Leo - Analysing Validators Performance in the Present and in the Future

Transcript

Thank you. Um, so thank you for staying here. I know I'm the last obstacle between you and drinks and food. So, uh, I'm trying to I'm going to try to make this light. I'm Leo, founder and researcher at Mikolaps. Uh, so um, basically what I want to talk today is a little bit is a kind of a walk through the memory lane and talk about how, uh, staking has been evolving since, uh, Genesis until today. So um the genesis was on December 2020. At the moment um you know we had set penalties the slashing penalty is very low because uh we were just starting with the beacon chain. Uh one thing that we actually took into consideration from the very beginning was client diversity.

From the very beginning of the uh work on the beacon chain we started developing multiple clients and we care about that. So we at Migalabs developed this crowler uh to uh look at the client diversity on the peer-to-peer layer but also at the geographical distribution of the nodes in the network. Um then one year later we had uh the first hard fork of the beon chain alter uh and we did uh we had uh sync committees. So now we had another type of rewards s committee rewards uh we increased the slashing penalties a little bit. Um and then we also started looking into the decentralization level across you know all the validator entities in the Ethereum network. So a kind of similar figure to what was

presented today by Hobbby um looking at you know what is the P the the amount of validators that each entity is running in the network. Then uh we had Bellatrix on 2022 uh and this was for the merge. So this is when we had finally consensus layer and execution layer coming together and then for the very first time on the beacon chain you had execution layer rewards. Uh at this moment we also increase the slashing penalties uh to the level that we initially thought about and so basically when you had an effective balance of 32 ether you divide that but by 32 and basically your initial um slashing penalty will be about one

ether. Um then uh now for the stakers what changed uh was that okay you have first of all more data to transfer across the network because now you have to uh to transmit the execution payload uh in addition to the attestations that you had on the consensus layer before. Um and then with execution layer you also had the arrival of me. MEV, you know, changed significantly uh the game at the time and then you had all these, you know, proposal builder separation.

Uh then you had, you know, flashbots and then you had the relays which which by the way the figures that you see here is we have been also tracking relays activity and the distribution of relays and so on and so forth. Um we have been seeing that actually ME has been decreasing in the sense that what validators get uh as ME rewards has been decreasing over the last couple of years since 2023 until 2025. Uh but I'm not going to have the time to present this today but maybe come to talk to me if you're interested on on on that later.

Now uh sometime later we have in 2023 uh Capella the Capella hard fork and with this we had the withdrawals. Now this was very important for staking because it was the very first time where you could actually withdraw uh you know not only the rewards but also your initial investment on uh on on staking on Ethereum right so this was finally closing the validator lifetime loop that it was open up until now so uh then you had finally exits and withdrawals and so on and so forth. Now it's important to take into account that you know this was designed several years ago and it was designed thinking about possible attacks uh in the peer-to-peer layer. Uh what

happens if there is a bunch of malicious validators trying to attack the network uh and these are these were the scenarios that we were thinking about but uh there are many things that have happened that we didn't think about. So to give you an example, what you see on the figure here is the amount of the number of days that you have to wait in order to get um you to finally exit uh your validators. For example, right now is in the order of 35 days. So if you exit your validator right now, you will probably receive your funds by Christmas. Okay? So this is a huge um you know, it's a huge weight, right? Uh the red arrow that you see here, the small one is uh because there was a centralized exchange that withdraw a

significant amount of uh ether from um a and then there was a liquidity shortage and then there was a borrow spike and then there was a lot of validators that exit um at the same time which is the first bomb that you see in the figure. Several months later there was um um hack in a API for Solena right uh in a Swissborg API and because of that uh the institution the staking institution that was responsible for this API decided to as a precautionary action to withdraw all the validators from Ethereum network and that's the huge red arrow that you see here and that's the weight that we have to um you know that we see right

now for exiting validators in Ethereum Now I remember a time we were envisioning you know possible attacks for the exit Q. uh if I came up and say well you have to take into account the case of a hack in another chain that is completely separate from Ethereum and that impacting the execute of Ethereum everybody will look at me like you are crazy this never happens right these are completely separate entities and there is no way they are correlated well it actually they are right and so it's interesting to see how uh you know the D5 world and and and all these aspects actually correlate each other and how mechanism that we develop years ago are impacted today by things that we didn't envision.

Now in [clears throat] 2024 uh we had the den hard fork um with this came blobs. This was not a very impactful you know change for stakers uh except for the fact that now you will consume more uh network bandwidth because you have to transmit block blobs. Uh so at the time I think the average size of a block was about 100 kilobytes more or less. uh and now you have to send multiple blobs between three and six blobs and each blob is about 128 kilobytes. So you you multiply the the um bandwidth consumption by factor of three or six and then you have to add a little bit of you know um peer-to-peer inefficiencies that duplicate data and so on and so

forth. Now the figure that you see here is the amount the number of blobs by different um layer twos and rollups. uh we also monitor that uh in our website but again there was not a significant change for stakers um so I think this was a quite easy fork for most of staking entities now we had uh pectra at the beginning of this year and with this uh we had consolidations now consolidations as uh EC just presented are quite significant change uh for staking entities right so um now you have you can increase the effective balance of um the validators from 32 to 248 by consolidating multiple uh validators

into one. The figure that you see here is how much of consolidations we see per day over the last uh couple of months. Uh what we and the color code is different entities that are consolidating. Uh what we have seen so far is that most like the entities that are consolidating the most are whales. um not much large institutions or large liquidity pools uh probably mostly because of the uh reasons that is just explained in his talk uh is not that easy to do but we do expect to see more consolidations in the next few months and in you know in the next year uh and so it's important to keep monitoring this now uh with PETA we also decrease uh the initial slashing penalty now we

don't divide by 32 but by 4,96 which means that Even if your consolidated validator with 2,000 ETH gets a slash, you don't get uh initial penalty of one ETH but rather hard ETH if uh well plus you know proportional u slashing uh penalties and other uh other stuff. So uh then we have Fusaka. This is the second uh hard fork of this year uh which was already presented by uh party and the band ops team. Um now with a pure dust what we have is we usually if you are running just a simple node a full node uh you reduce the amount of uh bandwidth that you need in order to

custody the columns that you need to custom because now you don't need to download the entire blobs uh you only download part of it right so this should reduce the amount of uh network uh consumption uh bandwidth consumption that uh you have for your validators. Now through all these hard forks not everything has been perfect. There has been uh significant issues and concerns. The uh image that I post here there are some concerns that have been raised and this is what you see here at the bottom of the slides. Uh for example uh on the way we access blobs for a rollup. You know we changes multi we changed multiple things with Fusaka that are that is making the life easier for some

of rollups. This is something that we need to improve and we expect that maybe with glamsterdam on the future hardforks uh we will be able to solve these issues. Uh and this is just some of them. There is also fractional staking uh and so on. So there are some concerns and and you know this process is a is an evolving process and we are trying to uh do better all the time but it's important to keep in mind when these things happen so that we improve for the next time. Um now [clears throat] we also discussed this afternoon uh what happens when all these institutional uh funds come into Ethereum. Uh right now we have about 35 million ETH at the stake uh which is

about 100 billion at today's price. Uh and this comes with a staking APR of about 3%. So the question is what happens if in 2026 we have another 50 billions coming into stake because we know that um the APR curve goes down with the number of validators or basically with the amount of it at the stake right so it's going to go down to 2.5 is it going to go down to two point 2% what is going to happen right and then the question is is there a point where maybe putting my money on a or any other compound um or or a strategy defy strategy uh is more uh is going to make me more yield than a staking, right? And

if that happens, is that dangerous? So um we look at that and we look at one side of the coin is rewards, the other side of the coin is risks, right? So if you look at risks in the DeFi world, there are many. Um just yesterday was presented that there was about 15 billion uh dollars in hacks since 2016. Uh just earlier this year in February we had about 200 million in liquidations uh in just one single day. Uh about a month ago we had these 19 billion liquidations in just one weekend which is I think the largest liquidation u that has happened in crypto since the beginning. And then

just a couple of weeks ago, we had the balancer hack uh which you know there was about 120 million that was lost in just one hack right and this was a uh smart contract that has been used for many years and has has been audited and so on and so forth. So these are quite important risks in the defy world. Now what is the risk on the staking world? So for that I want to play a game [clears throat] because there is a quiz after this. So, let's warm up. Um, so how much has been lost in Ethereum slashings since Genesis? Option A, 156 million. Option B, 4.82 million. Option C, 37.1 million. Or option D, 0.75

million. So, who says option A? One person. Who says option B? A few person. Option C. Okay. And option D. Okay. Most of the room say it's option D. The answer is effectively option D. There is only 0.5 million has been lost in slashing. So let me break down that to you. Uh there has been about 537 slash validators since Genesis and you can see them in this figure organized by uh month and colorcoded. So Genesis is in red uh alert alter is in orange, Bellatrix in yellow and so on and so forth until today. Now uh the big green bar that you see here was the largest um

correlated s lashing that ever happened in Ethereum. There were over a 100 validators that were slashed in a very short period of time. So then you only you not only have the initial slashing penalty but you also have the correlation penalty to pay for it. Right? And then the second uh the bar, the blue bar that you see with the second red arrow is the you the first time and the only time where a consolidated validator has been slashed. There was a validator that had over 2,000 ETH at effective balance and it was slashed. But because of the changes in the slashing penalties, then he only got he only lost about 0.5 ETH. Um now of course during this all this time there has been very different prices. So

it's very hard to calculate how much really was lost. Um and then there is also the fact that some of these losses were socialized across you know multiple uh operators and then in many cases also the not operator actually cover the losses. So it's very tricky to say exactly how much was lost but if you make the numbers and we did this is more or less the number that you will have as a result. So this in comparison to the D5 risk this is nothing. This is orders of magnitude much more safe than anything else that you will find in NFI.

So I think kudos to all you uh not operators and stakers and researchers and client um developers because it's extremely safe uh to stake in Ethereum. Now if we look at the future and this is very dangerous to do to try to predict the future but I'm going to just talk a little bit about what is coming. um we have EPBS um coming I don't know when but let's uh suppose that it's going to happen next year this is going to change how MEV you know works and this is going to impact significantly stakers uh we have also inclusion lists so for example right now we are uh there are some groups talking about um preconfirmations preconfirmations are now possible thanks

to um the look ahead but there are also people talking about reserving the entire block or multiple blocks uh in a row. Um but that's okay. But you know you have to take into account how that um you know plays with inclusion list for example. Uh if you are forced to include all the transactions that are in inclusion list then you are not really reserving the entire the entirety of the block because you need to include this transactions right. So there are certain things that we have to think about it and then in the future we also expect uh lean ethereum uh and today we had a presentation about how CK uh technology might change a testing and proving in Ethereum and so um that also implies

that maybe in the future stakers will have a completely different hardware that what we use today maybe because a testing will be much much faster and much easier. So maybe you can run thousand validators in in a Raspberry Pi but if you are a prover then you probably need a cluster with 24 GPUs and so it's unclear exactly how that will impact staking uh in Ethereum. So I guess what I'm trying to say here is that you know from genesis we have been you know evolving uh and we expect this to continue. Basically if you think about a staking as a constant thing uh constant thing that doesn't change uh is actually the opposite. The only constant is the change. It has been changing from

the beginning and it will continue change uh through the next years up at it some point where we think the protocol is uh where we want it to be. So um this is the work that we do at Migalabs. Uh if you are interested on uh getting more information about our work we kind of uh sit in the middle between the core research and the staking institutions and we try to bridge that gap uh between the two. So you have all the links here and yeah that's pretty much my talk. Thank you very much. [applause] >> We do have time for one or two questions if there's anyone in the audience. I see here one in the back row.

>> Uh thanks. Just an observation regarding the risks of running validators. I would agree that on the surface the slashing risk looks to be minuscule but actually I think the biggest risk of all is the supermajority bug risk which is a very very low probability in that it might happen but if it does the level of catastrophe is maximal and I think that's actually often forgotten about uh in the discussion of the risk of running validators. So until we get to a point where maybe 40% of the network is running multi- uh EL combos whether it's through the likes of a testant VO uh OAL

DVT or SSV DVT in some fashion only then can we say that the risk of the supermajority bug will have been sufficiently mitigated. So that's that's a discussion I think that needs to come up or a point that needs to be made in the discussion of the risk of running validators which everybody here is aware of and maybe sometimes we conveniently try to forget. Just a comment. >> Yep. Completely agree with that and this is why is important to pay attention to plan distribution and use vouch and all these kind of things. Absolutely agree with you.

>> Excellent. One more question. >> Hey there. Uh you mentioned you sit between the uh validator teams Ethereum and then the um institutional stakers right where do you see institution institutional staking going and what are they looking for specifically is it like distributed validators higher yield more liquidity if you speak to that >> okay um yes so I think uh institutional stakers what uh at least in my opinion what they are looking right now is first of all most of the institutional stakers that I talk with they already have a very very secure um system in place where they are using uh multiple clients. They are using they are

focusing on uh geographical distribution. They are even focused on ISP distribution. Um so on that aspect I think most of them or at least with the one I talk with uh they are very safe. Now what they are looking right now is to explore different opportunities like for example reststaking looking into AVSS what new services they can provide with the infrastructure that they are already running right and there are also options into uh mixing staking with DeFi right um you can actually because of the restaking you can actually now invest in new defy opportunities uh with the with the ST that you already have so I think they are trying to mix this and there are also some other way more riskier strategies like looping and stuff like

that, but that is just only for those that are, you know, very risk um tolerant. >> Excellent. Well, Leo, thank you very much for just an excellent presentation. [applause]