Eigenlayer's Blueprint for a Decentralized Trust Network

Transcript

imagine you go and put your money into a landing market and you're taking a 100x margin pos position now if the market price moves by 1% you will lose your capital and you have no control on the market price so you're taking a completely exogenous outside your control risk with i l when you stake and you opt in to promise to validate 100 Services you're taking an endogeneous risk which means you know that if you are not malicious you are not going to lose your eat it's in your control to run your operator and stake [Music] correctly hello and welcome to the

protocol podcast I'm Brad C here with my co-hosts Margo nykerk and Sam Kessler we're excited to dive into Today's Show with the latest news and developments in technology behind crypto and blockchains first please do not forget to subscribe sub cribe to our Weekly Newsletter the protocol on coindesk.com and let's dive right into it we've got a pretty uh pretty hot guest uh in the crypto world uh shiram Kanan uh the founder of Igan layer we were just looking earlier their tvl uh is now 12 billion I think you're second after Lio now uh and not even on mainnet um which is one of the things we want to talk to you about and shiram you know we've done a lot of research and

you know obviously been there's been a ton of reporting and you've been on a ton of podcasts about I ler um but you know real quick you know if you could just give sort of the elevator pitch you for people who might not know a lot about igen layer but want to know in probably the simplest terms but also the briefest of terms if you can just yeah absolutely I think the way to think about ID is you want what is you know and a solution is considered a blockchain or a crypto solution if it incorporates decentralized trust and decentralized Trust emerges from what there's a distributed group of notes and then they take a stake you know which is

some asset of value like e lock it in a contract and make a promise that they will run with decentralized network correctly this is the core of decentralized trust in any blockchain like AIA now what we do what I does is to enable you to offer decentralized trust create a general purpose Marketplace for decentralized trust so anybody who wants to build a new protocol incorporating decent trust let's say I want to build an oracle or a bridge or an AI protocol or a database or uh a multi party computation protocol anything now what I do is instead of trying to go and start a new decentralized Plus network for each one

of these protocols there is a common decentralized trust Network on which is a group of operators who stake in ethereum uh and a common economic stake you know Brad mention 122 billion so this basically forms the root of trust so now when I build a new service I can just directly go in and uh you know consume the the trust from these nodes so I don't have to build my every new idea doesn't need a new resized Plus network so that's the vision of I another way of putting it is if you are building a web application this is by analogy if you're building a web application in

1995 what you would do is you would have to build your own server you would have to build your own payments uh platform you would have to build your own identity login platform you have to build your own database and then you're building whatever application You're Building let's say you're building an Amazon bookstore you to build all these other things to just build the bookstore that you want in 20 24 that's absolutely not the case you would go to Amazon web services as a cloud provider you would use us like a payment service like stripe you would use a identity service like o you would use a database service like Mong be and maybe even you have a rapper for creating your e-commerce website like Shopify and then you just build and deploy your thing on top

really really easy now you know millions of people build and deploy e-commerce websites this took like 20 30 years of work in the web application space what we want to do is bring the same amount of you know ease of use and Innovation into the crypto Space by providing a common framework of decentralized trust now anybody can create and build new infrastructure services and then if you are a consumer application you just consume these infrastructure Services you say I'm using this bridge this AI service that uh execution environment this data availability system and then now you bundle all of these together and then create an and user application so this is the vision that we willing so concretely the way that Igan layer works

and we've talked about this a lot on on our podcast and you've talked about about it a lot more on a lot of other podcasts is people can restake ether essentially that they've staked on the ethereum blockchain to secure it to secure other systems called actively validated systems abss and one of the critiques that you've gotten for over a year now since you first introduced this whole concept is um you know that this looks a lot like rehy hop rehypothecation um stuff that we've seen in traditional Finance I know you've talked about this but can you kind of lay out for our listeners you know your viewpoint on that critique this is you know it's not new even inside crypto this is called Shad security like Cosmos has this thing called Cosmos subub where

you know anybody when you build a new chain can go and borrow security from the cosmos H this is not new even inside ethereum ethereum is a common framework and then now anybody can come and build a new dap on top of ethereum imagine you build like a new stable coin on top of ethereum wor hundreds of billions of dollars then are you overleveraging ethereum stake okay so there is a foundational thing that very few people actually seem to understand which is how staking and security works and so the first thing about staking and versus all kinds of other things is endogeneous risk endogeneous means you control the risk and the way you control the risk

versus re hypothecation or anything else that you know in in traditional markets you are not taking a price so imagine you go and put your money into a landing market and you're taking a 100x margin position now if the market price moves by 1% you will lose your capital and you have no control on the market price so you're taking a completely exogenous outside your control risk with i l when you stake and you opt in to promise to validate one 100 Services you're taking an endogeneous risk which means you know that if you are not malicious you're not going to lose your eat it's in your control to run your operator in stake correctly so this is a foundational

difference I think a simple example to like tether this is imagine there's a big mall and there's a central store in the mall and the mall says you know when you come into the mall you have to put up a $1,000 deposit so that you know you don't steal something from the mall and you know the main store has to deposit system and then the other stores are saying Oh but you know they could come and steal from my stores and then now like they all form a coalition and say that when you're put in $1,000 deposit you promise not only to steal from the main store but also from all the side stores and not stealing is in your control it is not an exogeneous thing it's not something that somebody else needs to do that's that's on the Staker side so the ster if they know what

they're opting into and make sure that they're running it correctly they're taking no additional risk on the Services side there is a potential to uh to over leverage and imagine there's $1 billion State and then there are 100 applications now each of them if they think oh I have $1 billion stake I have $1 billion stake and they all account for their own economics like that then you know it is chaos because now each of them think they have $1 billion dollar but they're actually only in total $1 billion but this is just a matter of accounting sharply and igir has the sharpest system of cryptoeconomic accounting we have what we call attributable security there is something called pool security which is instead of

you know to to take simple numbers instead of each of the protocol having $1 billion separately state if there was1 billion do common Stakes Tak across 100 protocols it's better for all the protocols why to attack any one protocol now you need 100 billion rather than needing one billion so this is a big kind of benefit this is called pool Security in addition to pool security security IG also offers attributable security attributable security means an EVS a service that's built on IG can come in and say out of this h100 billion that is totally kind of stti I want a particular attribution of $3 billion because you know on Bas I have a lot of transactions whatever so I'm going to say I need $3 billion now it's item

there as a protocol's job to make sure that even if all the AVS is simultaneously PR I will be able to redistribute you know the the amount three billion to this base are years so this is a kind of like a complex accounting kept in the idland L well so we just went over a lot there um you you hit a lot of the questions we were going to ask um but the the big thing um that you mentioned you you give this example of we'd rather have a hundred million securing 100 protocols rather than you know 100 protocols each individually secured by a million dollars that's the whole point of igen layer now there's two sides of that there's the benefit of it would now

cost the entire pooled amount to attack any one of those protocols but then there's um it gets to the you know the leverage things where there's also leverage risk risk in a sense where if one of those protocols is attacked there's the contagion risk that's kind of wiped across all of the protocols that are in that pool you mentioned attributable security which is the idea that you can kind of add an insurance fund to each of these abss so that they're each you know um you know attached to a a a a part of the pool but that's kind of where I'm actually confused like how do you make sure that you know if one protocol leads to a mass slashing event drains the entire igen layer pool leaving you know nothing

behind other than this Insurance Fund how do you make sure that the Insurance Fund is solvent um to to still support protocols yes so so what happens is you know there's a lot of details on this but I'll give a high level picture so when there's 10000 billion do stakes and that are these 100 avss any one AVS can only slash and redistribute their attributable security so if there's an EVS which has only5 billion attributable security they can only slash and redistribute .5 billion if no other avas ever trigger slashing and you know that1 billion worth of Staters have done and stuff P the period where other people haven't triggered the slashing you can

actually the remaining funds but you are not going to allow that ABS to take the remaining funds and then go and redistribute it so the high level idea is the benefit of pooling is coming from to attack any one service I still need a huge amount of capital to even enter the system become malicious and then go and attack it but if I do attack it that service can uniquely only slash and redist so you can think of like after you slash what happens something is redistributed and something is burnt right and so that's the two things that that can happen redistributed to the AVS so the attributable security is the amount redistributed to the AVS and if nobody else comes in and claims their portion of atable security by triggering

their slashing then yeah the remaining funds can be burned so you know when you're like you're getting simultaneously the benefit of pooling and the benefit of attribution so this is a really really powerful idea and you know normally any simple way to do this would either get you one or other but not both and that's really B like we have to take a lot of care this you know I'm wondering you're talking about this this sort of insurance for for these abss but like how do you figured out the details on this like what are the specifics behind this yeah so I think like calling it an insurance you know initially I was calling it that but I think it's not the right way because it's not when you're insuring something like let's say you're

insuring against like fires right uh like What's happen if there's an exogeneous agent the fire which is C in the the damage and you don't know you're not going to go and extract from the fire the damages and then pay back whereas ID layer is a system completely closed system and this is why I think when people talk about insurance there's this other idea of slashing insurance which is some other person is coming in and underwriting risks right that's not what's happening what's happening is there's a common pool of stake already like a stater is coming in the contract that they have with ID is I'm putting in my money and I know if I'm honest I will not be slashed right like this is the deal that a Staker has to

title have and if I'm malicious then all my money may be gone that is the contract you have with on the one side on the other side an ABS comes in and says that hey you know I want to use the pool security but I also want to make sure that if I if my AVS goes down I should be able to redistribute x amount of money to my har parties Okay now what's happening is out of this 10000 billion pun that is sitting inside i l now what what it would do is it would say oh this Ben slashing happens this three billion portion is reserved for this areas to come and Trigger you know a redistribution for so so the redistribution happening is a closed

system you're snatching from the bad guy and giving to the harmed party which is very different from exogeneous party coming and I'm coming in and I'm underwriting risk of some third parties like what a normal Insurance Market is and an insurance Market is usually statistical which means if all cars in the United States crash today Geo doesn't have money to cover it but that's not how IG layer Works IG layer is combinatorial and unconditionally correct security which means all the abss trigger slashing simultaneously I has a proof of solvency which makes sure that you can redistribute to each one of these simultaneously so there is no leverage no over pting no statistical uh like

Risk multiplexing there is no extrinsic party it's a close system of kerma if I can just try me that makes a lot of sense concretely how do you calculate how much you don't want to call it Insurance attributable security should go to a given ABS how much are they going to pay this is a market where you can come in and buy like let's think of like there are various stages this Market will evolve let's think of the simplest possible Market there is a fixed APR and you know you would come and say oh to acquired this insurance I have to pay a s you know this much is that how you're doing there's a mechanism we have not published publicly so we'll we'll we'll have that in the coming months and so I guess the question then becomes just timeline wise um where are you guys at um I know

you're kind of you've got the economic security which is the pool security stuff we're talking about then there's decentralized trust which we've talked about where are we on the Ian layer road map and when are we going to see more details around attributable security the slashing all of this stuff um I think a lot of people are wondering yeah no that's a great question so the right now we are uh we have a public test ring in this public test you do the following things you can stake you can opt into an AVS uh you you know on the on the AVS side you can build and deploy an EVS uh there's sters there's operators there's AVS all these three sides can

can build a deploy services on the um uh igen da which is the first ABS that we are building it's a data availability service that is also running on the system what is not live in the system is slashing so everything else is included so there's no slash ability in the first version which is sitting on Pam which is for many reasons for abundance of safety right like you don't want to have a flashing event you want to turn it on slowly uh it it is a complex system with attributions you know so we will have a phase road map by the next phase will have slashing um so all of that so that's on testet and that's what's going on mainnet in the coming month so we have

you know we published a post called the road to main net uh you know there's a phase Road mapap there we we going to bring in operators and then we're going to bring in IB and then we're going to bring in abses all of these things coming up on the mnet but the exclusion still is slashing so there's no slashing on this m let's just take come in you can St you can you can offer services and then the services are the avas can use these services to do other things but the core slash ability is not uh we need to just take a quick break here um we'll just go to the break uh and we'll be right back with Sham Conan calling all developers score a consensus 2024 developer pass for just

$19 but actually fast only a limited number of these passes are available you may have heard that consensus ain't for devs but here's why you're wrong consensus is the only place you can fully immerse yourself in a multi-chain environment and learn directly from 20 plus chains including arbitrum chain link salana and more enjoy 3 days of intensive learning with technical talks 40 plus expert speakers and 20 plus in-depth workshops including dedicated half days for ethereum and Bitcoin and three full days of programming on our protocol Village stage consensus 2024 is happening May 29th through 31st in Austin Texas don't miss your chance to

network at curated developer meetups find new career opportunities and explore hundreds of side events and hacker houses around town grab your $19 developer pass today but remember this exclusive offer is limited visit consensus. coindesk.com now to secure your developer pass before they're gone explore the epicenter of blockchain innovation at consensus [Music] 2024 and we're back with shiram Kanan um

so shiram you know I believe you've put out there that you're targeting main nut for early Q2 is that still on track that is on try how do you define early Q2 it's it's in a month from now like you know be April maybe latest May everything will be live on Main um the the set of services we're talking about I I think as as I was just mentioning you have uh stakers you know can delegate to operators operators can opt into avss avss other people can build and we have like 10 avss live on our testnet and many of them will be going to mainnet our own areas ID will

be on mainnet so a lot of the ecosystems kind are coming together the exclusion like I mention SL I have a question sham you know related to this Market is that also up and running on on testnet yeah so the the market means basically as a Staker I can choose which AVS to op yes you can but there somehow you know that and that's a market mechanism that that causes that rate to go up and down up or down yeah so this again there is some some details here each ads basically can announce how much they are paying and then like you look at it and then you look at the

risks that we are taking and then you say okay don't this make sense for me or it doesn't make sense for me and then you opt in or don't opt in they just fix their rates and then you decide yay or nay or there's like a make sense to you or not got it and and people will people be adjusting their rates or the these are emergent things we have to actually see when things go on in the main right so does it sound a little bit like the kind of like the corporate bond market in terms of like pricing you know credit risk maybe uh these these all have some dimensions in they're really good and some Dimensions that they're uh not very good but essentially like what you're

doing as a as an operator and a ster is you're looking at all the avss and seeing what am I getting for opting into this AVS I'm getting some on payments what am I putting at risk okay what am I contributing I'm actually contributing some amount of compute power in the normal more right so that's that's something that you're giving to the network the second one is you're taking amount of in liquidity on your token so if you want to go sell your eat or whatever like you don't you not going to be able to do it instantly because you're committing it for some amount of time and then the third one is you taking a risk of slashing so you you know because that ABS may have may

trigger slashing for you and if the system works correctly if you audited the smart contract you should never be slashed for if you are not malicious but you know you TR trusting the contracts to actually work correctly like when you're putting your eat into a or something else you're trusting that the smart contract work I think that's super interesting sh Ram is just you know and maybe we'll move to the next question but how you know how this is so also completely brand new right like people are gonna have to figure out how to trade all this stuff and price all that stuff and you know and a lot of it might be end up being mispriced right I mean absolutely possible yeah so you know there's the initial price Discovery is

going to take some time right it is not uh it is not at all obvious that on day one you'll have the right pricing on on the system it may be overpriced it may be underpriced people may say I need a 20% a or they may say like actually I'm already Stak and I only need like a 0.1% so there a huge amount of variance here and we don't know what what is going to actually emerge so that's why we are we are taking an Adaptive approach here which is lot it see what people do learn from it and then update the market structure so we're not submitting to a very rigid Market structure to begin I want to Circle back real quick because you know you brought up Igan da as one of the you know first things that are going to go live and you know we've

spoken a lot about da in the past Celestia had its main that launch few months ago and Avil is also coming out with their own one but I'm wondering specifically because you know we are talking about igen layer and about reaking and about your technology will IG da like use poed security yeah yeah I da will use poed security so the the the way we think about uh you know how I Works versus all the other systems is they are asking the question so other blockchains like you know Celestia for example asked the question what is a minimal layer one what is a minimal

layer what is the minimal set of functionalities for layer one that maximizes other people to come and build new things on top and you know just have a core data availability system and a consensus protocol and and a new network and then now other people can build other things on not so that's the so it's an answer to the question what is the minimum minimalist minimalistic blockchain the question that we ask is very different the question that we asked is how do we scale it right it's a narrower question in the sense and and also it is rooted in ethereum so the question is oh you know you have ethereum ethereum already has a consensus protocol it has validators it

has you know uh all this other infrastructure but it doesn't have a high bandet of data available like how much data you can write to eum is very small it's you know 30 kilobytes per second with the reccent upgrade 4844 that bent live like one or two weeks time you know the the crazy thing is even 30 kilobytes per second is not getting filled right now this is because lot of rollups were hyper optimized for even more congestion but now we have 30 kiloby per second what IG da is doing is launching at 10 megabytes per second so the architecture everything of I da is centered around this question how do we augment ethereum to scale and

so that means how do I reuse the etherum stake how do I reuse and that answers your question about like tool security right how do I reuse the ethereum validator set how do I reuse the ethereum's consensus we don't have I doesn't have a consensus protocol ethereum already has a consensus protocol so why should I build another consensus protocol it just missing in data bandwidth so I just we just have a system of you know nodes where each node only downloads a small portion of the data but together they download and guarantee the availability of all the portions of the data so it is an shed system just to clarify one thing you're you're going to use poed security but slashing doesn't exist yet the attribut

security hasn't been defined yet IG da is going to launch so concretely it's G you've talked about decentralized trust in the next couple of months when IG da launches what is that model is it just going to be a bunch of validators are they going to be economically incentivized to be honest or is it going to be this kind of training wheels like we've seen in rollups like it's a totally normal thing but I think people are confused about that yeah there is definitely like a training wheels SP for a where you know even though the full slashing is not live you are we are going to run this system in you know whatever beta and make sure that people are able to use it you know maybe have redundant you know backups and other stuff right just like right now you have

single centralized sequencers you know you can use the system uh with this this model without you know slash ability of e but as slil of he comes in I also becomes you know economically secure so that's that's a road map for I gotcha so and do we know when we're going to make it we we still don't know exactly when we're going to get the pulled security for Ian da though that's on the road map but it's not there quite yet yeah I think that this is also going to launch in faces like you know there is the day of launch and then there is you know we turn on like incentivization and then we turn on like pool security and then we turn on the so we going to see all these

phes and at any phase we should be very clear what is actually running in production then absolutely and then like the um another thing I wanted to ask about as we're closing up is just to kind of go back to to the beginning of igen layer um you've decided to kind of Base things on ethereum ultimately and on The Ether token it makes complete sense it's got a lot of value relative to what all these other ecosystems would be able to prop up on their own if they're trying to create proof of stake but I know like at the very beginning you guys considered a bunch of ecosystems and I I I actually saw a deck too your initial idea was like cardano so can you kind of explain what your kind of history is what's your path from cardano to ethereum initially we had

this idea that now I can use SK to run like other networks and our initial idea was I can do virtualization virtualization means I can run a different execution engine on a different chain so this was in 2021 and when we looked at the ious things so initially we were thinking I L will be a multi-ecosystem like project like oh you you you can go and bu it on every ecosystem and we looked around and asked like where is virtualization going to be most useful and virtualization is most useful when you cannot do something on the layer one by itself but actually there's a lot of economic value and particularly we were thinking about evm

virtualization I want to run evm on in ch and okay if you want to run evm on any chain which chain is most interesting and we said we narrowed down to a few like maybe it's salana maybe it's near maybe it's K and so we were going to build evm virtualization on all of these and we initially actually have a version working Ono because it was the hardest smart contract system to build on like you know the harder it is to build on the base one the more people going to build on the virtualized one so that's that's how you it didn't contract at that point right that tends to make it pretty difficult yes but but you know that was not the reason we didn't uh continue to

build on it I think we realize that the scope of what I L is is much bigger than virtualization you can buildt pretty much any technical feature on top of a common framework and so what that meant is we we fix the Sing Source and then now you can add the features from every ecosystem into common place so that we can interoperate you know it is in the same place we don't have to go and acquire state from 30 chains like there's a single chain single source and now we bring all the Technologies into one place so they interoperate they become kind of this like new cloud like structure so that's that was an evolution path and I think you know I

was at e Denver and I really resonated this 2022 and that's the first time I interacted with the ethereum community say I interacted with ether researchers for a long time I found the Rance with the values and like you know that there is the corale value of transformation inovation credible neutrality all of he ass with decree than fre Rob uh that we're gonna wrap it there thank you so much I mean just the I I I heard you talking about the comparison with Bitcoin merg mining and how you sort of pivoted off of that towards proof of stake which was super interesting well that's it for uh this week's episode uh Shri Ram thank you so much uh for coming on and answering all of our questions thank you for listening

to the protocol podcast if you have any questions about any stories or comments please reach out to us at podcast coindesk.com subject lineth protocol you can listen to US Weekly on coindesk podcast Network or wherever you get your podcasts also please subscribe to our Weekly Newsletter the protocol on coindesk.com thank you to our producer Michelle muso and uh and Mel monz our Booker uh CN next week