Distributed Validation as the antidote to Staking Centralization - Pablo Villalba

foreign [Music] if you think well hey everyone I'm Pablo from Diva and today we're going to talk a little bit about how distributed validation can be a decentralizing for us removing elements of trust and of centralization in the staking system so it finally happened uh the merge went through smoothly and now we have an equivalent of ethereum is taking representing 30 billion in assets which are yielding a real yield of five or six percent a year um and because the if is deflationary this is the first time that we have a massive real yield in crypto but with this switch from proof of work to proof of stake now security is determined by capital and that's unfortunately centralizing so to look at what's driving that trend I want to ask why what are we asking from our validators So today we're saying anyone can run a violator as long as you have 32 East which are about 40 50k today you will need also a reliable server and skills something that you can keep online that can that can run a full node but we're also asking you to be willing to accept the liquidity and potentially lower returns from solos taking so this is not a great deal from we're asking and one question is who's left when eth reaches ten thousand dollars well we know the answer right um the ecosystem has been changing and a lot of people have been instead of seller staking which would be ideal joining centralized exchanges because they give you a deal where you have no minimum if you have sometimes liquidity but the best solution to date has been liquid sticking because you have no minimum ease you can use an Unchained protocol get your liquid stick ease which you can compose in D5 and it's definitely a winning option but it's not something that you can do on your own so because of that today's ecosystem looks like this we have one major player later controlling about 30 percent of the stake liquidity we have the next three or four entities controlling more than 50 percent and then rocketpool is doing a great job decentralizing and somehow a bankrupt entity uh and yeah it's not looking great I think we can agree so what I want to talk about is how do we get here and how can we change it so when you look when we look at traditional staking the word that comes to mind is stressful you either need to trust yourself that you're gonna do a good job or you need to trust somebody to do that for you and some of the risks are if you lose your keys or you go offline you could potentially look at a loss of 16 eth plus a 10-year lockup by the time you get that if what is that worth at that point you also have slashing risks which are deterring people from casually setting up validators but the most important is it's very hard to delegate a validator because you need to trust the person that's running it for you if the person running your validator wants to Blackmail you or do an economic attack they could do that they could say hey I have your key unless you pay me something I'm gonna go offline that person could get hacked so it's not the person you're trusting that could do it but somebody else could do it or they could make unilateral censorship decisions so clearly we can do better one thing we could do is we could say well let's not trust one party let's appoint a number of operators and let's get the most professional people we can find but the issue is even if we have a larger number of operators what we're doing is we're making collusion harder and we're making attacks a little bit harder but if we're not asking for collateral we still have a zeroed if cost of attack collusion is still possible even in small groups since the entities are known um and today for example with Lido the the risk is real right each operator is holding one percent of state if um that's at the lighter level but if we look at the network level they're also quickly approaching the 33 critical threshold once you have one player coordinating a third of if a number of cartelization issues start to become more obvious um well one way would be to have platforms limit but clearly that's not the case so we urgently need different solutions uh to balance the ecosystem so another thing you could do is while rocket put proposes saying look we'll ask everybody for a collateral we'll ask for a bond of 8 to 16 eighth per validator and that means that if somebody misbehaves we'll just take it from their collateral so what we're doing here we're saying there's people who are kind of senior and protected the Staker is more protected there's people who are doing work they're getting paid for that work but they're putting some guarantees so that model works but there's still margin for improvement in this model what we're seeing is each operator is custody in one private key so you're not fund custodial but you're still dearly custodial because somebody has that key second issue is that bonds can be prohibitive and the current state of Rocket pool is the deposit pool is full and that is because there's more demand than Supply so we need better mechanisms to balance this so one thing you could say is well why don't we do distributed validation and I would tell you that is great because you remove the custodial element from the private key but you're also changing the model right because now collusion could paradoxically become easier since you only need two thirds of the network to do it and the other thing is distributed validation on its own is not the answer you need an economic model to make sure that it's not economically rational to attack the network because even if you're Distributing you could still have a similar risk of collusion so questions like how to reward how do we reward good operators that's not built into the model how do we ensure good faith how do we attribute losses and how do we eject Bad actors all these are questions that today are not solved so we need more resilient Alternatives I like to say the future of ethereum is at stake so we went back to First principles we went back to the Whiteboard and thought what if we could rewrite the rules for validation what if we could design validators that are fully replaceable what if if somebody goes offline somebody else could take completely that responsibility without any loss of funds whatsoever what if we could reduce messed attestations because we could distribute the ability to be online and just like if an employer doesn't come up to the office the office keeps working but we found that we could reduce messed up stations by 500x and I'll show you that what if we could have smoother incentives that are socialist rewards what if we could eject actors quickly undecisively to make sure that that doesn't take 10 years to get kicked out so what we said is well we want native Unchained machines that can't be evil so that's about what we've been designing and I'm going to walk you through some of the decisions we took so the way to understand diva is you have three design spaces you have the liquids taking space where you have light or Frags stakewise you have the economically trusted space whereas what Solutions are there out there where it's not economically rational to attack them so the only solution that comes to mind is rocket pull here and then you have DVT but as I said DBT still has this and answer questions how do you pay people how do you reward people how do you take uh fines from them uh mobile and SSV are doing a great job at coordinating DVT but it needs to be connected to liquids taking and it needs to be connected to Economic Security so whatever is creating is we create a two-sided staking pool if you're a Staker you're the passive part You're simply is taking that you could do a neither and you get a liquid stick token there's no minimum if so you get the full advantages of a liquid sticking solution now if you're an operator we want all of you to be operators by the way what we're doing is we're boosting the rewards by allowing you to get up to a 2X boost if you're running a node and as we said it's not just the capital that it takes to run a node it's also the difficulty so we try to make that experience as simple as possible so how does diva work the infrastructure looks like this if you're a Staker you can send your eth get your div eth back that goes to a Smart contract and what the smart contract does is every time it can it creates a validator and the difference is we're not innovating at the liquids taken level we're innovating at the distributed validation level so what we're doing is everyone in this room could have a node and if you're running a node you're playing a lottery every time that there's a new validator and you might get a chance to join it now to join that validator you need to post a little bit of collateral but there's not 32 it's not 16 or 8 it's one eighth so if you have one if you can be an operator on ethereum mainnet thanks to Diva the way I think about it is if you have a Macbook you can probably afford a validator so how does a validator work is basically an execution client plus a consensus client we support any diversity of clients and what we did is we built a very lightweight client the diva client is just a few megabytes that sits on top of that it consumes minimal resources it's just a very light peer-to-peer Network by the way we're not a blockchain we're just a peer-to-peer Network so it's much lighter it's much easier to coordinate it has much more capacity last point is unfortunately the way ethereum is built today you need an oracle to read rewards but we've been trying to minimize trust there as well and we are prototyping already zero knowledge oracles that allow you to verify the balances trustlessly even without the EAP that we're still missing for ethereum so how does it compare you have for more decentralized to more centralized honestly it doesn't get better than solo staking for decentralization but it's a trade-off right so what we're doing is we're saying we'll give you all the good things that you have with liquids taking no minimum a liquids taking token the compost ability the smart contract will give you variable rewards that are competitive with everyone else but where we're innovating is we're the first solution to have node redundancy well to bring it to production that means that because every node every validator is handled by 16 different nodes up to five nodes can go offline without any impact to the network and that has a huge impact on performance the other thing is because we're splitting the bond between different nodes we can reduce that to one each one so we're trying to build the most decentralized and the most trust minimized Network for staking the same principles we have is for a solution to be a market leader you need to have Max economic utility so you cannot compromise an APR you need to give the best results we need to be integrated with the major platforms of a Euler maker liquidy you name it because useful everywhere you where you can use if you could use a state eth and that would be better for you because it's yielding so we need to work on all these Integrations to make sure that this is the most useful the other same Principle as we said is decentralized and inclusive running by only one if means a lot more people are able to do it but we also made the simplest node operator out there with Diva you don't need to take care of private Keys it works very similar to BitTorrent when you're running Diva nodes discover each other negotiate kids with each other and it's completely transparent to the user just like when you're streaming files to your friends difference is here you're streaming economic attestation there is lastly we have a design principle of resiliency to make a collision resistant and have the best risk management out there so I'm going to Showcase a few decisions we made for example the distributed validation key we're using dkg threshold signatures which means that the key never comes together nobody ever has that key which means that it can only be reconstructed with the consensus of the network we have another feature which is a self-healing network so many networks when you start losing nodes Lively Nancy grades so we have economic we have incentive models to self-heal the network and recover 100 liveliness one way to look at this is what happens if you have a very bad validator set people who are five percent of the time offline well if you only have one then you have 95 uptime but as you keep adding sets on a two out of three consensus the downtime risk goes down dramatically so we're improving it by more than two orders of magnitude with a set size of 16 that we've chosen one way to visualize that is we have this really terrible set of nodes which are constantly offline one out of ten blocks are messed by them like one one of 20 other stations sorry but what you can see is even if this set of participants is honestly terrible it never reaches the threshold where the network is offline so in this example even with that spot the availability we still have near 100 uptime this is something that directly translates into better APR and better resilience in the network next thing is we like to say we're the most decentralized network but if you look at the big contain there's only 5 000 nodes out there on 40 of them are in the US and we feel like the main reason for that is because it's prohibitive economically it correlates very strongly to GDP so with Diva we have a mandate of uh enabling enabling diversity we made it super easy you can bring your adapt node to your own machine you can install a Docker package with zero config and in a few minutes just like you would install an Apple TV or something you can be up and running and you can be validating on ethereum and again we incentivize that very strongly by giving you a boost in rewards for anybody running a node operator that means that we have created a crypto economic comparative where the incentives are making sure that there's always enough demand friends takers and enough demand for operators one way to think about this is if you're uber you need to fix a closer problem you need to bring drivers in the street you need to generate demand sometimes there's search pricing to make sure more people come online so we have something really similar where it tends to our Network equilibrium where everything just works so we started with this I feel like we have good answers for the economic part we have good answers to make it easier we have definitely answers for giving better liquidity but that's not all even if we solve this still these three problems there are still trust elements that we're trying to remove so one element that we're working on is removing trusted oracles using zero knowledge proofs this means that thanks to zero knowledge technology we will have one of the first use cases which is using ZK snarks to look at the state of the consensus layer report that to the execution layer and be able to calculate rewards in a completely trustless manner this solution is trust minimized and it means if you have at least one honest participant which you can run if you want to make sure then you'll be fine this is some technology that succinct has been developing called telepathy and we're looking to bring this to production so um what's exciting about all this uh for me personally I feel like we're working on many of the major challenges for ethereum this is the roadmap that vitalik posted a few months ago uh many of the unsolved challenges or how do we get to distribute the validation how do we adapt to a post withdrawals world how do we deal with Mev and the threat of Meb becoming a centralizing for us proposal separation all these issues what they have in common is that validators are a base layer where value is created and where security is built so building the right Solutions and back in that by the right crypto economic incentives is fundamental so that's my talk if you'd like to learn more we are about to release public testnet in March we are looking for people to play with it so if you're interested in playing with it please scan that you can find that node operator form where like um opening gradually this until it becomes public we're also looking for developers who are interested in the project so if you like to work in golang client uh smart contract development cryptography crypto Economic Security this is your project also looking for TVO Partners who want to join the network early in in exchange for like being like kind of founding members co-founding members of this project and of course node operators so thank you so much uh I think we're gonna turn down to questions so [Applause] do we have any questions there hello hello okay um right here in front of you yeah uh I'm just wondering so obviously you guys have a one eighth bond to start a validator what is the difference between like let's say that you have eighties uh would you be holding a key Shard for eight different validators would that be the the best in terms of economics or would you just have um you know one key for eight eats of of a validator what is the kind of trade-off between lowering the bond for yourself if you have more eth sorry can you repeat the question yeah so let's say that I have like AE um but you guys obviously only have a one e Bond requirement yeah leaders okay so what happens is when you have eight if you're eligible to join up to eight validators and then there's this Lottery where if you're eligible then you can receive one of these validators and in principle there's no limit to how many validators you can run but of course we want to make sure that we have a very diverse set to make sure that it's very well balanced geographically and across availability zones and so on but economically it's pretty much equivalent whether you run them or somebody else the other questions have one question there hey yeah this has been a great talk thank you um is there a road map to zero eth requirement or like very minimal a roadmap too to getting to even lower than one eth Bond requirement it's technically possible we feel like the benefit goes down very quickly under two four eight and we found that one is a pretty good amount once you go more than what you start to have is a lot of network gossip and chatter so I feel like we could lower it but there's not much of a point what we're optimizing for is making sure we have enough capacity and it feels like 1 8 is the perfect point oops I think I think I'm out of time but uh my whole team is here so happy to talk to you thank you so much for coming foreign [Music]