Creating distributed validators in the wild with Miguel Prada

hello everybody so um I'm here to present uh how to create distributive validators in the while and how we are doing this in Diva uh so I don't know if you ever heard of Diva but it's a project about creating liquid is taking in combination with distributive elevators and trying to decentralize as much as possible the evaluation um divided in 32 meters into lower amounts so how is Viva working so Viva basically combines the ethers that are deposited by his takers with the heaters that are deposited by operators so stickers only want to receive some yield on their ethers but they don't want to run a validator they just want a passive seal and operators are actors that wanna run the technical part but they also want some extra rewards they they are not conforming only with the ethereum use and they want something extra so in order to participate in a decentralized network people can participate in Diva and join the community to actively decentralize ethereum so uh this is great but there are always risk pros and cons of everything so liquid is taking on DVD are great but they always come at the cost and it must be we must be very careful when introducing any kind of DVT and liquid is taking into the protocols there are many attacks that can happen in ethereum but the main ones are reorgs uh double finality or finality lace which is basically in the case of ryorks that you can if you control enough validators you can always reorder the transaction the blocks avoiding the creation of some blocks that were already proposed and creating some double spending in another place something like an exchange where you could buy something or when you're buying a product you could do double spending thanks to these rework attacks so depending on the months it can be costly then you also have double finality which is the case in which you have two versions of ethereum progressing at the same time this attack is rare because it requires a lot of Capital to perform to be executed but we'll see how this can be reduced when using liquid sticking or or DVD and finality means that you stop the chain from progressing by actively being inactive that also requires platform Capital invested but it can be reduced again so um the main drawback of liquid is taking is that it has an implicit leverage so you're giving the power to an operator of controlling 30 to each by only depositing let's say 16 feet of collateral so you have basically uh a power of attack of 32 each with only 16 East at risk which means that someone that deposits half of that is risking only 16.5 percent in the case of 16 to 32 ratios while controlling still the 33 attack which is a benchmark for the previous stack that I mentioned before so um this basically makes everything cheaper depending on the collateral and must be considered another thing very popular is the DVT so DVT has been talked as the uh sent real of uh decentralization and validation and it's great however there is always a cost and we can see how when we have a group that instead of a single node is validating so let's say you have a threshold signature like a multisig that is validating on behalf of a validator then with DVT you need only the majority of the group to validate on behalf of the validator this means that you have a set and you have to very carefully choose the size of that set and we can see how the bigger the set the higher the resiliency because you have to have downtime in more nodes to have the validator has a detected such as downtime um there are some particular points in which is more efficient to to take the the set size but the catch up that uh I want you to take from here is that resiliency always comes at the cost it's not free and the cost here is that attacks become cheaper because you only need the majority of the set to perform the attack so the collateral you have imposed to the group is not anymore the rule but is the threshold of the collateral of the group the one that you need to attack the network right so the more validators you control this way the higher the risk so we can basically do an estimation a rough estimation of what will be the protocol size that we could consider secure and what would be the cause of attacking ethereum um so you can have an amount of collateral divided by the uh by The Ether so you are depositing uh and how much will be to perform this 33 attack based on the DVD threshold and if we put some numbers on this really simplified version because again there are many other things to take into consideration like you have to have control over messages stuff like that we can see how uh with DVT and without DVT number changes and an attack that was costing um half of it with collateral with DVT is even cheaper because you are reducing that to only 11 so just for example if you have a collateral for ethers um and uh you are not using threshold signatures you have a cost of around 4.13 percent of the ethers to control the 33 attack and if you add DVD to that only with 2.75 percent of the ethers of that set you can perform a 33 attack so those are numbers that yeah have to have to be uh carefully uh Chosen and um the protocol should limit themselves depending on the size that they can achieve and the roles that they impose to their stakers and operators so it's not all about uh I don't want to be here the the the one that is uh complaining about DVT or liquid sticking because both Technologies are great they have value on providing liquidity they have volume providing resiliency but they the risk must be taking into account and as long as we are aware of the risk and take actions to reduce them it's okay to use both Technologies but they have to be considered so imagine you want to choose four operators from this set and you want them to to run your evaluator which one will you choose the random people the FTX logo over there probably not right so you want to choose carefully and you always will have a tendency so we're speaking the ones that you like the most so there is a natural tendency to over centralization towards choosing the ones that you trust and this is risky so how do we solve this in diva in Diva all the wickets and all the pro all the operators are chosen more randomly we don't we don't select from all the DVD node operators we don't select who is going to participate and we don't let these takers choose as to Asaka shouldn't choose they should uh they should they should not have this um load on them having to pick one or the other it should be safe no matter who you choose and the protocol should do all the hard work to give you a proper node operator to run your ureters right so you can have the liquid sticking and B just believe that it will work fine so in Diva we have a coordination between the notes that allow us to create these evaluators and then we can perform the duties and evaluations that are executed uh um from the point of view of the beacon chain um I don't know how much time I still have but basically uh the liquid is taking as a wrap up and DVD are great but they come at the cost uh centralization forces uh must be taken into account um activity by Design and yeah only cooperation as we saw on the panel a few days ago uh can fight the centralized the centralization forces so we might actively work towards providing a centralized way of uh of validation um I also want to invite you to our Workshop we will be on Wednesday uh sorry on Friday Wednesdays today but on Friday we'll do a workshop testing it is coming really soon so you'll be able to play with the notes you will be able to to also um stake on Diva at least for now ungodly so yeah you're very welcome and invited to participate with us um and that will be all if you have any questions I would rather talk to you and engage in an active conversation so please feel free to us thank you [Applause] any question question question okay I'll be wrong anyway you'll see the logos and the white jacket so if you have any question please come to one of us we are around as for Diva uh yeah looking forward to having you on board in Diva very soon thank you