An Economic Model of the L1-L2 Interaction with Fahad Saleh | a16z crypto Research Series

Transcript

Hi everyone, welcome to today's A6Z crypto research uh seminar. Very happy to welcome Fad Salah uh who's sort of uh has just seemingly infinite energy and has used an awful lot of it to do good work in the crypto space broadly. So very happy to host him today and he'll be talking about something that's actually been on the mind of a lot of us here um recently which is value capture between L1's and L2s. So Bahad >> well thank you Tim and uh thank you for giving me the opportunity to hang out over here and and and the opportunity to present uh it's been a lot of fun and I'm hoping to get uh some good feedback here. Uh let me say I'm actually going to try to achieve two purposes uh through this presentation if it's not too ambitious. one is I do want to uh

provide you sort of the perspective of a financial economist on um whether L2s are beneficial to L1's and that's the focus of the paper though one of the important pieces um where I think sort of the the economics comes in is that in the analysis I'm going to be showing you I'm going to be thinking about things like total value locked indogenously so based on economic principles um and so I also wanted to spend a little bit of time after I get through the introduction kind of talking about how economists more generally think about this which will also sort of relate to the approach that's in the paper. Um so I guess one way one way of saying that is if you find the actual subject of my talk completely boring maybe you'll get something useful out of it in terms of um uh some background on the economics

literature that you might find useful for instance for your own work. Um but please you know ask questions um throughout um I I I really prefer to engage rather than to turn this into a monologue. Um but okay with that let me uh dive in. So in this audience uh I don't think I probably need to provide a lot of motivation but you know I have the slides so I'm just going to quickly go through them. Uh so here uh I got the ether price and I got the ETH plus L2 uh TPS. Um I suppose TPS is probably not the metric this audience wants but these are general purpose slides so you'll have to bear with me. Um and so I'm showing it to you at two points right?

So one point is the day before the Deacon upgrade. Um and the other point is uh the day after uh sort of I guess the beginning of the vibe shift in the Ethereum community um when the EF sort of started putting out announcements about um kind of uh prioritizing things like L1 scaling, right? And so this I think is reasonably stark, right? Which is that ETH prices before the Deacon upgrade were close to 4,000. Uh you fast forward a little more than a year uh and they're under 2,000. uh despite frankly I think the political environment actually being pretty good for crypto.

Um and so Bitcoin was probably near all-time highs for instance at the second data point. Um so ETH prices collapse but at the same time uh the transaction activity in the Ethereum ecosystem goes up which in some sense was kind of the point of the beacon upgrade. So it's kind of like uh I think more than anything what I want to sort of draw out which is probably you know familiar to this audience anyway is that there is sort of a gap between Ethereum the blockchain and ether the asset right uh and certainly Ethereum more broadly conceived when you include the L2s there's a gap there and so it's not totally obvious for instance if you're scaling through these L2s um successfully that uh that's going to be good for uh for the valuation of the asset ether right so uh just in case

people are worried about you know me picking two data points out um uh I'm showing you here the time series of actually here this is ether market value uh valued in terms of bitcoin so essentially I wanted to take out the crypto market factor right um and you know this is not quite a line straight down but it's pretty close to it and again this starts the day before the deacon upgrade and it goes up until the day before uh the Ethereum Foundation posted about sort of rep prioritizing and and and and making scaling the L1 kind of an important piece of that. So getting more to the paper, right? Uh what I really want to emphasize here um is the core economic channel in the economic model I'm going to show you

right now, right? Um so I think you know the idea that for instance having blobs and and thereby reducing transactions costs for data posting that that's going to accelerate transaction activity on the on the layer 2 that's you know that's the point essentially right. Uh but the the thing I want to emphasize is that uh if you are going to drive transaction activity onto the layer 2s, that's not the only thing you're going to drive to the layer 2s. Why is that? Well, think about the incentives of somebody to provide liquidity on layer 1, right? So for instance, what is the incentive for people to provide liquidity at unis swap right on layer 1, right? if all the transaction activity is migrating to layer 2, there's going to be a natural shifting of the uh investment capital of the TVL also to

layer 2, right? And so that actually is the part that I want to emphasize because I think everybody sort of gets that um sort of the the things like the Deacon upgrade should shift transaction activity, but uh it seems to me the part that's missing in the conversation is what exactly that's going to do to the uh to the liquidity um on layer 1 versus layer 2 and then what that might have to do uh to uh valuations of ether. Now, one thing I want to uh point to here is that there's a very strong assumption in the model I'm going to be showing you with regard to the relationship between TVL and uh and the value of ether. And the reason that I'm going to make a very strong assumption about that is that I I kind of think what we're assuming is

more or less capturing the the sort of the reality of the situation, but I don't really know how to think about the deeper forces underneath that. And so, it's kind of hardwired in the model. So this is one of those points where actually I'd be you know I'd be curious to get feedback on um which is um it seems to be the case that if you look at you know TVL on the layer 2s versus layer 1 um a lot of that TVL on layer 2 is in terms of ether but it's less in terms of ether than it would be on the layer 1 or than it is on the layer 1 rather right so for instance like synthetics of bitcoin say coinbase wrap bitcoin on base um seems to be relatively more popular than it is than say wrap bitcoin is on layer on right

and so um if you in if you envision a world where uh people hold ether on the layer 2 to the same extent that they hold it on the layer 1 then probably you don't really have to worry about the kind of the channel that I'm describing here at least to the extent that you care about ether prices because you could say well fine the TVL will migrate to layer 2 uh but it'll still be basically all in ether and so the value will still be there for ether and essentially layer 1 becomes like a data posting layer or something like that right nothing really going on in terms of transaction in terms of uh let's say DeFi. Um but you don't necessarily have to worry about ether prices collapsing in that context. Uh but to me that doesn't seem to be the case, right?

Practically speaking, it does seem to be >> quick question um there. So or should we think also of this as being like very it seems very financed. So, if for example like 90 not not that I think this is actually like a a likely near-term outcome or maybe ever outcome but if like 90% of use of all of these layer 2s and maybe of the Ethereum network itself was consumer application something that like you know sort of utilized ether more as a medium of exchange than a store of value like would we expect to see like a deviating story like it's not so much about like value is more derived from use value than TVL?

>> Yeah. Yeah. So, I think that that you're you're kind of putting your finger on on the most important thing here and I don't know how to think about this. Right. So, basically what I'm saying is um to begin with it's not even clear to me why the Ethereum separate from the proof ofstake mechanism and stuff. It's not clear why people need to need to be using Ether for instance for their main trading on unis swap and it's not clear why they need to use it for payments or as opposed to any other token even an ERC20 token that I might make up right now. Right. So um the only thing I can think of about asymmetries between layer 1 and layer 2 is that you know ether is born on layer 1 and it's a synthetic on layer 2. Um bitcoin is born on the

bitcoin blockchain is synthetic on layer 1 and synthetic on layer 2. And so one of the things I've been thinking about is like particularly given the sort of mainstream uh entrance who favoring bitcoin maybe when you push to the layer 2 you sort of create a situation where people >> Yeah. where the the synthetic synthetics are substitutes for each other in a way that that the the native is not it was is a not a the synthetic is not a substitute for the native on the native chain. Yep. >> Yeah. So that's exa exactly and then and then and they're not they're not even perfect substitutes, right? Like my point is like Bitcoin actually has an advantage outside of this ecosystem which maybe we see more on layer 2. And then there's another piece here that makes it even more complicated which is Bass doesn't have a native token yet but

if they created one and they have Coinbase that that seems to be another problem. I I really think like you're putting your finger on the dial there, but the answer but the thing is I don't really know how to think about this, which is why I've kind of hardwired this idea that the that that the TVL um on layer 1 will be in ether and you know on layer 2 less so. Uh but you know to the extent that all of you have ways of thinking about it I I want to but I want to avoid things like totally trivial coordination effects right like you can write a model where you sort of like nobody uses it so nobody else uses it kind of a thing. Yeah, absolutely.

>> But but I don't know how to think about that. And so in some sense, the the more robust point that I'm going to be talking about here is about the TVL, right? Like the that that that holds a respect of the assumptions about um kind of what people are using on the on these chains. And I will also say like I think there's a deeper question here that the Ethereum community doesn't take seriously enough, which is like what is the value in the asset separate from the chain either? And I again I don't have the answer to that. Uh so so anyway the punch line is essentially that um with this with this strong assumption here about the relation between TVL and ether that an increase in L2 activity can in fact reduce uh ether prices. So just an overview of the paper right uh it it

this is an economic framework and what do I mean by an economic framework? Well, the really thing the really sort of I suppose big differentiator relative to maybe the way computer scientists think about things is that TVL which is the investment capital is going to be determined uh as a solution to the model. It's not going to be taken as given. Okay. Um and it's going to be determined due to economic principles. Uh similarly, ETH prices will also be derived indogenously although again there's a bit of hard wiring around it as I noted. Um now what am I really or what do my co co-authors and I really think the benefit of this kind of a framework is uh well it very clearly highlights something that might be anyway readily apparent to to many of you but but I don't think to everybody which is that if you want Ethereum layer

1 to be more than just sort of a data posting layer it is necessary to maintain non-trivial transaction activity right um and in particular how you do that specifically is the L1 has to remain sufficiently competitive relative to the layer 2, right? It doesn't have to be better, but it ha but it can't be neglected. Um uh and and part of the reason I'm even framing it that way is that in some sense the Deacon upgrade I mean it was about improving the layer 2, right? It's it's it's the layer 1 sort of developers improving or helping the layer 2, right? They essentially were subsidizing the layer 2s whereas they weren't really improving the layer 1 experience in what they did there. And that's that to me that's a bit weird, right? like we have layer 2 developers doing things to

improve the layer 2 directly. Um when when the Ethereum found when Ethereum does an upgrade, I would hope that there are things in there that are sort of directly useful to people who want to directly use the layer 1. And I don't think that was the case in the Deacon upgrade. Um and and so this paper is a bit pushing back against uh the idea that that that those should be the the biggest priorities. And so again, helping the this is not a negative L2 paper, right? So helping the L2s is perfectly fine but that that shouldn't be at the cost of improving uh the user experience on the layer 1. That's kind of the point right and so I will show you a formal result that will basically highlight this point about as long as the sort of the the development of the layer 1 is not sufficiently far behind

or is sufficiently close to the development of the layer 2 then everything will be fine but if it falls too far behind then then everything won't be fine and the channel again is what I was describing which is sort of the TVL. So, this is the Ethereum Foundation announcement. Um, I'm sure you're all familiar with it, so I'm going to skip over it. I'll just point out that the first thing they did list was scaling uh the L1. Uh, they also had improving the UX as a third one, which I would say is also kind of additive. With that, before I sort of conclude the introduction, I do want to point to this post by Justin Drake. So, incidentally, the entire idea of this paper uh came out from uh something that Tim organized, his CCE conference this year.

Uh so uh you know I I watched the the panel uh which I suppose initially had Justin against Dankrid essentially. Um and uh I suppose I agree with Dank I guess is what I'm saying. Uh and and and and so you know like let me just point here to those of you who don't know Justin's position on this. Let me just point to what he has to say here right he's saying 10k TPS on Ethereum L1 10 million TPS on Ethereum L2. Most of the activity will happen on L2s. Most people won't know they're using Ethereum. Ethereum is for everyone. believe in something. Now the point I want to actually uh he's obviously talking about you know sometime in the future but the thing that I want to fixate on is what he's describing as happening on the layer 1 right so below the 10ktps on

ethereum L1 he says highv value transactions for all so you know the it seems to be the thought process is somehow it won't just be a database posting layer layer 1 the really really high value stuff will happen there and there's less of that stuff um so in in raw numbers there'll be less sort of transaction activity, but but the high value stuff, that's what you want to do on layer 1. Now, the thing about this is if there's no liquidity on layer 1, and I'm taking that obviously as an extreme case, but just just just to get the intuition right. If there's no liquidity on layer 1, and let's say let's forget about, you know, the security through proof of stake having to do with market values of Ether, etc., let's just take for given that layer 1 is 100% secure,

bulletproof, but there is no liquidity. That means you don't have to worry about security, but even small trades will have essentially infinite costs. Who would trade on layer 1 in that situation, right? Like you need some liquidity on layer 1 for highv value transactions to make sense on layer 1. Um, and the high value trans and the liquidity is not going to live on layer 1 if there if there's not enough transaction activity, right? And so that that step seems to be kind of missing as to what is the response of liquidity providers to um to to what uh the Ethereum Foundation is doing in terms of you know scaling through blobs and so on. Um yeah, >> I think I agree with you, but I Justin would say in response to this that like

the high value transactions are actually going to be larger trades and so even if there are fewer number of trades like the total volume, which is really what matters for that spread equation. Well, I don't know if Justin would say that, but that's what I'll say is like like there is a counterargument that these are going to be larger trades and therefore there's going to be more liquidity because they need to fill those, you know, retail orders that are huge. >> Perfect. Excellent. So, as an economist, the way I'd frame it is you've described a trade-off, and it's not clear which way the trade-off goes. So, I very much think of what I'm going to show you as because partially, frankly, when I saw that that that argument, and I remember Max, you got involved too, right? Like they had a few extra chairs and people

sort of got on there. it sort of rather than kind of converging to a nice point, it kind of like got even more all over the place. And so as I was sitting there, I was thinking, wouldn't it be nice to have a mathematical framework to start to think through this? So I think it's a fair point and and you know, the way I think about transactions here is relatively reduced form. probably doesn't properly appreciate heterogeneities, which you're alluding to. But the way I would think about it is that it's not at all clear that the channel you're describing, which in fairness, I think you're you're saying like let's let's think of the best counterargument. It's not at all clear that would be sufficient to get around what I'm describing. And I would love to either, you know, take feedback and and and augment the model in the direction

that people in practice think is, you know, important um to really get at the question. Uh, and I'd love to have other people pick it up and and explore it, too. Um, but, uh, but I think that, so let me put it this way. Um, to Justin, I guess I would say, uh, if that's the conjecture, I would love to see proof. >> Or another way of saying it. I have a model and he doesn't, right? So, okay. Um, all right. So, now let me kind Yeah. >> Can I ask another general question? So, uh, in the end, the layer twosh hinge on the layer ones for security, right? So if the Ethereum price decreases that means that security also decreases potentially and will it like the model takes it into account like the

developers might move in response for that from the layer two to the layer one to sustain this security >> so okay um there are a couple things I have to say about that so uh there are externalities so just just because so there's one world where you say the L2s are all good faith players and they actually would hate for ETH cases to crash to the floor because then proof of stake basically becomes insecure and the L2s become insecure and uh and and and so they would want to try to work to avoid that but the fact that they might do that doesn't really make clear the mechanism by which that would actually happen right so so there's one piece there in which I think like particularly in the Ethereum community they tend to

take this sort of idea for granted they'll say things like look um ether prices have to be high because it's securing a lot of stuff but actually it's exactly backwards. You write a nuclear model, it's exactly backwards. What that sort of reasoning would predict is that you will suddenly have a massive attack and everything will be worthless, right? Um uh and so there is a need to think about how you if you believe that vision that you were sort of describing where the L2s are all in good faith that there is some some way to actually capture value and the current mechanism just doesn't have it. Now I guess you could do things like offline like you could literally have I don't know I guess L2 labs paying buying ether or something like that. I don't know. um but there's nothing in the mechanism right now and so there's in

some in that sense there's nothing that that is there to model but the Ethereum foundation should be thinking about this the second thing I want to say about that though is um you know think about the incentives of Coinbase here is it obvious uh that that that they are invested in the longr run vision of Ethereum so you know one of the interesting things here is that it's very hard to beat the first mover. But what if the first mover helps you take part of their audience?

Could you just basically become the new first mover? Like, okay, you know, let's say you wake up tomorrow and on Ethereum layer 1, the total TVL is 1 billion. And on base, the total TVL is 150 billion. Okay. And what if Bass decides at that point, you know, we don't want to be in L2. We will be even a centralized blockchain. And they say, yeah, okay, fine. can trade on Ethereum with the 1 billion TVL or you can come to us as a centralized chain and you can trade through us. Now I would argue that the success of something like Tether suggests that you know at the end of the day a lot of people are like whatever just give me the thing that works the best from the perspective of let's say transactions cost etc right and so I don't know if you want to put yourself

in that world um uh so it's not even clear to me and and that's another reason why I think to the extent that people are going to make these arguments about uh about this value capture you do have to find a way to embed it in the protocol and in some sense it get back gets back to what Scott was saying which is you like, yeah, why why do I really need Ether as a user? I get that if I use it, there's an externality to the security of the system, but I'm not just going to go buy a bunch of Ether in order to secure the chain because I don't have that much money anyway, right? Like that that's a very odd reason to ask people to buy Ether, right? Um, so yeah, it's kind of a long answer. Hopefully that got somewhere close to it. Um okay, let me provide a

little bit of sort of uh context um on how economists broadly see uh uh thinking about things like indogenous investment because I'm going to have to do this in the model. So I want to give you some some background um uh just broadly on the economics literature about it and about my our particular choice in this paper about how to how to think about this. So there are essentially two approaches in uh in the asset pricing literature which is the main literature that this would be relevant where where this comes from. uh there are two approaches to think about how uh how how we think about indogenous investment decisions. One approach is called the general equilibrium approach.

I'll tell you now, we're not going to use this approach and I'll tell you tell you why later. But but this but essentially the way the general equilibrium approach approaches things is it says okay look there are a bunch of agents in the economy. They have preferences particularly over their consumption over you know what they eat. um and they decide how much they want to consume and how much they want to invest in order to optimize their their utility structure. So I'm giving you a very sort of simple static context. The the thing I'm going to show you here doesn't depend on the static context. I I can generalize it um uh and has been generalized. This is just to kind of make the point. So uh there are a bunch of sort of hidden assumptions um which I maybe I'll briefly point out uh which is

usually what economists do in the setting although they don't always make these assumptions uh they will assume that your utility structure is state separable and time separable. What does that mean? So I could say I have a general function that takes in my consumption today and my consumption tomorrow. But that's not what I'm saying. I'm saying I can decompose it into the the val how I value the consumption today plus the expected value of the consumption tomorrow. This is a very common assumption. and it's in almost all of the literature but not 100% of it right um so I have those state time separable preferences um but how is consumption determined and this is where the investment part comes in is so imagine I have some income and so I'm using y for income so y z is income today that income is going to be

allocated across consumption and investment okay and I can have multiple assets here it just would complicate what I'm talking about but essentially I have this consumption investment problem where with whatever I have I'm going to split it between consumption and investment right now Why would I forego some consumption today? Well, because tomorrow this whatever I have invested in will pay some return and I can basically use that to consume tomorrow. So I can use essentially any asset as a mechanism to transfer value into the future but it might be risky. So this R01 corresponds to the uh the gross return on that investment which it's a random variable, right? Um and I might have some some new uh wealth some new income tomorrow. Uh okay. So where would where do this approach go? Well,

basically you say well I want to find the optimal amount of investment. So essentially you know if you know if the if the U is increasing and strictly concave and everything is well behaved then it boils down to a first order condition. And so there's just a little bit of math here right? So if you take a first order condition with respect to a z you get the the marginal utility today uh times its effect times the effect on your consumption today of of investing a bit more which is negative one. Right? So every unit that I uh choose to invest is a unit that I don't consume but I value it according to my utility, my marginal utility rather. And what do I get for it? Well, I get a little I get I get um more consumption tomorrow. Um which I value according to my marginal

utility tomorrow. And so it gives you an equation a little bit like this which is really famous in asset pricing. Um and and so there are a couple things to that I want to highlight about this. So the first is if you play with it a little bit you can rewrite it the way I have here which essentially what it then says is that the return on any investment risk adjusted must give you the risk-free rate. Okay. So what I'm saying is you can think of this this ratio of marginal utilities as a random variable and particularly as a pricing kernel. Okay. Uh so that so I'm basically normalizing it and that's what I call this Z thing. Um, and so now I'm saying if I if I use this Z thing to price the return, I get back the risk-free rate, right? Um, so that's a very general

principle in asset pricing. And so, you know, this this idea uh that that uh that returns are supposed to once risk adjusted equalize does come from something deeper than that, which is essentially agent optimization problems. >> Maybe just comment. Yeah, >> it might help. I wonder if it's worth connecting this back to your motivating story like who is it who's making an investment decision? What does investment mean? What does consumption mean back with respect to the L12 story? >> Right? So actually that's that's a nice pivot to my next slide. Um because in some sense I'm going to punt on your question, but I'll tell you why I'm going to punt because it's actually important. Uh so the this approach which I'm calling the generally clear approach is never used as far as I know in financial markets in practice. So I've worked as a quant and I've worked as a

trader briefly. I have never I never saw this in industry. Uh and they're very very good reasons why this doesn't show up in industry. Uh so one is you know what if investors are not rational you know maybe people are not taking first order conditions right. Um another piece is sort of assumptions about the utility functions and so on. Um uh but but maybe but getting a little bit more to maybe where where where Tim was uh was going. Um even if you took everything I said as given that you know investors are rational and they have state time separable preferences. What the hell are the utility functions? And more than that, what if we have different utility functions, right? Um uh there are aggregation results I should say and if anybody has heard the phrase representative agent um uh there a lot

of times what macroeconomists will say is that it's as if essentially there's one person and so there's only one utility function and and there are some theorems that tell you like given uh heterogeneities in the economy how that would map to uh sort of a representative agent util utility function but I will tell you like personally I never find those things very satisfying and so ultimately part of the part of the the the point I want to make is so Tim was asking well like who are the users right? Um the truth is this framework I'm not sure it really helps you think about that to begin with. Uh because it's making in some sense very strong assumptions. Um and uh and and so so it's it's it's the truth of the matter is like exactly the let's call it the

microeconomic heterogeneities. I don't think there's a good way at getting at that. uh if you go talk to people who are sort of themselves um you know using these kinds of models they will argue to you that the reason they're using those models these models is specifically because they do want to say something about the welfare of the underlying agents but I think there are a lot of people myself included who think you know this doesn't you know this doesn't comport very well with the data and and that actually is is frankly the reason why I'm not going to be using it here because one way to think about it is the scope of the paper is trying to think about an aggregate quantity which is total value locked And we can for instance find a way to model that effectively um which is in some sense

without loss of generality to what's going on at the microeconomic level. Um if you model it this way you then could say something about the microeconomic level but then you have to take very seriously what you happen to assume about those micro about that microeconomic structure. And the difficulty I have with it is I don't think we have a good way of actually um uh of actually specifying the microeconomic structure um in a credible way. Let me maybe uh I I'm remember I'm I'm now reminding myself what I have on my slide. So this is actually a perfect uh next slide. Uh so this uh what I'm screenshotting here is one of the most famous papers in economics and certainly one of the most famous papers in in in equilibrium economics, right? Generally equilibrium economics. Um so you can see the title is the equity premium of

puzzle. If you've ever heard of the equity premium puzzle, it comes from this paper. One of the co-authors on this paper, Ed Prescott, is a Nobel Prize winner. I'm just going to read you the first part of the abstract. Uh, restrictions that a class of general equilibrium models place upon the average returns of equity and treasury bills are found to be strongly violated by the US data in the 1889 to 1978 period. This this result is robust to model specification and measurement problems. What are they saying? They're saying we specified a general equilibrium model. We checked it against the S&P. it doesn't fit and we're quite sure that's not because of other like things like measurement problems right so this this is kind of a classical paper and you know as as a graduate

student I found this to be the oddest thing because you know my thought was like if another graduate student had done the same exercise his adviser would say well what do you think you just specified specified a very unrealistic model it didn't fit the data you know specify a realistic model but you know it was Ed Prescott who was doing it so it turned out it it created an entire stream of literature trying to solve the so-called equity premium puzzle. Um, and so, you know, the point here though is that these models, they do have the ability to provide us very useful qualitative insights, but if you really want to start talking practically, I don't think these are the sort of models you want to use and then they're not used generally in practice for that reason.

>> Wait, that that that seems like a pretty large leap. So the reason we don't you you're saying the reason we don't want to use general equilibrium models to reason about like classical GE models to reason about practical allocation is because of the equity premium puzzle, right? That basically >> No, no, no. That's not what I mean. Sorry. Okay. No, no, no, no. I I'm just using as an example. So the the point here is that you know most people don't have your economics background. So I'm just trying to provide some context on some of the struggles that some I would say this is a fairly prominent you know class of models the specific one the CRA preferences that we're talking about here right but so I'm just I'm just trying to highlight for because I think most of the room doesn't know anything about the lit this literature and I just

want to clarify >> that there have been decades of struggles to try to get it to fit even sort of relatively simple data and this is just to make that point. This is not supposed to generalize across like every single >> Okay, great. Sorry go ahead go ahead. Yeah. Yeah, fair enough. Uh, okay. Uh, so now let me talk about the approach that I'm going to take. I'm going to take an approach which is called which is sort of known as the no arbitrage approach. And and essentially what this says is that if there's no arbitrage in the economy, then there exists a pricing kernel. There exists a risk adjustment. So you kind of get that that random variable without having to specify sort of the underlying preferences. And what it really tells you then is that structure we were just talking about uh applies across a very large set of uh of

situations. And so you don't have to necessarily specify that stuff. I should one thing I do want to clarify is what I mean by no arbitrage. So the setup here essentially is that you tell me there are a bunch of assets and so they're characterized by random returns and you impose that there is no arbitrage in the sense that I cannot come up with a portfolio that is costless and earn and and then the the sort of the random return of the portfolio is strictly positive with positive probability and weekly positive with probability one. So essentially I'm saying I create a random portfolio. It costs me nothing and I'm guaranteed to make uh to never lose money and sometimes I will strictly make money. So if I if if you if you impose this you can basically use sort of more

technical results like whatever the separating hyper plane theorem to just say that there must be essentially this this this random variable. Um anyway, uh we can't say sort of how it relates to utility functions and so on because we haven't specified anything, but nonetheless, the pricing kernel still exists, right? Um and so if you if you really just care in some sense about thinking about sort of um aggregate quantities, then this tends to be sort of a more useful way of going about it, right? Um, one thing I I tend to do in the paper um is I tend to use what one stream of the literature, the quantitative finance stream does, which is it sort of invokes this measure change. So you'll see this like Q superscript, right? So that's literally just a change of measure and it's and

the change of measure is based on the normalized version of the pricing kernel. So if you care, you know, kind of what I mean when I when I say this expected value with the Q, I literally am talking about changing measure with the normalized uh pricing kernel as you see here. And the important thing is it gets you again back to where that GE thing had sort of left us off which is at the very bottom of the slide here you see it says that under this Q measure the expected return of any uh asset must be one plus the risk-free rate right so riskadjusted returns must give you the risk-free rate is was where you get back to so let me just kind of uh sort of summarize the differences and again I use no arbitrage in this model um so general equilibrium implies no arbitrage

because if you think about the sort of portfolio construction I described it would imply if it didn't hold that the agent optimization problems were unbounded. So you wouldn't get a solution. So if you have a generally clear model uh then then no arbitrage does hold. Um and so one way of thinking about it is that um it's just that I that if you are committed to a certain generally clear structure then then I can tell you explicitly what this measure changes and I can tell it to you in terms of the underlying preference structure. So the advantage of no arbitrage models I don't have to specify risk preferences explicitly. The disadvantage is I can't talk about welfare, right? Right? So, I'm not going to talk about welfare here because I can't because I'm using a no arbitrage model. And just to give an analogy, I'm

sure most of you have heard of the Black Scholes model. If you look at the Black Scholes model, right, um uh at least the one that's sort of uh most famously known now, it doesn't say anything about the underlying microeconomic structure of the economy because it doesn't need to because it's it relies basically on creating a synthetic call option and saying that that call option or a synthetic derivative and saying that the synthetic derivative must have the same uh price as the actual derivative. And that's that's no arbitrage. So let me get to the the paper then, right? So as I said, it's going to rely on uh a no arbitrage framework. So I'm not going to make any explicit assumptions on investor preferences. Um implicitly what these models always assume is frictionlessness. Uh which is to say that I can sort of scale up and down my

portfolio without having any price impacts. That was that was assumed both in the general equilibrium setup and in the no arbitrage setup. That is one thing to keep in mind though. For instance, like when we're studying let's say uh can I say shitcoin? uh when we're studying those sorts of things um uh you uh you don't want to use these sorts of frameworks because these frameworks are essentially assuming that you can scale up your positions without having uh price impacts. Um but anyway I think for this setting it's probably okay. Um now when we talk about indogenous investment what are the sort of relevant uh pieces of what are the relevant investments in this setting? Uh well specifically what I'm going to be thinking about is DAP investment both on layer 1 and layer 2 which are going to be competing with each other. Okay. And I'm also going to

be thinking about staking investment. Uh to be honest, the staking investment is just there because it's supposed to be a proof ofstake layer 1. It it's kind of a sideeshow to the model, at least in terms of the channel. But anyway, it is there. Um and I have a short-term risk-free bond also specified just because uh when you when you when you equate things to the risk-free rate, you have to specify the risk-free rate. Um but anyway, okay, let me keep going. Well, let me also just uh emphasize that the fact that these investments are modeled doesn't imply that there aren't other investments. It it just doesn't matter. Um, okay. So, uh how do we think about uh layer 1 DAP activity in this setting? Uh so, uh here again, I think to to Max's

point through Justin uh earlier, um we're we're it's it's a relatively kind of high level approach to thinking about this. It basically uses kind of a construct from economics which is like production functions to kind of think of um transaction activity as downstream of the investment capital. Right? So the idea for instance is that the more investment capital there is on a layer the more transaction activity it's going to generate. Um and this is kind of just using a familiar functional form to generate that. Uh so we are sort of specifying specifically that the transaction activity on layer 1 between t and t+1 will have a particular form uh and and and so there are three things on the on the right hand side here uh that

I want to sort of take a little bit time to unpack. So first of all we're assuming it's sensitive to both the TVL on layer 1 and the TVL on layer 2. Okay. Um I'm going to come back to that in a moment. Uh but for now all I want to say about that is the TVL is indogenous. It's something we have to derive. Right? So if you're thinking about this more from a probabilistic perspective as to what is the guts of the model, what's driving the you know what what what defines the probability space, it's not the TVL1 and the TVL2 because they have to be derived. Okay. So what's going to be driving the the probability space is two things. One of which is on this slide and and that one thing that's on the slide is this A1 process. What is the A1 process? Well, at an intuitive

level, what I want you to think about it is I want you to think about let's say the quality of the technology directly. And so one way to think about what I'm what I'm framing here is imagine you wake up tomorrow and that Ethereum has all the same technology uh both in terms of the design of the protocol and in terms of specific smart contracts on the protocol, right? But there's no liquidity whatsoever on the chain, right? Um now my point is I think in that case the total transaction value is probably going to be zero, right? Because even if you have excellent tech, which is the A is very high, the the the liquidity is not there. And so there makes it makes no sense to trade, right?

And so part of what this is supposed to capture is that transaction activity is not just a function of like how awesome the tech is. There probably are many blockchains out there that have very good technology but don't have any activity and therefore don't have a lot of trans don't have a lot of liquidity and therefore don't have a lot of transaction activity. And so this is supposed to capture that by essentially specifying this multiplicative relationship which again is the same relationship that we typically use for production functions. But then again, if you flip it around, right, and you say something like, well, what if there's a lot of TVL, but the technology isn't particularly good, right? Maybe gas fees are insanely high due to that. Then again, transaction activity shouldn't be particularly high. So, we have the dependence both on the exogenous sort of

uh quality if you like of the chain, the tech quality of the chain, and then also this dependence on the indogenous investment capital on the chain. Um uh so the usual assumption about this is that you know the is that there's some concavity and so we have that imposed here which is that in other words uh investments uh produce or the more and more you invest you get sort of decreasing marginal returns right uh that's the beta 1 uh between being being between being between zero and one last thing I want to deal with is this TVL 2 right that's the TVL on the layer 2 okay um and now what we're assuming here is that the they are competitive with each other. Okay, so we're assuming that the more TVL there is on layer 2, the less likely you are to trade on layer 1, all

things equal. Um, if you think that's not true, if you think that more activity on the layer 2 is going to cause more activity on the layer 1, then to be honest, you shouldn't really be concerned about this. Like the entire idea of layer 2 is actually is is horribly unconcerning, I think, at that point. Um, so, uh, you know, in terms of solving the model, you could, we don't really need this assumption, but I kind of think it's true and I don't actually know why anybody would care about, you know, we'd be concerned about layer 2 if it wasn't true. And I don't know how to reconcile what we see in the data over the last year without this assumption. Um, uh, so I just want to kind of be kind of upfront about that because I think sometime sometimes people sort of think I'm trying to hide this

assumption. No, it's like I I I I don't see why it would be such a big concern if you thought that like more activity on the layer 2 would then be get more activity on the layer one. I think precisely that's probably not true. So it's it's sort of like a core assumption from the perspective of the of the worldview I guess of the authors. Um although you know I I'm I'm I'm happy to be disproved about this kind of assumption through the data for instance. Uh I just don't think I just do think they are competitive with each other. Okay, where do we go from here? So I've now specified transaction activity, right? And so the idea here is that the transaction activity uh translates to some uh revenue to liquidity providers. uh so some fee you can think of it as like the transaction

activity is transaction activity of trades at unis swap and there's some proportional fee fi 51 right so so this term is capturing sort of the return over and above your uh the capital the the the sort of asset uh P&L okay and so whatever whatever uh fees this is generating is shared prata by all of the investors right so that's why it's being divided by TVL12 but then you're holding risky crypto and so there's some uh P&L due to that I should say in the draft we allow for you to invest in other assets and it doesn't have to be just ether. This is just really for simplification on the page. Um really it it so for instance you can have wrap bitcoin etc etc. Um you do need some investment in ether. Um but uh but the but the broader point that I want to make here is that

when you think about specifying returns in this setting it's not just about for instance the fees that I earn from providing liquidity at unis swap. It's also about the fact that I have a risky portfolio position and that's what the first part is capturing. Okay. um and and the prices will be determined indogenously although as I said a little bit of hard wiring there okay um so so so actually let me clarify so here I've specified the the return uh as a function of uh specifying the transaction activity um this is all I need essentially to sort of to apply the no arbitrage approach because now I'm just going to basically say this thing risk adjusted has to give you the risk-f free rate and that's going to allow me to basically derive uh uh uh things like the total value locked and so on so I'm

going to move on to the next investment thing which is staking. Um it's not really the folks of the paper so I don't intend to spend you know too much time on it. The only thing I will say is that one of the places where I think the CS literature differs from the finance literature particularly is a lot of times people tend to model the returns to staking as sort of exogenous whereas there's a convention in the finance literature at this point to model it sort of as inflationary which as it is right and so so this is adopting that structure of making sure that it is inflationary in the model right so so this is the return in terms of the ether but you have to multiply by that by the return on ether because if the price of ether goes to zero you actually get

nothing because 5 * 0 is zero right so this is this is the staking investment it's just inflationary from from the value of ether um now in terms of uh determining pinning down the prices well this is just market clearing okay so what we're saying is if you stake you have to hold ether if you invest in in in in the DAP you have to hold hold ether and again in this case I showed you they're actually investing fully in ether but in the paper in the draft um it it's just it's to share. So it doesn't need it all to be like 100% in ether. It could be 20% in ether if you like. Uh that'll be perfectly fine.

But the more important point is there are sources of demand for holding ether. One is taking the other is DAP investment. Okay. And that has to equate with market supply. And that's how I can induce uh and solve for uh uh a number of indogenous characteristics including uh the amount of stake, the amount the ether price and uh the TVL. Okay. Um so DAP investment on layer 2. >> Sorry M there is just the circulating supply or something. What is capital? Okay. >> So M is the literally the number of ether in circulation at time t. Yeah. Okay. So L2 DAP investment is specified along the same functional form um as L1 DAP investment. So again transaction activity on uh layer 2 has this multiplicative form. It depends on if

you like the tech on on layer 2 which is this A2 process. So A2 and A1 are the two underlying stochastic processes that drive the whole model. Everything else is indogenous. Um again it depends on TVL on this layer in a concave way. So sorry it should be pointing there. Uh it depends on TVL on the same layer in a concave way. It also depends on TVL on uh on layer 1. Um we are agnostic about you know whether you think that this is going to be uh have a whether you think TVL on layer 1 will have a negative effect on layer 2 or vice versa. Um, I've heard it argued both ways. Um, I don't think the layer 1 and layer 2 are perfectly symmetric. Um, which is why I could see for instance people using like for instance all the people who were

using Ethereum to begin with when the layer 2s weren't popular and the layer twos became popular. It seems to me that you know that there was a positive effect there. But you know it's we don't need to take a position on this sign. Um, so we're not going to uh some people do seem to have a strong feeling about it. So I'll just I'll let those people adjudicate it. So let me keep going here. So again the the the L2 DAP returns uh this is just sort of accounting at this point right which is there's some transaction activity some of it is paid in fees uh it's given prata to the people who are providing liquidity and then there's some P&L now so here by the way is like the the really hard strong assumption which is to say um again the version I'm showing

you is a simplified version in the draft you can invest in other assets but importantly even in the draft we don't allow you to invest in ether when you're providing TVL on layer 2 which is supposed to be the hardwired, which is supposed to be capturing this thing that we were talking about at the start, which is for some reason it seems to be that people when they do things on layer 2, they're less likely to use ether, right? And and and and so the this is really just a a hardwired way of capturing that in an extreme case. Um uh the the real question is, you know, do you think that what we're seeing in the data right now is actually reflecting something deeper or that we could live in a world where essentially people treat either on layer 2 the way they treat ether on layer 1. Um if we could

live in that kind of world, then probably the solution is like well let's figure out how to get to that world and then we don't need to worry about this. But again, I actually don't know at all how to think about this. And so to the extent that you guys have have thoughts on it. Uh it would be very useful. Uh because right now I'm just kind of like saying, yeah, like the long run trend seems to be bad in the sense of, you know, ETH usage on the layer 2s. Um I don't know why. Uh but I'd like to know why and then I'd be happy to put it in the model, but until I know why I don't know what to put in the model. So you know, so again, I've now shown you basically the returns from these different things. And so as I said

earlier, all I'm going to really do is I'm going to require that when you do a risk adjustment, you get one plus the risk-free rate. And the reasons for it is is what I had uh been you know said earlier in the in the part when I was talking about indogenous investment. So this is just a model solution. I won't you know bore you with the sort of the details of it. The only thing that I'm going to sort of emphasize is that so I mentioned I don't specify risk preferences and there's sometimes people get confused about this uh in that I am not saying that the risk preferences of investors do not have quantitative effects on things like investment. I am saying for the model solution I'm giving you I can subsume that. Um and in fact for the qualitative results I'm giving you it doesn't matter. Uh but it does

have quantitative effects. And so I want to just take a moment to to clarify that and to show you how because particularly if you were taking this data to taking this to data you would see the quantitative effects show up. So in this is the solution to the total value locked on either layer right and so everything on the right hand side is exogenous. But one thing I want to fixate on is this mu star q thing which is defined down here. Now one thing that's important to realize is where risk preferences matter in any economic model is in pricing shocks. Okay. If you don't have any shocks in your model, then who cares about risk, right? So I told you there's like two stochastic processes driving everything. One is the uh the A1 process and the other is the A2 process. These epsilons correspond to

the innovations to those processes. So those are the only random things. And so the way risk aversion enters into this model is if you tell me that investors are more riskaverse, what it what it implies is when I price these shocks, it's going to be lower. Okay? And the consequence of that is you if you follow it through is if you were to tell me that investors are more riskaverse it implies quantitatively that this TVL is going to be lower. So there is a quantitative effect. It it's not that it does not matter quantitatively how riskaverse investors are. What the point of using no arbitrage here for instance is like you could literally sort of estimate the aggregate risk parameters rather than worrying about you know what preference structure should I use for all of the agents in the economy. Okay.

So risk does matter. It it matters quantitatively. Um but it's subsumed by the by by the solutions of the model that I'm giving you. This is indogenous staking and and the and the market value. So everything is in closed form. Let me now sort of give you the the punch line. So I mean in the draft there there are a few different results, but you know at the end of the day it seemed like there was really one thing that that that was worth kind of uh talking about uh or at least the one thing that we felt was the most important thing to talk about. uh and it relates to what I was sort of saying at the outset uh which is so first let me clarify this condition I have on the slide so mu a1 corresponds to the drift of the log uh

a1 process so this is kind of getting at how much does the technology on layer 1 improve over time on average okay mu a1 is and you can the way I would think about it is think about this as the ethereum foundation's development efforts for the layer 1 specifically for the layer 1 so if they're doing things sort of expanding blobs and so on then it's not helping mua1 It's not it doesn't mean mua1 is high. Uh but but to the extent that they're doing things specifically to improve the L1, um that's MUA1 high. To the extent that they're not doing things, it's MUA1 low. Okay. Um MUA2 is sort of the same thing for the layer 2, right? Although it can be affected by the Ethereum Foundation's efforts beyond sort of layer 2 developers because of course, as we've said, the Ethereum Foundation can or

Ethereum can do things on layer 1 that actually are are essentially subsidizing layer 2. But the point is uh mua2 captures kind of if you like how uh how uh the tech on layer 2 evolves and if mua2 is high that means that the tech on layer 2 is evolving you know positively uh whereas or very positively whereas when it's low it means it's not evolving quite as positively. Uh and so the point is you can take the ratio of these two things and if it's below a particular threshold value right um then it turns out you get kind of the horrible scenario that you don't want right which is the TVL goes to zero um pointwise or almost surely right and the market value goes to zero now again I want to emphasize like the the TVL going to zero

is really the more robust result and the second result is coming from this sort of hard wiring um which goes into thinking about you know what assets you want to uh be holding on layer 2 and why I do want emphasize that the model assumption is that mua1 is positive which is to say that the layer 1 is productive in even if equation 5 holds the layer 1 is productive. So in some sense this is kind of like the the saddest outcome right which is that you have a technology that is fundamentally productive but due to competition effects outside of it uh it ends up basically becoming useless.

Um and and another thing I want to emphasize here is uh notice this condition is about the ratio of mua 1 and mua2 right it's not about the absolute level which which goes back to the point of like what I was saying about like it does matter kind of how much development is happening on layer 1 relative to layer 2 right these these these things are competing for against with each other for transaction activity so again the channel is is fairly straightforward right so if the L1 is insufficiently productive relative to the A2 the L2 that is mua1 is sufficiently lower than mua2 Right? Then transaction activity is going to migrate to layer 2. So the transaction activity on layer 1 in this case is actually going to go to zero while the transaction activity on layer layer 2 is going to go to infinity uh

almost surely. And you know to some extent you know we certainly you like the idea of the transaction activity on layer 2 going to infinity but the more important point is kind of what happens to the TVL's and the TVL of course follows the transaction activity. the TVL on layer one in this, you know, will will basically go to zero almost surely, whereas the TVL on layer 2 will diverge, right? Um, again, there's a bit of hard wiring on this last step. Um, and and the market value will also collapse. It turns out that if you flip the condition I gave you, you get the exact opposite result, right? So, if the same ratio I was describing is higher than that threshold, what happens? Well, everything is fine, right? That is to say that the TVL goes to infinity, the market value goes to infinity. And this

is I suppose the the rosy vision of of layer 2 scaling right which is that given that L1 is sufficiently productive relative to layer 2 which I'm noticing now I shouldn't have this is this this statement about mua1 a lot greater than mua2 is wrong uh I I don't mean that because in some sense there should be a constant here where mua1 can be smaller than mua2 still but you know what I mean like if it's sufficiently close if mua1 is sufficiently close to mua 2 then both transaction activities will diverge and both tls t TVl will diverge and of course the market value will diverge right what I'm going to conclude with is so I want to again clarify this is not a paper that is negative about layer 2 the point really was that when I saw that debate between uh Justin and Dankrid I

really felt like some structure in the conversation could be useful and so related to that this is also not thought to be like a endall paper uh as I've said there's a particular part of the model that I personally would like to improve but don't know how to improve it because I don't really know I haven't really uh figured out the underlying economic relationships and interactions well enough. And this is again how we should think about using Ether on layer 2 relative to using say synthetic Bitcoin on layer 2 as compared to layer 1. I still don't know how to think about that and it's obviously important to thinking about what the long run behavior of ether prices are. But I do think this is sort of uh a properly formal way of showing that there is the possibility that layer 2s could be delterious for layer ones. Now when you

when you put that into the same context as that price graph of ether that you all were probably already familiar with right and you put that into the context of the timing of this stuff I think it becomes hard to say that you know it's we should be completely on autopilot and just presume that if the L2s thrive then the L1 will thrive and ether prices will follow etc etc like I even if you don't buy this story I then I have a question for you which is like how do you rationalize what we saw Right. Um uh and so I think you know what the Ethereum Foundation is is is doing now in terms of like reorganizing and and emphasizing the importance of scaling the layer 1 and so on is is is very useful but uh but it's not clear they've actually

fixed the problem more than just sort of identified that they concede that there is a problem. And so I still think there's quite a lot of thinking to do related to some of the questions you all have had. And I'm hopeful this will be part of the conversation that can help us move forward. But again, I'm I'm very happy to think about you know enriching and improving the model uh particularly along the dimensions of people who are sort of closer to the practice would think useful. Um and one last thing on the no arbitrage framework is that framework does I think take relatively well to the data. So one thing I have also thought about is that if people are interested on the empirical side that is a direction that my co-authors and I uh could explore. With that said I think I have about three minutes. Um, so I will

stop there. >> My main issue with this approach would be the kind of hardwiring of TVL and the value of the underlying asset. Um, I think that's like >> it's my main issue too, man. >> I don't know. For me, it's like what does TVL give you? Zero. Uh, is my intuition there. I can't say. >> So, what do you mean by that? like what does TVL in something that is not the native asset on the L1 or on the L2 give the holders of the token?

>> Oh yeah I so I I agree with you. I agree. In fact you could you could argue so like look if they if there were if you woke up tomorrow and all the ETHUSDC liquidity pools were actually wrap Bitcoin USDC pools that's even without the layer 2 that's already a problem just for the coordination reason. So I I totally agree with you, but this goes to like a deeper question that I think you've been touching on and Scott was touching on, a few people have been touching on is what the hell is the point of ether to the user, right? We all get that there is security generated by the market value of ether, but that isn't actually a useful reason for an individual person to hold ether, right?

Um, it's kind of like if you think that, you know, I'm sure everybody would agree that the government can provide useful social services. Even if you don't think some of the social services provides provides actually are useful, I'm sure you think that there are some that it either does or could provide that is useful. That is not a reason for you to voluntarily pay a massive amount of taxes because you're not going to yourselfund anything, right? So, the idea that, oh yeah, look, but we all get that we need to prop up ether is a really bad economic mechanism to solve this problem. >> Okay. Yeah. And then my second point would be I think you talked about it but it's not in the model which is that the size of the L2s the relative size of the L2s to each other >> matters a lot. So for example base is

doing like 120 TPS arbitrum 30 optimism like 18 or something >> and if they were all doing like 50 >> then that's a different story than if bass is doing 200 and everybody else doing one. >> Yes. So, so I I think I'm exactly on board with what you're saying. I do respectfully I do think that's a different point of inquiry. What you're getting at is sort of the market structure across L2s which essentially like we talk about, you know, these systems are permissionless and so on, but um but that doesn't mean there's not market power owing to network effects, right? So if all the TPS is happening on one layer two, that player has a lot of market power. Even if you can create it

in layer two to compete with it, but nobody's going to use that layer 2. So they have market power, right? I I think that's actually separately a very important question. Um it does dovetail a bit to to to this question in the sense that if you think about the incentives of a layer 2 that ends up in that position and is looking at ether prices collapsing now maybe it becomes very much like yeah why don't we just you know essentially uh you know uh become our own chain or something like that. Uh but but I I would say it's a little bit beyond the scope of the current project. I was just trying to understand the reason for the asymmetry between the layer 1 and layer two in the final conclusion. So is the reason due to two factors. The first one being that the externality of the L1 innovation is

positive on L2 whereas it's the other way around. L2 innovation is negative on L1. And is the second factor that on L1 the return on investment is true like not only the fees but also through staking whereas on L2 it's only on fees is like how would you quantify like the difference? >> Yeah. So there so there are a number of asymmetries that are in the model. Um, but I think the one So, for instance, like there's staking on layer 1 that's not on layer 2, but but I think the one that's really driving what you're asking is is not so much that we've so it's about the cross interaction between layer 1 and layer two through through the betas. And it's not specifically that we uh imposed anything about how the the layer 2 responds to layer 1, but

we did require that the layer 1 sees the layer 2 sees the layer 2 as competition. that that hard restriction plus essentially if you think about the parameters themselves uh they uh they're sort of embedding the other side of that restriction. Um uh so uh to be clear again we we're not saying our our results do not require that the that the uh the other cross beta is also negative, right? They can both be negative and our results are still true. Um but but the point is the fact that we require that the effect of the layer two upon layer one is negative and are agnostic on the other effect that's already kind of an asymmetry and then it's just about like yeah if the mua2 is

sufficiently high relative to the mua1 which is a symmetric effect but applied in that context essentially generates this there there's sort of one equation I can show you um that would probably completely clarify this but it's it's very hard to communicate in in words I kind of need a a pen and paper but it's it's exactly that that piece it is the cross elasticities Early last year, there was this breakthrough in the bit Bitcoin ecosystem which is called BitVM and it will enable uh roll-ups on top of Bitcoin. >> Uh so I was thinking it's interesting to to see maybe the differences between rollups on top of Bitcoin and roll-ups on top of Ethereum because for instance u rollups on top of Bitcoin will have a technological advantage. They're going to probably run EVM so allow for quasi

turing complete scripts and smart contracts and defy and so on. >> Y >> uh so they will probably uh yeah gain some traction. On the other hand, we have Bitcoin users which most likely want to hold their Bitcoin. So maybe they're not going to transact uh bitcoins on the layer one, but they will have a reason to have bitcoin uh in the first place. And I wonder if um yeah, this is going to be uh in the light of what you uh showed today, if it's going to be a push for the Bitcoin community to uh increase uh the scripting capabilities of Bitcoin to have more equilibrium on uh yeah, on the technology side. I think it's going to be interesting to see how that's going

to play out because maybe there are going to be fewer transactions on Bitcoin and incentives change. I agree that that's actually a very interesting point of inquiry. Um I and and I I think also in terms of what it's going to show us about what's going on here on the Ethereum community. And here's here's what I mean. Like the the hidden elephant in the room is precisely Bitcoin synthetics, right? Like so for instance, imagine Bitcoin synthetics didn't exist, right? Like imagine Bitcoin didn't exist and this world is happening. You know what really are people doing on the layer 2 in that world when they're holding risky assets that are not Ether? Because right now it's not, you know, it's not Arbitum token that's really competing with Ether on Arbitum. it's wrap Bitcoin and on base it's coinbased wrap Bitcoin and the

significance of Bitcoin there comes just from you know Bitcoin being the OG and being kind of the thing that that that people who don't know anything about crypto think about when they think about crypto right and so in fact like in the Bitcoin context I would be actually quite surprised if they had the same problem from their L2 succeeding because over there like what are the are people going to hold ether like a synthetic ether over there like I don't see that happening I don't know I don't see why like native layer 2 tokens in that because I'm I'm conjecturing here, right? But my point is like one thing that is sort of the hidden elephant in the in the room is that there is sort of a network effect around Bitcoin and that's a problem for Ethereum, but in the Bitcoin layer 2 context, it's not

right. In fact, it's it's it's it's it's it's supporting it in that context, right? So specifically because we don't see the native layer 2 tokens being very sort of successful in terms of becoming the main risk asset on their chain. I'm almost I if I were a Bitcoin maximalist, I wouldn't be that worried about proliferation of layer 2 in in the Bitcoin uh community or in the Bitcoin ecosystem. But of course that's that's just complete conjecture. Like clearly I have I have a I have a view on this that's that's beyond my model. Um but uh but but in some sense that would that would test my theory, right? I guess.

>> All right. Thank you for >> Thank you.