Epicenter - Learn about Crypto, Blockchain, Ethereum, Bitcoin and Distributed Technologies - Elias Simos: Rated Network - Reputation for Machines = Transparent Blockchain Infrastructure

Episode Date: September 28, 2023

Data immutability and transparency are key features that define a blockchain’s public ledger. However, while transactions are indeed transparent, information surrounding the blockchain’s infrastructure layer is not readily available. Since Ethereum’s Merge to PoS and the introduction of staking delegation, actionable data on the validator and client status quos became crucial for the wellbeing of the network. Rated Network aims to build a reliable ‘reputation system for machines’ forming the infrastructure layer, through better data curation and improved transparency.We were joined by Elias Simos, founder of Rated Network, to discuss data interpretation in blockchain infra and what conclusions can be drawn about validator performance, client diversity & much more.Topics covered in this episode:Elias’ background and founding Rated NetworkProviding transparency in blockchain infrastructure layerTapping into Ethereum’s data. The subjectivity of performanceThe consumer profile of Rated dataEthereum’s credible neutrality & client diversityEIP-7514 & managing Ethereum’s state bloatGovernance: dev council vs. on-chainExpanding Rated to other PoS networksEpisode links:Elias Simos on TwitterRated Network on TwitterThis episode is hosted by Felix Lutsch. Show notes and listening options: epicenter.tv/515

Welcome to Epicenter, the show which talks about the technologies, projects, and people driving decentralization and the blockchain revolution. I'm Felix Lutch, and today I'm speaking with Elias Simos, who is a co-founder of Rated Network. Rated is providing node operator metrics and ratings for the Ethereum network. Hi, Elias, and welcome to Epicenter. Thank you, Felix. Thanks for having me. Huge fan of the show since I joined the industry. So very, very glad to be with you today. Awesome. Yeah, yeah. I'm also really excited. to have you. We have a long history of like working together in in the staking space and yeah,

I've been really interesting to follow your path and now seeing you build your own project with rated so which is where we're here to talk about today. But yeah, let's start with the classic basics of, you know, how did you get into crypto and ended up where you are today. Cool. So first touch with crypto started in 2013. I used to live with a really good friend of mine. He found out about Bitcoin and he started talking about it nonstop. I started building stuff, talked about it even more. Initially, I thought he was kind of nuts and didn't really get it. But then the more we talked about it, the more I got it, but didn't really pay attention too much. when it really clicked for me was in 2015. I'm Greek originally. And so in 2015 was the worst part of the decade-long crisis effectively that Greece was in. And in 2015, we had capital controls come in.

Huge referendum. Should we stay in the European Union? Should we break away? That means also like leaving the monetary union, issuing our own currency. And so capital controls was this. really gut-wrenching periods, if you will, for Greek society at large. It's like crippled the economy, all the young people left, but like there are there are really visceral images that I still have sort of in my mind of very long lines of pensioners around each ATM that you see like on the

on the street driving around talking about, you know, 50 people, 100 people, blocks like whole blocks worth of lines, waiting to get their weekly ration of money. And so at that point, like I had the sort of the light bulb moment regarding Bitcoin. I was like, okay, I get it now. Non-state money, you're not beholden to this idea of, you know, institutions and the way institutions work in the modern financial system. And I really found that appealing. So then started researching more, but again, not being very involved. All sort of came together in 2017 with Ethereum for me.

And this whole idea of applications that you're able to build on a platform that has like the properties of Bitcoin, but then can extend this logic sort of arbitrarily, right? like the vision of the world computer and so on. So spent the whole year just researching stuff, trading, trying to build things with friends. But by the end of it, I looked back and I was like, well, you're having like so much fun. And you resonate with like the whole mission of self-sovereignty and just building something better than kind of the alternatives, which is kind of what is the status quo. And I decided to commit myself full time to the space. So I got a job with a fund called the Central Park Capital.

They were just starting out back then. I was the first hire that they made as an analyst. I stayed with them for three years, made a bunch of investments, built a pretty expansive data platform while at the fund when, you know, Dune did it exist. Token Analyst was like one of the earlier data companies that were looking at blockchain analytics specifically and help them raise a $75 million fund too

and then I left and I joined the startup called Bison Trails which at the time that I joined was I think I was employee number 20 or so I was a protocol specialist there I think it was the second ever person to be called a protocol specialist in the industry although I know you

you have been doing like very similar work in your in your history in the in the space so the first was victor was my colleague who hired me in basically and buy some drills as he was validators as a service right that's what we were building we ended up building a pretty large platform i think at the top of the market it was you know north of 30 billion dollars on on platform a year later we're acquired by by coinbase and then i stayed there for for another year before i branched out on on my own found rated with my co-founder ours, but I'm super happy to talk to you about, you know, the internals of the story there. But I want to let you ask whatever questions. Thank you. Thank you for that background. It's really interesting. See you like witnessing, I guess,

first hand in Greece and how it meander to where you are now, like seeing blocks of pensioners and now you're seeing blocks on the Ethereum chain being full, hopefully. So yeah, I guess, you know, what stands out to me is like your, you've been always like in this sort of space around data and obviously that's sort of what rated is focused on. So maybe just to start, can you explain to us, you know, what is rated and sort of what are the products you are building? Sure. So the whole thing that we're building, we fit it under a one liner, which is reputation for machines. And this is like a really charged term. You can fit like a lot of scope in it.

But really like the mission of what we're doing is providing transparency into the infrastructure layer of blockchain. And we've started with Ethereum. Now, where this is coming from, like the why are we doing this, ties actually in pretty well with, you know, 2015 and those lines of pensioners and so on. I'm here. I'm doing what I'm doing. I got involved in the space because I really do think that we have sort of an opportunity to build something better, something more compelling, something more transparent by default. but also transparency is not handed to you, right? It's like, sure, the source material is open. Anybody could theoretically just go and access the data,

but without the interpretation layer, you're really not improving much, right? And the beauty of blockchains is that actually they allow you to go and get the data, but obviously, you know, interpretation doesn't come out of the box. So on a long enough time frame, I think we're working on systems and we're building systems that are by default a lot better than what the financial system, for example, is running on. And I also do think that it is a matter of time until things migrate to the systems that we're working on. Now, how much time that is going to be? That's another story. It could be 10 years. It could be 20 years. but I think the fundamental properties of the things that we're working on are undeniably

orders of magnitude better than the alternatives. So if you then take that to be true, then I also don't want to imagine that world where we don't have transparency and visibility into kind of the layer that guarantees it all, the layer that actually packages transactions, intense, whatever that is and transitions it, you know, from a want to be done state to a done state, there is a protocol, there are rules to be followed, but also like rules can be broken. And actors that operate the protocol and the base layer could act selfishly. And acting selfishly means that you impose a negative externality on everybody else that is not only everybody else that's around you and that's honest,

and that's acting as they're supposed to be acting or expected to be acting, but also everybody that transacts on the top layer, which are really like the most important part of the whole equation. So that's what we're largely here to do. That's the mission. This is why reputation for machines, right? And how you get to reputation is by indexing and contextualizing and contextualizing and building and dispelling the subjectivity of what is necessarily good and what is bad,

but also what is, right? Just merely providing the context of what is happening, what has happened, how what is happening fits in the context of what has happened. And is this behavior like expected or not? Or what even constitutes a behavior. That's all the work that we're doing with data. So it's at the moment we're operating on Ethereum. We started with Ethereum because it is, I think, the most consequential piece of infrastructure on blockchains out there by users, by developers, by assets that are powering the infrastructure and so on. And we currently host a network explorer where the index is not the block. It's actually the operator. So we're, uh, basically contextualizing how Ethereum infrastructure works,

and then you can be as granular as the validator index, but then we're providing sort of abstractions in terms of, you know, node operators and different pools and how these pools are composed of node operators, and then all the way up to sort of the global network state, with the ability to actually, like, zoom in down to like the most granular unit of account, which I suppose is the validator. index at this stage. We also have an API which basically serves the data that you can sort of

see on the Explorer, but also way, way more to build interfaces to power financial products that serve the infrastructure layer. And we also recently released an Oracle, which is a gateway for us to bring the contextualizations that we curate on chain. And actually, be able to power a suite of products on the Ethereum mainit execution layer. Yeah, that's super interesting. I guess most people, the block explorer, is essentially like the portal into the crypto space, right? Like many people's interactions with like when they first use crypto, I mean, aside from

the wallet, it's probably like looking at ether scan and looking at their transaction. it's quite powerful because like yeah like you said right and initially everything is open theoretically accessible but like how do you actually do it in practice so um i guess ether scan was like kind of the first wave there to like really um do it on the transaction level and maybe did the application layer and from what i understand you are very focused on on this infrastructure layer for the proof of stake like chain how does it work on the on the very bottom layer and And is that going back on your like sort of experience in bison trails or, yeah, could we say that? Is that how you?

Totally. So like the origins of rated go all the way back to, I guess, October 2020. So I had joined bison. It was like six months in or maybe a little bit less. And then the beacon chain was launching at the end of the year. and the Ethereum Foundation put together a hackathon and the brief was do anything with all the data that the Medasha test net produced

and that's the last test net before the beacon chain infamous for like a prism bug with like an update that they pushed that had a cloud flare clock to run side by side with like the validator's own like sense of of time and then it just like wiped out, I think one third of the of the networks validators at that point. Great that it happened in a test net. And I guess that's that's what testes are for. So myself and my friend Sid Shaker, he was back then, I think the CTO of token

analyst and they had just been acquired by Coinbase, if I remember correctly. So we sat down and we basically went on a two-week sprint where we just really took all the data that the network produced, and we did like a very expansive report on anything that happened from like, you know, client sinking times, like even off-chain data to storage, to how kind of the clients and the nodes behaved over, over like a span of like two weeks, which is basically like when we run the research to slicing and dicing and diving super, super deep in all the on-chain data. that we got.

So you can, I think you can still find it on eth2 data. Dot GitHub.io. It was still called ETH2 back then. So we did this hackathon. We won silver prize, I think.

And we learned a ton from that. We learned that there's like so much wealth of data about the network. And there is so little definition of actually how you, can make this data useful. We learned things like, you know, performance, externalities, risk, but like primarily performance and, you know, the risk and externalities thing was a, was sort of a

realization that came a little bit later. But it's very subjective, right? Like people were measuring no common language on what performance actually means. So is it uptime, rewards, is it something else. No, no clear answers. And then kind of over time, as I kind of went about my work at Bison Trails and, you know, we were operating on north of 30 networks, I think 40, 45, 50, maybe more. I kind of saw this issue everywhere around me. Like, it was a very narrow slice of value in the world that I was looking at, but as kind of someone operating in that, industry, I didn't know how all we're doing altogether. I didn't know how we're doing versus our network neighbors or competition or whatever you want to call that. I didn't know how

the network is doing. I didn't even know how to articulate or, I mean, at least I had like some sense, but in general, it was like, again, pretty hard to even articulate what it means to do well and what it means to not do well and compare it against what. And then I started thinking, well, if this thing is going to be super successful and run incredibly consequential things on top of it, 1% of the world's transactions and finance, commerce, media, I don't know, you name it. That cannot be the case, right? It cannot run on something that is amorphous and not very well articulated, right? And then at the same time, you know, you know it as well as anyone. Proof of stake grew from, what, some, I don't know, hundreds of millions of assets staked

to the height of the market at some, like, over $300 billion at stake, which is like a massive number in the span of like, you know, three and a half years or so, and a massive increase. So you're like, there is so much at stake quite literally and so little. data about what is at stake and so, so little understanding. And then also like extrapolating forward, then you're like, well, you know, that, that stake element and nodes in general are important. They are important because they help these networks run. If these networks are going to be valuable and they are to different degrees, then security at that layer is important. The well-functioning of a network is important. How do we, but also like the assets, the

themselves, the nodes are valuable because they do jobs and they produce future cash flows. They're like, you know, it's inflation rewards, it's transactions. It's like all kinds of things. So we can actually like then you think you can actually build products out of that. And in order to like help the industry move forward, you need credit for those things. You need insurance for those things. You need like, you know, potentially even derivatives, right? You can do lots of cool things to hopefully not gamble,

but provide levers for people to operate businesses in that space and actually contribute to running infrastructure for really, really, really important constructions, right? And then I started thinking, well, you need like a third-party independent guardian of all that data to actually help power all of these products. that you foresee in the future. Because, you know, accredited underwriters, for example, job is not to wrangle blockchain data. Their job is to underwrite credit and do that type of type work, right?

Equally with insurance. So this is, this was kind of like the initial impetus for, for starting rated. I was also like relatively good with, with data historically. like I built this data platform at the fund when not much existed out there through that thing I ended up being like one of Dunes earliest users before starting until

I started the company I think it was number three on the you've been dropping now on the leaderboard but then I got like I got eaten alive you know it's like you got to run just to stay still and you know go to work closely with with Frederick and maths there as like one of their earliest users feedback, blah, la, la, all that stuff.

So it felt like a natural sort of extension of my work, putting infrastructure and data together and seeing what we can build. So I guess here we are. Yeah, it's super interesting because, yeah, like from experience, many operators, like, sort of struggle maybe. So you have like these different users, I guess, right? like the operator itself wants to understand better how they're doing, maybe serve their clients, data on their rewards,

and like might end up trying to build some of these systems themselves. But then, you know, you know all the other operators have that same problem. And you're not the press, like, project, obviously, like, because you might not want to share that data. You just want to do it for yourself. So like, obviously there is also a need for a third party. And then I think, so you basically have the side of the, like, if you think about the users,

for rated right one is like the stakers or the operators or can you like sort of talk about it in those layers you know who is the consumer of rated data um and what are they using it for definitely i'm going to talk to you first about kind of what what i see in like very very high level and then and then go down to like the nitty greedy presently who's using us how they're using us and so on. So I sort of see us in the middle of a possible network, right, of networks themselves, node operators that run these networks, capital, and applications. And there are, like, if you think about things today, without rated in the picture, these are all connected, right? these are all sort of nodes in a broader ecosystem that are connected with one another.

And what they sort of exchange between one another is distribution, legitimacy, and data, right? And we can sort of exist in the middle of that. We can basically help facilitate a bunch of implicit value transfers, implicit in the broader sense, and make them explicit, help them actually materialize easier, more systematic, and thereby, hopefully, just inviting more of that. Now, all the way back to where we are today and the products that we have on Ethereum, we can and we do serve node operators,

and the use cases there are anything from, you know, rewards accounting to performance monitoring or just being sort of a second source of truth to your internal monitoring, for example, to performance benchmarking, and understanding how well you're doing versus your peers and so on. And I think increasingly more so, there's like this idea that we have of just becoming a data co-op effectively, maybe explicit, maybe implicit, but, you know, a shared cost center

effectively for node operators where your job is not to wrangle data necessarily. But also, like, you know, there are node operators out there that have these capabilities, But it's sort of awkward to think that, you know, now they're actually powering another node operators, reward accounting or performance benchmarking, or even like the credit products and the accounting products and the insurance products and so on. Because there is like a large surface for moral hazards there. And if that sort of actually transpires, then we're no better than sort of the mistakes that that have been made in previous iterations of the matrix, right?

The other part of our user base is pools. And more broadly, it's capital allocators, right? Because in proof stake, as you know, stake is one part of the equation. So how does capital get allocated into these node operators? So we're working with several pools like Lido, Liquid Collective, we're working with Stater. and the things that we're helping there is with the tools that we've built to better manage the active set

both in terms of deciding who is onboarded onto the active set and who might be offboarded one day. I don't think we've seen offboarding from active sets yet, but it's natural to think as that this industry matures, these things are not going to stay static, right? I mean, specifically in Ethereum, you couldn't even withdraw up until April 2023, So for the first two and a half years of the beacon chains existence.

So even just evicting someone from the set was really like not possible. But I'm sure that eventually we'll see those things and then, you know, compositions of active sets changing. As a manager, as a custodian, as a steward, as a whatever you want to call it, this could be a Dow, this could be a company, it could be, it could be an interface, whatever that is. you have to make decisions about the well-functioning of your set. And so understanding what that well-functioning is and making decisions is all powered by data. And this is what we're helping and seek to do more of these pools with. Thirdly, you have applications.

These are applications that basically reference the infrastructure layer to create new value, right? And that could be credit, for example. So think in terms of, we haven't really announced it yet, but we're working in partnership with a credit fund to actually trial out uncollateralized credit for not operators based on their accounts receivable. So you think of it as pipe for validators. There's a predictable stream of rewards and transaction fees that they earn, and thereby you could actually monetize that revenue,

add a premium to sort of the interest rate that you earn on the beacon chain. There are interfaces that were powering with our API to help capital that sits on the other end, make better decisions along the lines of, okay, where am I allocating my capital to which operators? Am I putting my capital to monitoring an ongoing basis, revisiting that decision and so on. And then you'd think that, you know, oh, yeah, just like, I just want some APR, give me the maximum APR, and then, you know, see you in a year or whatnot. Sure, that's how things really started, I think. But then the more the industry grows, the more the industry matures,

the more other factors are going to come into play, right? It's like risk versus reward. Then you have things like eigenlayer. You have more risk. You have like, you know, risk starting to sandwich on top of like preexisting risk and so on. So without like a well contextualized data world. in that paradigm, you can't even articulate those things. And I like to think that we're playing like a virtuous role in the ecosystem by helping

people actually like articulate those things. So far, it's been it's been a pretty exciting journey to get here. That's definitely the case for me. I think I also like, I mean, in the end, you see a lot of times the rated data, the aggregate data being used in like these sort of discussions around. decentralization maybe right like now we talked a lot about contextualization maybe mostly on in terms of like performance and you know is the operator doing well there's like these objective things but there's also the wider goal of like I guess credible neutrality of the

network and maybe geographic diversification yes which you are also looking at that like the end user being more the the wider network right which I think is like a very very nice outcome, right? You're kind of getting maybe money from the people that want to, like, optimize performance, but also providing this, this public good of, you know, what is the state of decentralization in the network. And I think, you know, there's like really, really interesting things to see there. If you'd visit graded dot network, which we'll also link in the show notes, you'll be able to see that for Ethereum, you know, like all these, uh, things that, that you try to keep Ethereum neutral or like decentralized, like, you can like actually

check. Is it actually happening? So maybe thinking about that, maybe can you like, yeah, talk a little bit about how you think about it and how you see the current state of Ethereum in terms of like its goal of becoming credibly neutral. Is it actually happening like from the data you're seeing or where are maybe like things that need to like be worked on? Since I guess you were like probably one of the people that has like the deepest insight there, I think that would be nice to.

focus on that dimension. I think Ethereum is on a really great path. It's obviously messy and hard and sometimes undefined in terms of like where you're at and how you're

tracking in terms of your goals. But I think undeniably it's like on the on that right path. Right. And I'd like to also think that we're helping just push that conversation forward. Like I still remember when we were starting out, when we first launched, I think we first launched the website in February 2022,

then March, April, around that like two months period, it was like a big conversation about client diverse. So I think Prism at the time had 80% dominance or something of that nature. And we had just launched. The first feature that we delivered on the front end, which is based off of Michael Sprow's work on block print, which is basically an open source tool that fingerprints, the proposal patterns of validators and works backwards

to identify which client they're wearing effectively. And so we actually push that out on the front end, and then we started getting shared all over Twitter. It's like client diverse, the client diversity. And now here's like a way to actually merit. your client diversity. Since then, client diversity has massively improved. We're now, I'm just looking at the slash overview screen on the Explorer and we've gone from like 80% prism to like 40% prism, lighthouses are like 35% tech was at 17%. And then the smaller clients, Nimbus and Lodestar,

are actually like improving. We, I'm not claiming we were responsible for it. But I like to think that, you know, we had, we helped. And, and that's all we can, we can really hope to do, right? Just give people the tools to make, make the right decisions, whatever the right decision is for, for their objective function. Also, like, in terms of another stated goal of the theorem, I guess is resilience, right? And just surviving like a catastrophic event in terms of censorship, in terms of war, in terms of like nuclear Holocaust and, and so on. And, and, and, and, and, and, and, and, and, Since the beginning, Ethereum had like a very strong focus in solar staking, right? Staking from home, which is kind of antithetical to like the whole proof of stake thing from when you look at things from bottom up, right?

Because proof of stake is part capital or like one half capital. Capital is dominated by power laws. So like just crushing those those power laws is very, very difficult. But, you know, if you transport yourself back in time and you think, you know, when the beacon chain was launching or even before it was launching and what the conversation was shaped around is like, you know, validators are going to run from fridges and everybody's going to run a validator at home. And so it turns out it's like it's actually not that easy, right? 32Eath is a lot of money these days for the average Joe or Jane. It's the interfaces, I mean, with things like DapNode and so on, like tremendous progress in. making the whole solo staking thing more accessible, but it still is pretty daunting. Like what happens if my valid is like, am I going to get slashed? I'm not going to get slash, like education around this whole thing. It's not like I buy a pack of candy, no pun intended from, you know, the kiosk or or whatnot. It's actually like it takes like

a pretty long learning curve and you either have to be very committed to some higher order goal to be part of it. or you're just incredibly interested in sort of the nuts and bolts of it for whatever, idiosyncratic reason. Again, like these two things don't make for like a hugely available market sandwich on top, like the capital requirement and everything actually like is a pretty small segment. But we have found that be that as it may, Ethereum does have about 6.5% of all the validators that are active on the beacon chain being solostakers,

which is a tremendous outcome, right? That's like in the billions of dollars running on people's homes and so on. And more interestingly, even like if you now change the denominator from amount of stake that is running to beacon nodes, which are sort of think of it as the box that is running those validators. And you can run like many validators on a one box. It's probably like our best estimate there is 25% of the network. which is tremendous for like a proof of stake network where really important stuff runs on.

Sure, it might not be 50%, it might not be 70%, but I'd contend that even if it was 6% of the boxes, that's an amazingly resilient long tail of operators where if the large operators that have a lot of capital and are visible companies and so on are easier to shut down in an event of extreme censorship, that last line of defense, which is running in like, you know, random homes or basements or I don't know where all across the world is an incredibly sort of resilient backbone that is demonstrably helping Ethereum like achieve its goals.

I think the last thing, the last stronghold that remains is really, you know, geographic distribution and increasing that and execution clients diversity. So Geth is still the dominant client. Something goes wrong with Geth and the network will experience, you know, sour times. Like I think it was just this week or maybe late last week when Geth had an issue with block production and then together with how Prism actually decides where to build the blocks when the validators are boosted, they actually kind of,

we saw the network experience as lowest effectiveness since Chappella, which is not really a frequent phenomenon. Like, Ethereum is running really well for all intents and purposes. And again, like it was, you know, a two sigma. event that did like no one bat an eyelid on the execution layer. If you were transacting on Ethereum, you wouldn't know that this thing was happening. So, you know, there's still like work to do from like a network perspective. You know, we're seeing that the network is actually like pretty heavily concentrated, you know, in North America and in Europe, particularly

when it's relating to what we've sort of contextualized as professional operators and looking only at that slice of the network, but then, you know, you have things coming up like DVT. So, you know, Oval, SSV, devise working on a DVD solution of their own. I think there are a few more. So the idea there is that you can, you know, now separate sort of operation keys and custody keys,

and then you can have these like fractional sort of staking schemes, effectively so you can, you know, You can airdrop adapt node to someone in Africa, and they can get started with, like, a very small amount of collateral, you know, one-eath, half an eth, whatever that might be, and then someone else basically puts up the rest. So I'm generally hopeful. Now, is that going to be enough to quell basically the power law?

Maybe not so, but it doesn't matter. right? What matters is how strong that backbone is because when when push comes to shove, this is going to be the last line of defense. And that's what I think matters the most. Yes, super interesting also. Just to see here about the practical issues right now, I think, yeah, there's so many layers that you theoretically want to have decentralized, right? And then every one of them is like on a different level on the spectrum. Sometimes it's even subjective, you know, is it you centralized many people, been like criticizing, for example, also like Lido to be like centralizing.

But then again, right layer below, there is like 30 different operators operating in Lido, which can be overlooked if you just see like the 30% on even on rated, but you guys have like the drop down right and showing like the different operators that operate Lido, for example. That's right. or even on client distribution, right? Like, so I think chain safe that is the developer of Lodestar, which is like the fifth client that's available to run Ethereum on validators

and also like the one with the least penetration, they are now running some like 10,000 validators under Lido with Lodestar. Right? That's, I believe, something that kind of Lido as the pool manager, the active set manager in that sort of neighborhood of Ethereum. And it's like one, one third at the time we're recording. I don't necessarily think that this would have happened. Like this increase in participation of Lodestar in the mix of clients would have

happened if a pool manager like Lido didn't actually design for it. So there are like benefits. And it's a really hairy subject. I generally don't have strong views. I try to be, you know, just the facts and like here's here's some data and you can make your own mind up. I don't necessarily have like a strong view on whether, you know, having limits on pools and so on is like the right thing or the wrong thing.

To some degree, it is the market that will eventually decide, right? and I don't know that even having, like, at least, you know, for my own disposition, that having, like, an opinion even matters. I see, I see positives. I see negatives. I see opportunities and I see, I see risks. And I try to quantify them as best as I can. And then also, like, you know, offer it to people to help them sort of make, make the best decision possible for the network.

Yeah, extremely valuable. I think we can, building off the solo staker sort of discussion and I guess the wider sort of state where Ethereum is at right now, I think one very relevant discussion as we're recording this is around EIP-7514, right? So basically the amount of growth of staking that the Ethereum is experiencing and sort of the design with 32Eath pervalidator is like impacting. a little bit the network performance. From what I understand, so I think it would be helpful maybe if you could explain a little bit, the background of what this change is about and why it's happening. And then maybe we can discuss a bit,

you know, what the implications are after that, but maybe we just give like a bigger overview of what this is about. I think it's been like in the news now, and I think it is quite relevant, you know, for many network participants, including like the operators, of course, and stakers themselves, right?

Perhaps we should better set some context for listeners that don't have it, right? So in Ethereum validators are basically the unit of account in consensus. It is kind of an instantiation of a watcher, consensus participant that basically attests through the state of the network. I see things around me, the flattest most dumb. sense possible, I see things around me, and I report what I saw, basically. And then in order to have one of those virtual watchers or participants of consensus and so on, you need to have 32Eath, right? 32Eath enables one. Now, that doesn't mean that that one validator is also like a one

machine. In fact, you can run like many of these validators on a one machine. Now, I think the design originally was as such to help solo staking happen, make it a little bit more difficult for concentration to happen. As we've seen in delegated proof of stake, these constraints don't necessarily exist. And so it's much easier for power laws to actually instantiate. Right. So design decisions made at a time in the past for reasons and arguably the right reasons, right? Or at least the right reasons for Ethereum. Now, the downside to that is that when you have a lot of interest in staking, you have people wanting to participate. You have increasingly more validators joining the network, these digital instances of consensus participants, which means

means that you have a lot more data that the network needs to handle. There's, you know, data that's being exchanged in the P-to-P layer of the network, which is, you know, all those validators talking to one another and saying, I saw this, I saw this, I saw this, I saw this, everybody saw something. Everybody saw something. And then there's like another consensus actor that's not like really explicit on chain that takes all a clear. What did everybody see? Okay, let's put it together. Like, what is reality? And reality tends to be what most people saw. But now you have like so much more of these messages being exchanged. Really like it looks like the people that were on the, you know, the withdrawals are

bullish and sort of an enabler rather than a, oh my God, like all the stake is going to flee the beacon chain. People were on the right side of history. So we've basically seen like a 50% increase in active stake in the last six months or maybe less. It's April. Now it's we're recording at the end of of September. So I think at when withdrawals came, that upgrade came to the fore. It was 500,000 active validators. Now we're at 800,000, right? And we're talking about 300,000 extra validators, which is more than 50% of what were active back then, that joined the network in the last six months. It took the network

two and a half years to get to 500. So if you're like, I mean, that's a concerning kind of trend, right? Like concerning from like a network load perspective because then you see kind of like a straight line, but then you see that line accelerate and the curvature change. That sort of fits in an exponential curve. That's concerning. Because then there might be like an undue load for the network to process, which means that the network is going to underperform because it's not set up to actually handle

this type of load of messages, then a bunch of like useless effectively, or let's not call it useless, but the marginal value that this extra message I saw this brings is tiny the more of, you know, the more of these messages you actually have. So in order to basically prevent the network from experiencing hard times in terms of infrastructure and networking and state bloat and all of these things, a decision was made to actually cap the validator activation limit to eight validators per e-book, which I think was the number that the network was at before withdrawals activated. It was seven or eight. I don't remember which one was it was exactly.

I think right now that number has gone up to 11. And basically, Ethereum works in a way that creates like a bottleneck on the way in and on the way out, the same rules that apply to the activation queue roughly apply one-to-one to the exit queue as well. but basically the bandwidth scales together with demand, that bottleneck becomes wider and relaxes as more demand sort of comes to the door effectively. And that bottleneck exists to also manage part of that whole data clutter and so on, but also to control stake,

in terms of it leaving as well, right? And suddenly Ethereum not only losing a bunch of useful information or maybe not so useful information in terms of consensus in the well function of the chain, but also like a lot of security in terms of stake. So this EIP, as I understand, is a sort of short-term solution

to actually give the network some breathing room so that a more permanent decision can actually be made as to, okay, well, we have this real issue, you know, too many attestations, too much load on the P2P network, a bunch of like redundant information that just like bloats the state of the chain. What do we do about it? And so there are, again, like as a person that's trying to be neutral and like look at, look at just the facts, I don't necessarily have like a strong view on whether this is good

or bad. I can see like arguments on on both sides, right? And the argument on the pro side is like, well, let's buy ourselves some time with this, you know, not so intrusive change to the protocol, roll back to like a few months ago the state of the queue two months ago and allow stake to trickle in more slowly so that we can buy some time to figure out what we're going to do eventually. But then on the other end, this creates at least a, when you're looking at the two states, right, like, you know, 11 validators per epoch versus eight validators per epoch, two states of the world moving from one to the other, you are

disproportionately with that move favoring the status quo, such that if, for example, there was a strong need or willingness of existing stake to be reallocated or new stake that on board to go to places that are not dominant in terms of their representation of stake in the beacon chain. Now they have less of the window for them to actually kind of catch up, if you will, is narrower, which effectively buys time for whoever's in the lead. I don't know if there is like value in really kind of diving a lot deeper into that or really ruminating on it on it that much because it's it's all

very subjective it's all in the eye of the the the beholder but the the problem is is is still like existent right like from a network perspective from like a well-functioning network health perspective like that's a problem that we're going to have to solve eventually right and there's it rhymes well with like another proposal which is like okay let's aggregate um the state why do we need to have 32EE per validator, when we can just collapse that and do it, I don't know, 20 times, 50 times, 100 times more, and just allow validators and validate operators that option. Not make it mandatory, keep the minimum at 32, but then you can add those 32 increments, and instead of running various instances, having many of these digital agents of consensus

that send all these messages, suddenly instead of 100 messages, you can just have one. But that also comes with like a whole world of tradeoffs and largely like all of these things are, ends up being like a political thing among other things, right? Which is, you know, people have made design decisions.

People have built technology and infrastructure that actually matters the state of the network as it is today, which might or might not be a competitive advantage. So I've made all of this effort and build that advantage and built all these capabilities. And now maybe they will be deemed completely redundant, right? Or I will need to like retool a bunch of the things as I've worked for them.

There's also like very credible, you know, arguments in terms, okay, how do we handle slashings? Does the penalty sort of increase proportionately? Which means that, well, I might fuck up the same way. Excuse my language. but I might fail in the same way that I do today, but instead what is at stake, and the slashing penalty today is one-eath, what's at stake is much larger.

Whereas now, for example, like a correlated failure will be staggered and it will be one validator after another after another, which basically buys you time to actually address the issue before it kind of like causes contagion in like that one box or that one cluster of boxes that is running like all of these validators and so on. So, you know, real issues that don't really have clear-cut answers. And that's both kind of, you know, the beauty of it. And it's also sometimes what, you know, slows progress down, right?

And it's, again, it's a world of trade-offs. Yeah, definitely. Thanks so much for elaborating like this on it. I do think it's very interesting how. you know, these dynamics that you mentioned are like sort of like one, on one hand, like, favoring in a way the power loss by like sort of limiting how many new can enter. And that's like sort of a result of this design choice to that actually tries to kind of achieve the opposite, right, by like helping solar stakers to stake.

So it's a really interesting trade of space. And I do think maybe the solution could also be, right, like, because like we're saying, It's a temporary solution, so like something needs to change. And it seems like, or from my perspective, one of the only things I guess that I have seen is that sort of increasing that limits or aggregating the validators in some sense. So I do think what is also interesting is that maybe you can do that, but then also you have like solutions like DVT that maybe like can still maintain

like the ability for solo-sakers to sort of participate even if they don't have, because already 32E like you said, it's already maybe too much for many these days. And actually so if it's even higher, I guess maybe DVD can be the layer that sort of solves that and it wouldn't like blow the Ethereum state so much. But

yeah, very interesting field and yeah, I guess also cool to have like all these teams like working on different parts of it and you sort of, you know, making it all transparent and like contextualizing and I think that's like a very cool spot that you're in.

We're trying our hardest. But just to harp on what you, where you left off for a little while longer, like the flip side of having all of those teams work together in like loosely or less loosely connected ways in no sort of central planner that just explicitly says, like, you know, this is what should be done. And it all kind of like basically gets done in, you know,

a network of influence. effectively is that you have the bloat situation, right? Or you just keep adding features to something because that's sort of the natural impulse to do. But then you have like a strong contingent of like client developers and so on actually advocating for a cleanup fork. Let's not do like all the sort of improvement stuff and add this. And then you know, DVT comes into the picture.

And then other components come into the picture. And then complexity is just like, piles up because again, that's the natural impulse of humans to build features and not not to actually reduce features. Reducing features is painful. It means that you have to like roll back your decision that you made earlier. It means that you have to like accept the mistake, admit defeat. And also it's generally more more painful to do that, but very necessary. So it's just so fascinating, right? Just seeing humans just work together and like, implicit ways at this level of scale.

Yeah, it's just really fascinating. Yeah, and I guess like one spicy topic we had there before we started recording, I guess it's also like, yeah, how are these decisions then made in the end, you know, like Ethereum is known to be like sort of a advocate against, or like, I guess the Ethereum community at large and maybe who even does that right, but like I guess also Vitalik and some of the core early people sort of against this sort of notion of on-chain governance, like making the tokens decide. But then we can definitely see in this case, for example, that these choices are made by some of the core developers or in some way, right, the decisions are made, but they're potentially

not like taking into account the people that this change is impacting, right, in a way that maybe other systems would. And I think that's also very interesting now to see how these things play out as many have critiqued the on-chain governance but actually like here i feel like it's a very good um example of why maybe this other path isn't like working as well either right so you yeah i i guess if you want to comment on that like what how how are you thinking about this the next chief governance there are no good options we live in a world of like bad options uh when it comes to this right like it's a look at governance like at governance like at outside of chains or whatnot.

It's not like things are rosy there or that like we've solved that problem as like the human civilization. So we just have to live with like bad options. And it's also like there's precedent of on chain governance actually like not being a solution. Like there was a TESOS upgrade,

I think was their sixth or seventh upgrade maybe like already two years back. There was like a big debate about an escape hatch feature. And then the upgrade wouldn't like actually go through. It's a very political process. With Ethereum, as you mentioned earlier, there is no explicit on-chain governance. And I don't even know that explicit on-chain governance is the answer.

So governance happens in a rough way, right? The Ethereum community calls that rough consensus, where ecosystem participants call it the client teams, the Ethereum Foundation, the operators, and so on. They're all in like, you know, three or four different forums, not including sort of all the private chats and IRL places and so on. But I guess it's ETH research. It's like a couple other forums.

It's like a couple discords and so on. They discuss those issues at large, right? Like upgrade issues. What are we doing? How are we moving forward and so on? What ultimately happens is that if the operators don't decide to upgrade the software, even if like all the client developers, for example, agree and like the protocol researchers and protocol.

developers agree to like push the change forward. If the operators don't upgrade their nodes to the new version, then the network doesn't upgrade, right? What is like a little bit under-talked about, perhaps, and also maybe a little bit spicy, is that these operators have a lot of stake on the line, right? Like they have their actual stake. They have businesses that they've built on top of, that they're effectively like, not custodians, but they're agents of the stake of others, right? So there is like really like a lot on the line for them.

And so ultimately, if they did have like, even as a solo staker, if you did have like a deeply rooted belief about like a specific feature, which, you know, even saying it that way sounds like a little bit ridiculous, but, you know, so be it. It's like if this infrastructure is as important as we all think it is and as important

as it will be, like being religiously militant about features, actually makes some sense. So the point is that even if you were vehemently against, your stake is at risk if you're in the, you know, if you're in the less representative side. So, you know, what happens like if you follow, right, right, right, right. You know, the other chain, Ethereum Classic.

or what we're for state classic. Like, you know, what happens to like your, your, your, your, your, your, your, your, your, your, your, your, your, your, your, your, your, your, your, your, your, your, you know, again, really dense, not only technical like, in fact, far, far from uh, uh, issues that have no, no clear, uh, answers. But I, I will say that, you know, Ethereum has gone pretty far further than most, if not all. with that model of governance that it's got. So you have to acknowledge success when you see it.

Right, right, right. Yeah, definitely. Yeah, it's a super interesting discussion. Like you said, it's far from technical, and it's something that we haven't solved as civilizations anywhere, really. I do think, you know, yeah, you have that extreme option, like you mentioned, the Forking. but obviously that's, yeah, like not really a tool for like day-to-day operating like this sort of network, right?

It's like in a very rare situation. Maybe that will, you will like sort of play that card. I guess, you know, some sort of formalization of this governance, to me at least, feels like it is potentially like a direction that is maybe other explored in Ethereum also. I do think in other places maybe it was formalized too early and not well enough, but I guess being some more effort there, I feel like there is a big area that one needs to talk about more, like, build or like discuss what could be done. But yeah, so thanks for this like sort of expedition into that direction.

I guess we're already talking quite a bit. So I wanted to end it on the notes about radio. again. So we've been talking a lot about Ethereum. Obviously, you guys are very focused on Ethereum. But the broader vision, right, is very reputation for machines. So there's, like, obviously, other machines, aside from Ethereum validators. What's your, yeah, sort of what's the future for rated? How are you thinking about what you're building and how you're expanding maybe from Ethereum?

Or are you even doing that layer two, is who knows if you agree to just, yeah, talk to us a little bit about that, then we can wrap up. Awesome. So we are, as we've discussed, present on Ethereum today. This is where we started from, but the vision is much larger than that. It's potentially like as large as Ethereum can and will get and even bigger. So it's just coming from the background that I did and having worked with people across many different networks and having kind of like interface with.

these networks, I see value in like plurality, right? And I do see a future of multiple different networks that have different core value propositions of different layer twos. And I'm, you know, multi-chain, multi-network maxi, I suppose, although this might be like a somewhat contrarian view to have at the time that we're recording because it's like the deep of the deep in the bear market and you know, it's hard sometimes to see kind of like the light at the end of the tunnel. So what we're planning to do in the coming quarters is actually expand to multiple proof-of-state networks. The Solanas of the world, the cosmos of the world, the polygons of the world. And thereafter, to expand to actually covering multiple agent sets.

So not only validators, but there are many node-dive. that actually have very similar properties. They participate in networks. They produce useful work. They earn rewards or a fee stream. They participate in a fee stream for the useful work that they produce. They are agents in networks and they're entrusted with sort of missions. And they might be fulfilling their missions successfully, not so successfully, not at all.

they might be inflicting positive externalities in the context of their networks or negative externalities by taking and do risk and so on. So, you know, I think by just doing what we're doing in multiple networks and going into all these different environments, not only can we help these networks just be better, achieve their goals better. Like one more, like little anecdote, I ran the numbers at some point and I looked at the historical effectiveness of the Ethereum network on the whole and it turns out that apart from the merge

which was a very volatile month effectively the network hasn't had a lower effectiveness month than the month that we launched. Now, again, I'm not like saying we are wholeheartedly and uniquely responsible for that outcome, but I do think that we've had a small part to play, right? By contextualizing, by surfacing

information about performance transparently and in an easy to access and understand way. So I do think that there's benefits in actually having that unified view, common abstractions in terms of, you know, what is performance, what is rewards, what is risk that are virtuous for these networks. There's a lot of work to do.

There's a lot of scope. there's a lot of complexity. It's a big infrastructure build. And then while your ability of adding the marginal network might improve over time and the cost of doing that drops, then there's another more insidious curve that you don't get to actually realize until after some time, which is like maintenance and cost and scope. So these are all very challenging things.

And also there's another challenge in that, there might be, and I think there are, more networks and more agent sets than a one company might be able to cover, right? And then you end up being in a situation where, you know, you're, you as the company or the organization or whatever you're in that boat. And then, you know, there's three holes. And then there's a fourth hole and there's a fifth hole. And then you have like to cover the fifth hole, you have to just open another one, right? Just move your hand from like one to the other and then exactly it's the it's the whack-a-mole analogy that we we talked about before before they're recording so um that's going to be a challenge figuring out how to solve for that

is going to be a challenge but it's a challenge i'm super super excited um to be to be taking on because because you know we see the positive impact of of the work that we're doing and and that is sort of the biggest reward of all, if you will. So just being able to have like a small contribution to things moving forward,

to things improving, to arming people with the right of in arming might be like the wrong word, but giving people the ability to access information and empower them to make good decisions for whatever their objective function is is something I can

you know, get behind and continue doing for a very long time. So I'm excited, I'm excited to be on the path that we're on and I'm excited to explore where it takes us. Awesome. Yeah. Thank you so much with I ask for the inspiring conversation. You'll, you'll die on this hill. We get that. After this. I will die on this. I will die on this hill. Right. Yeah. I mean. Appreciate it, Felix. Again. Thanks so much for coming on Epicenter. I hope this, yeah, very, very interesting episode about Ethereum, about infrastructure. And at least we'll add to the show notes, like a bunch of the things that were mentioned in this episode.

So, yeah, our listeners can find out more there. And yeah, hope to see you soon. Really appreciated. Thank you so much for having me. Huge, huge fan. Thank you for joining us on this week's episode. We release new episodes every week. You can find and subscribe to the show on iTunes.

Spotify, YouTube, SoundCloud, or wherever you listen to podcasts. And if you have a Google Home or Alexa device, you can tell it to listen to the latest episode of the Epicenter podcast. Go to epicenter.tv slash subscribe for a full list of places where you can watch and listen. And while you're there, be sure to sign up for the newsletter, so you get new episodes in your inbox as they're released. If you want to interact with us, guests or other podcast listeners, you can follow us on Twitter. And please leave us a review on iTunes. It helps people find the show, and we're always happy to read them. So thanks so much

and we look forward to being back next week.