The Impact of Distributed Validators on Staking

Transcript

[Music] foreign [Music] [Music] nice to meet you all my name is ashin I'm CTO and co-founder at over labs and today we're going to be talking about the impacts of distributed validators on staking um but before I get into distributed

validators I wanted to ask a few questions about solo staking um hands up who thinks solo staking is easy today no no hands at all who thinks it should be easy in future nice sweet um so yeah that's where I'm going to start and talk about you know some of the reasons that solo staking is not currently easy and there's quite a number of them one of the main ones is it's you know prohibitive on a cost perspective unless you're you know into ethereum for a long time or getting involved in a liquid staking pool there's other things like being on call 24 7 365 is a massive drag on the solo stickers more or less without fail every time I leave the country my node will

die um and even just you know trying to wake up in the middle of the night and stuff or being a one-man band keeping a solo stick online is not fun um there are other complexities like having you know internet bandwidth not everyone has unlimited data I have burned through all of my like fair use bandwidth in the internet and gotten throttled before um Hardware regularly dies most solo stickers don't keep you know a full ready to go second machine and the meme I have here caught me this week my nethermind fully filled up my disk and I was like cool I can prune it and then it was like no you need 200 gigs of free disk to prune so I was kind of completely screwed and just had to like

kill the machine and was offline for 20 24 hours till everything to note [Music] um keeping up versions that's also important but even just generally it's quite technical to get a node set up and even more technical to kind of keep it high up time you know Auto restart self-healing particularly you know people sometimes try and run keys in two places and get flashed please don't do that but generally Solo's taking right now isn't as easy it could be and I'm hoping that distributed validators will change that um however before I talk a bit about distributed validators I wanted to briefly kind of tell you a bit about how I kind of got here um I joined consensus in 2018 and around Devcon 4 I had kind of started to

appreciate how much energy was being used by proof of work and I was like oh what are we going to do about it and the serenity design came out which was the original eat proof of stake design and that had a 1 500 ether minimum um and slashing it wasn't clear but it was pretty obvious you could lose all of that if you got slashed or you know it was still being worked out and at the time 1500 each was like half a million dollars and with the risk of losing it all I wrote this article saying I don't know about you guys but most rational people would Outsource this if it's like a half million dollar risk that you know it's not your core business you're not going to do it and my big fear was that you know sooner or later there would be

you know 90 of stake running on a small handful of providers um that was kind of my fear up until about a year later at Devcon five uh these two gentlemen Don grad and Carl from the ethereum foundation um over the that year ethereum decided to move to the BLS signature scheme for eth2 and what that allowed is mainly allowed the state to come down to 32 ether but what the EF researchers were also aware of is you could do signature edition or signature aggregation with BLS which is not something you could do with the previous elliptic curve and this allowed them to you know give this talk on what they call trustless

validation I have moved away from calling it trustless validation because there's still plenty of trust involved in staking of all types and now we call it distributed validators um but this idea kind of gave me a bit of hope that this might make a bit of sense so fast forward to kind of today where are we at now so I've been worrying about proof of stake since 2018 and um you know trust was validated this is a community it was established after Devcon five and all ball is a project we've been on the go since early 2021 um we put we did our first test net in early 2022 around devconnect and we've been running on mainnet with one

validator since the end of 2022 recently this year we ran our second public test net for Bia I think some of you that took part are in the crowd and we've recently announced our like Alpha mainnet phase so capped at one validator you know risk it if you have the means and we're now kind of looking at like beta where we kind of take off that one validator cap and we say okay you know this is getting a bit more stable let's give it a try let's kind of take the numbers up a bit and you might recognize italics roadmap slides vitalik has had distributed validators on them since day one and he now has them close to being you know fully ready to go which is quite cool to hear um and yeah one of my questions now is you know did my fears of centralization

come true oh okay yeah on Google Slides I skipped some of these but alas um so so far it's not at the point where all nodes are it's not at the point where we have three nodes running it this is what we currently look at of breakdown by market share we have 54 across five entities if I think if this is if you count Lido as its own one and then even in the staking types we have LSPs make up a very large amount solo stickers according to rated is around six and a half percent which we're hoping we can kind of do better about um so yeah um yeah probably gonna have to skip a couple of these slides that I had on to be disappeared but how do we improve

staking first like solo staking and it's decentralization for me uh I'm starting to use the word squad is taking a lot bankless and David Hoffman came up with a good Meme and we're trying to kind of spread it more widely and get more people running in squads as opposed to solo staking on their own so what are DVS and how do they make solo staking more accessible um a distributed validator the easiest way to think of it is like a multi-sik most people in this room I assume know what a multi-sig is particularly it's usually one with more than two-thirds need to sign something and what's novel here is you do not need all of the nodes online all at once this is something that you know will allow people to run at home with fault tolerance have

machines die but still have you know hopefully three nines of uptime um what sort of thing advantages does this present and this allows people to have you know high uptime and high Effectiveness regularly solo stakers don't have super good Effectiveness because they keep their peers kind of small and that can impact their inclusion distance and sometimes raise the chance of like missing attestations and even worse missing proposals when you have multiple nodes that all broadcasted at the same time so far the data is showing us that the inclusion distance regularly works out very favorably and that you have a better chance of actually getting included when you're running in a DV than running on your own and it also you know massively reduces the maintenance pain I kind of

gave you my example of my node dying I was running a you know a solo validator on that it was off for 24 hours the DV was not it just did online and so General and then coming on to the kind of the liquid staking tokens or like liquid staking protocol side of things um liquids taking protocols would really love to have lots of solo stakers and to have like a really Diversified operator set but so far you have to give your entire stake to one operator and they can kind of go offline they can kind of withhold there's quite a lot of like Risk and Trust so a lot of the liquid staking protocols either stay with curated operators and only pick you know people with kind of cash flow name

recognition are likely to be around for a long time or they have very high collateral requirements and that's the kind of other work around however with distributed validators and you not being like a total person running the validator you're just like one piece you can then instead you know have a lower risk of taking in like a solo stake or to take people into running a cluster and you can also you know split the collateral requirements or even just kind of reduce them more generally because anyone you know machine going offline shouldn't be too fatal okay so you know is this all tier yet or where are we at in terms of you know seeing this being adopted and seeing the impact of distributed validators on the stake the first one I want to talk about is Operation solo sticker so operation

solo sticker is an initiative by etherfi dapnode avado and opal and the idea is to get um more solo stickers taking part in LSPs so so far I think we've shipped about 12 machines to people and there's about a hundred more on the way if you have interest I think they are open up sign ups again there's an event on on Wednesday to go chat about it but the idea being that um solo stickers are going to run distributed validators in groups and etherfi will use them as they're like node operators and the goal is to kind of you know push the limits of decentralization and you know have a much more Diversified operator set than some of the existing LSPs in a similar manner Lido have also been

working on bringing solar stickers into their like operator set and they've recently upgraded to version two which introduces the idea of modules and these modules intend to have kind of different purposes and pretty high up the list is having a distributed validator module that brings in Solo stickers over the long run the goal is to have you know permissionless bonded access to the Lido operator pool but before then what I alluded to about mixing you know professionals and amateurs in a distributed validator is kind of what they're looking at first and recently we can we finish the second test net with them in this regard we had 41 home stakers or Solo stakers or like new

entrants which is more than 100 of the existing Lido operator set they were partnered up with 12 of the professionals to run 10 plus 5 10 clusters I believe and they ran for about 60 days and the performance was net net indistinguishable from kind of some of the best operators on Gurley and yeah so far so good we're now kind of in talks about you know planning how we put this on mainnet and how we kind of start to like actually bring solo stickers into Lido um and then another example we had recently of you know distributed validators keeping things online in about a month or two ago now there was uh ethereum last finality there was an issue with kind of

really old messages screwing up with some of the clients that kind of caused them to lag horribly and like miss their slots um at the time with a small number of main net validators on and none of them missed their attestations granted we had a bit of a kind of an over-representation of Lighthouse if Lighthouse had an issue it probably wouldn't have been as well but the idea of being able to hedge your client risk in a DV you know was at least somewhat shown there and I'm hoping that we'll get more of these data points the more you know their kind of bumps along the road hopefully no bump too serious so one of the other things I wanted to talk about is you know what are we doing to hardened EVS get them already I mentioned that we're in Alpha so you know what's the kind of like the

Securities of this like um I'm happy to announce that we've recently finished our first audit with Sigma Prime of the Caron code base it'll be published recently I think there's a QR code there for you guys if you want to investigate it early it'll be going out in the blogs fairly soon um very happy with how that went it I think uncovered two critical five highs or three highs it's their five highs um and yeah that's all been merged all of the changes are out in our current V16 release we just haven't published it yet and as well as that we also have quite a number of other security related initiatives going on um one of the things we've done is um a review of our software development life cycle with the Ethereal Ventures team the software development life cycle

and also the employee opsec is ones that I spend quite a lot of time on a lot of people kind of really focus on finding bugs and clients but the vast majority the hacks these days are you know compromising some devops who has access to keys or you know getting into like the supply chain that way so we do a lot of time you know bolting things down making sure that there's not a lot of harm that can come to it there's an interesting blog on our side around our architecture where Caron does not actually have access to the private Keys once you're up and running so even worst case scenario our software is entirely compromised you should end up with a liveness risk not a safety risk because we shouldn't have access to the private keys the other things we're doing are solidity Audits and should be announced

shortly if there are any Auditors in the room that want to come chase me please do and we're also pen testing our launch pad which some of you may be familiar with this is the UI where people kind of create their distributed validators so this is um how we're hurting them and the other thing I want to talk a little bit about is how do they change the economics of staking so tightly this talk is the impact of DVS and staking and this is in the blockchain economics track and one of the things that we've spent a lot of time on and before I would have started over I would have worked for Block demon building their eth2 stack so I'm kind of very conscious of the cost of running staking it's not as cheap as you might think especially for the solo stickers

and if you're not at like kind of huge scales so what we've done recently is we've worked with the meager Labs team we went out onto the internet and bought a load of the cheapest bare metal servers money can rent and we I think put 21 of them around the world and 10 or so locations and we ran three clusters on them a four node cluster a seven node cluster and a 10 node cluster and the goal is to see could we get them up to a thousand Keys um I had intended for it to like break somewhere along the way and not make it all the way to a thousand but all three clusters went just fine and the like headline conclusion of the report is none of them were more than one percent off of their reference nodes so they all performed nion identically to a solo

node and then after the testing completed we wanted to you know push it to breaking and we put all of the 3000 keys on you know one set of four nodes and it was still fine and stayed online still look good in fact the performance went up slightly um so from an economic perspective we are able to run an an astronomical amount of validators on you know forbearable servers um whether you know we want an astronaut to validate us we run by a small group of people not so much I would much rather just be kind of run by solo stakers and we kind of further subdivide it but it is something that's super interesting particularly for the node operators in the room that have you know High Cloud costs and are trying to figure out how they kind of adjust now that prices have come down APR has come

down competition has gone up there's quite a lot of node operators that are like getting feeling the pinch and yeah this is one example of our performance I won't dwell on it too long because it's on the test net and earlier I'm going to skip both of them because I'll we'll show you more data coming out too and I don't want to undermine some of our cool reports that are coming out but I will say you know the intention is to make staking particular DV sound easy but I would say that it's you know not as easy as it could be yet you know there's still some caveats you still have to you know run a validator keep it online learn how to manage it but what I would say you know there's a little bit of communication and coordination over it with your team but the real benefit

is having a team to run a validator there's a lot of people that would be scared to run a validator solo they've been you know not so sure if they have the technical shops for it not so sure they'll be able to fix it if it goes down particularly when your validator does go offline it's quite a frustrating time to try and get it back online but when you're in a group in a DV you have allies with you that are quite happy to work with you they'll let you know it's down they'll you know figure out how to get you back online if you need to kind of like connect to their Beacon note for a while you can kind of Leverage them and in general the hope is that you know your squad will have your back um but the real question I want to kind of

leave this with is you know where will staking and solo staking more particularly be next year that is you know the kind of phase of this that I'm really hoping will kind of take off some of the last two years have been heavy in the r d you know getting to Alpha and stuff and at this point we're now kind of really expanding it and trying to make it super accessible for people we're doing a lot of work with all of the launchers so the kind of dap nodes avados theorems all of these type of uis to make things easy and we're generally hoping to you know get more solo stickers involved and to lower the barrier of Entry to staking and if you guys would like to kind of help us with this particularly over the coming year there's a number of ways you can get

involved depending on kind of you know what you like to do at the you know top end if you have the means to run a main ad validator and you would like to take partner Alpha I would encourage you to do so come ask me questions if you would like and Beyond then you can join our Discord and this is kind of where we keep most of our communication and we have people going you know taking part and getting involved you can find people to run DVS with there if you you know don't know a group yourself we have recently launched a community Forum which is where we're starting to do some of our more like Advanced research on a distributed validator protocol a distributed validator protocol what that is is the idea of being able to provide succinct proofs of who is online and offline in a

distributed validator with the intention of eventually being able to kind of reward and punish people individually within a distributed validator rather than collectively right now it's just a team game you keep the validator on you can like spit them equally but there isn't quite a way to kind of prove to a Smart contract hey I was online but they weren't so I should get the rewards they shouldn't and we're doing a lot of good work with nethermind on this netherminder also working with the Lido on kind of permissionless validation so if the kind of more research you end cks vdf Byzantine consensus all this kind of stuff interests you I would recommend you to take a look at it and then more generally um what I alluded to about the education

side is something that we're you know really trying to make a push for at the moment there's quite a lot of kind of you know communities involved in staking and ethereum more generally regularly there's you know the East taker Community islaram could keep you know shifai there's you know groups all over the place that are trying to kind of promote ethereum for you know their Community or their special interests we would love to work with you guys to make TVs accessible for your kind of group the goal is to kind of you know train you to train others and the hope is that we make running distributed validators less scary for people and that we have kind of groups that you know are all involved all keeping validators online with one another and lastly for the kind of developers in the audience we have recently soft

launch an SDK for making it easy to create um DVS programmatically it's in the advanced section of our docs if you want to go find it it's still a bit ugly and Rough Around the Edges so I'm not sharing it too hard just yet but if you are eager to give something like that a try I would encourage it and other than that yeah I see thank you very much and I would like to see if you guys have any questions for me [Applause] go on someone throw up a hand there take the solo sticker good question I think the only

organizers might be around we have a picnic well we have a talk uh Wednesday evening at the solar Staker event um but if you search etherfi operation Operation solo sticker you should be able to find a sign up form there I don't think we've had anything on the opal side right now but uh yeah I don't think that works oh it does work now um do you foresee any issues if say 100 of solo stakers are running on mobile um software is that an issue do you have like a spec that others can run yeah if you hit the nail in the head there with the specs so that's what the effort would never mind is to make a distributed validator protocol spec um if a hundred percent of well if it was only solo stickers solo stickers are

a small percentage right now I don't think it would be too big a deal but yeah generally if you know 100 of all stakers ran with only one client and I've talked about how Caron is the name of our software client it doesn't have access to private keys so even worst case scenario like software is totally compromised you you know Karen shows flashable messages to the validator that's connected to it the validator shouldn't sign anything slashable but you will end up offline so to reduce the liveness risk that's still there we are working with you know making a spec and the intention is to have ideally four interoperable clients over the long run and that way if there's you know zero day or a massive failure in any single one of them literally nothing should go offline but I would also say that Karen is like independently versioned each

cluster you know is isolated from one another so you don't have to upgrade versions at the like simultaneously and you don't quite have to kind of you know All or Nothing it so if there ever was you know a dodgy version you could roll it back you can upgrade some of them slowly so yeah there can invariably be liveness risks but we are working on having multiple implementations to kind of get ahead of that as well yeah any last questions Speak now or find me after okay I think oh no last minute yeah I I hope some of them will make

themselves known afterwards and give us a bit of info about what it's been running so the idea my understanding is that ether fire the ones that are like swallowing the cost of like sending out these machines and that you would gradually like earn earn your way out of that by you know they keep the rewards and that's kind of how you kind of get in and you start like earning your keep and um yeah that's effectively the means is ether fire like they have to eat they're looking for node operators they want solo stickers so they will send you a box if you kind of prove you are a solo Staker and that you're you know credible as for actually what it's been like I think iridium might be around somewhere but he's been working with the teams getting them set up they have avado and

dap node packages and I think it's they're at least running on girly I saw one proposed a block recently so that's a good sign and I think the hope is to start spinning them up on mainnet fairly shortly I don't know if there's another hand or if that's just a wave no probably not anyone else last questions okay won't give you some shine this time thank you very much all you can come find me afterwards [Music]