Exploring EigenLayer and The Future of Ethereum Security | S8 E2 | Sreeram Kannan

Transcript

oh maybe a proposer doesn't sign off on a block by itself but a proposal's key is split into like you know like a DVD type scenario and there is a committee of like 50 people who have sted much much more and they actually have to sign off on this transaction and if they sign off on a wrong transaction they're going to get slash not 32 e but now 1 million e we have all the pieces kind of in the in the space floating around we have DVT floating around we have Economic Security with Dion floating around it's going to be a whole bunch of fun work to make this all work but it's possible to make it work at much higher economic

Fidelity than 32 e because that is needed to build a billion dollar decentralized table or a hundred billion doll decentralized table hey everyone this episode is brought to you by Kinto the safety first L2 which is accelerating the transition to the onchain financial system they've got some very cool features like user owned kyc and Native account abstraction which solves a bunch of the onboarding and bottom back problems that many l2s suffer from so you're going to be hearing all about them later in the program but if you want to learn more and better if you want to join their launch program engine GN you can click the link at the bottom of this episode go check them out tell them I sent you 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 all right everyone welcome to the first interview based episode of this season of Bavon re taking I'm joined by my co uh trusty co-host miles O'Neal Ry and of course tram Kanan of Igan no one that would be rather starting the season with tram so thank you very much for joining us thank you so much Mike miles is really excited to be here yeah we're pumped um now a little disclaimer this this episode is

going to assume a certain amount of knowledge about Igan lay the protocol and reaking generally just because we have so much that we want to cover with shiram and for folks who don't have this uh this history I would highly recommend um the three of us did an episode about a year ago on bell curve the intersection of liquid staking and reaking we'll link that in the show notes where sham does a magnificent explanation of what reaking is so I would highly recommend you go back and listen to that but I want to jump in and get a little bit of a state of the union so to speak on the ecosystem of reaking and if we rewind the clock a year I feel like so much has happened since the three of us last talked publicly on this podcast so since then we've seen over 15

billion and counting in deposits flow into igen lay igen layer has obviously launch its main net and token we've also got a much more diverse ecosystem of AV V s's um the current estimates between 30 and 50 something like that um which is super impressive uh we've got liquid reaking which was just becoming an idea the last time three of us talked but now there are many lrts that exist um and on on the horizon there are a couple of competing uh platforms alongside uh igen laay these days so we want to get into all of that but maybe just starting at a really high level from your perspective um you know I guess uh knowing what you know now about this last year um what do you think has gone well in terms of

reaking in the ecosystem generally in igen layer and then maybe um knowing what you know now is there anything that you would uh change do differently improve on Double Down on give us a little bit of the State of the Union from your perspective yeah um the um the adoption has far outpaced our uh our own bullish bets battery would be in 2024 May and uh I think uh whether it is in the um quality of um people building avss whether it is in

um the emergence of liquid reaking we were thinking this is like a minimum 3 four years before we see anything like this the emergence of um the um rollup ecosystem being built onda we the amount of stake and interest in agon layer all of these have definitely outpaced our like you know upper confidence limit so which means basically that if we rewind the clock um if we knew that this is where things will be we would have def Ely done

things differently that's just because you allocated more resources to like for example engineering is lagging you know adoption on our product road map like that's a big big thing and part of it was you know we are conservative in allocating resources before we know that is like Market fed instead of funneling all the resources into it and then knowing like what's going to happen so that would that is definitely one thing we would dial differently if I go back in a back a year be much more like aggressive in growing our engineering and shipping even faster and more

aggressively the public Communications is the second one which I think we would have done differently is for example we went from like uh nobody knows who they are startup to like something pretty Central so fast that we hadn't internalized it that oh if you make a small change now like these lrts have to make a change those avss have to make a change this rollup has to integrate differently it's just a lot that is kind of Downstream of this and we had not designed our processes both engineering or communication to kind of match that it's kind of like ethereum has this all core Dev call and

like we need to get everybody on the same page seems like we need to move to a model like that so that's not something that was on the horizon like now people think it's obvious like four months back it's not obvious that we'd be anywhere near here so it's a very um fast rate of change so we are adapting uh as an organization to all these like emergent changes um in terms of just Market structure I think certain things we got right and I think the first thing we got right is we said Ian L is an objective platform we are not building an LRT everybody thought this year last time that I is

going to do their own LRT or like something and compete with Lio and this and that and he had stated that we wouldn't do it but like you know people had to wait and see it to believe it so because it came from a kind of like a foundational view that protocols which make object subve judgments let a lot of other like subjective judgments flourish on top of it and so that that's the view that we took I think that view proved more right than we would have imagined it there is an alternative universe that would be like we have this and nobody builds uh you know an LRT or anything on top and maybe it is a fact that like without an

LRT on top of I this ecosystem doesn't work it could be a possible world view and that's something I think we did right in terms of Market structure in general that's been like our principle of building ion layer is build the minimum minimal coordination systems that let other people build everything else like anything that's deadlocked by coordination we try to build and anything that's not deadlocked by coordination we try to externalize some people think we should or even in our own companies people think we should be like oh maybe we wouldn't have built da if we knew that we would have like 40 avss at this time there some truth to it

but there's also some uh some misunderstanding to it and I think this is not well understood that you cannot really have a system of cryptoeconomic security without a powerful scalable da nearby it it's because crypto Economic Security comes from slashing and slash ability to do the slashing the inputs required to trigger slashing should be publicly available one could take a narrow view of cryptoeconomic security and say oh only things that published on ethereum you can slash for but that makes the use cases so narrow that it's not you know broad enough so you know a system of

crypto Economic Security without an adjoining system of data availability is actually like a big hole and I think we did the right thing by building da along with Ian Lair and having this very ambitious scaling road map for IA which we continue to execute on makes a lot of sense and I um I'm curious what has been the most challenging engineering problem and it it feels like your decision to support native stake maybe represents like 90% of the work right because we've already seen a few um know new like emerging competitors come out with only supporting liquid assets and feels like it's pretty easy to spin up um and so

you know going back in time like is that a decision that you're still you're you're proud of and you see as like a core differentiator that's going to be hard to to replicate um and and maybe just speak to a little bit of the challenge I know I kind of answered that for for the audience for if you could also just Define native stake and what that yeah so there are two ways to participate in the I protocol and one is using you know you can deposit an LSD uh or in general any rc20 token the second way which we particularly made for ethereum is the ability to do like you know native reaking which is you

stake in ethereum normally when you stake in ethereum you set your withdrawal credentials to your Hardware wallet wherever you want to withdraw the money later but instead you stti through I layer you just add a step in your withdrawal flow that the withdrawal goes from ethereum to the igon layer contract back to your Hardware wallet it's kind of like a step and so that's called native reaking and I the it is true that native restating is much more complex than like you know the other part of um actually uh just building an lstd uh LST

reaking the reason we did it is the reason we do a lot of things in igen layer is kind of complying to the ethereum methos and one of the big things you want to we want to support is the uh ability for all stakers who may or may not trust a particular like LS or a third party to participate in the reaking economy this is kind of like a flash Bots type view of everybody should be participating as an equal in this economy and if you don't have native

reaking it basically breaks that Paradigm it basically says oh you know I can do s or r e like but I can't be a home Staker I can't be an Institutional Staker I can't be Co base and like the you know stake participate in I I think this entrenches certain players relative to other players and it was super important for us to do this what we did not know is most of the Native reaking was going to happen via native liquid reaking this is not a thing we know and I don't think this is the natural like consequence in like a five to 10 year time period um because you know as

institutions come as we get more home stakers we don't know which side the the fork may go But whichever side the fork goes we are ready for it and I think that is a very important thing this is particularly important for things like proposal commitments where you're trying to do something on top of ethereum L1 right like I'm I'm an ethereum L1 and I'm modulate like what I'm doing on top of it I think for those things this matters a lot because otherwise if we if you are able to do proposal propos proposal commitments are very very powerful you can start doing lots and lots of different commitments and if only Lio can do it then like relative so the whole like

architecture of how ethereum designed for meev neutrality across all the different players just gets broken so there's like so I think of like anything that has native reaking is decentralized reaking and anything that entrenches only a certain kind of lstd or a couple of lsts is centralized reaking so that's y One Way thinking about it makes makes a lot of sense it's like looking at the future and saying this is going to become you know larger and larger percentage of the overall stake is going to participate in reaking if there is you know a barrier to anybody outside of those using um something like Ste right or another LST

it just you know further moves the incentives to to to stake with them right um and that's not a healthy outcome for for ethereum I think we can all agree on that um so makes a lot of sense the challenge of course is that the state of the beacon chain is you know very different than the exe or hard to communicate with the execution layer where the contracts live right and so your these proofs being sent back and forth is really kind of like a a new engineering problem um that I think will continue to be refined um you know yeah that's right and one of the views we took there is also a little bit more hardcore than what most lsds lsds have the same problem is moving this information they said either we have to

have a ZK proof or you know use andrine Beacon State and make proofs relative to that rather than using an oracle which like you know Lio and and Rocket pool were using for some time so these decisions definitely made our life more complicated and we there were like when you make these complex protocols you take a step and you realize it's a misstep and then you to turn around and then do something else and then like ethereum has like a EIP which now brings the beacon State directly now you to do proofs against that rather than so it's it's a whole thing so I agree with Miles that it did take up a lot more effort than if we didn't do it like that for sure it's a moving Target and it's a difficult accounting problem as well um

but yeah I think you guys have identified that and you know it's kind of one of those non-obvious things that happening under the hood that um you know unless you're an operator you probably don't feel that pain dayto day but it's super important right um so yeah really appreciate that and maybe just to move on like a a little bit I've been um so we talked about the State of the Union you know igen layer has this first mover advantage that's pretty significant in terms of tvl on the supply side and number of projects building on the AVS side and kind of think about okay how how large is this first moover advant and I think that it um really comes down to to two questions uh you know how significant are the

economies of scale on the supply side right um is it a dynamic where the more tvl you have the cheaper the security you can offer right to the demand side of ABS is um and the second one I wanted to ask you about which I think is arguably more important is you know are there Network effects to reaking protocols and specifically demand side Network effects meaning the marginal you know ABS that joins the system makes the system you know even better right um and yeah would love to just hear how you think about that and then if there are how do you position igen layer to really unlock these Network effects um you know maybe being slightly opinionated about

what you build yeah um the term ABS is actively validated services so any service that wants to kind of consume the Shad security and decentralized Trust from ier and we called it abss actively validated Services rather than actively validated chains or anything else because I think that was what was natural to pretty much everybody the idea is oh you got stake you're going to run a chain like that is the obvious thing to do and and I think we did not think that in fact I would

say Igan L is entirely a demand side protocol it was constructed for distributed systems Builders it is not constructed to like maximize tvl it's kind of like an accident of history that we got some amount of tvl right now and I think you are right in that tvl is the most like uh temporary Network effect compound head $20 billion tvl like three and a half years back and now it doesn't and so on and so we don't take that part for granted that you know you have TV all now they may not exist in you know even the token announcement like one of the biggest things is what happens before and after and like know one of

the nice things with what happened was basically there's no change like it just goes you can't find where the token announcement happened which is I think a pretty crazy outcome but that's these things are not be taken it for granted they'll change in 6 months like so I don't think there is a significant Network effect from TV itself so going to this idea of actively validated Services it's it comes from a certain world view of what needs to be built and like the starting point of this world viw is I think what we call in crypto modularity right modularity is like different modules do different things but it's a specific spefic view of

modularity which is not that there are three modules because there are three maybe like you somebody would just build those three or like there are only three winners right and you don't need a platform to just get three winners like it's not a stable equilibrium the world view is that there are going to be thousands of interesting services and it kind of comes from modeling what happened in the cloud right in you know if you look at like Computing you had the Mainframe error which is you know analogy to the Big Chain error and then you have the personal computer or a personal web server each each company like website

runs its own web server and that's like a app chain error and then you have the cloud error where like you are horizontally like figuring out what the functionalities are and each one lives in a certain like software as a service or whatever module as then now you just kind of integrate all these different things to make a useful end user application so igen is built for this third worldview which is lots of services and end services are then concatenating end applications are concatenating a bunch of these services to make useful productive crypto applications so it's built on this like view of the world and so that's why AVS

is actively validated Services because you don't need these to be chains and you know them being chained is actually kind of slightly counterproductive because each chain has its own ordering function rather than it's a service that is called for by any application and then it returns that answer and if it doesn't return the answer correctly it's going to kind of be slashed and so on hi everyone this episode is brought to you by Kinto the safety first L2 with user own kyc and Native account abstraction that is accelerating the transition to the onchain financial system now we've been talking about Larry thinkink on this program and how he's been shouting the good word about tokenization to all of his buddies on Wall Street this is happening and the very cool part about

Kinto is that they are built to accept a whole bunch of assets a whole bunch of users from the traditional Financial system which current l2s have done an okay job but there's a lot of room for improvement for onboarding so it features very nice things like user owned kyc and Native account abstraction that solve some of the biggest blockers around mainstream adoption namely security and user experience you can become a founding member at engine. kin. XYZ if you believe in this onchain financial system join Kino's launch program and thank me later cheers guys and the link bottom of this episode click it now could could we um get into some of the more almost like squishy like branding go to market uh considerations that some of these uh

Services might be having because I think as we've seen in crypto time and time again it's very nice for some of these new projects to be able to buy into a narrative right um and so we saw this with eth rollups and obviously there are many concrete advantages to being a rollup but an enormous amount of chains that were sort of existing in their sort of like maybe they were full stack app chains or l1s that never really got product Market fit they kind of like Hey we're a rollup now and Everyone likes rollups and it was it was part of their goto Market strategy and so I could see projects uh positioning themselves as an ABS not only because of what igen lir allows them to do but also it's like hey I'm part of this ABS club and everyone's

excited about this and they want to learn more and it's kind of buying into this brand and narrative on the other side I could also see uh you know maybe another reaking platform taking the opposite um you know taking the opposite approach and maybe there's some big projects out there that are saying hey I actually think whatever the construction that I have now maybe I'm a smart contract app on ether maybe I'm a roll up this ABS thing that trium is talking about IG layer this sounds great right but I already have a brand in and of myself and I don't really want to be in this like ABS Club I want to be my own thing and I want the infrastructure that reaking allows allows for me to just be a tool that I use something behind the hood something like that so how would

you compare and contrast those two approaches like what are you hearing from the abss on the ground yeah I think um to the extent that this squishy Narrative of oh I'm an EVS exists it's not intentional from our end Our intention is to be like the cloud so which means nobody no consumer application is saying hey I'm I'm an AWS uh cloudbased like you know app nobody nobody cares such a crypto thing be like ABS let's go let's make out of thing it's really

funny it is a crazy crazy idea that that's a like a differentiat or anything like you know the but there is one one thing which is concrete and real about it and that is when there is an economy so to build something specialized you need specialized counterparts so let me explain this right like if I want to be uh a data availability layer let's say I need an execution layer which is just an execution layer you can't be a data availability layer if there is no counterpart of the specialized stack and that's the same thing with igen layer is

what it's doing is creating an economy of services and you can be one of the services because there are a lot of the other services and just a service by itself doesn't have like an economy to stand on itself an AVS could exist only because other avss exist so this is a different kind of like a this is like creation of a kind of evolutionary Niche so now that there is this family of abss and avss are supposed to be super specialized I'm the X thing which then like is used by other avss I'll give you an example like a proof of location service okay it's a service that says hey you know I'm going

to measure each of the operators where they are and give you like a certificate of location now what do you do with this service like if you have it outside like it's not clear what you would do with it but inside an AVS economy it could be like every AVS comes in and says hey when I give operator rewards I give like a decentralization bonus for more decentralized operators so it's a service that CS to other avss and because the other avss exist in a in a kind of like a Cooperative economy you can be an ABS so if I just take the narrative part out and then ask what is the kind of like that would be something where you're thinking consciously about the the

remainder of the evolutionary ecosystem and then say Oh I I'm there's this big bucket to slot in or to create inside of that and and that I think is like something concrete that could happen and we see happening to an extent of ready that when you open up specialization other speci existence of other specializations means there is opening of other super specializations to show up and so that's a that's an effect where it's goes beyond just simply branding it it's the same reason why every chain needs a DEX right like you're a chain you need a DEX you need a learing platform you need this you need that and the first one obviously needs like that with a higher like adoption momentum

than the the next one so being the first kind of like the like the reaking platform and being Builder Centric and thinking a lot about what abss should be what should be the right structure how they will interoperate we have a little bit of an edge in actually empowering and creating these things so one thing I find when like talking about evss is I L is not only premised on a certain model of how the ecosystem will be built but also in a certain model of decentralized trust which is things like attributable security things like you know you should have insured Economic Security right where you know how much security you

have and you should be able to slash and redistribute that portion these are new Concepts this was not how any blockchain was built before because they're new Concepts and we invented we kind of shered the avss to think about it like that so imagine you are an AVS which is building on Unis Swap and you're doing this much transaction volume on a weekly basis make sure to have more than that amount of attribution and economic security on I layer so that itself opens a niche because it's a new trust model because other there may be other like co-processors but not thinking through the trust model carefully so now that opens an opportunity for somebody beyond the marketing these are the different

nuances that actually exist in building anagon Aras some people are building them carefully and as in any ecosystem lots are kind of like building whatever they can today and we'll figure out the better options over time I think that yeah makes makes a lot of sense so maybe you can even split up igen layer's like demand side into the avss themselves so these Services um and then consumers of those avss right and a really easy example of a like AVS consumer is a rollup um a rollup needs lots of different you know infrastructure Services all of them can be specialized you know they're not going to operate their own Bridges they're not going to

operate their own co-processors right and so is the is the idea here that you can you know for the customer it's kind of like Amazon where you can kind of you come into one platform you get the let's say I da is like the the core thing you need right but then hey there's all these other you know services and if they're all here I might as well you know look at them and uh they all kind of have the same you know uniform security Properties or at least guarantees right so I'm not fragmenting you know my services across different security and then for the for the avss especially like these stat list uh Services um you know it's a distribution play right where they they're on the platform they get eyeballs um and they

have the ability to to you know I guess capitalize off of maybe some of Ian DA's adoption right and so is that the way you're thinking about it are you kind of um proactively I guess helping put these pieces together um when you have let's say an igen da customer um say okay here are these other avss that you could that you should be looking at um yeah it's a great question I think the way we went about adoption are are going about adoption is like a spiral so what's the spiral so you start so stakers as operators there is AVS developers and then there is AVS consumers things like rollups and the spiral is you you look

at which is the site that is the lowest friction to start and that's stakers just move money okay that's stakers start from stakers then go to operators because once you have stakers they need operators operators show up once you have stakers and operators then AVS developers show up so AVS developers are showing up now and then once AVS developers are there then AVS consumers show up and particularly we being our own like one major AVS developer as IA lets like AVS consumers show up for that and then then we build tools to make uh AVS consumers integrate across many AVS developers so

that's the four-sided like Spiral like just go from Staker to operator to AVS developers to AVS consumers to then get more stake and like just play the flywheel off that but we are right now like one of the things you know principles we have is make it possible then make it easy then make it scale right like so make it possible like first like is it possible today to be an AVS consumer and consume many AVS is yeah it's possible is it easy no it's not easy is it possible to be an AVS developer possible and like it'll chew a little bit of glass to be an ABS developer and

yeah it's not is it easy no it's not easy so our principle is first make it possible we are in in the make it possible phase once we make it possible you then make it easy have more sdks e EAS to build abss the same thing is true with EAS to build AV a consumer of an AVS like is there a rooll of development kit which comes in with like you know 20 buttons and say oh you know this Oracle this da this uh you know Ai inflence and then like you are a go that is you know one outcome state that we are interested in creating the other you know pass about Network effects and I think there are a few

Network effects of like IG ler and and in general systems of Shar cryptoeconomic security but I specifically designed for this the number one is the pooling effect the pooling effect is instead of having 100 pools of $1 billion security for each protocol if there was a hundred billion of security shared among 100 protocols is the same fees but it's more security it's just better so there's a kind of like a coordin problem of bringing these 100 things together that's what we have to solve to be at the center of that like you know pooling Network effect that's the first one so pooling makes things just like

better the second one is um what we call the uh elastic scaling of security so elastic scaling of security basically means imagine I'm a bridge and I can re ire a transaction my weekly transaction value fluctuates between 10 million and 200 million if I were to provision my own security I have to provision 200 million of security because you know I may need up to 200 million but because I'm drawing from a common pool there might be a way that oh there is this 20 billion and then I draw 200 million this week but next week I only pay for 10

million because that's all I need so you can pay for the like by being so this called elastic scaling of security you're scaling the security to the amount that you need rather than like having your own tool of security this is the reason why cloud is a thing in our view cloud is a thing because it elastically scales compute to how much you need there's a big pool of compute everybody draws like just in time how much our compute they want so there's the elastic scaling of security is a major thing for I layer actually sham uh just to one of the I want to do a little bit of a callback because one of the things that we're

exploring this season that I think Igan lay intuited correctly because Igan lay is not the first iteration of shared security is Miles and I went back and took a look at uh polka dot and uh some of the earlier shared Security iterations in Cosmos and actually there's there's an analogy as well to um some other protocols that have subnets like Avalanche but there's this very consistent trend of being creating this like on demand or very elastic demand for compute resources and I I just want to give credit where credit is do I feel like Igan layer understands that very intuitively and I want to underline again for the audience this is something that earlier iterations of shared security uh didn't get right and I think aenir has improved on so just wanted to

absolutely I I think there is still a lot of work that we are doing to get the market structure are correct the pricing and the and the fees correct for this kind of a model it's not a easy thing to get right and like one of the things we saw is for example the bundling of security price and computation price is uniform in all of crypto everywhere you don't have the Nuance of separating I'm paying for more security versus I'm paying for more compute these two are bundled in some complex mess so a single ethereum transaction with

transacting 10 billion will cost way lesser than like you know like a big gaming transaction which moves like 0. one right like but the first one is drawing a lot of cryptoeconomic load could lead to more me could lead to reorgs could lead to like all kind of crazy things but it's not paying its due because you know you're paying for security should be different from paying for compute and somehow these two axes were tied together and you say oh poker. auction is giving you like a slot of compute and a slot of security like it's just bundled in like there's no Nuance in

separating the amount of security you have from separating the amount of compute that you want and I L's road map is to have a pricing function to price these Dimensions correctly from first principles and then like separately price for security and separately press for C could we talk a little bit about some of the different maybe moving on to this next section I know miles has a lot of specific questions on this but just on the different like uh topology or different Arrangements of security that are possible like maybe miles I know you've got many more specific questions on some of the parameters but I can turn it over to you here yeah sure um so I think we you've just described like a bunch of different types of security and and um and and maybe like there's even

different more clicks to go I guess in how these things can look and so before we the show started we tried to take a crack at like if you were to pick the major let's say parameters or like what are the topology structures of security and then how can you mix and match them right in a way that makes the most sense for your use case and so before getting into like the maybe that that side I would I'll run through kind of what we were thinking about so one is just the asset selection so what are you using as the collateral asset is it native Stak eth is it a liquid token like an LST or even a stable coin is it a special kind of liquid token in igen right um that is

meant for specific like types of work um so asset selection and then I think related to the igen piece is what kind of work are these operators doing and how you know is it purely objective faults right very easy to handle or is there this inter subjective piece AKA things that you know two humans would be able to very clearly see one is right one is wrong but the chain cannot see it itself and so it's not something that can be completely programmatic right um and then I think the last one is uh there probably even more but what happens when there's a a slashing of it um is it slashed and burned or is it you

know seized and then distributed to to the harmed users um and these things you know whether it's pulled or I think that one's specific to isolated right but pulled versus isolated maybe a fourth category so I'll pause there asset selection fault supported uh resolution process and yeah I think like value at risk models is that are there is there anything else that we miss there or would you agree that yeah I think all of this already assume Economic Security is the right right thing so maybe we start with whether yeah yeah let's take a step back people saying people saying Economic Security is a meme so maybe we could start who here who are these people

up to is a good example of a person saying uh Economic Security a meme recently tog and a few others you know both from the ethereum community are like oh maybe it's true it's actually correct so I think maybe revisiting what what's the principles on which these systems you know the the source of trust okay I see really like three different models of trust the first model of trust is what I would call Purely decentralized trust like oh you have many nodes and they're all different and like therefore they're trustworthy this model of trust so so firstly like maybe even zooming out one

more level we see that the whole of the crypto economy is premised on decentralized trust you you remove decentralized trust from it the whole crypto economy becomes Cloud economy there's no difference so there's some element of decentralized verifiable openness whatever you know different aspect of trust otherwise this whole crypto economy nothing makes sense it's just you just go around on a computer anything you can run on a blockchain you can run in a computer so that's right really fundamentally that's what it is okay so the first one is like just many nodes decentralized validators many many nodes are coming to consensus so maybe it's good the problem with this model just by itself

is who are these nodes how do we know these nodes are any good and then the usual answer is like oh stakers chose these nodes and therefore they're good but if stakers are choosing these nodes but they just choose bad nodes for you know Mal you know willfully or otherwise what is the recourse okay so decentralized trust by itself is especially in a pseudonymous world maybe actually much worse than centralized trust because AWS if it gives you like a wrong answer to JP Morgan's banking calculations you know goes and manipulates the server to do it no basos is going to jail like there's a concrete slashing condition that is built into

the AWS centralized model which I think people kind of like you know we in the crypto economy try to minimize but it's real and it's actually the most powerful slashing condition is Criminal liability right like centralized models with the reputation and legal liability actually have like a pretty strong skill in the game to get it right usually they they don't go wrong by making provable errors they go wrong by exclusion like they don't commit errors of inclusion they commit errors of exclusion they exclude somebody from building something and just like sneak it under the carpet shout out low carb Crusader who who figured that out this week yeah exactly right like I you know go God knows if we

would agree whether local Crusader did a protocol action like a sand you know sandwichera the protocol not not clearly defined okay so the first one pure decentralized trust without accountability seems like you're getting the worse off the kind of like the real world because you don't have identity and legal ability and you don't have anything else so the that one doesn't seem to work at least in our view the other one which people point out is oh maybe we don't need Economic Security it's all social consensus that's the other view that I've also seen and this

view has I think there's something reasonable to it which is that oh if the block was wrong in ethereum we will reward the block and if you know there were two blocks you know or data was unavailable like you know that block will be forked all this kind of stuff it's true but social consensus operates at a slow time scale at the time scale of let's say weeks to actually get anything done because you know I publish a header and then I don't publish a block for three hours and then I publish a block afterwards is this block acceptable or unacceptable like there's a lot of thing that you have to socially negotiate is because it's it's not an onchain

protocol so social consensus operates at a slow time scale and is only like perceives faults at a slow time scale not at like oh I didn't publish this block within the first like 400 milliseconds and I published a conflicting block these kinds of things cannot be resolved socially so social consensus is the ultimate back stop for the blockchain economy but it doesn't work in the shorter time scale I don't want to pay for coffee and then wait for social social consensus to occur before I drink my coffee it's not going to work so what's happening is economic security is the one that is transporting letting you transport value in the short time

scale and economic security is like adjudicated and like resolved using social consensus at a slow time scale so this is a correct decomposition of the blockchain stack decentralized validation for non-attributable Falls like oh I I don't know if I can attribute censorship so try to get as many distinct nodes as possible but there are attributable FS like you know double signing data non-available all of these things and I try to like incur slashing but slashing will only be incurred in a slower social consensus kind of time scale so that's the kind of trust model that's why Economic Security

is needed so I think you know I I don't any alternative to build an onchain economy without this intermediate scale of economic Economic Security sitting in the middle so that and what we figured is that people haven't really thought through this carefully because if you think through this carefully immediately becomes clear that the total amount of value transported within the social Horizon should be smaller than the total amount State it's a very simple equation once I say it but you don't see it anywhere else and like we first wrote it in a paper called Stak Shure that this is actually what's happening is that that's what really is being transported by Economic Security and we need a

mechanism on chain for matching these two numbers the amount of Economic Security should be greater than the amount that is value that is transported within the kind of social consensus Horizon and so that's a core principle so now then we can dive into exactly what is the type of Economic Security what is the resolution mechanism and all of that that's yeah that's a a great overview um and before igen layer really the only kinds of security that you could acquire is objective and attributable um which is a pretty limited surface area of I guess um issues that you can catch and pretty very limited and and i' would say like

um not actually very good at resolving them because it's often the you know not the malicious operator whose deleg who whose funds you're actually slashing right um so now we have okay Ian layer introduces a way to programmatically or or a combination of you know onchain and offchain coordination um to support inner subjective faults so this is not needing you know slow kind of um I would say unstructured social consensus to uh to support something like uh an oracle providing the wrong price feed right and the problem with that is even if you do reach social consensus on the Oracle right that provided the wrong price feed

you don't have actually recourse to to reorg ethereum right where the users were impacted um okay so now we have a way to actually punish The Operators and stakers of something that is inter subjective and then the last piece that you just touched on is this idea of attributable security or or really Insurance um and won't go like through the whole stake sh paper but it's it's essentially it's a a subset of a subset of a subset of value is actually at risk at any given time and it may be small enough um and that is actually feasible that it could be insured right um meaning that if it's just let's say

Bridge transactions or you know um one side of the transactions on chain the other side is coinbase paying you USD to settle your withdrawal right these are things that uh today you just end up you like coinbase right just Waits a long time before it actually lets you settle it um so those are the three types of security now available um and there's a lot of work to do to guess put in place infrastructure that allows you know this Market to kind of develop for it but really what I want to get your sense of is how do you coach avss through all of this right it it's not just you're introducing New Concepts it's okay these

are all the tools at your disposal um what what do we you know how how maybe you can't be like totally prescriptive but if it's just us talking you know how do you how would you advise them to Think Through how to use them um because it it does introduce new complexity right and so but it it also offers unlock so that's a really really important question I think the way we think about this is AVS Builders are accessing raw decentralized trust so they need to understand the dimensions of decentralized trust and compose them in ways that make sense so it is an important and non-trivial job for to build an ABS so AVS building is is done with this like hat of hey I

need to understand the trust models I because I'm providing a service with a certain amount of trust guaranteed okay so one kind of generic prescription that works for any any fall that's at least transparent off chain is the idea that if I take an amount of insurance greater than the value that I'm transacting within this like social consensus p period Then I am safe right so it's a very simple kind of a model where you're saying so the API that you know um an

AVS needs from Igan L how do we simplify this as much as possible that's the kind of thought that we put into it and the simplest one is saying hey I'm going to transact like $100 million in the next couple of weeks and like that's the time for like reorg or social consensus so I'm going to say that I need $100 million of like worth of this uh igen redistribution or E3 distribution to to match it this is I think a simple enough model for all the fults for which you can actually like slash and redistribute and and and I think Miles mentioned briefly not only each staking in I layer has this in the road map even IUN

staking has this in the road map where like you know when you stake iGun and if there is a f that you know is attributable from the people outside and what happens is the POR don't burn all the money of the malicious like stakers you redistribute portions based on onchain equations and then like say that oh this guy gets this much this guy gets that much this ABS gets this much and so on so that's the prescription which I think is relatively general purpose right the prescription is what is your value at risk within this like uh you know couple of weeks period and then now you state that value to I make

sure you get like that amount of insurance now you have the correct amount of cryptoeconomic security in fact what you might also want to do is have a nonchain contract for your application it says hey I'm holding this much amount of like attributable insured security so don't let more than that amount of risk be transacted in the application this is like a kind of like a formalized circuit breaker people have like right now for example put some kind of a circuit breaker which says okay you know don't want to take too much risk so I'm going to bound the total amount of bridge volume by X but this is exactly specifying like a tight counterparty saying that oh the the stakers have promised that they'll do this correctly till this much amount of economic value

so I'm actually going to let only that much amount of economic value pass through so that's a general API that we want to achieve it's a single variable to communicate like how much trust are you kind of eating from this common pool and this is not only this regularizes the specific guarantee you get as an AVS it also regul regularizes the payment and pricing mechanism because you are paying for attribution in a non-leveraged manner like if if I'm telling you I'm giving you $100 million of attribution from the igen L protocol that means that $100 million is not also attributed to somebody else so which means that you have to you you have your

incentivized to pay for it otherwise it won't be there so that's the uh it regularizes the payment mechanism it regularizes what you're consuming and then the question about whether one way to get like attributable Security is to have segregated security say oh you want $100 million of security just $100 million of stakers are running your application that would be like a segreg model and the thing with the segregated model is you lose out on the benefit of pooling the fact that there's a big pool you don't you don't benefit from but the way IG ler works is instead of trying to ask each Staker to say like how they're

allocating across all these different applications one way you can think of is a Staker participates and says that hey I'm validating all these different protocols so if I violate the commence of any one protocol I may lose all my money but how the money is burnt and how the money is redistributed to a AVS the igen protocol maintains protocol level calculation to do that so instead of making these so the end API is very simple for a Staker a Staker opting in is saying hey I'm running these 10 Things I I have to run all of them correctly otherwise I'm may lose my deposit from and and from an application the API is oh you have $10 billion of pool security and $100 million of unique

security so that's the trans so make it simple as possible and do the hard work kind of in the protocol to make sure that the solvency of the protocol is maintained irrespective of who may draw their attribution at what time so these these things can all be they can play together and and you can you can mix and match and choose a combination of you know based off of what you're doing um and so maybe like to let's just make it really clear-cut example like um how about a bridge uh that's an AVS right so what is um a large amount of pooled eth security get you um well that makes the cost of double signing attacks really

high right and it creates a even larger downside risk to reputations if something goes wrong um what does you know uh I don't know actually if a bridge would have anything inter subjective but like if it was an oracle right um you could pair that with igen right which then is also pulled let's say there's a a large pulled number as well so that makes the cost of attacking you know via inter subjective faults very high and now within each of those two security tokens you could measure a second subset of of security right or like a smaller let's say it's whatever maybe your your pulled amount is 1 billion and maybe let's say on a day-to-day basis you actually have $10

million of value actually at risk like meaning these hybrid transactions that flow out of the system that there's no recourse for reorgs um so you would pair that you would also purchase a much smaller number of uh of I guess Capital that if something went wrong could actually make the users whole right so now you have put it all together you have a much higher cost to attack your your protocol in the first place um and if you you know even if there was an irrational actor right that decided to attack your protocol uh even though like basically what the the profit right even if it's below the cost the reason you

need attribution even if you had a large amount of pooling is you don't know who else you are pooling with right okay because when the pool security is attacked they will attack not one protocol you just attacking your protocol is not going to be profitable but attacking 100 other protocols simultaneously is maybe going to be profitable very difficult for IG ler to know and maintain the cryptoeconomic balance to make sure that the total profit that an attacker can extract from any combination attack will be smaller than the total amount they have liable to lose this is a hard you know non nonprotocol verifiable calculation but if you took attributable security for 100 million just for yourself or 10

million in your you know use case number then you would say it doesn't matter matter who else triggers slashing simultaneously I will be able to slash and redistribute 10 million to my like Bridge users no matter what like it's a very high grade of security which is even if even if it turns out that the total profit you can extract from attacking all the protocols is much higher than the total amount at stake you will still get your 10 million back like it it is a unconditional correctness for each AVS that they will be able to slash in and redistribute their attributable security so this gives a mechanism for

like the uh avss to have a guarantee independent of making assumptions about all the other avss yep and why is that like actually important um longer term for the future of crypto I don't know if you agree with this but like the you know crypto's inability to reverse transactions is is just scary to to institutions to to end users to to anybody right um you know worst case scenario you could have you know like even if it was like a swift you know transaction you can have government step in to reverse them um that's not possible yet on chain but you're saying in this future that's why it's so important to have this actual

like redistribution of collateral um and that you know that will open up or at least make crypto a lot less scary longer term for for like you know you know these institutional participants that are not used to be like you know this this new assumption that if they send it to you know something goes wrong there's no there's no recourse yeah totally and I think like this was one of the things that I got asked is like hey who is going to like nobody seems to care about these kinds of like hard Notions of like security and like you know pretty much every chain run like they all been fine like why do you need any of these kind of fancy primitive and to me the answer is typically

consumers are not great at doing tail risk calculations yeah you need the institutions to come in and they'll be like Oh I'm now deing at scale I need to make sure that risk is Quantified correctly and you know if you let people decide whether they want to take auto insurance or not like they're probably not taking like right car insurance but because it's kind of mandated at where levels and so on it just happens and in the same way like an exchange operating on like a blockchain would be like absolutely I want to take like Insurance attributable security otherwise I won't operate and like big players moving their money around would want to take

insurance otherwise they won't operate whereas you know that should also be space for small players to say yeah you know I'm just doing a small transaction I don't need to care about all this stuff like just have enough buffer in the system so that they can also operate so that's kind of like our guiding philosophy that in fact this does the right thing like that you know you know we in the crypto space like which is let the bigger institutions pay for the premium of being a bigger institution let the smaller players like you know operated a little bit yellowing into the system and have buffer so that they can operate and so that is I think the right kind of scale whereas we don't see that today that the coinbase is paying more than a normal user they're moving a lot

of money and they actually pay much much lesser on a kind of like a you know bips basis than like all the other players exactly exactly um okay I think that gives a good overview of the so what to to why H these pieces uh should exist and I guess last piece on the the security side before we move on to I guess the segments of of use cases that we see for abss there it you know does beg the question okay we used to have this native staking token uh as kind of the you know standard for all sort of either you know app chains or offchain networks and you know whether it's right or wrong the idea of having a staking

token has been beneficial to the you know market caps of many projects and so there's a project that's looking at this and saying okay well I was I knew I wanted restake eth right I'm not sure about igen with with the do some research there but what about my own native staking token um and you know I know that Ian Lair at the moment at least wh lists you know the coll acceptable collateral assets and so it's not realistic that every single ABS you know token's going to be in there but um how does how do you how would you respond to that like to to folks that might might have concerns yeah totally I think there are like from the AVS point of view one of

the kind of core principles we have is open Innovation we want to maximize what you know Innovation happens in the space and if you if you take the view you say okay you know an AVS wants to use their own token in addition or in you know in in varying proportion to e or igon or whatever else that should be acceptable so you know over time the collateral types will become permissionless and people will be able to State their token for sure okay so on the other side then like are they leaking value by you know also adding an you know sending a portion of the fees

to eat Oran or something else as opposed to keeping all the Val for themselves and I think the there are several factors here the first one is bootstrapping on on day one you know do you have a token maybe you don't even have a token and you know you know you want to build an a useful application before you have a token it's powerful you want to do it you can do it on i l so that's number one number two I have a token but it's just starting out you know it's a it's a $20 million token but you know um chain link is 10 billion and I want to provide a useful Oracle service not capped by my own like size

there are use because I figured out how to run this much more efficiently people have to pay for it or maybe integrate permission LLY into rollups whatever my new feature set is that's attractive security should not be a limitation for supplying the service and this is one of the main reasons we created ion lay is that security should not be a bottleneck for innovators and supplying services so that's a second reason for you know you say oh yeah I get 20 million Economic Security from my token but also I get like 10 billion from e or ID or whatever okay the fourth one is um the igon layer you can think of as pure optionality like how much e taking

to use how much I taking to use this is optionality over time you may say that's zero and that's totally fine like as a innovator you figured out like how to bootstrap your thing you know you know it's enough to like stake your token just do it it's totally fine so because the AVS token is going to govern it's not you know on the governance layer it is not going to be e and like a or whatever else deciding how much to allocate security between the ABS and and eth andon and so on you can say like over time as you figure out you know your own economy you say no I don't need I I don't need like e and I and so on that's totally fine too

so because IG layer is pure optionality because I layer is pure optional it only increases the value of the AVS token if you couldn't do this whatever value that you have you just have an optionality to turn on and use it to bootstrap to insure to like do whatever things you may not be able to do in your system have this additional degrees of freedom that doesn't decrease your value that just potentially increases maybe doesn't increase but it doesn't decrease your value for sure so that's the first thing to realize is I I because of the structure of how we built this it's incentive compatible for AV is to say oh my token value is not going down because of this it's if at all it is augmented

because I'm able to bootstrap myself so the there's another thing that I think people haven't fully understood which is what is the right collateral for like various like Economic Security and people think of usually volatility as an important metric and e or BTC or USD may have less vol volatility but it's not really true pure volatility volatility that matters it is the cross volatility between risk and the economic security device right imagine I'm moving 1 million e every week in my bridge now my risk is measured in E like so if I can

get insurance of 1 million e that's just better than getting igen or getting like in AVS token because then I have to measure out for cross volatility risk within a week or a two week period and then like buy much more to be sure that I'm I'm covered for my downside risk there so that that reason also gives you like you know another reason for people using like you know the uh e or i on finally so then the question comes back to us to say oh maybe all that will happen is a will bootstrap on I layer and then leave and the first thing I say to that is is if we bootstrap the whole

open Innovation economy we'll be very happy but the second thing we say is we have to figure out how to provide value added services that make it useful for abss to stay in the economy rather than thinking like oh you know they're stuck with us and they're going to we can keep extracting rent from them and I think this is just not true for anybody you may think you build an application built on a chain and you know if you're extracting too much rent the application will move out of the chain so this is just like the value added nature needs to exist for it to be sustainable so we think a lot about is it elastic scaling is there an economy of scale that when you have many AVS as each AVS supplies a

certain kind of service and together there're more than the sum of the parts is there a switching cost is there what set of things that actually make it much more reasonable for abss to want to stay over long time scales and that's a problem for I can not a problem for the abss to solve so sure if I maybe to summarize a little bit um what we've talked about so far and move into our sort of final section of questioning here but you know we kind of did an overview of State of the Union of raking as it exists today the amazing amount of progress that igen lay has been able to make building out this ecosystem of AVS is and it sounds like you know where Network effects uh exist at least on the demand side of

this Marketplace uh that you've built sort of has to do with providing value to your abs ecosystem such that they want to stay there over a long period of time and obviously there's been you know there's been an enormous amount of discussion about like what is the right way for these abss to procure security and just all the different decisions that they're going to have to think about I think one something that would be interesting to get your take on is what are the different subsections or sectors of abs and what's the right way for listeners of this podcast to think about that that and one way that Miles and I have been trying to uh to reason through this and it's a little bit tricky because it's still so early but a lot of these words and Concepts just aren't out there and don't exist is you

know a lot of what we've described today in AVS land is what we describe as like external um a facing AVS so maybe the equivalent of to use a really basic example like a cosmos chain that might have existed over in Cosmos land and now it's using restake e security and there's a you know native token yield that's coming to E3 stakers um another another I think large burgeoning use case uh is actually using reaking more endogenously to eth the protocol itself and using eth resters to improve ethereum the base protocol there's a bunch of different sort of middleware services that can be used to do that so this is sort of what people are excited about with pre-confirmation you were

describing that earlier in the episode and earlier miles and I talked about the intro of this season like almost an outof band uh solution that could have been described as just boost which is also getting a lot of uh traffic today I saw donrad and um uh Peter soaz going back and forth about that but in in maybe if IG was around earlier that could have actually been facilitated as as an ABS so I think people are really excited about this idea of almost like endogenous using uh some of the uh allowing proposers to commit to these additional things and actually improving uh parts of ethereum the protocol itself and I think that gets lost in some of the discussion of oh we're going to secure all these protocols and bring all this yield to ethereum stakers but I

think one of the really exciting things that doesn't get talked about enough is the potential of reaking to allow us to accelerate the growth of ethereum and to enable really cool things like uh base sequencing and solve the fragmentation problem on the rollup layer uh it you could uh you could accelerate the adoption of single slot finality uh things like things that would if we were to do this through an EIP would take many many years but now we can almost iterate on them in in a sted way and then maybe if they're successful enough they could find their way into the protocol can are we thinking about that the right the right way and like how how would you define that whole category yeah absolutely I think the reason we haven't I know I give a talk on like me boost Plus+ in in the Stanford

blockchain conference in 2022 September so we've talked a little bit about this before the reason we kind of like stopped talking about it over the last year or so is this problem that when you know $1 billion is a lot of Economic Security for pretty much any application but $1 billion is not enough is is only like 1% of the block proposal on ethereum so this like e is such a big like ocean liner that to barely make a dent on it now we are starting starting

to be at a scale where we can say okay you know proposal commitments may start to make sense because you have 10% of block proposal something like that so that's a very important like thing that ethereum is such a big thing that like getting enough block proposal to actually participate in reaking so then you can start doing these kinds of services which is starting to be at scale to do it so one of the reasons we haven't focused a lot on this in the past we're going to start to focus a lot in in the coming years right like we are a very ethereum native project and improving ethereum is very important and you know a huge part of why we're doing this and

so we'll start to see like you know many things happen in this proposal commitment landscape but to give a sense of what kinds of things can you do right because you have this ability the first thing is you can start doing you know pre- confirmations right like I I have a rollup and I want the rollup to give a confirmation to users the rollup is using ethereum itself for doing transaction ordering but if it's using ethereum for transaction ordering you have this big loss which is that you have this like really amazing like instant confirmation in arbitrum and so on where you're used to hey I got confirmed in like zero time and now I

have to wait for ethereum Block it takes 12 seconds and like you know you need to be sure that the block is in all of that it just uh like spoils the ux completely instead if the block proposal was restate and they already can make an extrinsic promise that hey I am going to include your transaction later that can happen in 300 400 500 milliseconds and then when they include it they include it and if they don't include it they'll get slashed so reaking actually gives us this ability to add features and change the time scale of the ethereum protocol in a way that like you couldn't do before and it allows us to do it because like I mentioned before in because I has native restating built in allows for

Universal participation in in that kind of an economy it doesn't I don't need to be like in an LST or something because you know that's one thing that is inevitable is without an optionality like this like major exchanges or major node operators can make offchain promises about various things and they'll be credible because they're big and huge and they have the ability to make those kinds of promises but smaller operators and like you know you know permissionless Global operation is not going to be possible with that you need trust and reputation and you know aggregation and centralization to

actually make those kinds of promises just like me boost was a kind of an attempt to make it transparent and Democratic permission list for anybody to participate in that native reaking allows anybody to participate in let's say pre-c confirmations what else can we do we can start doing things like partial block auctions I can promise I'm selling a portion of the block to Mike and a portion of block to Miles and this claim will be credible because if I don't do it I'll get slashed you can start doing block space Futures by saying that whenever I get Block you know participation in the future I need to include ABCD if I don't do it get slashed I can start doing uh EV driven actions that whenever a liquidation

shows up I have to include the Liquidation in my block if I don't include it I'll get slashed so we think the space of pre-confirmation is pretty huge it's like we're barly getting started with this it's one of the places where we find that both big players can participate really well in in a market like this and small players can participate because the native raking and stuff the big players can participate because they have a little bit of an economy of scale so imagine I'm a big player I have like 5% off or 10% off like you know slots in ethereum now I don't need to restake the 10% of all the all my proposers to participate

in this market I might just put enough value at risk say put like oh I put 10,000 e well that's a big enough number and say that whenever I do this whenever I get these slots I will do my actions correctly and you're making a promise for a huge amount of like eat or or slots but you put up like you know this 10,000 e and because slashing and like redistribution can happen at a faster time scale than I'm going to deplete that 10,000 e it's good enough so big players have like this I don't need to restake everything but I can still participate in it small players have an advantage that I can just re nativ stake and participate in it so everybody can

kind of participate in this proposal commitment economy once we have like Rich expressive standards so that's a direction that we want to take going forward where like all these different players can come together and participate in a universal like you know propos a commitment economy I'm honestly I I think I might be maybe more excited about this category than um even the the native or the external reaking um I don't know it's pretty pretty even but this is what I'm I'm really happy we're talking about it because I think earlier this week we saw Lio um you know they announced that they want to support pre-com pre-ms um and I guess if you think about the characteristics of this

category um you know I guess its usefulness to your point is a function of participation right so it this does these things don't even really become possible until until you have you know a good portion of the um or not not possible but useful until you have a good large portion of the set that's opted in right um and then two you know these are only things that interact directly with ethereum consensus and these are basically plugging into the whole evm value chain um I guess the last piece that I'm a little bit unsure on um between in terms of the difference between external and Native is external reaking it's typically The Operators

that are actually running new service right they're running software Hardware to secure to actually operate these Services whereas today's you know me value pipeline or workflow right we have this idea of proposer Builder separation um and so that's just one example of proposer worker separation um do you see uh all of these different pre-commit or let's say just you know proposer commitments broadly whether it has to do with me or or giving an L2 a precom right do you expect to see the same sort of Separation uh or is it you know actually going to be uh the proposers

are running a bunch of software here too great great great question um I we've thought a lot about this from for many years now and one thing we saw we started seeing is people started talking a lot about things like Builder commitments oh you know a builder can give you like commitments that they will do like this and that I think this is a very bad Trend if we get a lot of Builder commitments the reason is Builder is not an ethereum protocol role and ethereum has done a lot of work in trying to keep block proposers decentralized so now like if you say Builders are opt you know making commitments firstly it centralizes the

builders maybe you say it's okay but the the problem is Builders don't have like monop power even temporarily in the protocol they have like very weak kind of competitive Monopoly Powers not like enshrined Monopoly Powers proposer is the one that actually has protocol level Monopoly power that slot belongs to that proposer right like that's a and it's the role that ethereum and you know flash Bots and me management has worked hard to keep decentralized okay so given all this how does proposal commitments fit in there's a bad world where we say oh you should not do builda commitments we'll just do

proposer commitments and proposers are now burdened with running like very complex additional software okay so that doesn't make sense Builder commitments doesn't make sense so what makes sense is our theory is proposers make commitments that Builders keep and relay verify proposals make commitments that Builders keep and release verify proposal is the the one that's in the protocol they better like have a say in what you know it's not a censorship commitment that they're making it's a inclusion commitment over liquidation it is a POS because you know there is dystopias abound in this kind of proposal commitments and you we want the

wisdom of the crowd in adjudicating what is the right things to do which actually like bring right value to the ethereum protocol so because you could very easily build a censorship Market on top of you know proposer or Builder commitments you take the three top top Builders pay them money and say don't include Mike's transactions and you know it you'll get liquidated and then I'll take rude rude uh sham I've got a so I've got a question for you on that and we did a we did an episode on pre-confirmation actually uh with Hart last season we had preev and chainbound on and they were actually it's it's interesting the market might be taking care of some of this dystopia because so Prime have's original angle was starting with Builder

commitments and they're what one of the prerequisites of getting builders on this uh getting them on board was that you actually needed the buying from the proposer because they didn't actually have the they didn't actually have the Monopoly power over the block and so I I think actually there's a little bit of market conditions which are this is one of those nice situations where the market conditions are kind of giving us the beneficial outcome here uh which is extremely nice but um I know miles you got a question on just like the relative Market size here but but could you just walk listeners through r that exact instance so let's just go through what is an example of like a commitment that a proposer makes that a builder executes that a relay verifies just walk us

through that like one example yeah uh let's say I have a um protocol a liquidation protocol that says that you know it says here is a pool of Liquidators and whenever like you know this price is below this according to this onchain verifiable state trigger a liquidation right now like you know when you look at for example what when and how a decentralized stable coin is viable one of the problems is the over collateralization requirements of of the how much you need to deposit relative to like you know uh what your backing and the over collateralization ratio is dependent on the time to liquidation because if you can liquidate

every block is a different over collateralization ratio from if you can liquidate in like you know 50 or 100 blocks so it reduces the over collateralization linearly like the the the over the additional Factor goes down linearly it's really crazy powerful thing okay so if you look at the um so processing liquidations consistently and you know within a block or a couple of blocks is a huge advantage to these protocols because they can now enshrine powerful parameters rather than requiring like some EXT external uh um assumptions so what this does is uh okay so now a

proposal commitment is whenever this protocol is under I would do trigger ABCD transaction this is a proposer commitment that let's say 80% of proposers made so you know like pretty much every blog somebody has to come under it but who's doing it to trigger this I have to check all the protocols that we have opted into check whether any of them are underwater and then include those transactions as a kind of forced inclusion in this block that's what the proposal but that's a lot of work like I have to have additional software running I need to have awareness of you know compound and a and maker and everything and like just headache so that's a commitment now a proposal says oh you know liquidation for a maker good things I'm getting like

some a tokens maker tokens or like some like regular drip of fees something I opt in and like that's what the proposal is doing then the Builder comes in and the Builder there's a signaling language that is already agreed on and the Builder comes in and says oh that's what the proposer has agreed to which means this block will be invalid if I don't include those liquidation transactions so now I'm constructing the block I already take into account the constraints of the block you know in my me like auction whatever other thing I'm doing see this is the constraints of the block this is like three con transactions these transactions have to be included I included now auction the remaining block construct the Block in

whatever other way but this has to be included and I do this and then I construct the block send it to the relay relay knows the grammar relay knows that like these liquidations have to be processed checks it signs it off and then sends it there's a lot of questions about exactly what this grammar is what is the language does it work with an optimistic relay in the me pipeline all of that like so we have to sit down and spend years decoding all of this stuff which is the kind of fun we like to have but uh that's that's what the kind of generic structure that we envisioning okay that is really really helpful and I think there kind of two two questions off of that um who puts up the capital at risk if they do not uphold their

commitment in this case is that something that a proposer could actually Outsource potentially or is that something that um you know would come from the proposer and maybe there's some sort of like you know recourse to the Builder another way I'm not I'm not actually sure okay great great great question so number one you know you could say proposer is restak right so they they restake so which means like they putting at least their collateral at risk but in a scenario like this the total amount of Economic Security you have is bounded by 32 e like now you know you can say oh you know every proposal will do it but I'll say let me eat the loss and take 30 and like now not process the liquidation take a bribe

greater than that whatever so it's possible so how do we go to more rigid kind of commitments like this is the uh the next step and then you can think of actually oh maybe a proposal doesn't sign off on a block by itself but a proposer key is split into like you know like a DVT type scenario and there is a committee of like 50 people who have sted much much more and they actually have to sign off on this transaction and if they sign off on a wrong transaction they're going to get slashed not 32 e but now 1 million e so it's like internalize an

economic committee into your like you know proposal process like create another level of much higher Economic Security without like you know interfering with the consensus we have all the pieces kind of in the in the space floating around have DVT floating around we have Economic Security D floating around it's going to be a whole bunch of fun work to make this all work but it's possible to make it work at much higher economic Fidelity than 32 e because that is needed to build a billion dollar decentralized table or1 billion doll decentralized table yeah go ahead Mike no I was just going to say this is super you know you just answered well do I'll talk about

this in the debrief but you just answered a massive question for me that I've always had which is how how are these proposers going to actually enforce these commit like I'm thinking about the solo sakers right which are already like sub 10% of the network and when be two people talk brought about like Pepsi or the or pre-com I was just like how are these people going to do this uh so you just unlocked a massive that was a massive unlock for me and I also think there it's interesting that ethereum is going this direction maintaining the Monopoly of the proposer on the Block where it seems like salana is going in this direction of async execution and breaking that Monopoly up not a question just an observation that that like very diverging paths and I'm super curious to see how both of those that is a great great point one of the

ways that we think ethereum at least ethereum ecosystem inherits some of those properties is IG da for example is a no proposal Monopoly system so what you do is if you want to disperse data to a DA anybody who has data just sends it to their end nodes and they can get receipt it's a completely asynchronous Lock Free system operating at the absolute Max throughput possible so everybody can send it parall and then like you get all these receipts and then these receipts are serialized when they returned on ethereum so you have the ethereum which is this underlying system which can maintain this locked system where like one proposer Monopoly actually like arranges all these receipts but you also can have like outside systems this fractal like nature

of ethereum is going to be like so so interesting to watch how it pans out because having a very rigid strong base and then letting like the market kind of innovate on everything else like you know what we're trying to do with ag da it's actually like a super interesting view that yeah like can you get the benefits of you know it kind of reminds me of like the block size Wars can you get the benefits of bigger blocks without well retaining a lot of those same properties um the properties that you want right I and I guess my last question on this is I find it super interesting to think about um just to contrast this kind of reaking to uh what all avss are today really like external reaking and I think

you know from a a value like uh flow standpoint it's potentially like a more unconstrained Market size because it's more like similar to meev boost a lot of these commitments are going to be forms of Revenue share to the proposer rather than um with with external Rec reaking R it's it's a cost for the AVS that is you know uh Now new found revenue for for the for the supply side so the cumulative Market size of external reaking is always going to be constrained to what the you know total willingness to spend is for security versus this category You could argue as like you know it's much more unconstrained because it's just how much

is available actually to to profit or you know um to in some ways benefit from the ability to uh pay proposers to make like these additional commitments um May an interesting question uh I don't think it's that simple the reason I don't think it's that simple is that the net willingness to pay on exogeneous Services is the net exogeneous economy which is still getting verified on ethereum yeah and so the question is how much of economy stays on like R is uh fundamentally under the RIT of the ethereum L1 versus how much of the

economy is much more free floating and under the RIT of the L2 economy right imagine like an L2 sequencer can offer L2 M services on igen layer without having any of these other constraints so the it's basically that question rather than the question of like you know U cost versus benefit I guess like that also there's some benefit but except those that there is no like Monopoly power to to extract that benefit whereas on ethereum L1 there is a monopoly power to extract that benefit so that's the maybe that's what you're alluding to there in the

sense of you know hey you know we have this system which is already transporting enormous amounts of value like what kind of things can we build on top of it but our view is that we want to move Cloud to crypto like it's just where we start and so that means like I can't move all of that etherum L1 so we need all these other other things and over time these other things are going to be like a massive economy that's the bet y I agree so maybe that's more of a an observation rooted in like this point in time right uh versus where we know the space is going is it I've always thought of Von lay as a way to horizontally scale

ethereum um where rollups of kind of the answer to to Vertical scaling is that maybe just kind of like a not precise but it's is that a fair way to look at it um it's a good good call out I think the the model that I have is you know I wrote a tweet on this once which is ethereum is the verifiable internet rollups are verifiable web servers uh ABS is verifiable SAS and IG layer is the verifiable cloud so the idea is basically like just like the internet allowed anybody to put a web server and

they can kind of talk to each other through the internet but when they talk to each other on a web server the other Server doesn't know that the original server did what it's supposed to be doing right and ethereum is this very fiable internet so I plug my web server in and like I can prove to you that I did the right thing so that's a rollup rollup is basically like a web server that's kind of plugging into this like verifiable internet and so is there an advantage of having verifiable internet yeah absolutely because I know when you're telling me something that you're telling me the right thing and that's a big deal and you know the obvious place where it matters is money right like you tell me how much balance I have and I know that's the balance I have rather

than hoping that you're telling me the right thing but the so this is the model rollups are veryable web servers and then like when you think about web servers like web server used to do like pretty much all the application logic was sitting on the application web server then all time you find yeah but identity is a universal service it doesn't have really much to do to the Amazon book service it's a universal service it's the payments is a universal service like uh maintaining a database a universal service and going from each application maintains its own entire application logic to abstract out portions of things which are common like Bridges is common Oracle is common a

database is common like this is common AI inference is common create these as like SAS Services that's what happened in the internet is you have the softw or service layers that actually then service those things in the IG world view we think of avss as abstractions of things that worked well in a particular role R up or a chain context and then abstract them as a s as an example let's say osmosis has threshold encrypted mols in in its road map and it's such a useful productive service but really if you ask you know is it specific to osmosis or it's just a useful service for a lot of people just a useful service for a lot of people and the market structure of an app chain kind of ties the the things together in a very

tight manner extricate that as a service run it as an igen layer AVS you have an encrypted mol that any body can now subscribe to so these so avss are verifiable uh SAS so you you have an encrypted mle but you know the encrypted mle is supposed to be doing what it's supposed to be doing and now all of this runs on like I can L decentralized trust and staking and so on and so that's the kind of view and it's a completely complimentary view to the the um rollup era like it's actually like support put in the roller pair a lot more than it is kind of conflicting with it and it's turbocharging it because for example oh

a rollup had to be by Fab had to write data at ethereum doesn't scale create a service I da it scales now you throw all the data on that and now you can interoperate among all these other things so you can just like abstract out each of the services that are needed to make this ethereum the universal verifiable internet and just like funnel more and more into it so that's our our kind of view so it is true in that sense that rollups are basically like these web servers are parallelized and like individual nodes running whereas there's a group of horizontal services that are across all these web servers that help do things that are common to all of these things yeah um sure that was I love that that was a great uh helpful

for me for actually just like a mental model to think about how to view the ABS first Ro I know we've kept you for a long time I want to end on on this question here like already got the uh the the subtext or some of the back stuff that's going on in the background of this entire conversation and the success of reaking is a debate around issuance on ethereum and this has primarily been focused within the liquid staking ecosystem there was a phenomenal podcast I've referenced it many times already of on Uncommon Core John and hu and Casper and ancar uh where they outline basically there is a proposal in the works or early stage proposal maybe it's going to happen maybe it's not to Tinker with ethereum's issuance and the way that the issuance works right now is that there like there's kind of a curve

that looks like this where the fewer stakers there are the the higher the rewards uh like the more issuance goes but as we get more and more stakers on the network the rewards tail off after a period of time and there's a proposal out there to actually Target a certain percentage of stake wherein the rewards would Trail off to such a degree that they would go to zero or actually negative uh to basically Target a certain percentage of stake for the network and there are a couple of different reasons for this a rationale and different camps and one of the rationals that hey you could really view stakers as a external service providers to the network maybe the the amount of security that ethereum the protocol needs right is maybe we think it's just as safe as 25% as it is at 60% so if we

have 60% we're overpaying right and we should redirect that elsewhere namely to the holders of ethy asset the other thing is if you can have something like Ste which is even just as liquid as ethereum itself and it Bears yield then there's like a preference that many people will want to hold steth instead of eth it gets yield it kind of earns some of the money and maybe we're preferencing the LST and forcing users into taking these trade-offs security trade-offs that they don't really understand instead of e the asset right now one of the other interesting points around this is what does this do to the validator set of ethereum right because if you uh Target like set a certain Target then uh the argument goes that you're actually only allowing the

largest most economies of scale profitable valid to operate which would basically be Lio and so there's an argument um that I I think hosu has made uh convincingly that you actually are limiting the amount of solo you're you're making it much tougher for solo stakers and I would just be curious I haven't your you really comment on where you see reaking belonging in this issuance debate so I'd love to get your perspective on that I would also love to get your perspective on where you see how you see reaking impacting the validator set of ethereum that was a lot but pause there okay there's a lot of like you know prolific and like well thought out people who have spoken on this so we've kind of kept our mouth shut for the most most

part but here is a kind of like I I don't want to wear my Ian layer hat to answer this I'll wear my you know Professor or researcher had to answer this question um I think the starting point of the problem is how much security is enough and what are we paying for what are we paying for what are we getting for it this is number one like is 25% enough is 50% enough what is this thing like it's a fundamental problem that is unclear in the current setting all human made cures and we're going to fight

about all human made CS day and night mhm number one number two ethereum as a protocol has needs that it does not know how to enforce okay controversial but I'll say it ethereum as prot as a protocol has needs it wants censorship resistance and to get censorship resistance we need right now the only way you get sensitive resistance in ethereum is that you need a decentralized validator set and ethereum doesn't protocol doesn't know what a decentralized validator set is how to incentivize it how to give more rewards to it so it wants a property it wants to guarantee a property that it doesn't

know as a protocol how to guarantee I narrow down the whole debate to these two problems fundamental problems number one how much security is needed number two you want properties that you have no way to enforce as a protocol so my instinct is let's go solve these two problems not saying whether like changing issuance good or bad in the short term it's up to the ethereum folks there plenty of brilliant people to think about it but my questions are just these two what do we need to do to like make the amount of security a market driven variable rather than like you know a researcher protocol researcher drawing a curve number one number two what do we need to do to make sure

censorship resistance is not a property that we get out of social badgering instead censor resistance is a property that a protocol knows how to get directly that's the two questions that I'm most interested then I have written an answer for the first one that's called stake sure which is try to solicit the amount of security that you want in protocol and then send it to it but by itself it's not sufficient the reason is even if you applied Stak share and let's say the answer is we need 1010 billion of State security in ethereum that $10 billion will either be coinbase or Lio that's what it will be and like we'll be un unhappy at that because oh you know these guys are whatever I think

it's just like bad principle for how that that is done because it's coming down to the second problem which is how do we enforce sensitive resistance without resorting to social badgering okay the the answer to the second one at least the starting point is I gave a talk at the censorship. WTF Workshop called rever rier from the American hero who kind of like drove on a horse to to intimate of the coming British troops and it's kind of like hey censorship is happening how do we intimate how do we make it observable to people that censorship's actually happening and so we need a lot more mechanisms but it's a

starting point but my dream ethereum would solve these two problems at the foundational level and get rid of these issu and discussions on once and for all because you're kind of like trying to socially engineer some properties that you can't engineer in the protocol level and I just don't like it I don't like it we have to negotiate with Lio and coinbase and you know me and like I I don't want to negotiate with the protocol it's a protocol just tell me what the protocol is yeah I uh completely completely agree with it's it's funny it's just like regular monetary policy it's you're it's off like a proxy action and it's kind of like taking a um a uh a hacksaw instead

of a scalpel or what I'm chainsaw to a butter knife that type of thing um whatever pick your analogy but uh and I guess uh like final question there I know we're we're really over time here but just do you see reaking impacting the like the ethereum validator set as like as well does this this actually a tool for solo stakers so that they can earn more and it's more attractive to be solo atome stakers do you see it putting some kind of impetus on solo stakers overall does this favor large stakers versus small like H how how if at all do you see the adoption of reaking impacting the ethereum validator so is TBD and I think the way to kind of think about it is what have we done to like minimize impact on the validator

set so the first one is we built IG da when we built reaking this is not it's not at all accidental there are many many reasons one of the reasons is if you build a system of Shar security where everybody has to run everything and the only way you can scale is by everybody running everything then that system has massive C ization pressure so any other reaking protocol which is not built with like a scalable da at its base layer has this big problem which is that you know you're putting a lot of pressure for everybody to run everything and if everybody runs everything then you putting centralization pressures that's number one so trying to make sure

that there's at least like a horizontally scaling substrate on top of which you can you can throw more data everything is I don't need to have everybody download everything that's number one number two uh preference expressivity for abss to specify whether they want more decentralized versus more centralized validat sets if I only care about how much e is staked then do I really care if it comes from a home Staker or not maybe not but if I care about like like slot shares then I want of course like anybody who has a lot that to participate but separately if I care about sending secrets to various nodes

and to gather there storing a secret I better find a pretty decentralized Valor set so there are different use cases some use cases going to have be very happy with like a lot of e some use cases are going to be like just give me the same as ethereum and other use cases are going to be like I want even more decentralized than ethereum because I don't want 30 notes to control it and I want 3,000 to control all these three are going to exist what is the relative Revenue share users preferences we don't know the thing that we the other thing that we did with Native reaking is to give the ability for you to be a home Staker for ethereum but to Delegate for restating it's an optionality right like

I'm a home Staker for ethereum but I can I don't care just let like con based on it's okay so that's the sets of things we've done you know trying to make the protocol as like positive sum to ethereum as possible awesome tram you have been extremely generous for your time um miles and I still had a bunch of questions we didn't even get to which is crazy even at the end of this but uh yeah M we just really appreciate you coming on and it's always just a treat to be able to get to talk obviously at this point folks will be very familiar that you are the uh founder and CEO D layer but if folks want to find out more about the work that you're doing or if you're listening to this and you're

interested in becoming an a ABS or somehow getting involved with the IG layer ecosystem what's the best way for people to do that yeah um the um reach out to like DM me at like ataman is my Twitter uh handle um I'll connect you to the right people in the team to like you know work on ABS is there's also a Research Forum research. axyz where if you have ideas problem s or like hey this is a thing I don't know how I can it solving or like here's a proposal for her to solve it or here's something that that can work well with I super happy to engage in those discussions please post that awesome well Sham this was a blast man thank you

so much for coming on we will have to do it again sometime soon thank you so much thanks Mike thanks miles really really enjoyed this chat one of the absolute best yeah that was awesome thank you