of DeFi application and you're seeing a lot of these spam like transactions that are coming through, maybe a lot of them are failing, you still want to charge the failed transactions that are eating up this block space, eating up this limited resource, and then rebate. That's the r in PRA to who you think the good users are. So like, one naive implementation here would be, I'm just going to charge a dollar to everyone who touches the state, and then if later down the road I decide like, oh, these were like real users, people interacting through the UI, their transactions succeed rather than fail.

I'm just going to rebate that back to them. So their effective fee paid for the state is zero or negligible, whereas these spam bots. I'm going to make this opinionated decision to say, this is like a bad use of the block space. I'm not going to rebate you. So it's really a question of like, sovereignty over the block space.

Does the network own the block space? Does the network own these, like, account hotspots or is that something that belongs to the application that created the hotspot? So, like, who sort of deserves that economic value? That's actually something that canto experimented with rebating the gas to the application developer for the, or some portion of the gas that their application use. I think it's an interesting model, and it is kind of a philosophical debate at the end of the day of who owns the block space and deserves the economic rights to it.

Lucas

Well, it is a philosophical debate, but I feel like it also is sort of an important debate that might end up coming sooner rather than later. And one thing that at least I've started to hear sort of like rumblings about is the possibility of Solana L two s as well. And I actually saw, I haven't actually confirmed this with the team, but I saw that, like, zeta markets very quietly just launched or announced that they were launching an l two on Solana, but I'd be curious if what your guys thoughts are maybe less on, I don't know, like general purpose l two s on Solana don't make an enormous amount of sense to me, but maybe app specific l two s for exactly the reason that you're describing there. Eugene, less of a, you know, in ethereum we have l two s because gas fees are so high on the l one that you needed basically to move a whole bunch of that compute essentially off of ethereum main chain up to this other sort of area. But maybe the other sort of reason why you might want to do that as well is you just have a lot more granular control over your block space and the economics that you create.

And it does feel like there is an incentive maybe for applications on Solana to do the same thing. And they'd be motivated not necessarily by cost or speed, which is already really great on Solana main chain, but maybe I could see app developers that are interested in having more sovereignty over the economics that they're creating, or, I don't know, additional aspects of control, something like that. Have you guys heard any rumblings about this, or think it makes sense or you wouldn't be supportive or what do you think? I haven't seen much beyond the Twitter discourse about Solana L two s. I think it's more of a gut reaction to it seems like there's not enough block space, and one way to create more block space is through an l two, I think.

Eugene

Well, I haven't thought about this too much, but my initial reaction is it doesn't make too much sense. If you are going to go build an l two, you want a really good settlement layer that gives you guarantees around things like when I need to post my transaction, I can post them. You need the network to be up, like actually 100% of the time instead of 99.9% of the time. So you can post your new state routes in your state transitions.

And there, I think because Ethereum has been, well, really for quite a few years now, focusing on being the most robust settlement layer, it probably makes the most sense to either settle there or settle to bitcoin, which also has a lot of pretty strong guarantees, very easy to reason through compared to Solana, where if you're building an l two, you actually don't need the settlement layer to be particularly performant. There are some cost implications for sure, but Solana's huge advantage today is on performance and speed and sacrifices, some other things to get there. And for a settlement layer, I don't think you need the performance and speed that Solana operates, that Solana provides. Can I challenge that a little bit, though, because one thing that I have at least thought about is right now, one of the things that people are very interested in on Ethereum is this idea of base sequencing. And I think the reason, I think there are a couple of reasons to be interested in that, which is one, I think there maybe is a little bit of like, worry from the Ethereum, the asset side of things.

Lucas

Like, hey, no one uses us for da anymore. And is settlement, is that really enough to number go up technology for eat the asset? I don't know. So the idea of there being additional lock in by using Ethereum itself as a sequencer. But I think the other reason why people are excited about it is because Ethereum is like, there are twelve second block times, right?

And interop is a massive challenge from l two to l two, that it actually is. It would be nice in terms of fixing some of these interoperable problems, but the problem is Ethereum itself isn't super performant, whereas something like base sequencing on Solana would be. And so what you end up having. To do is these like, shared sequencer networks. And like, you could make this argument that you're basically creating another ethereum on top of Ethereum.

And it's like, man, there's so many layers to all this, right? Whereas honestly, in Solana, I could kind of see it like, well, actually the block times are really fast, right? It's like 200 millisecond block times. Like, honestly, you could probably just route something right down to Solana and back up to another l two and it'd be hardly noticeable. And maybe actually liveness is not the thing to be optimizing for.

And instead, the performance of the l one is kind of critical. I don't know. Like I, this is just something I've, you've definitely thought about this more deeply than me, but I could see that argument happening as well. Yeah. The interop problem for l two s is really cursed right now.

Eugene

I think it depends a lot on if you're a zk l two or an optimistic l two because of the time two finality. So, for example, if you're an optimistic l two on top of Solana.

And. You still have some fraud challenge window before which you commit to the state being finalized, you're not going to get any of the performance benefits doing some sort of interoperability by going back to the settlement layer, because you're still going to have to wait for the state to be finalized before the other l two will accept this, or before the other l two should accept this. And the only way you're going to get that is through some sort of shared validity, which is basically like taking multiple l two s and making their validity depend on each other, which is basically like building one big l two.

Eugene

And this is going to be different if you are able to zk everything in a more performant way where. Yeah, once you are able to post back to the settlement layer, then everyone can treat it as finalized. But I think the. The tech to do that in a performant way is just, like, not really there yet. The primary bottleneck is the prover cost.

So for an optimistic l two today, you basically pay for two things. You pay for Da and you pay for settlement. And then. Yeah, like, the sequencer costs some amount to run, but let's call it negligible. And the DA is the biggest cost here.

And with the ZKL two, you also add in this prover cost, which is linear in transaction. So it's actually, like, relatively expensive. And those numbers are coming down, like, really, really dramatically. So maybe in a couple of years those will be ready. But if the goal is to increase the amount of block space available to Solana, like things, or things that need access to the Solana state, the optimistic version is just not going to work too well.

Lucas

Yeah, well, I feel like it's still early days. Very maybe nothing ends up amounting to anything there, and there's just been rumblings about it, but I've just started to hear people talk about it. Was curious to get your guys opinions. Yeah, it feels a little bit more like a narrative play than like a tech play that makes a ton of sense. Yeah.

Yeah, you might be right. All right, fellas, I know we gotta start to wind down here in a second. Any, I don't know, closing thoughts on anything that we might not have gotten to on just salona fees, mev, in general, I don't know, anything that we missed or stuff. You want to something that you want to leave the audience with? Yeah, I got one.

I think the idea is still being developed. There's a new SIMD out that I think is pretty interesting. SiMD 130 dynamic block limits. And basically, like, the idea is like 48 million compute units is kind of like an arbitrary number. I think the math was like, a compute unit's roughly like 33 nanoseconds or something.

And there's four threads, and, you know, it's pretty hand wavy. But this proposal is pretty interesting. It's basically like you can potentially increase block space if the utilization is.

If people are using more than, I guess this proposal says 75%, choose a number there. But if the network's being used a lot, then you can potentially increase the block space on the network. And basically, yeah, it's pretty interesting proposal. I think there's another proposal that would probably, probably be nice here, which is timely vote credits, which Santetsu at Shinobi systems has been leading up a lot. Basically, you get rewarded more credits if you are voting on a timely manner.

And so you can imagine like it could be. I think it's pretty interesting proposal if you like, start to increase the block size and basically increase it until like some percentage of the network can't keep up. And then, yeah, it would be a good incentive for validators to upgrade hardware. And just kind of an interesting discussion. I don't know if something like this has been proposed other than like the gas increasing the block size on Ethereum.

Lucas

Well, shout out to the author of this particular proposal, who is caveman lover boy. So yeah, Mister KV is a legend in the Solana community. I agree with the high level here, which is, again, because the demand exceeds the supply, we should do whatever we can to increase the supply of block space. But clearly there's some max amount of block space that the network is able to provide at a while maintaining whatever guarantees you want to maintain around like hardware requirements or whatever else. And then the natural progression would be, okay, we'll increase the block space if it's highly utilized up until some point.

Eugene

And then the question is, why do we need this controller in the first place? Why don't we just increase the block space to that and leave it there? Unless the goal is revenue maximization rather than Ux and fees. I think like at a cultural level, Solana really prioritize user experience. I think that's Solana's biggest competitive advantage over every other blockchain today.

And Solana should never toss that out.

Lucas

Eugene, what do you. One of the. I don't know, maybe color philosophies or design choices or whatever that I see. One of the biggest differences that I see between, I actually think Celestia and Solana are similar in this instance, and it contrasts against Ethereum versus bitcoin. It's kind of like to really drastically oversimplify, like the small scarce blocks versus larger, more abundant blocks.

And this kind of, this idea, like this was famously what the block size were in bitcoin were fought over, right? It's like keep the blocks extremely small. And there are some desirable properties that you get out of that. You can keep really low hardware and bandwidth requirements for the node operators on Ethereum, or the node operators on bitcoin, or the proposers on Ethereum. But the downside of that is fees spike through the roof because you're artificially constraining the amount of space that you can write to, whereas you could just take this totally different approach and say, honestly, we should be striving, the fees shouldn't be the dynamic thing.

We should be targeting extremely low fees, and we should just build out our networks of computers to accommodate for more block space. Why wouldn't we do that? So I guess there is some theoretical limit, right, to the amount of block space, but why shouldn't we just be? It's almost like the inverse of the way that Ethereum and bitcoin think about it, which is like, keep the block size relatively static, right? It is dynamic a little bit, but keep it relatively small and constricted.

Let the fees move. But you could just like target really low fees and just have the block space sort of modulate, which is what it seems like this SIMD is saying. I don't know. I think the elephant in the room here is the value capture question, where the revenue maximizing thing for the network is not maximally increasing the block space. And then, yeah, the other piece that you touched on is the hardware requirements, where I think Solana and Ethereum can't really speak for celestia or bitcoin have very different approaches, where Ethereum has somewhat explicitly, or parts of the Ethereum community have expressed a preference for Ethereum being world war three proof.

Eugene

And, yeah, that means you're not going to be able to have or rely on like data centers, for example. And then it comes down to, okay, so what should be the minimum hardware requirement to be able to participate on the network? Is it like I should be able to run this, like run a validating node from a ti 84 on the moon? Or should it be like, if I have a ten gigabit Internet connection and $20,000 of hardware, that's the guarantee. And like, you know, if someone is sufficiently motivated, they'll be able to acquire that.

Lucas

Okay, but let me, I guess this is, this is, I've had a really hard time articulating this thought, but all right, let's just play this out, right? It's World War three resistant on ETH main chain, but you get all these new super powered l two s, like mega ETH or whatever, right? And everyone builds applications that are based on having extremely abundant, very cheap block space. Right? Like the unit economics of all these businesses that got get built on the l two s are based on l two economics, not based on l one economics.

Then World War three ends up happening, and all of these centralized sequencers that are getting run out of data centers have to shut down. What the hell do you have? You have a bunch of businesses that don't make any sense based on the network. This is what I don't. I don't actually think if you were to, like, really imagine a world war three scenario, I don't think the current system actually supports that.

Like, yes, you would probably get some base functionality. You'd still be able to transfer ETH tokens and stuff like that, and that wouldn't be disrupted. But theoretically, the entire value of the network would be based on the apps that are built on the network, and they'd all be built to a spec that no longer makes sense in a world war three scenario. I don't know, this has always bothered me. I feel like the logic doesn't foot.

Eugene

Yeah, I think no matter what, during world War three, some stuff is not going to work as well as we hope it to. Yeah, that's true. I think that ethereum folks would say here, well, yeah, some of the l two s are going to get into trouble, but if the l two is in a roughly finalized state, you will still be able to do things like force, include transactions through the l one. You'll still be able to.

Yeah, like transfer ETH, transfer assets anywhere around the world in not quite the speed of light, but in like 12 seconds. And there's a lot of value there still. You still go to, like, use uniswap on Mainnet. Yeah, maybe the fees will be high, but it'll be better than having no system at all. Okay, that makes sense.

Lucas

Let me address the point that you made about revenue and that sort of tension. How do you guys think about revenue from that? Like, I guess, revenue to who? Because, I don't know, there's kind of. It's unclear to me that we should.

I've seen a lot of people try to jam, like, l one transaction fees into a p and l format, and I'm coming down pretty strongly on the side that that doesn't make a lot of sense. I think there can be intersecting sets of p and ls around a network. Like, I think a validator. P and l makes a ton of sense. Right.

Like that is. It's a. It's a business. They're a service provider to the network. But sol ETH, bitcoin, these are commodities.

You wouldn't say, what's the p and l of oil? That's a nonsensical statement. And so I guess when you say it's unsustainable from a revenue standpoint, from whose perspective are you talking about? Because I'm actually not 100% clear or certain on the idea that we need to be running this positive p and l from a fees perspective for the network. Yeah, to be clear arguing for that either.

Eugene

But I'm just saying that the revenue maximizing thing for the network is not to maximally increase the block space.

But, yeah, I don't think, like any of these assets, any of these crypto assets today are valued based on some DCF analysis. That's like the first, most mid curve thing anyone does when they get into crypto is like, okay, what is the revenue? Does this make sense? What's like the, the pe on Ethereum?

And, yeah, so I think none of the coins trade like this today. It's unclear if that will be the case in perpetuity.

And you also sort of see this pivot from the narrative side that Ethereum has been pushing for a while towards like, ultrasound money narrative, which is effectively trying to turn Ethereum into a meme coin, or at least get valued as, as a meme coin rather than based on the fundamentals.

Why do you think it not. Why do you think it's not revenue maximizing to increase block space? Yeah, I didn't. Well, there's, right now there's 50,000 transactions that want to get included, but there's only 50,000 /second but there's like 1000 or 2000 that are actually getting included. So it seems like, I'm an engineer, I'm not an economist, but it seems like if you were to include those, then validators would make a lot more money.

Eugene

Yeah. So it depends how much you want to increase it. So, for example, you increase supply to infinity. Then the price becomes like epsilon, unless you have some. Well, yeah, you have some like, base fee mechanism in there as well, which would have to be enforced by consensus.

But at the end of the day, it depends on like, the elasticity of demand. And then there's going to be some sweet spot between, like, price and quantity that is revenue maximizing, but it's almost never increases supply to infinity.

I think it's very possible that if we doubled the block space today on Solana, that the validator revenue just goes up. But I think if you 100 exit, assuming that was possible, it's not clear whatsoever. Do you guys, I know that this. So just also, I'm not a huge fan of the ultrasound money meme. I do think my framework for a long time has been.

Lucas

I think these l one assets look like something like commodity money type things, and I think there's going to be a basket of ones that people ultimately end up holding and using. But do you guys, I know there's less even appetite in Solana to talk about things like this, but do you view soul as potentially really a long time down the road, like some sort of store value, unit of account money like thing, and then, I don't know, it sort of feels like the. There's like two ways that you could look at the business model of and l one. And one would be like, you're an operating system for applications to build on. The other would be you're an exporter of a commodity money thing.

And in theory, those two things should be really linked. But they're kind of not, are they? I mean, you can just look at bitcoin, which has basically been ossified for years and years. Very few technical upgrades have been made, but obviously the export of the money has been pretty good. Just look at what bitcoin's done.

And it's been pretty divorced, frankly, from much technical progress. So I don't know. I mean, I know it's kind of a high level question, but how do you guys, what's your sort of mental model and framework for looking at these l one s in general, I don't. Have any view on what is money, what is not money, certainly not in the long run. I will say one.

Eugene

One thing that is important for these proof of stake l ones in particular. Is. The economic security of the network does depend on the value of the network token. And so there is some incentive from the network perspective to do what it can to pump the token, whether that's meme it into existence or generate revenue and return it to the token holders.

Yeah, I think I would agree with at that point.

I think that Anatolia kind of has an interesting view on this, where a lot of the stuff, it will be somewhat commoditized in the future, like block space and things like that. So I think viewing it from that angle is certainly pretty interesting. When you compare Solana versus Ethereum and the high fees versus low fees, and trying to maximize the amount of block space on Solana. I mean, I think Sol right now, it's like access to block space and Mev. And I don't think that focusing on ultrasound money is a good thing for Solana to be focusing on.

Eugene

Hmm. Yeah. Fellas, I appreciate you've been super generous with your time, and I think we covered a lot of really interesting ground. If folks want to follow either one of you, find out more about the work that you're doing, like, what's the best way for people to do that? Yeah, you can follow me on Twitter.

Buffalo Cheeto Labs as well. And, Lucas, what is the origin of Buffalo?

Lucas

I will put you through that if. You don't save it for you. Gotta buy me a beer to get that out of. Yeah. All right.

We'll get permissionless this year. Eugene. Yeah. You can follow me on Twitter. I'm X Shittrader, and you can follow ellipsis Labs as well.

Sweet, fellas. This was a fun one, as always. Maybe we'll do another one of these in a year. We'll give the big update on a lot of Mev, make a manual thing, but appreciate you guys.

Solana's MEV Inflection Point: Users vs Protocol | Deep Dive | Eugene & Lucas

Primary Topic

Episode Summary

Main Takeaways

Episode Chapters

1: Introduction to MEV

2: Technical Differences

3: Philosophical Debates

Actionable Advice

About This Episode

People

Companies

Books

Guest Name(s):

Content Warnings:

Transcript