The Merge (Part 2) - with Stephane Gosselin, Danny Ryan, Tim Beiko, and Hasu

Transcript

welcome to uncommon core where we explore the big ideas in crypto from first principles this show is hosted by suzu the ceo and chief investment officer of three arrows capital and me a crypto researcher and writer [Music] welcome to the second part of our conversation with danny ryan and tim baiko on the merge which is ethereum's upcoming transition to proof-of-stake proof-of-stake touches on so many parts in the protocol it's such a complex upgrade that has been in the works for so long that we really couldn't you know just do it in one episode so in the first part

we covered you know how consensus works what's the difference between proof of stake and proof of work and the potential trade-offs between the two but also what has the road to proof of stake for ethereum looked like how does it affect important topics like client modularity and diversity and how did all of the atm client teams test for such a massive event in part two we're going deep on two topics that are also very closely related to proof of stake the first be mev and the second lydo and the rise of liquid staking so why lido now that anyone who has eve can in theory participate as a staker on the beacon chain there's a huge demand in order to

delegate that stake to someone else who will you know run a node and validate the beacon chain for you with a very high uptime so you can turn your eve from a non-productive to a productive asset and what lido has done is give your eve to these professional node operators but also give you a derivative that represents this eave that's currently being staked on the beacon chain and allows you to to do stuff with it in d file and on centralized exchanges so it greatly lowers sort of the cost of capital on staking and basically turns when just regular eve into a productive asset however lido is the current market leader in this segment

their rise has been very strong so they make up currently 30 percent of the stake that is on the beacon chain and to be very clear like they they delegated to over 20 node operators and they don't stake any of this themselves but danny ryan and and who has been sort of one of the most outspoken critics of lido and liquid staking he has uh in the past sort of want people to you know exercise some caution maybe to put it mildly and and also ask glider to to self-cap their growth and so we are very excited to hear from danny ryan today and you know dive deeper into some of these arguments you know what he sees as the risks of liquid staking

and uh you know what he would like to see lido to do also what he would like potential competitors of leiden to do and what are maybe like arguments pro and against self-capping and other um sort of techniques the second topic that we are going to explore is the one of mev and mev affects almost everyone on ethereum so if you have ever traded on a decentralized exchange before then you know you've likely gotten front ran by board operators if you trade on a centralized exchange and you cause a price impact then you know there will be arbitrage sort of bringing the market price back in line if you use the lending market before then maybe you've gotten like liquidated so mev definitely you know interacts

with regular users or financial markets in in many ways but it also causes uh it has this you know almost existential risk on blockchains in general as it threatens to centralize block production that is because a lot of skill goes into extracting mev and if some valid errors are better the extracting mvv than others then and this has basically a very centralizing force on you know how the distribution of these validators can look like and this is extra important after the merge because what happens at the transition from proof of work to proof of stake is that the current node operators the current validators on the beacon chain take over the ethereum execution chain and they start earning

the fees and the mev that currently goes to the miners in proof of work and paired with this you know reduction in the block subsidy that is sort of also the result of the merge now mev plays an even bigger role in the revenue that let us make today and so it is even more important that we have a way to democratize access to all of the mev to all validators the company that sort of pioneered this approach of running a public market for mev extraction via an auction system and that makes these mev blocks available to all miners today there is flashbots and so flashbacks is where i work and i i'm very very excited to

for the first time in his career bring in one of our founders the founders of flashbots and also a steward and architect of many of these systems stefan gosselin also known as the go step welcome stefan all right super excited to be here i think this is an amazing episode uh i can't wait to talk with both of you and tim and danny um yeah i'm stefan so i'm one of the co-founders of flashbots also steward and uh sort of the lead architect behind some of the products that we've launched so um mev geth sort of earlier on in the flashbots journey writes software to

help miners deal with mev and introduce sort of a separation that allows them to outsource i mean searching and then more recently sort of the the product that's ready for the merge mev boost which will allow uh validators to do mev and outsource med production on on proof-of-stake ethereum so super excited to talk about all this yeah this is the reason that i really wanted to have you on for this episode because of the intersection between proof of stake and mev and how the merge is going to change proof of stake and how um you know flashbot and the atm foundation are thinking in terms of addressing the problem of mev

yeah so why don't we jump right into that could you give a small overview of what mev boost is and how it relates to the status quo on ethereum today i think the way that we can think about the mev problem is a problem of unbundling the activity the economic activity that um that a blog post does on uh on the blockchain so in proof of work ethereum before flashback sort of came around miners picked up transactions from the transaction pool and then they put them into a block propose a block to the network what we saw and the reason why we sort of wanted to introduce a new project here is we

saw miners starting to want to compete on having access to or private access to to hash rate so miners developed private endpoints that allow them to receive transactions directly from from bot operators in this case and include those into into their blocks construct them also in ways that didn't necessarily follow the traditional ordering so there's a traditional greedy algorithm that orders transactions by the gas price that they're paying to to the miner to the coinbase address and they were starting to experiment with with new approaches to ordering so we we wanted to introduce a new abstraction that would help users of ethereum sort of express their preferences in a more

complex way while also maintaining the ability to communicate with the majority of miners at the same time so that's what mev geth sort of introduced an ability to outsource the production of blocks and introducing new transaction types for miners what it kept however is the job that miners have to aggregate all these transactions in a way that's optimal for for block construction so with um with a lot of the research that we did following the launch of this we sort of discovered that block building right block construction remained a fairly sizable centralization risk because of

the ability or the amount of specialization that goes into block construction and we saw some of the miners also starting to specialize in this block building so mev boost along with the merge sort of does a further unbundling of of this role where it separates out uh the job of block building or block constructions completely from the job of proposing a block to the network and validating a block and you know this is sort of the idea and the the basis behind the concept of block builder block proposer separation which um has been introduced on the on the research world map for for ethereum and correct me if i'm wrong there's another um

goal of mevboost is that the existing way that uh the flashbox architecture works with my with the mining today is that there's only so many participants that you could handle because of the the requirement of kind of policing the theft of mev and that mev boost also allows for kind of this relay actor to serve as this kind of coordinator and help bypass some of those scalability problems yeah for sure i think it's an improvement in another way which is like it's more decentralized than the way that flashbots works today because like the system in which flashbacks works today requires flashbots to operate as sort of a watchdog

over the the transactions are sent to miners and make sure that the miners aren't abusing you know the what they view inside of those transactions in the mf boost setting the validators don't see the content of the payloads that they receive from builders at all they only see the the hash of the content and so you don't need to do this policing anymore there's some other sort of trust uh requirements involved but at least it's not like a policing one and it actually like shifts the onus of monitoring and of verifying that the system is healthy to validators so individual validators become responsible for auditing the performance of the

bundles that they receive the net impact of this is that it allows any solo solo staker to participate in receiving optimal mev blocks right just like large pools would and eliminate sort of the incentive for for teams to specialize on performing mev extraction at the pool level and and instead everyone has access to this open and sort of permissionless marketplace for blocks okay so i have another just question for you uh steph if you can expand on on the booth uh just in the naming you know the first version of this was me v geth uh which is basically an extension to the get co space and mvv boost was designed in a bit more of like a client agnostic way so can you kind of walk

through just basically how the the diff works in practice the the current clients and how that's changed from muv geth yeah this was a really fun design process right like we've sort of enumerated some of the features and the properties that mev boost have um these were sort of all outlined from the start and were like requirements for how to design a better system for proof of stake the other one was to support client diversity so mev geth was implemented as a as a fork of death right and what happened in early 2021 sort of overnight the majority of miners all the sudden were running this like third party fork of death like eighty percent of the

miners or something which like i think hassou has referred to it in this way before like this is kind of like a minor activated soft fork right like it's a different client that's running and producing blocks to like avoid this going forward like the goal was to produce like a client that was completely neutral and agnostic to the client that's being run so it's being implemented as a sidecar that just communicates with the standard consensus clients um over sort of an api that's defined and very generalized for outsourcing block construction yeah i think if we didn't go down that path we could uh wreck all the progress in client diversity overnight as you said i take issue with the soft fork why

i don't know if it's a soft fork because it doesn't change the consensus rules like any node running is just going to see these as normal blocks why do you why would you say soft fork yeah that's a good question i think because there is an outcome to the system rules for most participants uh even though they didn't opt into them and i think that's like one of the core hallmarks of what a soft fork is so for example if you're if you're someone who's transacting on a decentralized exchange then the miners activating the software fork would make it all of a sudden cheaper for you to be signature attacked this would be one but also other sort of behavior became possible right the ability on the flip side to get transactions mined privately if you also

communicated with miners through the use of this client so and that so i think it's a bit like torment tongue-in-cheek because no transactions became invalidated as a result of this but it definitely sort of happened uh without the explicit sort of opt-in yeah of users it like restricts and it changes kind of the rules of how most blocks are produced but if someone produces a block differently it wouldn't be restricted out of the consensus so it's it's i see that i see the similarity and the difference there yeah well yeah there is this one weird thing though when you get a hundred percent or close to it of hash rate running mev geth you've implicitly changed the rules in that like the majority of the hash rate and

overwhelming majority will consider a blocked mind by me v geth as valid if there's a bug in it whereas previously we had uh open ethereum was actually the second largest mining client and because of a bunch of reasons that like stopped being true so there's this weird thing where like if you had a bug in mev geth it would it could be enshrined it could be there's like a much greater probability that it would be enshrined than in the previous status quo where you had just say geth and parody um both mining blocks yeah which is more to the client diversity question how putting this inside of making this inside of a client kind of incentivizes the reduction of client diversity yeah which some of the second order effects is also you now have like a piece of software

that's maintained independently of the client like mbb guesses and it's actually probably a much more kind of safer in terms of software production and release cycles and stuff for this sidecar to be um kind of a third-party piece of software as it's an mev boost i'm much less stressed about making mev boost releases than that i'm much less stressed about you making those as well to any listener who appreciated sort of the nuance on the miner activated soft fork there's also this other i think nuanced idea which is that i don't know how many people know that in proof of work there was actually also an implicit form of commit reveal scheme happening between block builders and

miners in the form of pools because flashbacks was only able to scale in the first place with this role of the trusted minor because we only had to reveal the block content to the pool and then the pool only has sends the hash block header to its workers and then these workers they hash you know the block header again uh with the golden knots trying to find the golden nonce right so that's that was a function of how these pools already worked right out of convenience and you just kind of get to leverage that yeah yeah yeah really fortunate right because uh in proof of stake every validator just falls like it's almost like they fall back to being their own builder again and you don't have this abstraction in the form of

pools right in in proof of work someone who you know has uh you know a mining machine at home they can opt into a pool and just sort of do work on the block that someone else built without knowing what's in it i mean most miners that hook into pools probably don't run a node at all like in you know full implementation of the protocol they're literally just running mining software exactly yeah and uh this this wouldn't be possible in proof of stake in the same way right if you run an ethereum client at home and you wanna like have 32 if and want to run your own validator without math boost it wouldn't be possible for you to perform you know the same kind of work on the block that someone else built so this is you can think of math boost as like fundamental pool infrastructure as

well in a sense i think it might be interesting to talk about the kind of members tries to create proposed orbital separation without l1 without layer one protocol kind of enabling it maybe it's worth talking about some of the the downsides to this design is like bolted on rather than kind of integrated natively and i think one of the big ones that comes to mind is that i as a validator i sign a block knowing what the value i'm going to get out of that block but without knowing the full block without knowing all the transactions and then i now have signed this thing that i can be slashed for if i sign something counter to that thing and now

i'm beholden to someone else revealing the full the full block and so you you have some of these kind of withholding attacks like potential liveness attacks maybe you have to deal with reputation a bit more uh because of the kind of the nuances around signing a blinded block as we call it yeah it's not like a perfect solution it's not you know but what has come to be referred to as entren pbs because of these limited trust vectors i think a lot of the design of map boost has been to limit the impact of it eliminates liveness risk by having like easy ways for the client to fall back to normal block production for example yeah being critical for having

you know safety properties that that work for ethereum yeah i think another interesting part is is not just falling back locally but attempting to have it be attributable if some sort of relay or builder fails and is messing with people you know so not only are validate or signing thing but things but likely you know the the op the other side of the the aisle would sign things so that you know there's some sort of like attributable nature to repeatedly not getting your block out if you're a builder or something like that yeah exactly so maybe i can i can try to like describe a bit where that attribution comes into play so sitting in between builders and validators in the map boost model there are relayers and relayers are sort of

permissionless entity anyone can spin up a relayer and start accepting payloads from builders the job of the relayer is to simulate that bundle so or score it say like how much is that bundle worth sign on top of the header of the bundle and how much it's worth and then submit that to the validators that are sort of accepting uh payloads from this relay so the relayers all sign and say like this bundle i've simulated it it's worth this much and they also escrow the content of that payload so when the validator returns back a signed lockheader it can then like reveal the block to the network and the entire network can see both who is the relay that signed the payload

was the value that they claimed who is a validator that accepted the payload and was it revealed successfully to um to that network so that attribution allows to have sort of collective data that you can use for building like reputation of the relays and help validators inform like which relays are high performance and trustworthy yeah so like the the amount of value from the bundle that's extremely attributable if they signed something that's false the like timing around whether it was released on time and stuff is probably a bit more hard to get in aggregate but i think many will be observing so i think those reputation metrics are are tractable as well yeah steph you know you just mentioned relayers and i think one thing that

might be useful for people who are not super familiar with the design is can you like walk through the incentive of like the different participants to actually fulfill their role if you look at just from like the proof of stake side like the validator you know they get paid or they get slashed so like we have like these in protocol incentives for them to to submit a valid block but if you go from like okay they're like the proposer but like what are the incentives for like the builder what are the incentives for the layer who do you see in practice like as actors filling up those roles just because it's like a very abstract design and i think it helps people to get like what's the type of entity that sits in each slot basically and why are they doing it yeah definitely so right

now builders really are are miners right they're miners who who run either vanilla geth or you know maybe guests that have a bit more profitable block construction in a mev boost setting anyone can run a builder anyone can start building doing block construction and they can pull transactions wherever they want they submit it to relayers so the relayers uh i think act very much so as rpc providers do today so everyone who's like developing on ethereum is familiar with companies like infuria and alchemy who provide sort of a routing service and a sort of a centralized rpc to route transactions to the transaction pool and get data from the chain like a similar

thing happens here where you have a rpc provider who's specialized in doing simulation and routing payloads to uh to validators from builders in terms of incentives right like relays are building a reputation and they want to have the highest data availability possible so that the issues like danny mentioned of late reveals aren't a concern so they're focused on high performance infrastructure high throughput simulation and uh and uptime and they'll charge for that service in one way or another typically probably through like api uh request fees in terms of builders so builders are doing much more of an

mev job of optimizing block construction so they'll say like based on all the set of transactions i have available to me what is the best way to combine them to create a block that has the maximum amount of value there's multiple different types of builders so some builders might like set constraints and say like i don't do certain type of ordering like anything that could be considered front running is not included in my ordering or a builder could say i only do first in first out ordering so i include transactions only the way that i sort of receive them by by the time that i receive them or they could say like i only want to use the vanilla ordering algorithm that um that exists

within geth today which is like i just ordered my gas price and that's the only thing so builders can order in whatever way that they want um but really sort of the way that they can monetize is if they build a block that's more profitable than the next builder they can charge sort of a fee on top of of that like a spread in the value that they've created by creating a more valuable ordering i kind of see builders as and this analogy does break down but the role of kind of these these high frequency trading firms that have they're always kind of looking for the next strategy they're always looking for a new edge they're always looking to understand the markets in a way that for some amount of time they can make something more profitable happen in trading and other things than others and

builders aren't necessarily well the builders can be traders they're not they can insert their own transactions uh but they're i imagine they're they're these highly specialized entities that are you know have they're doing a lot of research they're doing a lot of like development to get transactions out of the network get transactions out of private sources and stuff and kind of inserting and recombining them in different ways yeah something i'm super excited to to see is in a way this bundling of builder and minor together has limited innovation right like limited innovation in the types of block construction that you can have and the types of products that you can offer users of ethereum by having more sort of control over block construction because miners or mining pools have historically been

focused on doing other things so if you open this up and make participation and block building permissionless and like pure competition how many more features are these builders going to be able to provide to ethereum developers and users to improve you know the usability of the system i think is is a super exciting path of development here yeah this to me this goes into like a similar discussion as we had before which is should we do meth boost outside the protocol or should we go straight to an enshrined version of pbs and i think so if you see if you saw barnaby mono's talk banabe from the robust and sentence group of the ethereum foundation

he gave a great talk at meph day where he talked about exactly about this trade-off and the way that he thinks about it is he wants to first let the market discover what like very good like market outcomes and market structures are like and let these actors kind of sort of find the best way themselves and then let us see how we can pull them into the protocol make them more trustless if possible yeah i tend to agree i think something that will likely there's something that's kind of not necessarily orthogonal but very very related and any sort of l1 scheme that's going to deal with builders in protocol is also going to probably need to address this

and is certainly not addressed in mev boost is the censorship resistance if we claim that builders are going to be this highly specialized entities that are firms that are doing lots of research to optimize and it's kind of hard to maybe break into this market there might be only a handful of them that are really like the strong market participants at any given time and thus there's much more ability to knock on a handful of doors probably to uh censor things and so i think there are there are these constructions that help enhance any sort of pbs whether it's extra protocol or in protocol but certainly when it comes in protocol that try to layer censorship resistance back in even when you assume this kind of external builder actor yeah there's i

think super interesting research to be done here around could you actually build like a censorship market on top of a builder construction right like what does a builder who offers out a transaction type that says you can pay me to censor someone look like and when you when you like take this slot experiment to the limit i think you start to discover some of the beautiful economic properties that blockchains have of censorship over multiple blocks being like prohibitively expensive and that sort of any party that engages in that in a sufficiently competitive market is going to to be doing so at like significant cost so it's i think it's going to be a question mark uh like what

are the mitigation techniques that can be implemented at the protocol level for this it seems like there's like a super promising area of research and the other one is like just economically how how big of you know a cost is there for performing um or performing any sort of sensors right but it's not always about cost right if you have three entities and they live in maybe one jurisdiction cost maybe doesn't matter depending on who's asking for the sistership exactly so in that case right i'd sort of describe building as not being competitive enough right where there isn't the ability for for someone else to come up and and compete but in practice though like it does seem like that is one of the biggest risks like i think there are obviously like you know call it like

blockchain native censorship which is like yeah i you know i bribe these builders to censor these people specifically i doubt that like regulators will engage on that front right like they're not regulators are not going to reach out to builders and give them an east denominated payment to censor you know some country that they sanctioned so literally they'll just make it illegal to have economic interactions with that country for example and even if you do have like decent jurisdictional distribution for builders i do think that's like a major risk because you can imagine a world where like 80 90 of jurisdictions can agree on on something like a sanction for example

and i think that is something that like at the protocol level if builders could be almost like unable to discriminate at that level where it's a very difficult problem yeah you know there's there's probably two avenues one would be that they have to they can't discriminate and they have to like you know utilize the mempool as as it is seen to some extent but then that requires coming to consensus on the mempool yeah there's no medpool right this might not be obvious but like everybody has their own it's funny how we we rehash sort of conversations that have existed like six years ago in like bitcoin cash and like about sort of zero transactions right so

if you could have consensus on the memphis right then you wouldn't need purple rock mining which you can't but you can probably have like kind of these alternative mechanisms where validator a at some prior slot makes a claim about a list of things and and such that list needs to show up at some point in in the next few blocks if not you know the blocks become invalid or something there's there's weird stuff with incentives here because all of a sudden you might if you if you make these constructions uh naively then that list just becomes for sale similar to the rest of the block but i think that there's an interesting design landscape there at least before we get sort of rabbit hole here i wanna zoom out a little bit and then

double click on these concerns that we all have about builder centralization so stefan could you outline what is this risk of builder centralization what are the forces that the economies of scale that sort of drive you know one or like a few large builders possibly to emerge sure so i think block builder centralization is one of the big open questions of mev in in general how much does the like natural economic forces of mev lead towards single entities being sort of winner-take-all in the system and how can you prevent these entities

from having negative impact if they do emerge having a negative impact on the rest of the system so what does a negative impact look like i think you mentioned censorship i think that's definitely one that's on top of mine i think another one is just entrenchment so doing vertical integration up and down the value chain such that it becomes increasingly difficult over time for competition to emerge and sort of take over this this winning position so i published a talk and an article called the mev supply chain which looks at sort of this this value chain from user to validator in a mev context and how it can sort of end up in two

distinct futures one of them which is sort of more dystopian of there being a single block builder that everyone interfaces with to be able to get their transactions mined on chain or more utopian where it's actually a network and sort of a market of independent competitive actors that are all bidding for uh inclusions in one way or another and these are sort of the two potential futures i think we can be headed towards and when we make a lot of these decisions on designing solutions we have sort of a a moral imperative to think about is this decision leading more towards a utopic or more dystopian future does it lead towards more builder

centralization or more builder decentralization in like the short to medium term because a lot of these questions of how the ecosystem develops aren't answered i think this is where social norms come into play and actually play a very big role in picking between these two futures so not only do we want social norms that say like as an ecosystem we will favor solutions that lead towards more decentralization i think that there can also be specific social norms in the type of activity that actors are engaging in that is sort of deemed helpful or less helpful towards decentralization and having conversations about those is super

important one thing that's also affected by the merge is instead of a lottery determining who the next leader in block production will be we now have we not know the next i don't know how many leaders in advance so is there any impact on mev as a result of this certainly i mean people are exploring the domain of multi-block muv and trying to leverage this which if unmitigated or regularly abused can have you know centralization pressures could have block timing you know pressures such that i maybe if i have two blocks in a row i don't release block one until i

release block one and two at the same time um so i think there are some in protocol things that are under exploration for example if that first block was really slate it would just be orphaned under normal considerations well that's a hard problem but uh in the ideal you would you would expect it to maybe be orphaned additionally yeah there's just there's a lot there i think as well with the proposer you can imagine a couple of additional things happening you can imagine one single secret leader election uh but also single secret leader election that does not have the predictability of say an epoch of 32 validators such that it's it's just in time single secret leader election um that might make things more

difficult but then again you know if that block reveal actually reveals the next one then you can abuse it as well so you know then then there's a question of well is it is it not the block that decides but maybe that's it you know there's there's a lot it's kind of uh the implications kind of ripple outward but it's there's a lot in our design tool belt that's being kind of explored to better kind of make sure mev is on on a level playing field that it doesn't disrupt the the kind of m protocol mechanics too too much do you think it's easier for applications to design resistance than multi-block mev than it is for single block mvv because you can imagine a lot of the naive strategies are like you change the oracle price in the last transaction and then you like include your swap in the first

transaction of a block and naive applications might have like some guards across you know don't do too many updates in the same block or something like that so like those feel like they would be easy you just say instead of doing you know one blocks the n where like it's very unlikely that many blocks are controlled but is that like in my view like is it easier to for an app to be resistant to multi-block muv than single block where they can't really control anything at the app level about where the different transactions in the block end up well i think it's always like it's relying on sort of on t-rob oracles for example so an oracle that sort of takes the average weighted price of the last let's say 20 blocks will always be more

robust to manipulation than just relying on you know the last blocks price and i think for any applications that use on-chain oracles at all um i think they have migrated to you know this kind of t-warp oracle that's provided to them by unisoft v3 i think that uh my my view is that that like the sort of this leader transparency and the rise of multiblock me we will definitely put us back in terms of the oracles that we can use on chain because i mean as you said so applications can run a lot of checks before executing actions but they we see in practice and generally you try to avoid even the most basic kinds of checks because this is just stuff that you have to like a math operations that you have to repeat

you have to perform on chain and repeat in every transaction and they really like do blow up your complexity and the kind of gas costs that users have to pay so they love to sort of outsource this to for example unisop which just performs you know that the t-bob operation once per block for every pool instead of every app doing it itself so in practice this is what we have seen and yeah i think it's um tbd uh like how usable this still will be after the merge do you all think that either the fair ordering protocols or some sort of like encrypted and then later revealed transactions like to just eliminate mev do you think those are actually like

fruitful design avenues or research avenues or are there kind of just roadblocks okay i have two takes on this i think the first one is one that i wanted to make earlier but i forgot and i think a cool nuance about meth boost that we didn't get to talk about is that it's actually completely neutral infrastructure that can in the same system house both builders that use fair ordering protocols and builders that use you know an mev auction or a sort of more opinionated mev option that excludes sandwich attacks and so on right and this is all so what math boost does is only provide this decentralized and more or less market for full block templates okay but you're saying that validators might choose to use builders

that have strategies that are fair yes okay because those strategies if i'm just looking at the price okay are likely not going to win yeah yeah yeah yeah so they would be there but if we have known participants and they have reputations and they always follow the strategy i got it okay i was saying that first to say we might actually see in the wild competition between builders who use these different strategies which i think can be really cool but sort of my view on fair ordering protocols in general is they can't eliminate sort of the pressure for mev nor like the existence of mvv itself right so even if you're fully private transactions then there's no front running possible if for example user as user i make a trade i create a

background opportunity because i like unbalance the liquidity pool in d5 then if this cannot be immediately captured by the searcher then these state differences will just accumulate and move to basically top of the block arbitrage in the next block and so what you get is you turn this game of bidding for mev opportunities via the gas price not the gas price like the fee that's paid to minor in the auction exactly you turn it into a latency war that replicates high frequency trading that we see in the traditional financial system and i think we think it's way harder to create a level playing field and low barriers to entry um for participants in the supply chain if they have to compete

on latency instead of price i certainly agree i mean with traditional um you know like new york stock exchange there's a dozen entities that play the high frequency trading game at the highest level and they have their fiber lines measured so they're exactly the same length so they're all playing by the same rules yeah we're not going to get there here and we'll see where it goes so i want to follow up on the conversation about multi-block mev to me multi-block muv is kind of a red herring like there's not that much more that you can do aside from some maybe lowering the cost of oracle manipulation in certain cases from an muv perspective but where things do end up looking quite a bit different i think after the proof of

stake change is sort of perhaps lowering the cost of reorgs like the impact that more frequent short-term reorgs might have on an muv market i think i'm looking specifically here and thinking about some of the other proof-of-stake chains that have sort of emerged over the last couple years and like noticing the frequency of re-orgs in those and like the types of mev strategies that bot operators have been engaging to make sure that they're able to be on sort of the winning reorg of the chain so you know the way that you can think of this strategy is if i control a large pool of the validator network i can propagate like through a subnet for uh attestations from these validators ahead

of at the stations from like other validators and maintain sort of a a second consensus like weighted chain that i can always like swap back to if i need to and have like more votes to exclude the block from some adversary um somewhere on another chain and we've seen these strategies be like extremely effective in other proof of stake chains as flashbacks we sort of made a decision not to build software that enables the development of or the extraction of like reorg based muv because it doesn't seem to be in line with sort of the ethos of a chain that just keeps moving forward and instead makes it move backwards um but i'm wondering how how you all think of this from from you know a protocol development perspective is this sort of

a major risk i think as a general rule there's like a there's many domains of these types of attacks around reorgs and subtleties um that without amyv you might say ah who's gonna do that but with mev all of a sudden you have like very incentivized actors for like each of these types of attacks and so um as you know protocol designers and researchers like these are at the forefront of our minds um you know trying to find the correct constructions and correct balance and parameters that make these hard or costly you know i think right now in the ethereum protocol crypto economic security is not accumulated on a per block basis it's accumulated on a

per epoch basis and so short range reorgs although might be hard unless you control certain portions of the validator set are not costly if you actually do them and so all of a sudden things like um single slot finality which i think is a bit of a misnomer it's more like cumulative cryptoeconomic security on a per block basis you know those become very probably high priority on the kind of two-year time horizon because they do make these things that are now probably much more incentivized to be actually costly which is funny you then have like you still have a bound right if there's a reorg and it's going to cost something then if you have a high enough muv then all of a sudden it's going to become rational to con to induce such a reorg

but without certain mitigations then you do have kind of the flashbots view of it in that it's not in line with the chains ethos and so you have you do have kind of some second order effects protecting you you know if i am a a pretty large uh staker and i am um doing things that are bad whatever that may mean um you know i might be affecting kind of the stability of the chain i might be affecting the kind of the implicit value of the chain and things like that so you do have second order effects helping you but we don't want to rely purely on those on the multi-year time horizon and get to a safer place in light of these types of attacks okay so that concludes our discussion on mev and i actually now want to transition to

probably what many listeners but also us here have been waiting for let's talk about lido so to give a quick introduction lido is a a decentralized staking pool protocol where as a potential sticker you can give them ethereum and they will allocate it to one of currently 22 professional node operators these node operators will stake the ether for you on the beaten chain and you receive what they call a staking derivative called stake if in return and this accumulates the rewards from staking and can later be redeemed for eve which is just basically a representation of if that's being staked on the beacon chain

and while ethereum proof of stake has been designed with making solo staking as easy and accessible as possible the outcome that we have seen in practice is quite a bit different right so lido currently has around 30 i believe of oil steak that that is currently on the beacon chain so i have seen you danny sort of come out with the tweet very recently i'm curious what is your thoughts on lido and how has your thinking on this evolved over the last couple weeks so the the thinking has evolved over many many months not this last couple of weeks i personally believe that such constructions um although do mitigate some of the classic

centralization risk so for example if coinbase had 30 i do believe that there are kind of inherent risks in the pooling of these assets at certain protocol thresholds call it 20 33 50 that kind of stuff um that we're not really acknowledging as existent due to smart contract risk due to governance risk due to well governance is i think a big one especially considering the ratio of the assets staked versus kind of the governance the value of the protoc the token that governs the protocol um also the number that govern the protocol you know then you add in pooled rewards and you add in uh mev and different things i think this

type of protocol becomes kind of a natural stratum for cartelization of the protocol and i would my personal view is that i wish there was better competition in this space 12 months ago which there's not lido's lighters made a great product but by all means like these guys have an exceptional product uh they got in front of all the you know any competitor by 12 months easily but you know i'd hope that there's one better competition in this space two an acknowledgement of the risks that exist by both capital allocators and the people considering protocol because i think there's risk there's two types of risks there's risks to those putting capital into a system at certain thresholds there's also risks

to the protocol i think that individuals allocating capital can will certainly price um and should price the risks of their own capital i think it's much harder you know if if the risks are purely to kind of the protocol and maybe cartelization it's much harder due to the tragedy of the commons for individuals putting capital into to price it but nonetheless i think there are many risks here um i think it's very important for people to discuss the risks it's very important for competition to emerge and um i also think it's important for those in the lido community to kind of consider and acknowledge what has happened in proof-of-work protocols for quite a long time pools have self-restricted you know easily in bitcoin and in ethereum there could have

been pools that uh did exceed the 50 threshold i remember a number of times in the past that you know pool would kind of get there and also there's a big conversation and people are like oh no and then the pool would self-restrict because of the kind of second-order effects on the uh the security of the protocol and the impact of kind of the view of the protocol to just uh hook into the question of like comparing mining pools and staking pools because i think do you think that they're actually comparable because in proof of work the mining pool is the one building the blocks whereas rs and lido sort of uh it's the node operators who all are individually incentivized to make good blocks and light over here yeah so i know i did note that i think in the normal

case that the distribution of node operators underneath such a lighter protocol is is actually it's better than if coinbase had 30 and well i'll get to that in a second the the mining pools are very interesting because the the uh the capital allocation is actually much more liquid you know so if i actually don't like a mining pool i can i can move over much more quickly um that that becomes more possible post-merge but it's not really possible at this point but i do believe that because of the the levers at play the governance lovers because of upgradability which i know is attempted to be minimized over time with lido but because of a white list of operators because of the reputation required to be an operator or not and the governance

levers there i do believe that in the extreme that governance can coerce this capital to operate as in a unified way you know whether it be deep attacks very likely not you know it's probably not like a one-third slashing event through a concerted attack to double finalize and double spend it's probably much more insidious in multi-block mev in short-range reorgs and that kind of stuff that due to the the way the white list may work and due to the kind of pulled rewards and penalties that i think that you you kind of have some insidious things that emerge tim do you have a sort of your own view or would you like subscribe to

to what danny has said i definitely like agree with with the vast majority of that i think the one thing that's like interesting to me as well is if you look at something like lido you know the two biggest problems are one you have unchained governance and then two you have like a white listed operator set uh and unfortunately they kind of balance each other out in that if you make the protocol more trustless there's less things you can do on chain to fix things and actually having like off-chain relationships with operators makes it easier to mitigate issues so imagine a world you know lido completely abandons on-chain governance and then uh there's a fatal flaw in the stf token

right and there's oh if there's only these 22 node operators that you mentioned the lido community can get together with the 22 of them deploy a new contract and like kind of agree that like this is the new sde and the other one is worthless something like that right so you can you can recover some scenarios because you have off-chain trust with the operators but that's another one of the problems that i know is you do have this often off-chain trust so that means that there's another failure mode where it's like it doesn't even have to be malicious to the protocol but it can just be they do not let other operators join the white list because you know they've established like a dominant market position and they make a ton of money and so why split it with more operators right like they could just

choose to restrict that scent so then there's like this other kind of critique where you would you would like some sort of less permissioned way for people to join lido you know maybe having a bond or having metrics whatever but then if you have that and there is an on-chain issue you don't have a guarantee that you can actually you know contact these operators in as efficient in a way so i think this is like the other thing i don't i would add like it's there's like a weird catch 22 there and like yeah if you solve the governance thing and like you make lido completely trust minimized then you're kind of going to rely on the off-chain relationships that you have as much more of a crutch and vice versa and i do believe that lido can be designed iterably and better over time um i don't

i don't believe that all of the risks of centralization can be mitigated nor do i believe that governance can be entirely um eliminated and i do you know say you put in a an upgradeability kind of time bound where if those that didn't want to participate in whatever the new rule set is could have time to exit uh which is which is good you know because in the extreme probably you can't have you know the theft ratio issue where you know a small amount of light of governance tokens and governance holders could steal all of the stake teeth you know can be diminished over time but i do think you don't really eliminate i think in the extreme you don't eliminate you you don't really have the uh the choice that that would imply you

know because if steak teeth becomes systemic then you know minor changes to the the way the rules are going to be operated and minor changes to how things that kind of compound over time you know don't really become reason enough to leave but you still have kind of this kind of extra protocol governance force on maneuvering and steering kind of things um by a small minority yeah on one point i'll add on that as well it's worth noting for this there's this little is multi-chain right like steak eath is the biggest product but there's like a bunch of other chains um and it is still the same ldl governance token that like acts as governance for what happens on these other chains um

the token lives on ethereum but if i was another chain than ethereum like there is this weird incentive misalignment where you know say that like the lino ethereum community wanted to be very us hostile to like the lido avalanche community or something like that they can pass a governance vote on ethereum and affect you know how st avax or whatever is like is managed and as an ethereum person you know i sleep fine at night because lido is mostly like an ethereum product but i think that the solution there is you definitely want some mechanism for individual chains to be able to veto or like block or extend a delay on changes

initiated by ldl holders i think that's actually a design being explored i think the uh intention over time is to study a better a governance model that can incorporate say steak teeth with eath upgrades or staked a whatever that token's called with their types of upgrades which i think makes sense and again i agree i believe that these designs can and should be made better um and i'm excited i know the lido community is like very interested in trust minimization but i think the lido community is very cavalier about the risks involved in the centralization and i challenge others that are getting involved with the protocol to question the proposed lack of risks so let's take a step back i like to play these thought

experiments i like to play them with mev where would we be if there were any flashbots today how would the market have played out otherwise and i think i would like to play the same thought experiment here where would the market be if there wasn't lido so i i think we can start from you know this assumption that users have eve they want to stake their eve and they want to earn yield but like 99 point whatever percent of them are not gonna run their own validator like we even know it would be deeply uneconomical for many reasons if it weren't for math boost if it's not for like pooling of rewards and smooth payout functions and all of these things right so what do you think would have happened

so a couple things there i do believe that you know 99.9 of users are not gonna you know self-stake even if you make it incredibly easy because it's just this is not what they're gonna do but i think it's very important part of the protocol uh to increase the amount of independent operators to increase the amount of independent nodes increase you know centers of resistance in in the extreme and other types of things and so people do that and i think actually if you look at some of the graphs early on likely there was a pretty high portion of kind of solo stakers whether they'd be staking 32 ether it's taking 32 thousand eighth you know even there were some large whales at the beginning but i think that's diminished over time because i

think there's kind of the amount that wanted to do so kind of got in and i think you i think you have some growth here and there i think you have new people come online but there's always going to be demand uh certainly for a significant portion to be professionally operated i definitely agree with that and so then i guess the thought experiments was to point to me point me in the direction of would it be well distributed or would it be naturally increasingly centralized over time and i think that without a lido product that you probably would tend to see centralization and under under a few large operators and that lido does for the amount of capital that shows up

distribute nicely to more operators underneath the hood i do think that based off of conversations and based off of historically what we've seen in different avenues that large exchanges and others that hold large custodians of capital would self-cap and i i think that lido gives us this kind of like middleware where we think it's okay to not when you exceed certain thresholds and so i think that's one of the one of the big difference here differences here is that lido and this this version of reality where lido exists we have a bit of a back door into increasing the capital under one system beyond certain thresholds and i i don't think that you get that as naturally in um in a

non-lighter world i do think in terms of in the thought experiment i think the lighter world or some version of it was probably inevitable because smart contracts exist and and the way that kind of capital can be put under the control of smart contracts here in the system um and so i'm in a lido world and i'm trying to make the best of it and i i hope that there's five more lido's and a lot of uh exploration and um competition system so i think if we like no matter what reality we want to live in right now we we live in a in a reality where light exists and also you know some people believe that the staking market is sort of naturally very centralizing

fortunately we have the ability to build reasonably decentralized like or progressively decentralized staking pools in order to sort of feather that risk which i think otherwise completely depend on what you describe as you know self-regulating or self-capping their stake which even though it has worked for for mining pools i have never like personally i've never been satisfied with that solution like we always knew that for example some of the asian mining pools had exactly the same owners and for most of the like history they probably operated way more than 50 just put a different label on it right and so i think that we should be

genuinely curious and excited for like ways to build like legitimately decentralized taking puts but also i think be like very very about the risk that exists and i think um i think lido has done like a reasonably good job at that sort of always coming out and saying so what are what risks do we personally pose for the protocol and like please hold us accountable and here's how we are currently thinking about uh addressing these i agree when i say attack i don't mean the bad guys you know i i do disagree with some of the approach taken here um i think you know spending millions of dollars a month on uh defy incentives when you've reached 30 percent of state capital is is maybe

a bit nefarious but you know i don't i don't think these are the bad guys i think they've they've attempted to design a very good protocol and attempt to to refine that and be open about some of the risks um over time i also don't think that self-capping is the only solution i do think you know stefan was talking about earlier with flashbots just kind of like the the assessment of risks the ethos of of uh the kind of community and the structure there can have a lot of impact over time and how this capital is allocated but i will also say there are in the event of you say a lido super majority is like in equilibrium

for some amount of time the uh you know the protocol is resilient to being attacked you know the protocol in the event that there's kind of in protocol attacking you know capital can be can be burned and in the event that there's some sort of nefarious activity that cannot be slashed in protocol you know the ethereum community and those users of the protocol which are not just those that hold east can ultimately fork out that attacking capital um you know that's not not something you want to get to but it is you know even similarly to uh centralization concerns that we've talked about for many many years of capital under central operator you know the same recovery modes exist for uh this and it it's actually it's kind

of nice in lighter's case it's extremely attributable so i i have a question about capping so i think right now it's a really bad approach because users of lido are basically locked in right until there's withdrawals on the beacon chain and there is a curveball but obviously that does not accommodate all the volume and it's not like a proper exit and behind the scene you know the the heath is still stinked with the same the same operators um do your views like change about a cap in a world where we do have withdrawals and it's almost a world where then you can abstract it on chain right like you can have like almost like a router which like will you know withdraw from the certain operators and

then move it to another operator yeah so this once we do have withdrawals like does that change your views on on on a cap i think that i think that a um a caps a personal issue within light of governance and those that operate that product i don't think that a cap on on their product really changes in in terms of like post merge or pre-merge i don't i don't see why that is a very important inflection point on what how they'd consider kind of the economics of such a cap but can you explain that again i think uh one possible explanation for why tim's question makes sense would be that like right now uh the switching cost between lido and these different providers is higher so in case lido ever

were to exert undo pressure against its node operators then right now the people staking with lido have no way to escape from the system but after the merge maybe if especially if sort of forced withdrawals are enabled then they could switch you know freely and there's really nothing that lighter could do about it except for the argument that you already gave if state if were sort of the de facto if then they would have to exit into a much smaller token potentially i think the other yeah the other thing that withdrawals change with caps so imagine a cap today there's two ways you can implement it one is you know you block deposits you know to the smart contract like you you just literally revert them uh past an amount

and then what you're saying oh you mean an in protocol cap like l1 no capping no no no a lie no smart contract right like so there's two things you can do one is the lytle mint function stops working right like or the stf mints function and so what you're saying then is only the users who have deposited before the cap can enjoy sort of the you know the service that lido offers which maybe is not the end of the world but it does kind of yeah create is two tiered i i think you yeah you would see a really fun scenario where staked eve would trade at a pretty big premium to eve right yeah that as well but and then the other the other way to implement the cap i want to get in let me in because because lido right now

socializes rewards across people who are also in the deposit queue as i understand it yeah so that's correct so then if you have a cap like the easy way to implement the cap in line though is you probably just leave people on the deposit queue kind of forever but then that's a griefing factor across all the current sd holders because you can just like park your eats there earn yield and deluded for everyone and and you can just kind of keep doing that over time so it's it's like a weird there's just weird situations you end up in yeah it's at least a costly griefing factor you gotta you know someone's but you also get paid though because you also get the you know part of the rewards because

you're going to be yeah so it's like sure but you would argue you could you could put it somewhere else and get more rewards yes yes yes but it's yeah it's an opportunity for in fact not i could call and to the to the the framing of like how a cap maybe is more beneficial now than later because they watch melody later i um i'm not worried about like on the the 12 to 18 month time horizon i i'm not worried about kind of the systemic you know types of attacks here i think it's it's more on the kind of three five ten twenty year time horizon that i think these attacks become increasingly um scary so i don't you know what's interesting oh sorry i mean i could eliminate it now if you're gonna

put the limit in but i'm just saying you know it doesn't really you know the difference between what happens now and in six months from now is important okay i i lose much more sleep over the next one to three years in the next 10 to 30 years i think like the odds that like we make it and this is true also for mev i think like there is like a perfect path you know five years away where we've like fixed these issues and we just have like this this great world and i think there's like a pretty good chance we get there and my fear is just like we just like [ __ ] it up sometime in the next three years that's because of like some bad decision and so yeah i'm much more i like i think we're gonna make it but i

think it's like we're walking this very very narrow road and like one you know bad stuff would just fall off the cliff and i i guess i agree with that sort of risk assessment so maybe i'm just worried about everything all the time what keeps me up at night is like is designing something that kind of works now because the there's the cultural norms or because you know everyone's generally aligned but what really worries me is like if in a few years this isn't what i want to do and i don't want to steward it and like it kind of ossified you know what happens in 10 years what happens in 20 years you know is it actually resilient to to those types of time horizons and the types of attacks and [Music] issues we might see in there so we talked about sort of um this idea

that you know lido could be throttling its its own growth and i think you know as a holder as a holder of stick thief i i think i actually want them you know to keep growing for a very simple reason that the more other people are staking via lido the more useful uh the stake if asset becomes to me right it becomes more liquid gets has finds more adoption in d5 where i can i can use it etc so um in a sense like liquidity begets liquidity yeah that's exactly right and i mean i i have a question though yeah do you think that the second order effects or third or fourth order effects whatever they may be that to the to the

ethereum platform for that to kind of meet its extreme do you think that that keeps eth in a valuable place or do you think that kind of brings systemic risk in that maybe challenges the value of being part of that and obviously you have you have kind of the uh tragedy of the commons as an issue here but just in the extreme what do you think i think you're hitting on the the exactly right point which is that it has very positive first order effects and i would say the effects are actually it's almost like we're creating a completely new asset here it's like if you imagine eve to be like cash today then the idea that you can stake eve and use it the same way that you use eve

today it's almost like sort of the leveling up from cash to a bank account where you just you know you trade in deposits with other people and can transact them seamlessly while also lending them out um you know and the capital gets gets sort of put to use in a productive way in the in the economy so i think it's almost sort of that moment for you here and just as a as a staker i think the first order effect for me you know it's great if that continues and i also want you know more people to sort of enjoy this benefit of having this asset and i want them you know i want to be able to trade with them so i just want to be like i think my clear like i think the first order effect so this are really great

and i think where it gets really very interesting is sort of the second and third order effects which down the road definitely if lido weren't too you know decentralized as we all hope it does then i think definitely then you get into a precarious situation there and one thing i think this is the thing i disagree the most would say you and you wrote this piece with georgios about liquid staking leading monopoly i appreciate the returns to scale like i get the argument that obviously you want light out to be on d5 you want you even if it's not on d5 you want to have the stf token have high volume and stuff like you want liquidity and that has a ton of positive reinforcing cycles um

what's not clear to me is you know when do those stop because it kind of feels like if you have an asset that's you know say tens of billions or hundreds of billions of dollars in market cap like you have liquidity for all intents and purposes right like for most people and so i a world where say there's like force taking derivative with like a 10 to 100 billion dollar market cap or something seems like one where like you do get those benefits but you also don't have the systemic risk so yeah i'm curious how you think about that because it's like i i get there there's like a level why you want liquidity but past a certain point it's like the risks to the protocol might outweigh the additional benefits and

as an end user like today i i like to think of it as stable coins you know usdc or usdt or dye all have like deep liquidity and you can trade in and out of them and and use defy with them right yeah that's right i mean i think it's it's still an open question whether um you know these debit coins can survive sort of in the state that they are or if there will be you know some more like centralizing outcome among them you know i'm optimistic that there are designs alternative designs not just alternative designs but but market valuable you know designs that are valuable to the market that lido has not explored you know and wouldn't be covered by their construction and that you know in the coming year

that we see a number of these pop up and uh maybe give a lot of run for the money such that you do have you know like there's a reason die exists there's a reason the usdc exists you know and there's and there's a reason like i think well yeah if you need more money you put more money obviously no no but what i mean is if you take like usdc and tether they have very different risk profiles but it's like great that both of them exist right um and i guess i don't know enough about usd recently i would not sleep well at night if like usdc was the only stable coin on ethereum like that would feel like a systemic risk to me and nothing against you know them in particular but it's like yeah i like that we have different

ones with decorated risk profiles yeah yeah i think that's valuable i mean i think it's systemically valuable i hope that it's valuable end users i have two follow-up questions to this first one for danny so what kind of designs do you think we haven't explored in the centralized stacking pool well one's very obvious and it's not decentralized you know it's the usdc of uh staked assets some sort of large centralized provider i'm not saying that's a good thing i don't know but like coinbase has that right it's just you cannot withdraw it from the exchange oh no on chain oh i see if you could withdraw basically if you could withdraw your staked east whether it's at coinbase or another place so you get the sde but there's no ldl token

it's like managed by a company right i'm not claiming that that's necessarily a good thing i'm claiming that that might be valuable to some users uh in a different way to stake these and i would be surprised if we didn't see something like that i do think that you can you know the the rocket pool collateralized uh version that just does not have the types of uh kind of network effects like you just can't scale that as quickly so that i don't think that's really competitive here um i do think that there is a new tool coming in the wheelhouse that i'm sure lida will be exploring but i think there's going to be a lot of like interesting things to explore for new companies as well getting in and that's dbt distributed validator technology just to more kind of trustlessly combine notar operators potentially lower the

barrier entry um obviously that can be leveraged by lido but i think that kind of marks an inflection point where a bunch of new people can maybe jump into the game to play with new things um but i you know surprise me i i think that there are there are very likely some things in here that um i'm not keen to and that the market will kind of play with over the next year or so i want to race like another build a bridge to something we already talked about which is proposal builder separation right so we have discussed that lido like the risk that lido could exert pressure on its node operators to you know include specific payloads which might be bad for the ethereum chain so i'm

curious like both staff and like you two as well how you think about you know lido outsourcing its entire block building to a decentralized block building market like math boost or pbs and if this would sort of alleviate some of your concerns i think that it is i think it's likely that large operators or large consortiums of operators actually become kind of mev powerhouses in and of themselves so i think that if they outsource naively then in the normal case maybe you don't have kind of the the cartelization and the maybe inter-block mbv or short-range reorgs that maybe having that much

capital would incentivize but i think a large enough actor is incentivized to do one of two things to have either special relationships with such a market where they're saying i'm the next three proposers help me out or to be large enough for them to not need the market and to do the mev research and the searching and stuff in the context of their very special situation you know and i think i think that that's one of the reasons that mev boost is very important is such that even if even a 30 operator you know isn't going to get outsized opportunity with respect to you know solo at home operator but in the extreme

you pass certain protocol thresholds especially you know 50 the the solo operator with the opportunities that begin to look kind of like attacks you know short-run resorts and stuff the solar operator even with a sufficient market isn't going to be able to participate unless they join the cartel of course is this really true stefan could we build multi-block mev into flashbots certainly the answer is certainly is this an information hazard question so you're saying it's a moral imperative in a sense that a builder gets big enough that they can extract multi-block mev and make that available to all validators i'll put it this way i'll say like if the blockchain requires like

the majority or like a cartel of validators to be benevolent for it to like continue moving forward it's probably not a blockchain anymore so it probably is not a good outcome to have like software that that requires that you know if we actually think okay there's only a bounded risk of short-term reorgs and we have epoch finalization anyways maybe it's fine maybe like we want to stress that's the system and see what happens if there is infrastructure for performing short-term reorg at scale that is inclusive um so that allows even solo validators to participate in the reorg game i think

that is like a design space that that could be explored well that's that's already a step further right from just multiblock mvb which is just forward looking cartels as a service yeah in a sense yeah i mean it could be yeah the market trends towards catalyzation for example between mining puts and searches then the only way that you can you know don't what you can do is like make this cutter public right and make it make it so that anyone can join it and make it accessible to all participants both up and down stream of the supply chain and i think uh this is actually the right way of thinking about what flashbots has done so far right and there's you know i'll go back to some of the in-protocol stuff at a certain point block slot which is kind of um

you know further incentivizes blocks arriving on time and single slot finale or single slot crypto economic accumulation these types of things i think on the two three five year time horizon are very important refinements the protocol to reduce the impact of cartelization as a service you know and the impact to users ultimately like you know that's that this is the thing for users and if you regularly have five block reorgs to optimize profits for those running validators whether they're in a cartel or running at home um you know it's important to that has an impact on users and that's

yeah that's the reason to remember that we would continue to refine and harden these things so we talked about mev and flashbacks and map boost we talked about lido and staking so i think i want to like for the final chapter take a step back and see so both flashbots and lido is kind of in your trend right it's protocol external projects that are playing very important roles in the protocol that are both very adjacent to sort of the the core infrastructure of the protocol and the and very important for its security both in the short and long term as the ethereum foundation how do you think about the existence or the rise of these protocols and how do you manage your

interactions with them we are looking for solutions uh that help make the protocol more stable more robust and ultimately uh valuable to kind of the end users and communities that use these so when we're thinking about say interacting with flashbots the ef isn't thinking about okay what's flashbot's revenue model is this going to help them or hurt them it's more of can we collaborate to find better solutions that help enable the markets that help stabilize the protocols and things like that and we found especially with flashbots that there's actually been a lot of really nice overlap in the goals here right you know i think flashbuck does want an open market

flashbacks does want a stable protocol and so in doing so when we're exploring both extra protocol designs like map boost or in protocol designs like enshrined pbs um you know we've been able to kind of collaborate very fruitfully it really kind of depends on the alignment of the project with the ethos of kind of stabilizing hardening decentralizing ethereum and if we generally have alignment there we usually can figure out how to collaborate yeah and one thing i'll also add is um beyond the ef the other like really critical people are like the client teams themselves um which again in case people are not aware there's many of them the vast majority every single one of them except geth is not within the ef and having them interact as well is like

a really valuable thing so i remember when like flashbots kind of first started off thomas from nethermind thomas is the founder of nethermine and he started getting involved in flashbots i had like a long call with him and like he had like reviewed like the mvp get codebase and that made me sleep better at that i was like you know if like nevermind is generally you know on board with this and happy and like and at first the concerns were a bit different it was like is there something in flashbots or like an enemy geth that could create like a really bad block in geth and like you know cause like a consensus bug on the protocol so it was much more i guess basic concerns in that way but having having thomas you know kind of review that and be part of it and like be involved in the design like

i think it was really valuable so i think that's like the other bit as well is like assuming these teams you know genuinely do want to collaborate and like build like a a good solution putting them in touch with like the the teams who write the protocol software is really helpful and a bunch of people are in touch with like flashbacks now for mbv boost the lido team has reached out to a bunch of client teams also to get some feedback and and potentially run some nodes so i think like kind of broadening the set of people to interact with so make sure it's not just like ef researchers but also like all the different teams who who built the infrastructure and this goes beyond you know also it's not just like a flashbacks and vital thing like i mean we talk with infer we talk with coinbase

we talk you know like or i don't know foundry like when these like infrastructure providers want to interact like you know just i think putting them in touch with the folks who like build the protocol and and you do kind of want to do some screening like because you don't want to like dos the people working on the protocol uh and this is where just making sure these teams are kind of aligned with what we're trying to build is valuable and then when they're not communicating why we think that and like you know there have been projects in the past that have built like protocol adjacent things that have some weird incentives and we've communicated why we thought it was a bad idea what we thought was broken making sure we we modulate the amount of interaction

based on how aligned we are and how valuable it is for protocol devs and researchers to spend time on this yeah it's like given that it's an open source ecosystem and it's a very open project you know everyone has a seat at the table you know the ef has certain things that we're attempting to do um and when we're aligned with any sort of ecosystem actor we're usually very happy to kind of work with them towards common goals should i talk about the other side of the table absolutely so i've been building on ethereum since like 2017 and i've seen like this playbook happen a lot of time i think the position of credible neutrality that the ef and like all ef researchers and builders are in

can be like at some times comical but at some other times like absolutely necessary so on the comical side there's like an obsession with like renaming things like anything that is like produced externally is going to get like for sure renamed before it gets any sort of uh rubber stamp from what's an example what's an example i think pbs is probably like an example of that but but there's like millions of others even at like the standards level everything gets like renamed a million times but that's just like part of the the joy of of collaborating in sort of an open source manner um i think what ultimately makes it really fun to work with the ef is just the intellectual honesty and the

commitment to like principles that they've laid out for the organization where you know if if you're working more in an industry setting and looking at partnerships and collaboration everything sort of ends up reducing down to like some type of like negotiation over power and economics and that just doesn't happen with the ef like it's always just negotiating on like the technical like what is actually the best way to achieve technical objectives which really makes it like a joy from like a design perspective there's obviously a lot of open-mindedness i think where in place that there has been frustration in the past for various different projects has

been that because you know it's not the ef is not like an industrial body it is like a bunch of like individuals that are collaborating on building things it can seem difficult to like engage with it can seem difficult to access and know who to talk to like how to build the relationships to be able to have you know to be in the conversations and be at the table and sometimes i imagine you have a conversation with party a and then like you have to rehab the conversation with party b because they're not communicating they don't know what's going on and they might have different views on things yeah the ef kind of for better for worse mirrors a lot of kind of the open source ecosystem it's a bunch of different teams working in

different projects that have certainly common goals that it's at its core but uh it's it's not like when you when you interact with an individual or team at the ef that you're interacting with the ef necessarily at large and almost in in a lot of cases team is a bit strong of a word like you can't imagine you know having a conversation with danny and one with justin drake and what with dankrad and getting three different with that group you're definitely gonna get three different answers yeah yeah um and so you know i sympathize with like if you're an external project coming in and you're like you know daddy then crowd justin what should we do and they tell you three different things and you're trying to figure out what the

ef thinks it'll be hard um so i i do think like high level obviously in terms of values and like vision for the protocol there's like a ton of alignment but it's also probably a bit hard given that there's not like a there's not like a party line on every single topic right like we do have a bunch of independent people who like think about this stuff and come to different conclusions fantastic i think this has been such a great discussion thank you so much danny and tim for participating and also stefan for you i think this has been your first podcast appearance yeah thank you so much for having me incredible we are so glad that uh yeah you chose i'm coming coy oh that's fun that's the the place to to make your splash your day fantastic so

uh yeah guys um have a good evening yeah i really appreciate it thanks for having us again yeah thank you both both parts were great bye-bye talk to you later [Music]