Ethereum Cat Herders PEEPanEIP, 27 Mar 2023

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: PEEPanEIP #103: MEV Boost with Alex Stokes

Description

Topics covered
--------------------------
01:10 - Guest's Introduction
01:59 - Presentation
02:14 - What is MEV?
08:58 - Why do we Care?
13:55 - MEV on Ethereum.
15:06 - Transaction Supply Chain - @thegostep
15:29 - MEV on Ethereum.
18:01 - PBS
18:19 - What is MEV-Boost?
18:35 - How do we implement PBS?
20:23 - MEV Boost- Network & Software
21:51 - How would it work?
22:25 - There's a catch
22:57 - MEV Stealing
24:04 - How to stop MEV Stealing?
24:46 - Democratization with MEV-Boost
27:27 - Enter the relay
28:59 - MEV-Boost Post Merge?
33:44 - Downside of using relay model
35:01 - PBS: What's Next?
35:19 - Longer-term for PBS
38:20 - Near-term for PBS
42:25 - Immediate-term for MEV-Boost
45:52 - What are the advantages and disadvantages of connecting to the multiple MEV Boost?
47:41 - Significance of relayer diversity
48:43 - Alex's thoughts on the Client diversity suggestion
53:01 - Any guidelines for solo & institutional stakers?

Follow at Twitter
---------------------------
Alex Stokes @ralexstokes | Pooja Ranjan @poojaranjan19

Resources:
-----------------
Slides - https://docs.google.com/presentation/d/1c4fwF94NG_ppXlTOJ5gcH7ilsRgwgydS0MyBDdKLhMs/edit?usp=sharing

MEV boost #shorts https://youtube.com/shorts/Vhff81Mxres

PEEPanEIP - https://www.youtube.com/playlist?list=PL4cwHXAawZxqu0PKKyMzG_3BJV_xZTi1F

MEV Playlist - https://www.youtube.com/playlist?list=PL4cwHXAawZxqntSgAaYHrV3NjAUMyzk7X

Shapella - https://www.youtube.com/playlist?list=PL4cwHXAawZxpok0smGmq-dFGVHQzW84a2

Questions for Guests - https://docs.google.com/forms/d/e/1FAIpQLSdsmujnH_ncIlSEJ23s9jzXAKwfpTGNeJqjREWPloBbkJVQTA/viewform

Check out upcoming EIPs in Peep an EIP series at https://github.com/ethereum-cat-herders/PM/projects/2

Contact Ethereum Cat Herders
---------------------------------------------------
Discord: https://discord.io/ethereumcatherders
Twitter: https://twitter.com/EthCatHerders
Medium: https://medium.com/ethereum-cat-herders
Website: https://www.ethereumcatherders.com/

A

Mev boost is a popular term among the ethereum validators community right before the merge ethereum developers engaged in the conversation of proposal, Builder separation to mitigate issues that could have happened had they been not separated. If you are interested in diving, deeper stay with us.

A

Welcome to paneer episode, 103 I am Puja ranjan with a special guest with me to talk on one of the most discussed topics in the Ethereal ecosystem in the next hour, or so we will have an overview of any boost. It's working and it's significant for the ethereum validators nav boost is an implementation of proposal. Builder separation, built by flashpot for proof of stake, ethereum to take us deeper into the woods of Mev boost. We are joined by Alex Stokes welcome, Alex.

B

A

Alex Stope is a researcher at ethereum Foundation. He has contributed in multiple ethereum related research work, it's highly unlikely, but for people who aren't aware, Alex has been graciously taking out time to share ethereum research work with the community. In the past he explained accounting reform with Altair upgrade eip-44 and the most popular EIP of shapeela upgrade EIP 4895 Beacon chain push withdrawal if you missed any of his earlier talks, check out p penny playlist moving on today's topic with no further delay. Let's peep in.

B

Great thanks, let me share a presentation. I have great okay yeah, so this peepinip is on my Boost. This is something I think a lot of people have had a lot of questions about there's like quite some drama.

B

Let's say around like the time of the merge- and you know now: it's like some months later, it's like a good time to check in and see where things are at and where they're heading so I'll start with, like some intro material, just about like what is this Mev thing uh give like a little example and then we'll like kind of drift more into PBS and that boost and all that. So you know we have this question everyone's talking about this stuff. What is Mev, uh you know initially was an acronym, so minor, extractable value.

B

The idea is that there are miners. There are people who are making blocks in our blockchain protocols, and it turns out that you know at least with the way that we construct blocks today. There's this degree of Freedom a block producer like a minor or a validator half and that degree of freedom is to like arbitrarily build the contents of each block right. So we have a blockchain there's blocks in the chain. Every so often there's a new block.

B

While this block producer actor is the one making the next block in the chain every so often- and you know they- basically they start to follow some rules of the protocol right.

B

It's a valid block um on ethereum right it like has like validity, VM execution, there's rules you have to follow, but there are some rules that just aren't specified, for example the exact sequence of transactions, and so what that means is you have things in the public pool and like other sources that we'll talk about later and then whoever is making the next block gets together from all these sources, and then they get to pick exactly the sequence of these things right so like rather than you know.

B

If I have like two transactions, A and B, you could imagine like a because before b- or you can imagine, B goes before a and like the whole point of the blockchain is to like sequence, these things um yeah, and so it turns out.

B

This is like actually a very like uh huge power and at least this whole concept of Mev, um because as I'll show in a second, it turns out that, like it can actually matter quite a lot, uh the sequence you know, the actual, the actual ordering of these transactions, so yeah I think I pretty much covered this stuff on this slide and here's an example right.

B

So everyone hears about sandwiches, uh the reason why is because they basically give you worse execution as a user of a decks on chain, for example, you know swap so let's look at this. um Let's say uh we want to trade, you know usdc and eighth right, so we have like our stable coin.

B

um Let's say we have like 16 56 of them, and we want to trade that for one eighth- and it happens to be the current, you know price and to meet a swamp pool. So because of the way these things work. uh There's this extra piece of information that goes with the universe. With your unit swap trade called the slippage, um it's usually a very small percent of the trade um in the example.

B

Here you know it's like half a percent, and what this means is that, basically, you know I'll accept you know around this price of 1656 I'll accept half a percent basically either way, and the reason we do this is because, like from the time I like can look at the chain right, like sort of get the head of the chain and see the state didn't get to stay, this unisaw pool if I want to interact with this decks there's. This is like this process, where I have to make the transaction.

B

The transaction has ended up on chain and that's going to be some time after I can look at the states right. This is just how blockchains work like this is.

B

In fact, like the whole point, um the reason we care about blockchains is because they order things for us and in order to do that, you have to have the things before you can order them and that's basically What's Happening, Here and so with examples like uh you know, decentralized exchanges again, like you know, swap it turns out that if you, for example, have some price uh the price can move before you can.

B

You know before you have your next uh enter your next opportunity to interact with the application, and so, if you had no slippage, then what happens is that basically the price can move against you and then your trade would like not be at the 1656, for example, and basically your trade would just revert or not go through, and so the idea right is I want to give you a little tolerance on your trade so that the order actually goes through um and now the question is like how much slippage do I set, and you know, there's a whole science and art to this.

B

You could set some like huge slippage, but then what will happen is well we'll see what will happen. So, let's assume you know, example here we've set like a small slippage and basically what it's just saying is that you know I would like to uh only pay 1656 usdc but I'm willing to accept you know paying up to 1664 and some change.

B

This is the the half a percent um and you know ideally I want to spend as little as possible, and so the way to think about this is that you know I'm accepting some worse execution and again. This is this trade-off between you know how quickly my trade can settle versus. uh You know how much I pay for it. So um you know, let's imagine that you know we look at this unisot pool at the end of some block and we see the state and we look at the state of the stacks.

B

You know this pool in new Swap and it's exactly these numbers um and then we can be the first next transaction that gets into the buck. Well, no one has interacted with the pool, so this is still essentially the offer right. The quotes that unisoft gives us, and so as long as this is the case, we make the trade and great we didn't even have to pay any anything extra.

B

So like I've been saying, though, there's this time in between when we look at the chain and when we can next interact with it- and this is just like a fundamental fact of these types of systems, and so now this is where some AV comes in, because the block producer is probably someone else, not just the unit swap user and they're figuring out how to assemble the block, and this is now openness.

B

You know this now opens our trade up to this thing called a sandwich attack um I, put attacking quotes because now this gets into like a whole different conversation around like the ethics of mvv and all of this um different conversation point being uh the sandwich thing goes as follows: someone an attacker uh what we usually call them Searchers. They see my trade because I just gossip it to the pimple or something- and this is a public pool. Anyone can watch I, see my chair. They see my slippage.

B

They basically put a trade in front of mine to drive the price up up until my my slippage, tolerance right. So basically they drive me to get like the worst execution while still letting my trade go through. They do that and then they do what's called a back run where basically, they put a transaction after mine, and then they essentially just sell the tokens they just bought for a profit.

B

um The thing here is that basically they're, essentially taking you know this like execution price, so to speak. That I would have otherwise saved for themselves is like the way to think about it. So that's not great um and has led to like a bunch of products that we might like talk about later. Flashbots has this like map protect thing, which essentially is a private mempool, so that this just can't happen um anyway?

B

This is the idea, and hopefully it illustrates the fact that, like the fact that you know these, um the sequence of these transactions is like very important and can lead to these. Like word consequences, when you have you know some application on chain interacting with the real world um yeah.

B

So you know zooming out a bit like. Why do we care about this and maybe stuff at all, and it turns out that, like most and like, maybe all because they get a different philosophical conversation, but maybe all interactions uh on chain expose, it may be in some way, and the question obviously then is like what do you do about it right so like with the unisol example? It's like. Do we just let the speed like? Can we think of ways to like change how the the application works on chain?

B

It opens up this whole design space. You know, medve, aware, protocol design, saying okay right, like we can still make our decentralized exchange on chain, but uh maybe we're aware of this, and we can somehow like use that as like a force for Grid in our application. So in the decentralized exchange situation, there's you know ways to think about essentially trying to recapture this Mev. That's otherwise leaked um and I'll say a bit more about that in a second.

B

But the point is: is that this is like a source of like very powerful incentives and again this is all super early, but I think what we're coming around to is like a space is understanding that, like Mev, is a thing and that we can actually use it. uh You know in our in our own application designs, which is like really exciting, so that was like maybe Broadway, there's Meb I mean basically everywhere.

B

um You know different blockchain systems, you can even say there's a movie in like traditional systems, again a different philosophical conversation, but we'll look at it today. So this is a pretty nice image that was made by the ghost app and essentially it's the transaction supply chain, like especially with respect to like how this Meb stuff works.

B

So on the left, you have a user they're like our unit swap Trader. They have some intent like you know, I want to swap some tokens for some other tokens. They usually interact with, like a wallet, a user. They you know want to do something they want to like swaps with tokens or some other tokens. They usually interact with the wallet which basically takes like their notion that they want to do this. um It basically makes a like protocolological thing like a transaction.

B

This then, is like gossiped or maybe it goes to like some private node pool. um We now have these actors called Searchers, and these are these like very specialized, almost power users of the chain. Who can do these types of like Mev type extractions, and you know partly this is because these types of opportunities become so specialized that uh you know it just makes sense for Searchers again to like specialize themselves, or rather I should say the opportunities become so competitive. It makes sense for Searchers to like specialize into these fields.

B

So you have this like distinct actor. We're then seeing the emergence of this thing called a builder. This is where the PBS comes in.

B

You have a builder, and basically the idea is the Builder is the one who's making a block, uh but they don't necessarily specialize in you know liquidations or, like you know, Dex Arbitrage or things like this, and so then they usually work with Searchers or, like a source of you, know, bundle flow or order flow to make the full block builders in need to work with validators, who are actually the ones putting blocks on Chen they're, like the the block producer in this scenario, so you have this like pretty rich landscape already and like one way to think about.

B

This is that you have like value kind of flowing across the whole stack, and the question is like how much value lands where which time you know. We would probably like to see a world where most of the values like push to the edges, so that you know, if users do something and there's some value attached to that action. You know they recapture most of it separately. I think we can argue that, like validators should also be sufficiently incentivized, because it adds to protocol security.

B

um So you could imagine like a lot of it like sort of accumulated the edges. That being said, you know, as like the market probably is like more immature. You could see things happen all over the place um within this like categorization, so like, for example, you could imagine, there's like cartels almost or like you know, just coordinated groups of Searchers or Builders and what they end up doing is basically saying hey rather than like bid. All the opportunity to the validator to you know ensure my inclusion on chain.

B

I could just agree to only bet half of whatever the opportunity size is, and even if there are many competing Builders as long as they all agree to the same sort of limits, then like they as a group like profits at the expense of say the other group. So that being said, uh I just want to paint this picture that it's like already quite a rich landscape and it's still pretty early. So this is just kind of to like drive from kind of things. I've been saying like Mev.

B

Is this very centralizing Force um in Parts? Because you can imagine going back here?

B

If you as a validator or like even a staking pool so like you, have like a good chunk of sort of the protocol stake, you can imagine vertically integrating with all these other actors like, for example, Searchers of Builders, and the thing about this, is that now you as a staking Pooler as a validator, you become more profitable and when you do that, you can basically reinvest and how efficient you are at building at searching and it becomes the Spy wheel and the reason this is not good is because it harms the centralization of the protocol, because you can imagine, then that you offer better returns as a sticking pool.

B

So everyone goes with your sneaking pool over some other. um You can just like directly reinvest so to speak into your staking operation and just like grow more values. It's not great.

B

So this was the motivation behind PBS and the idea is essentially well what it says so composer, Builder separation and what we want to do is like fix some way within the protocol to basically put the firewall in between proposers all evaluators and builders in this web ecosystem.

B

um That being said right, it's this, like very large design. Space I'll touch on this a little bit later, and we want to be kind of thoughtful about like how we do this, because if we like actually enshrine something in the protocol, it's like well one very hard to do do very hard to change, and you know there are consequences for getting it wrong right like if we designed something that like either is infused or like has unintended consequences.

B

That's gonna like just cause another negative sort of externality that we're trying to fix in the first place, or maybe we fix one but cause another right. So we don't want to rush into this and like what can we do about that? Well, this is essentially the idea behind that boost is rather than have this thing happen. You know in the protocol today we could just address this off chain in the meantime, and then now we have like a little breathing room to figure out. You know what is PBS and the protocol look like.

B

How do we enshrine it and answer all these questions? So I have this little graphic here and it's just to like Drive the points home. uh You have validators there's just like this firewall line, and then everyone else and the idea is that regardless, what happens you know out in this, like other part of the ecosystem? As long as we have like this, like nice subtraction of PBS, then uh these questions won't influence about our stats to centralize in the way I was describing a second ago.

B

So the rest of this talk will be about that boost and PBS, and what's going on there, I did just want to call out uh there's a lot going on all at all parts of the stock. In terms of like you know, user Innovation and like well, Innovation things with, like you know, searching and building and having that be done in more Justice fashion. There's you know I've every application of ethereum can ask themselves well. How do we minimize Mev? Can we like use it somehow in our protocol for good things?

B

uh To give you an example, there right, like the fact of like liquidations on all these lending protocols like maker like this, is it may be right, you're, basically like there's just like permissionless bribe to say? Okay, if there's bad debts like someone come and liquidate that Vault, for example, in maker um and so yeah, it's not the stuff somewhere strictly bad. It's just like a thing that you can hopefully use to your benefit.

B

I have some names here, so flashbots is working on their solution to this like broadly called Suave. It's just like all, encompassing uh crypto economic system to like, essentially have like these intents or transactions boil it down to, like um you, know, amab Utopia and this Talk's, not about Suave, but I'll. Just call it out. It's very interesting if you Google around you'll, find more information, there's other applications.

B

Dude thinking about this stuff so like rook and Cal, swap are different trading applications, and these are interesting because they're trying to like address this question of like map redistribution or like it maybe capture for our for for users of the system. um It's like Cal stops pretty cool.

B

What they try to do is basically clear, all trades off chain before they even land on chain so like, for example, if there's like a bunch of I, don't know, let's say like 100, people want to make like a thousand different individual trades for, like the same token pair.

B

If they happen to like cancel out, the cow is like coincidence at once so like if they happen to like cancel out in this way like if they can match the trades, uh Buy sell, Buy sell you do that before you even need to touch the chain at all, um then they have like liquidity. Buffers like on chain to trade against and the point being is: basically, you want to essentially strip off all the maybe you can before these things even are exposed to like a you, know the chain setting um and then from there.

B

If it like works well, you know so I mean there's like good sort of trade volume. Then you can imagine, like a lot of the trades, are settled off chain and then only like a very small number are settled on chain um where they still could be exposed to Mev. But you've had this like huge minimization step.

B

So anyway, again, that's like a whole different talk to explore landscape there, we'll now talk about PBS right. So PBS is like this big thing and it's like not clear how to like exactly you know, carve into the protocol. So instead uh we'll give ourselves some, like r d time to think about how to do and shine PBS well and in the meantime, we'll have this off-chain system called Med boost.

B

So uh what is it essentially? What we've been talking about this like off-chain implementation uh secured by flashbots, in collaboration with, like other ethereum researchers and developers, there's like a landing page here at spoof, stuff awesome.net. If you want to check that out so yeah, like stepping back we'll like walk through now, serve from like first principles. How would we even like think about building PBS in the first place and we'll see like map boost? Becomes this thing so again, PBS proposer, Builder separation like what's actually going on here.

B

You have proposers who may want to sell their block space and Builders who want to buy box base right. So, like that's kind of what's happening, is like proposers on their own. We should assume they're, just like you know, fairly naive, unsophisticated actor, and we want that. So there isn't. This, like. You, know, sort of this like point of Leverage for for centralization.

B

So you have this like fairly unsophisticated actor, who still realizes they have a valuable asset, namely like box base on chain, and so what can they do? Is they can sell it to builders? Who are these sophisticated actors? um And presumably you know pay them for their efforts and all this um and yeah. So you have this like market now, where you have buyers and sellers and now there's a little graphic.

B

The idea is that you have blocks in the chain and again this is like merged ethereum, so the blue, the blue bit, is supposed to represent, like the the consensus wrapper and you have the execution payload inside the execution bit. I have this little gold border and that's just to communicate.

B

There's like Mev Mev is this very, like spiky variable thing from block to block, and so there's like different sized supporters here to communicate that, and now we come to like the next block in the chain where proposer is like: okay, I have this thing: how can I? You know what am I going to basically put in there from actually payload um and I may want to work with an external Builder who's like specialized in some way at making a very valuable payload.

B

You know compared to what I could do on my own, so zooming out again, you know, proposers block Builders, there's like networks of them, and it's like. How are we even gonna like connect them? um I, don't know, and it turns out net boost is the answer to this thing. So it means like a couple different things. One is like this network because, basically, as we'll see we're just gonna, have this thing act as like: a rendezvous between uh block Builders and proposers, there's a protocol for them to communicate.

B

So you have this network, um you know today. It has a first price open bid, auction mechanism, so what that means is that proposers can sell their block space, but basically it goes to the highest bid. Bids are public and all this, and otherwise it's unopinionated um and yeah. That's the auction bit is interesting as we'll see because a big question of PBS is like what exactly does this auction look like and already I think we've seen some interesting data from like my Boost deployed that helps inform that question.

B

So that being said, there's also no boost, like the software project, which implements this like communication protocol that we're about to look at and if you want to you know, dig.

A

B

Sum there's the Builder specs, which sort of specify this this interaction, there's also the map, is repo and yeah. These links are here.

B

So right, when thinking about boost as this Rendezvous service, there's block Builders talking to proposers proposers, have this block space. They want to sell over on the right and block Builders. Have this thing they want to essentially buy like they built an extra payload. They need to buy the block space from the composer, so Med business is going to facilitate getting these execution payloads to the right proposal at the right time.

B

So, like again, how would it work um you can imagine each slots? A proposer basically just asks this network. It says: hey uh I'm, the proposer in this slot. Give me the bids, like you know, there's many Builders tell me like the best you could do. The proposal then picks the highest bid, presumably, and they include that in their own block.

B

That's what we have here. There's like different payloads, uh the one that's in the middle there is like you know. The idea is that it's the most valuable with respect to what it can pay the proposer um with this like very big gold border and that's what the proposal then puts into their block.

B

So there's a catch though, um and it's because ethereum is decentralized. uh We want to encourage snakers of, like all size in the valley are set, and so because of this you know anyone can become a validator because again, there's like a open permission set of these things, and they could include the contents that the Builder tries to give them and put them into their own proposals.

B

um And the reason this is a problem is because at least this concept of Mev, stealing.

B

Sorry, I just lost my controls there we go um right, so the problem here is the following: that you have this payload, it's on the left. Here we see like our payload. You know it has again this like nice, big gold border- and you know the way I think about this- is that Builders do actually provide a service, and this like refinement of the Box space, so like it's pretty natural to see in a Marketplace that, like they take some composition for their work, sort of like a tip um and pass on.

B

You know the rest of the value to the proposer again as like an incentivization or the bit. The thing is, though, nothing really is on chain.

B

um There's nothing really stopping like nothing's really been committed to at this stage, and so what then a validator or proposer could do is they could just basically take your work for the Builder basically take out this tip part and just take it off themselves and that's what this image is supposed to describe again, you have a situation on the left where, like there's like some, you know smaller amount going to the Builder, a you know quote: evil validator, you could basically just say thanks for the payload I'm going to like take out the you know the payment to yourself and just take everything in the block, and so that is already like an immediate problem with uh with this model so like.

B

How do we stop that? Well, the reason this came up was because this that was decentralized so like we could just say, Well only let Justin valerians do this, but then you again have this problem where you're like putting together this, like you know, specialized like special or like this enshrined matter set it's like almost cartel at actors who like have access to the special thing, and ideally, we went to like provide access to any validator regardless again of, like you know, stasher status reputation any of these things.

B

So then that kind of leave this to option too, which is like we have to like, remove the conditions that, like lead to the stress, assumption that we trust validators not to like steal Mev from builders.

B

And the way we're going to do that is kind of add this like commit reveal step. So the point being is that, in order for a like successful, you know I'll call it a PBS transaction or execution for this like whole thing to happen per slot uh for, like one, you know, one contract so to say between proposer and Builder to like happen.

B

Basically, the Builder gives enough information that the proposer can commit to it and then once they commit to the thing they reveal the rest of the block and they do it in a way where essentially Builders only reveal their their part of the bargain once they know the proposal has committed to going with that, and only that thing, so they basically lock in the fact that Builders will, you know, receive whatever they deem appropriate for, like their payment for their work and then validators you know, will take what's left over.

B

So there's the catch here with this commit reveal the commit. The the reveal part is very important and, like Builders, have to do this so again, there's like another little graphic I have this again. This is our same extension payload with like the green and the yellow, and the idea is that it's not blinded. Somehow with this like opaque rectangle on top, the builder then says: okay, so proposer says hey. Give me give me like I, think I can commit to the Builder sends over the splendid thing.

B

The proposer includes that in their block- and so it turns out at this point- uh there's a way for the proposer to give it to this uh in a way that you know if they deviate the penalty of slashing like basically what they end up doing is they say: I'm gonna sign this walk. They produce a valid signature both for this blinded payload and also the unbinded payload, and this is how you get this commitment function, so proposer signs the whole thing, even though he's still blinded to them.

B

They pass that over to the build the Builder and then once the Builder sees that uh you know the proposer is committed up to some slashing penalty. Then they know that they can release the rest of the block and there's no no issue with uh you know it may be stealing or theft.

B

So that's cool um and like helps a lot with this more very trustless model. The thing is kind of what I was saying, though, is that the builders now are trusted to reveal their selected block right. So there's the commit part which fix like this may be stealing bits, but now we've introduced almost another problem with this like reveal step because Builders now need to like reveal their block um and if they don't, they can grief. Proposers doing this basically by just like not doing the reveal part the issue.

B

There, then, is that a proposer who's committed to this thing. Well, you need the full box for it to go on chain like otherwise no one will know how to like execute the thing that was committed to, and so there's like this availability issue, and that's not good, so we had another actor, the relay that people have probably heard a lot about, and the relay is basically trusted to fix this last bit of this availability concern.

B

So we do have some trust in the model and basically it's encapsulated inside the relay so relays you know do a couple things. They basically make sure that Builders are only giving good bucks so that proposers can like trust in that push. Network um relays also ensure that this block data is made available so that again, like Builders, don't need to be trusted either.

B

So now you have this, like you know, pairwise sort of untrusted or you know, trust minimized model, um except for the relay and we'll see that there's some consequences to that as well and yeah. A nice thing here, though, is you, can have many competing relays and competition, usually, you know, drives out some of the negative externalities we'd see of having these more. You know trusted security models, so that was a lot putting it all together.

B

This is a nice picture from some of the flashbots writing and basically we see you know, there's like three Builders here, they're all working. You know particularly every slot they would every Builder since all of their bids to every relay they can find so there's now three relays here and then the idea is that that boost is this piece that multiplexes between all the relays- and you know, facilitates all these communications between the validator and all the relays.

B

So yeah there's a lot going on, um but yeah Builders realize that boosts down to proposers.

B

And yeah, so this the system emerges, you know with the move, to proof stake and here are some stats. So there's this really nice website metaboost.org. uh This was just a few minutes ago. Before this talk, so we see that our presentation is like around 90, so it's like very widely used, uh maybe concerningly so uh flashpot's dominant.

B

So this is basically looking at how many blocks same last hour or sorry last day came from the flashbots relay because they're the ones who kind of bootstrap the system- and you know something again- that would be nice for integrity of the construction- is having a lot of competition really at every part of the stack. And so we want to track like flash plus dominance here.

B

uh This is actually like, probably a low metric for, like you know, since the Verge um there's been a lot of work recently to like improve this which I'll get into um but yeah. So that's pretty cool to see you see, there's like 10 active relays.

B

So it's not just like you know it's not that everyone just hopes that this thing would work and like there'd, be relays that showed up like people actually did show up, um and a lot of them at this point are related to provide this as a public good, uh essentially to like help with this off-chang PBS, abstraction and um yeah. Essentially, in that way, it helps secure the network.

B

There's another nice website map boost.picks uh definitely check that out and yeah, just like tons of awesome data. Here we see on the left. This is like uh slot share per like Broken Out by relay so basically for every slot on chain that is, um delivers a block to the proposer for that slot we can see like which relay was essentially the one sort of you know, um providing or executing that uh that delivery, and so this is broken up by relays.

B

Here and again you see at the beginning, it was like mainly flashbots, but then even a couple months in we had a lot more relays come online, um and then you know even in like say the last month or two there's actually been. Quite a quite a you know, an improvement in like Market composition, with respect to uh competition here on the right, you'll see the same thing for Builders.

B

So, basically, rather than look at like the relay stage of things, you look at this Builder stage and uh yeah again, you see the thing was sort of essentially seated by flashbots uh with this I know Builders, oh actually, so that's actually I think local uh local block building, but that's something else. Flashbots uh is in the orange here and point being again, you see a similar story where, like early on, there are only a few players, but over time uh the market starts to mature.

B

You have your entrants come on board and it actually becomes quite competitive. So that's all really cool to see and yeah. So there was a question: I saw like yeah. Why? Why do validators? Even do this in the first place like as I've? Probably you know hinted at this point: is that validators? uh You know see this as a way to like profits, because you do have these like.

B

uh You, have this like very competitive uh game between Searchers and Builders, and the idea is that for them to get inclusion on chain, they basically bid away a lot of the value of these different opportunities upstream, and that can at this point, go all the way up to the proposer and yeah I just wanted to call this out. This is the screen cap from etherscan, and basically it's showing that um this is the biggest that Boost payment I believe to date you know almost 700 each and basically yeah. This is well.

B

It was a little funny so to speak, and basically there is a a bug in a search, robots and then I think another search or buyer too. Basically like we're fighting over some ARB that happened on chain and then it led to essentially an attack of like the first Searcher and um yeah point being. uh This is like a historical, each payment, but it went to in this case the lighter pool which is then you know, I, believe shared with all of their validators. So um yeah I mean there's like real money here for sure.

B

That being said, most uh sort of like Med payments or these blocks come in from the members Network. You know if you want to talk about, like average Mev for Block, it's more like you know, 0.1.2, maybe point three or four eth. um This is definitely an outlier, uh but it happens and, like you can kind of assume it'll probably happen again in the future. So yeah, it's uh pretty wild stuff.

B

So you know there is this problem of like Mev and it seemed to like threaten the protocol and we're like well okay.

B

This is actually quite a big problem if you think about it, but what we can do is just make sure that the core protocol itself is okay with this PBS thing, and so you know, even if we can't ship PBS in the protocol today we can kind of do an off chain with net boost, and here we are right so like everything's, good yeah, well, not quite, and it turns out that uh pretty quickly we saw downsides to like the security model of this whole thing, um especially in the form of relay censorship.

B

So another nice website, um if you haven't noticed like people, are, however, watching this stuff constantly, but another nice website, medwatch.info and they're. Looking here, essentially like a lot of these big relays, uh for example like flashbots, they follow effect compliance, and you know that in itself is well I'm, not a lawyer.

B

We're not going to get into this, but basically they've they've made their decisions to uh comply with these sanctions as they see them, and because of that um yeah there's, essentially some transactions that could be in the public Min pool that flashbots would refuse to like build into their blocks or deliver through their relay, and so now the concern is like censorship with the core protocol, and it's, like you know, aggravated by the point or by the fact that uh you know so much of the network is using this.

B

This system right this med Boost network. So that's not great um because again, censorship resistance is like a core value, I. Think of all this blockchain stuff. We do and yeah. That being said, Situation's improving, but it's definitely like uh you know, top of Mind research question and just Hot Topic generally like what do we do about this yeah?

B

So you know what's next, um you know that problem and others will probably you know at least the the the idea or the aim is that in China PBS fixes a lot of these trust assumptions that we needed off chain um by enshrining this into the protocol.

B

Somehow and like using that as this like trustless infrastructure, so you know longer term for PBS we actually do want to like and trying something in the protocol and again, a lot of this is getting rid of this relay actor so that now there's like a gossip layer, let's say for Builders talking to validators. uh There's not this essential single point of failure that can do things like sensor transactions or you know, have availability failures or different things like this.

B

There's a bunch of different ideas around how to actually do this um I think the leading one for the protocol would be this like two slot idea: there's an eight research post from vitalik and basically it's just saying, take them at Boost auction. You drop it into the protocol and that's pretty much.

B

It um there's like a pretty key difference, which is that, rather than have like one block per slot, as we usually think about it, you kind of need like two blocks per slots, or you know you need this like sort of like well, it's essentially like a um you know, you know bid phase and then a like bid. Selection phase and the idea is that basically uh Builders would like cross up their bids. You'd need the proposer to commit to one and then they'd essentially need that Builder to then like reveal the box. So.

A

It's a similar commit reveal.

B

Thing uh just basically done with the core protocol. That being said, I think there's a big open question around. What does this auction actually look like so, if you're, just taking that boost and like Drop it into the protocol, uh you know I think there's some questions even from today that you know our pause. You know our reasons to like you know, pause on during that to not just like copy paste.

B

The Boost at the end of the protocol and like one of them, for example, is a censorship question, um because you can imagine like well. Okay, yeah today, with Med boost it's the entire block that a proposer is selling, and so what that means is they have. They have no agency over the contents of the block. You can evolve into the Builder and you can imagine a world. Well, that's essentially where we are today and that leads to the problems we see today again with censorship and whatnot.

B

You can imagine different world where the proposer only gives a baby half the block or, like you know, some part of the block um and in that case, they're not auctioned off the full thing, but just part of it. uh You then quickly get to different, like option ideas rather than like auctioning off the block. You actually just auction off the slots and there's like some other off-chain protocol or some other protocol somewhere that like organizes what goes where and if you like, really take this to the extreme.

B

You get to this really cool idea of Pepsi from Barnaby, basically saying like what, if there's actually no option at all, it's just this like generic programmable commitment idea, um because if you think about it, the options like committing everyone to this, like one particular outcome, you can imagine this just being generic evm execution, and you know who knows what people will think of so you know that I think is like really cool and worth exploring.

B

The the idea here is basically saying, like you know, almost like four through seven with the abstraction, is like taking the notion of like what it means to have a wallet and like adding this like rich performability and expression expressivity to it. uh You can do this with PBS, so super cool lots of open questions, I think we'll see a lot of like really interesting r d come out over the next couple years near term. You know we have what we have today and we like need to make it secure.

B

So one thing: that's pretty interesting, like already we're seeing the emergence of these latency games uh in this message, Network. So what I mean by latency game? Is that essentially the faster you can, as a builder the faster you can get to your bid to the relay, essentially the better your chance of winning the slots or winning the auction? And what this means now is that the more I can invest my resources as a builder into like getting better.

B

You know, lower latency and, like better inclusion, um it'll, you know again lead to me winning the auction more and more. So that's not great because again you have this like this flywheel thing that centralizes the Builder set, and you know PBS. While it does remove that pressure from validators, it kind of just kicks the can right up the road to builders and that's not great either. So um there's some some charts here again from like live message: data which is really interesting.

B

uh This first one that left is from Casper who's at the EF working at rig and essentially he's saying like given the winning bids from babus so far, um I think this is one particular slot, but either right given like they're giving winning bids like when do they arrive, and you see if you look at like the distribution, they pretty much all arriving right at the very end of the slots, um meaning like the prior slot, meaning right before um the proposer would pick your bid to include as their block for that slot.

B

uh You basically want to be as close to that decision Point as possible. We see more data over here. uh We have Mike who just uh joined some of the r d efforts here and so he's looking at again a similar thing. It's just another sort of view on this data. It's saying like for all of these Builders.

B

um You know what do we see over time and it turns out that, like basically the longer time like the more time that, like Builders, have the bid gets more valuable and like intuitively. This makes sense because basically, Builders are just saying like you know, there was the most recent block. We now start this game to build the next block and what's happening is people are using the chain just like continuously, and so there's, like you know, every so often there's like a new transaction, even if it's just like an eighth send with like some.

B

You know some gas payment, that's still enough to like have an edge or even just like a couple way over, like the other, like the next Builder right, because, again, the way the auction format works is you're just going to take the highest price. Even if the difference again is like you know some some giveaway or something like this. um It's still enough for the software to pick you over someone else and yeah so because of that Builders are like constantly refining their bids and it leads again to this like arms race.

B

So that's not great, so the question now is like: can we do anything about that um and it turns out we can uh so like Mike and some of the people at the ultrasound Relay Team have been working on this optimistic, really idea, and the idea is that basically so relays do two things they validate. That blocks are good and also handle availability, and you know a lot of the latency from um essentially bid submission. Is this, like validation of the block, to make sure it's actually good in the first place?

B

So the optimistic relay basically says, let's think of a safe construction, to essentially remove that, so that there really doesn't need to validate anything and I mean again, there's there's ways to ensure that it's safe and that if Builders have lied somehow, like there's like a bonded, uh there's a bond there's some some collateral, there might forfeit point being. Is you actually remove this extra latency and it should improve?

B

um You know this essentially how much you can gain as a builder by playing this game and so there's lots of something to play in the first place and that was actually shipped just this last weekend, um they've been working on the like r d for, like maybe even a month or two like yeah, maybe even two months, um but either way it was. uh It went, live for the ultrasound relay of the weekend and I wanted to call this out. This is again from my boost.pics.

B

You can see that there is a spike that I circled in red uh for like relay inclusion of ultrasound, so it seems like it's working um and that's that's pretty cool to see so you know that's been going on and you know there's some desire within the Memphis Community. It's essentially like roll it out to all relays. Basically it's uh you know it does remove this, like pressure on builders, for them to like play this latency game and to the extent that we can do that it.

B

uh It improves like again the it lowers the rate engine to be a builder which is generally healthy for for that part of the market, you're looking even shorter term, uh we have the Chappelle Fork coming up. So, like there's upgrades for that for, like map boost just to continue working, um you have to really uh update things. Like my Boost, the relays, the builders there's like a whole bunch of software that needs to be updated. So again, this community's been like working hard on that.

B

We also have four for four, and you know the fork after chappella that's around the corner, and you know there'll be a similar thing. We'll need to do around it being all the software thinking about. You know if we add blobs for Ford. Does that change, how this game works? Different things like this thinking very much top of mind like because of the consequences of like the different assumptions of these models.

B

We have things like you know, censorship via the relays, and so the question is like: are there things we can do or deploy even today, to improve that situation as well?

B

There's some research around this idea of like an inclusion list where, basically the proposer would say Okay um you know, I I would like to work with you to like have this built execution payload, but you know I also need you to like include these transactions I see in my eventful that perhaps have been floating around for a while, haven't been included, and so maybe I suspect that they're being you know, excluded purposely.

B

um That being said, this idea is like kind of hard to enforce off chain, because it's like how do you I mean? How do you make quality incentives, line up and either way, uh there's like some pretty interesting research happening there as well, so I'll go ahead and wrap up. um If you want to contribute, there's, like a you know, a number of different places, there's like the Boost repo uh I, have an implementation of my boots myself.

B

uh Under my own GitHub mem RS, there's the Builder specs I called out plenty of them, maybe research on each research. uh We have a newer entrance to the space owner to call out Frontier research. uh Frontier.Tech, um that's uh you know: they're they're, building a research Community there and I think there could be some really cool stuff. Coming from that effort.

B

uh Over the next couple of months we have started a series of community calls, so the second one well I, guess the third one, because I have zero X index them, but the third one's coming up soon. There's a link here, if you're interested in stuff, please attend and uh join the conversation so yeah, that's it from me for now. I know that was a lot um but yeah.

B

If you want to have any questions or like figure out how you can like start contributing space here, I am on Twitter and yeah feel free to reach out thanks.

A

For the presentation, love the pre-context and the example of an attack, thanks for touching on the concept of Mev and PBS as well, Med ceiling was interesting to me. I love the architecture diagram and the graphical representation of Builder proposal, communication and relay for our viewers and listeners. We have recorded few episodes on different topics mentioned today, so check out PBS with barnabyman, not Mev and flashback with Thomas tanza and Builder API, with gajender Singh on p penny playlist moving on I have a few questions listed for you. My first question is related to relayers.

A

We did talk about release and I understand. There are multiple players. I came across a question like what are the advantages and disadvantages of connecting to multiple Mev boost release. At one point of time,.

B

Right yeah so relays are this actor that, basically, you know bridges over this like trust and compatibility between Builders and proposers? So we essentially want an open, Builder market and an open, Ballot or set.

B

You know, vis-a-vis the proposers, and so because we have these open sets like they basically can't trust each other, and so the relay isn't this thing that sits in the middle and like kind of like it papers over this, like trust mismatch, so there isn't, as a validator you want to connect to many relays um is essentially to get like a wider set of beds, because you could have different Builders, who, like only work with some relays and not others, uh there could be relays with different like features or like policies so like, for example, uh most people would say that this, like Mev, like sandwich, attack uh that I I gave as an example is like very harmful right because it's like you're, basically taking someone's trade and giving them the worst execution than they would otherwise.

B

um And so because of that, uh you could imagine, there's like relays, or you know, even Builders, but really either.

B

I could basically say, as a builder I like declare I'm not going to include front running transactions, um and that could just be a policy they have or you could have relays that say, I'm going to work with any Builders, but if I see that they're doing fun running then I'm just going to like discard their bits, and so because of this you know, there's like a pretty diverse relay landscape and there's like these different features or differentiations. They can have.

B

That might be appealing to like some validators over others based on either how they see Mev or their jurisdiction. Or you know a long tail of things.

A

That is pretty interesting to know and I also found out something related to the diversity of relays. So what do you think would be the significance of having diversity here.

B

Yeah I mean essentially just for the reason we care about diversity and, like uh you know, node composition about other side to all these things, so the more diverse that the network is and for pretty much almost any definition of diverse you're going to have greater resilience because, like there again won't be this like concentrated single point of failure, so you know again that's another instance here for this with relays where, like the more relays, we have it again just like hardens this construction, pretty.

A

Interesting, my next question is coming up from a survey conducted by ethereum cat herders on client diversity. We try to collect feedback from Community about what they feel currently about their clients and what they think could be improved. What kind of feature they would like to see in their respective clients. So uh there were some mention of Med booster I have a few questions around that. My first question is related to uh relay outbidding at the mempool.

A

So there is a suggestion, slash expectation with client, node provider. I would be curious to hear your thoughts um on the below suggestion.

A

The client should consider local mempool before accepting bids from the relay. It might rarely trigger any change, but just as a matter of principle, the relays, especially if they are censoring, should only be able to buy the block space if they outbid the membrane and the another comment is. There should also be a feature that says: relays have to outbed mempool by at least XF before the client will accept it.

A

I would be curious to hear your thoughts on this.

B

Right I mean yeah, so so the the thing this question is getting at is basically saying like if we do have sensory relays well like how can we get around this, and uh you know the like very obvious thing: is you just don't work with the relay or the Builder by extension right?

B

And so that's not where you have this issue of like or this this concept of, like you know, building from your local MIM pool having like local block production versus this like remote block production and um yeah, so actually I mean something in the client team's been working on is essentially like adding more uh sort of features into like the communications protocol of like the local node. To facilitate this. So, like one thing, that's happening in Capella is that basically the execution node will now pass the block value to the cons of this client.

B

So when it goes to get the extra payload from the local execution, client versus the remote one, you know in this Builder Network and then what you can now do is say: okay, even if I'm, using that boost it'll, give me some bid. uh That being said, if I can build a local block and a local block happens to pay more, then I would go with that, and you know this is in some sense a proxy like this question's kind of pointed out, uh for you know, censorship for things like this.

A

Another suggestion is about integration, so uh I am wondering if there is a possibility of integration of Mev boost with a client node running software.

B

um Yeah, so yes, and no probably I, mean again whenever we talk about these things, there's always like a host of trade-offs like Med boost today is like pretty lightweight, and so you could imagine just like taking that logic into inside this client, um and this has been like a conversation that different client teams have had.

B

The thing is: is that one it like now brings a lot more functionality into this client, and so there's just more, like you know, surface to secure um the fact that this whole thing sits behind this, like Builder API like and defined in the Builder specs. It's really nice, because again you just have this, like essentially there's just like a very small surface of like having to think about this stuff, and otherwise there could be stuff happening outside of it.

B

But like it's not really, your concerned as like a node, um so I think there's a pretty good reason to like keep uh this stuff behind this Builder API and moreover, it does let us like evolve. My Boost on the line without worrying about you, know, clients, building things out or including different features or stuff like this.

B

So like, for example, if there was like some you know, maybe we needed a more sophisticated protocol to do like partial block auctions, where basically the proposer could say you know: okay Builder build me like 10 million gas or something, but then you know I'll build the rest of the book. uh If now like this, this logic is like sort of you know. If it's included inside the clients, you know, need the client teams to do this. They have plenty already and so yeah Again by having that boost, be separate.

B

It like gives us more flexibility to think about how to iterate and evolve it.

A

All right and if I understand correctly, with PBS coming on protocol level, it will be a part of the protocol itself, so I hope that would be interesting. Future.

B

Right exactly yeah so like once, there is an instrument PBS like yeah. This is actually something that client teams definitely would, uh you know, would would build into their software.

A

Awesome um I understand this uh me: reboost will work well for validators institutional and solo, validated Sports, any limitation that either Community the institutional or the solo validator should be aware of, and also this is the first upgrade since uh the implementation of Meb was any guideline or anything specific that a valid editor should keep a note of.

B

Yeah I mean mainly I would just point out that it's still like a pretty amateur ecosystem and, like you know, things are moving kind of rapidly like um yeah, and so because of that, it's you know, I'm not saying it's Alpha software, but like maybe you should like kind of think of it that way, as you're choosing to like run this stuff, like yeah sure on paper, you're like, of course, I want these like 700 East payments, so, like obviously, you're gonna run this thing.

B

But that being said, you know how to think about, like okay, like what additional risk am I taking on as a node operator, throwing all this extra software and like work with this, like other network that, like doesn't really have any like incentive structure, respond it's in the same way. They say like the core protocol does um so yeah. It's I basically will just say, tread carefully uh and it's not you know everything has a cost. It's not just a bunch of free money.

A

Of course, it does uh all right uh it's time to wrap up any final comment, thought or anything that you would like to share with the community.

B

um I think really I would just ask people to get involved, um I think even this community effort around my Boost is like pretty new and like again to the extent that you're interested in these problems- or these questions like show up to the community calls and um yeah I mean let your voice be heard, especially with like solo secret stickers are listening to this. uh Some pretty interesting questions have come up on the existing Community calls already around.

B

uh You know this optimistic Greely idea and like how that changes, the market and how that influences Builders and relays and like what it means for validators um and yeah I mean it would be great to have like solo sticker engagement with that. So you know to the extent that you can attend. Please do Alex.

A

Talks we really appreciate you taking out time to provide an over boost. I hope it will be helpful for institutional as well as solo validators. We would like to decentralize the validity test Community as much as possible and people if you are interested in Mev boost. If you are or planning to be a validated, please join the community call. You may follow Alex Twitter handle provided in the description there. You can find the announcement of upcoming Community call.

B

Thanks for having me Pooja awesome.

A

On this note, thank you to our YouTube viewers for watching and podcast subscriber for listening to this Mev special episode.

A

You have any question on this or any other topic. Let us know what catheter's Discord next week, we'll be back with another interesting talk till then keep watching and listening and keep sharing your love with ethereum caterers. Yes,.