Ethereum Cat Herders PEEPanEIP, 17 Aug 2021

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: PEEPanEIP #44: MEV & Flashbots with Tomasz K. Stańczak

Description

Topics covered
-----------------------------
0:25 - Introduction of the topic & guests
1:07 - Guest’s presentation
1:49 - Flash Boys 2.0 (MEV History)
5:06 - DarkForest
05:36 - MEV explaining
06:00 - Flashbots introduction
7:50 - Ethereum 1 Flashbots Architecture
10:40 - MEV bundle & Mega bundle
12:42 - Flashbots technologies
12:58 - Flashbots action
13:30 - Flashbots data
13:53 - Flashbots research
14:38 - Flashbots tools
15:17 - MEV explorer
15:53 - MEV Geth
17:04 - Explore Flashbots
17:54 - Flashbots dashboards
19:17 - Flashbots research proposals
20:46 - MEV SGX
22:30 - Can I spin up MEV Geth and start receiving bundles of Flashbots on my own? Is there a process in place to receive the bundle?
25:15 - xDAI MEV
26:17 - Eth2 MEV
27:22 - Why is MEV in Eth2 important for validators to look at?
27:50 - How to make sure that this MEV is available equally to the small validator the same way it is available to the big validator?
28:11 - Before the Merge
29:22 - Proposer, block builders
30:30 - MEV flow diagram in Eth2 (family -1 solution)
31:48 - 2nd solution proposed flow diagram
32:50 - Solution 1/n proposed flow diagram
34:45 - Solution 2/n proposed flow diagram
35:50 - Carrier transaction
36:35 - Block builder recipe example
37:42 - Multiblock MEV
39:16 - MEV-SGX after Eth2
40:57 - Q&A
41:10 - Transaction sequencing is up to the miner. In a perfect blockchain, would there be an arbitrary transaction that a miner can defect? Is it better the way it is?
45:07 - Game theory
45:58 - Does the distribution of the validator affect in any way?
48:05 - How often speed is a factor in a bundle? How often do we see a bundle that needs to be executed really ASAP? Is that a significant percentage?
49:43 - White hat sweeps
50:56 - In Eth2, you know in advance that the validator will create block x. In case when tx needs to be processed ASAP, setups like that will cause bundles to be published later?
52:38 - ExplainTime-bandit attacks
56:00 - Fee-based forking attack,
56:55 - What’s the purpose and what’s the main drive for MEV? A summary.
59:36 - Tomasz’s message to the MEV community

Resources
-------------------
Slides - https://docs.google.com/presentation/d/18JxAugvT59aNoUzrFLib_-n3mDWRPZSz6eqHiIk32lU/edit?usp=sharing
Flash Boys 2.0: Frontrunning in Decentralized Exchanges, Miner Extractable Value, and Consensus Instability - https://ieeexplore.ieee.org/document/9152675
Rapid Rise of MEV in Ethereum - https://medium.com/etherscan-blog/rapid-rise-of-mev-in-ethereum-9bcb62e53517
Flashbots: Frontrunning the MEV Crisis (Medium) - https://medium.com/flashbots/frontrunning-the-mev-crisis-40629a613752
Flashbots: Frontrunning the MEV crisis (EthResearch) - https://ethresear.ch/t/flashbots-frontrunning-the-mev-crisis/8251
Flashbots GitHub - https://github.com/flashbots/pm

Follow at Twitter
-----------------------------------------------
Tomasz K. Stańczak @tkstanczak | William Schwab @William94029369 | Pooja Ranjan @poojaranjan19

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

Follow PEEPanEIP earlier episodes at https://www.youtube.com/playlist?list=PL4cwHXAawZxqu0PKKyMzG_3BJV_xZTi1F

Contact Ethereum Cat Herders
---------------------------------------------------
Discord: https://discord.gg/tzYmDmF
Twitter: https://twitter.com/EthCatHerders
Medium: https://medium.com/ethereum-cat-herders
GitHub: https://github.com/ethereum-cat-herde...
Email: support@ethereumcatherders.com
Website: https://www.ethereumcatherders.com/

A

Welcome back to be penny pen, educational series, where we learn about ethereum improvement proposals, upcoming features and important concepts. Basically, every key term. You want to learn about each and blockchain on this journey of educating ourselves today, we'll be talking about mev and flashbacks with our guest thomas stanzak, the lead developer of netherminecline.

A

In the upcoming talk, you may expect to get an overview of minor extractable value mev. What flashpot is flashbacks architecture mav before and after the merge of the helium, blockchain etc. Along with me, this meeting is also joined by william schwab and alita more from the keegan catholics. Thank you to mask for joining us on the show. I'm excited to hear what you have got to share on today's topic, so without any further delay. Let's speed in.

B

Hi everyone. uh Thank you, pooja. Thank you later, william. uh Thank you here and cut herders organizing this sponsor yeah. Today um today talk presentation, yeah, it's mostly to bring some more light to what mev is and also how flashbacks deliver solutions for this mev being uh much more aligned with the ecosystem and much better for ethereum.

B

uh So, as we will see in a moment, there's lots and lots of things related to maybe above the research, but also development and tooling, that is available, so maybe first a bit of history. So my first conversations about mev- I guess they started a long time ago, but there was this um very, um very important paper that was describing how exactly med happens.

B

What are the risks related to letting mvv being executed without any without any tooling, that is aligned with the ecosystem uh and it describes all the processes and how the gas prices behave, how the bots behave within the memphis, so the flash voice, 2.0 um that talked about both the opportunities, the mechanisms and and the outcome of the developing mav market, and we started seeing that probably like a year two years ago, uh maybe even before, but it started being more and more visible with the d5 projects um rising having higher tdl and exposing a lot of value on chain which is not related to the uh main ethereum eth token, uh the cryptocurrency actually the native token, uh but the value that was inside the erc20 and various uh d5 protocols.

B

So when we talk about mev very often we use the the dark forest term and most of the mev is inherently related to memple and why we talk about memple, because this is where we discover all the public transactions and the public transactions change the state of the um of the network whenever we change the state of the network. This is the moment when uh the prices on the defy systems and the prices of the tokens uh change, so they can become misaligned on different decentralized exchanges.

B

uh The some of the liquidations become possible for the d5 protocols that use the um under collateralization and various mechanisms suddenly become enabled based on the memphis transactions that are about to enter the block, so knowledge about what is in the memphis creates a lot of opportunities because we know what will happen after the block, but maybe even we can react within the same block by injecting various transactions in between the transactions that we see publicly.

B

So if there is some particularly big opportunity, a transaction that moves price on one of the decentralized exchange a lot but doesn't move the price and another decentralized exchange, then we have this misalignment and we want to be, as a bots builder, a mev searcher bot builder. We want to be the ones that execute our own transactions exactly around the transactions. That creates the opportunity- and this is not only the arbitrage in the sense of a very positive mev that leads to the price.

B

Discovery leads to the very aligned prices between the defy eco system, but also this uh negative mev, which is like all the front running the sandwiching and so on, uh and that's why we talk about dark forest. So, first of all there forest is about the detection and being visible, and then, whenever we are being detected, uh someone can can use this opportunity to play against us.

B

You know I like the in the stand, which brought sandwiching boats that take our um our trades from uni, swap, for example, uh they can wrap the transaction and it's just because we beca we are for for a moment.

B

We are visible in the memphis and they and they react to that, and uh mev is just the general description of all the minor extractable values, so ability of a miner to extract value from the blockchain uh from the misalignments on the in the pricing, from the opportunities liquidations and trades that are happening by inserting reording or censoring transactions and when we think about flash boards, which very often comes now in pair when we mentioned mev, uh the flashbacks mission is ready to mitigate all the negative things around time, ev and bring the network safety and align the ecosystem and the community around ethereum with how a movie can be delivered.

B

So flashbots is not a mov flash, but it's delivers research and tools around a movie uh that first of all make this minor extractable value, but more like community extractable values is something that you um that you make a bit more democratized, more decentralized and also does a lot of research. That says if there is any threat to the network uh that comes from miv, uh then how do we? How do we deal with it? What do we deliver to make sure that the network is safe and the network is protected?

B

Can we actually use the flash but stills to prevent negative mev to prevent attacks on the network uh to prevent hacks and so on and so on? uh So all of that is within the rnd mission of flashbots, and this is this famous great power recreates responsibility.

B

Nowadays, flashbacks uh provides the tooling for 85 of the hash rate, so 85 percent of the cash rates from the mining pools minor, solar miners all together uses uh either a version of a tool that is mavgef that allows for flashpoints to deliver the mev bundles or uses some kind of implementation of the spec of the officially available flashbot spec.

B

And how the how the flashbacks architecture looks for this delivery of mev in a more democratized way. So here we see on the left sides the searchers so who are the searchers? Anyone can researcher, and while this role becomes more and more sophisticated over time, even nowadays you you can hope that you can become searcher by spending, uh maybe a weekend on writing tools and experimenting and finding some opportunities, um but just playing with the tools.

B

Documentation, flashbots and that's uh extremely rewarding fun obviously can take a lot of gas payments and not necessarily brings an easy way to to extract value, but definitely it's simpler than it was before, because it, the the the very name of the minor extractable value points at the minor as the one that is uh ultimately allowed to extract all the value. And- and this is the truth- is uh true thing.

B

If you assume that the minor actually wants to do that and has the ability to do it more like a skill to do that, because the ability is there like miner is the ultimate creator of the block ultimate sequencer of the transactions in a block.

B

But the minor business is very often not very similar to the type of type of skill set that is required for searching mev opportunities, which means that you can simply outsource that to the scores of mev searchers that use the flashbots relay to deliver the information about how they would like to order the transactions, how they would like to insert their own transactions and how they would like to share split the payment between the miner and the searcher themselves.

B

And this is the architecture that's available nowadays. So we show here free relays. um The flashbots approach and the flagship mission is really to democratize democratize. This solution, not only on the serger side, but also allow anyone to run the relay and there's lots of research to uh make this happen to make the relays something that anyone can run and it can.

B

Anyone can run in a way that me researcher can trust it that it will not try to steal this med opportunity, because, similarly, as we're talking about detection detection in the dark forest, so that detection mechanism also works someday on the bundles, if they're not protected by some level of trust nowadays with flashbacks or some uh privacy algorithm solutions technology in the future. That says you cannot be detected if you're sending a bundle. Your knowledge about the mev opportunity cannot be stolen from you.

B

What, when I'm talking about bundles and transactions that we're sending to relay so what they really are? So it's a a mi v bundle. Is that simply a list of transactions that have to be executed together? And we assume that I mean the bundle has to be executed at the front of the block. So that's the kind of tooling that flashbots provides and the definition for the bundle.

B

So by reordering and inserting transactions, you say, okay, so let's leave the left of the block the way it is, but at the front of the block, we insert some of our transactions and we reorder some of the transactions that we see in the public. Memphis the.

B

If you remember this sensory component, which I, which is tracked out, it's not delivered in any way by flashbacks. The censoring transaction is not something that is desired, and it's not something that is delivered. The mev by itself does assume that the miner can actually keep censoring transactions within a block just to increase their value.

B

But this is not something that flashbacks provides. The mega bundle concept that comes with flashboth version, 0.4 that is just being built, is the concept of preparing more than a single bundle like merging and simulating multiple bundles verifying that they can run independently.

B

uh They do not affect each other bundle, profitability and verifying what is the most uh most profitable for minor way of executing them together, which of the bundles, provide the maximum value to the miners and because it's a flashbacks player when you think about the most profitable bundle, you should think about the most profitable bundle, also for the searchers and the ecosystem in a way because of this um targeting of the positive mev and the democratized media access uh flashbacks, apart from this main architecture, delivers the ideas of flashbacks options.

B

So this is the description of the set of technologies that provide the relay the the bundles.

B

uh Let me maybe uh switch on the line, uh so flashbacks auction is the set of the uh tools that allow for these bundles to be delivered to the miners. Why do we call it flush but auction because we're replacing the old style price gas auction, the one that was leading to very high gas prices very often on the network, the one where the bots were competing with each other by just simply beating up all the time on? What's what they were planning to pay?

B

We replace it with the sealed bit auction mechanism of the flashbacks. Then there is flashbot's data which is set of tools, dashboards um tools for inspection of the data on chain and mev data, post, factum or also real live data.

B

Real-Time data and this I'll show a few of those uh the examples in a moment, and then there is flashbox research, which is just the idea of collaborating with the top researchers in the field, with contributors in an open way, uh inviting the uh the speakers, organizing the uh mev roads. The flashbacks throws the what the fact sessions that describe the various important concepts with the top researchers talking about it for hours and hours straight together and dealing with the things that are hard to be really explored solo.

B

So all of that comes from flashbots and you can find a description of all of those at the documentation of last notes.

B

So those tools that I mentioned within those three categories- so you can see- maybe inspect so it's just being rebuilt with the python tools for exploring various main uh most most popular d5 protocols and detecting the liquidation opportunities. Misalignments prices, exploring what happens. What kind of events are happening there uh and this set of tools can allow you to build more efficient uh searchers. So you can have your bots use mev inspect to quickly detect uh information from the mempo from the from the blockchain state from the ethereum state.

B

Then there is mev explorer, so this is the way of exploring all the all the mev transactions and the dashboards, which shows the flashboards operations, like all the all the actions happening on chain as a result of flashback bundles, we have the ethers.js provider that allows you to send the bundles, prepare the bundles and send them to the uh to the relay to the flashbots relay mvvgf, which is the main tool used by miners for accepting the bundles from the flashbots relay so mavgaf itself.

B

It's a fork of gaf, it's a fork of gaff with just a handful of comments on top which are described in details and documentation. uh Second, review: it's open source: this is definitely a huge commitment of the flashpoint's research and development. Everything is open source. All the research uh documents there as soon as they leave the like note stage, they're available for contributors to review and contribute to- and as I mentioned, there is also the specifications. So miners do not have to use exactly this version.

B

Very often, miners have their own modifications to the gef nodes, either some optimization specific for their infrastructure, or maybe some modifications that allow them to inject their own transactions based on their agreements or maybe like even old style mev, this private mev that was not available for everyone, but generally they provide solutions that are consistent with the spec and consistent with what we have here in the midgaf repository.

B

The explorer flash ports is the med activity on the ethereum network, so this is not only flash button but generally exploring what happens. What kind of extraction is there on the network and showing you various uh statistics uh since the beginning of 2020, showing you that the daily extracted mev is higher the times of higher volatility uh and it grows with new d5 protocols appearing with the volumes growing on defy space?

B

And you see these numbers are pretty high, so they soon will be reaching one billion dollars of value extracted from mev.

B

The dashboard shows you the information about flashboards and, for example, here you have information on the number of blogs, including flashbots bundles, so this number was reaching at some point: 65 percent.

B

uh It means that, like two out of three blogs happening on, mainnet nowadays would have the bundles delivered by flashbots relay it's a huge number and, as you think, of this 85 percent of the uh of the hash rate out of this eighty-five percent of the hash rate, you see sixty-five percent of blocks uh being mined with flashbacks bundles. uh Why not?

B

Eighty-Five percent, because sometimes miners will decide to that they will not receive the particularly valuable bundles and I do mine empty blocks or decide for the vanilla block with opening bundles inside um also, when you think about the flashbots bundles as percentage of the ethereum block space. I really like this number is around two three percent, so it's not a lot of gas that is used by the bundles themselves. It's a lot of value, but they don't take that much of the block space.

B

So three percent, when you think about 15 million blocks, uh so this would be around 400 000 gas.

B

What's about the research, so flashbacks runs a lot of research calls and it's all documented in the flashbots mev research repository and you're very welcome to participate in that research and mostly it's related to all the important components of ethereum scaling nowadays of the ethereum future, mostly a few months to two years, future of ethereum. So we're thinking here about what happens after the merge what happens with etm2 mvv uh what should happen with the mev available on side chains like polygon or xdie?

B

What's the layer, two solution for mev, because we know that nowadays, mostly it's centralized sequencers? How can we decentralize this sequencing of transactions on layer, two, uh how they play together? So can you do the mev across chains or shards or side chains?

B

What are the solutions for the privacy of bundles, so one of the research area is meds gx that is run by flashbots and the actual implementation of the solutions being prepared, and then there is also crypto economic design for the forward decentralized relay.

B

That is one area of exploration that you can participate in as well.

B

Talking about me, vs gx, so this privacy of the bundles, which is very important thing, because nowadays you have to assume quite a lot of trust within the flash plots ecosystem. So, first of all, searchers uh trust me relay on flash plots that the bundles will not be leaked to the public or linked to um any any other participants really before they included in the block.

B

The searchers also trust that their bundles will not be dis disclosed if they are not included in the blog. So we assume that flashbacks does discord. All the information about bundles does not make it public if the bundles do not land on blocks and also in a way flashbot has to trust miners. That miners will not take the bundles away. How do you steal the bundle you take?

B

The bundle you analyze its content and you replace the addresses that are actually benefiting from the bundle with your own and probably uh send it again to the relay. So this is constantly being monitored and detected by flashbots to make sure that the miners follow the code of conduct, because miners participating in the flashbots relay operations.

B

They have to build a reputation. They do not want to lose that source of revenue. Then this works, but with the flash puts being in the center. You really assume some trust there and some very efficient execution of this monitoring and overseeing and one of the.

C

Main requests? Oh yes, sorry! um So that means that I can't like just spin up mev guess or something else's flashbots compliant and start receiving bundles, like on my own. I guess then, there's like a vetting process for who's able to receive the bundles.

C

B

So if you are a solo, miner or a small mining pool, then you have to uh you have to fill in the form and get in touch with flashbots and be registered with the relay. uh Otherwise, the relay will not deliver uh bundles to you now, for that simple reason that you could, but that, if you are.

D

Just becoming you're not trusted by.

B

Relay you can potentially steal all the bundles and just keep on the mev value for.

C

Yourself, I was wondering why that wasn't possible. Actually so yeah that did answer that question I mean I was wondering why it didn't become kind of just a second memphis, and then anyone who's tuned in to that would then be able to resubmit either with a higher like a higher price or whatever. The case was so that that doesn't yeah. That explains that very elegantly. So.

B

Exactly that's what's the permission network, so one of the important area of research and development in flashbacks is this sgx solution.

B

So you have here the full description of a bit more complex process so setting up the in case protected, uh protected execution environments uh that can prepare the block and in a uh in a secret way share it with miner within sgx allo miner to verify the value of that block choose the most valuable block, start mining on top of it, and only after finding the proof of work solution being able to extract the decrypted body of the block for publication, uh and this solution is being built and should be seen available.

B

So there are some technical challenges, but we are dealing with them one by one and we have first fantastic results. So there's also a lot of exploration of how it looks in the future. So it looks with with new versions of sgx how it looks in ethereum, two and so on.

B

So this uh detailed description of the solutions available at df's, if research forum so make sure that you go there and if you're interested in this either research or development of the area, make sure that you reach out to flashbots and one fantastic way of reaching out to flashbacks is by just joining the discord, server and participating in the conversation there and reaching out to any of the flashbots crew and starting contributing, that's extremely rewarding and very very exciting.

B

What else we built so together with the nethermind team. So we built the uh the mev plugin that enables the x-tie mev uh x-imev because of x-dia running with the pos-dow consensus. It's not possible to run it with med-gaf right. So none of mine is the only way to deliver the bundles to xdi, and this is just being finalized.

B

Now we tested first bundle delivery last week and I think there'll be some announcements next week on how exactly to execute that uh what xda relays to connect to- um and I think today, we've seen some tweets about ideas on mev around the dark forest game, so the zk game on xdi, so that you can execute some like searching or attacks on the planets uh within the mvv solution.

B

So pretty exciting, then a lot. A lot of research is done around ethereum, 2 mev. So some of the main presentations and discussions around the topic would be by vitalik. The proposer block builder separation design, and it was also discussed later on this uh eight hour block on mev organized by flashbots that you should be able to see somewhere there in the in the flashbacks documents at the repositories.

B

Then there is also uh analysis of how to build ev2 in the most basic way, with liquid, staking together uh and representation of the project from the uh if global hackathon so watch this one in just five minutes, and we also discuss uh in multiple places how a movie looks after the merge of ethereum one and theorem two now we show it with nethermind and flashbots, but also and some other configurations and some of the solutions I'll try to explain you a bit more now.

B

So why mev in ethereum 2 is important for validators to look at because it greatly improves the rewards of the validators.

B

The difference is so huge that uh it's simply a big thing and a great loss to ignore mv entirely when you run validators and that's quite increases also the volatility of your of your revenue within a year, um but simply the question. The big question is how to make sure that this mev is available equally for uh small validators, the same way as for the big validator networks.

B

So if I run just one independent validator node, how can I get the same quality mev as everyone else?

B

So when we talk to film about ethereum 2 mev, it's worth to discuss how the merge looks now in the current design, so the selection of transactions and block building looks entirely the same in the current design. So whether whether it's before the merger after the merge is the old style team, one client that listens to the mempo collects transactions, sequences them and prepares the block adjust.

B

That block is either immediately added to ethereum one on the proof of work by the miner that created the block or after the merge it's passed to the beacon chain as the beacon chain that publishes the block proposes the block. So that's why, in after the merge, we talked about the separation of a block builder and a block proposer, and if there is a separation, it means that we can have various dynamic with multiple different builders competing for the effort that slot the proposer.

B

The validator has, or also block builders trying to build the market after a quality mev and entering into agreements with proposers and as proposer blockbuilder separation. It assumes that, first of all, you don't have to have a trusted builder to deliver mev there. So so you can say: okay, the design of the protocol says that if there is a block builder, they don't have to be trusted and known to the public uh to be equally uh verifiable and equally like in a trustless way at worthy of the of miv bundles. That would send there.

B

Similarly, with proposer, we don't want them to be trusted because it improves the centralization of the network. Anyone can join, be a proposer and receive mvv in a in a trustless way and deliver that mev to the validator weak proposer friendly. So not only um you don't have to be trusted unknown to the network, but also you don't have to have a lot of validators to get good mvv, uh and then we want those solutions to be simple and also avoid the bundle stealing.

B

So that's what vitalik mentioned as a set of requirements for the design, and there are two families of design for consensus, embedded mv in ethereum ii and those uh include a solution where you have a single, a single header selected by the block proposer from the set of headers that block builders create so block builders, propose headers of the of the mev bundles and they commit to the body and they prepare a payment for the block builder from the blog proposer and then, after the blog proposer, selects the most profitable one.

B

It either receives the body or not, so the blog builder can disclose the body or not. And after that we have the fork choice, role that may allow you will allow us to create attestations and vote for blocks depending on whether the body is available or not. And the payout of the of this payment from the builder to the proposer can be split between between them, depending on whether the body was available or not, and there are some detailed edge cases that are worth to explore.

B

So make sure that you read a longer conversation between vitalik and multiple researchers. On the topic, the second family of solutions is similar in a way. But after you have the headers published, the blog proposer, selects a subset of headers and signs them and says that I will publish one of the bodies from from these headers and then the blog builder collects the bodies and once again selects one of those the uh pre um signed before right.

B

So there is a bit like two-step conversation between block builders and proposers and because there are many of the block builders. So in the end, uh you don't have to think about the fork choice rule because you expect always to land with a header in the body of one of the blocks and just sometimes the block proposer may end up having to send their own block solution, but then properly, probably with some some punishment if they don't select from a broader set of bodies provided now.

B

If we think about the solution that doesn't include the proposer and block builder separation in the consensus level, and then we may look at one of the solutions. The naive solution. That represents a very similar relay and minor approach that we had in. In theorem one now so that we have in ethereum one now so look at uh flashbot solution now and we say how it would work in the simplest way in ethereum two and that proposed solution from the.

B

If global hackathon shows the situation when you have a staking pool which can be liquid staking like lido, you have the validators that work with that staking pool so they're registered to be the block. Validators and proposers of the blocks I have ethereum. One node operator so might be item might be someone else that prepares the blocks. So that's a block builder, so it is separation, but not on the consensus level, with some additional tooling on uh on ensuring the privacy of the bundles. It's without any of those.

B

uh So we still have very similar architecture where emmy researchers send their bundles to relay relay provides those bundles to the ethereum one node ethereum onenote prepares the the best bundles connects the block and during the block processing you splitting the payment that in the past you would split the payment between the searcher and the miner, and here is splitting the payment between the validators between staking poll and ethereum, one node uh and this nave solution.

B

The simple solution suggests displayed by everyone receives one third of the pool, but it can be really any split, depending on the strength of particular participants of all the market participants and then, after the block is created, your own two clients simply broadcast that to the network for attestation voting.

B

The second solution talks about something: that's uh jason, brian, your flashbacks suggested of using uh peer-to-peer maple for delivering transactions. So it's not exactly solution only for ethereum 2. It can actually work in ethereum 1 as well, but it was suggested as part of the research for ethereum 2.. So we started talking about it in the context of ethereum 2 mev. So we have mav searchers that create bundles. They encrypt them with a a key that the validator of the particular block suggests as the as the key that they accept.

B

So they use a public key that they can decrypt later and they kind of nominate block builders in the registry that they would accept blocks with from uh the bundles are being delivered over the maple to the block builders. Blog builders can decrypt them if this bundle was directed by the for them and then they publish the blocks to uh to the validators and validators and block proposers broadcast that to the network.

B

So there is one solution here which is called carrier transaction, and this is just to not require a special transaction type for this solution. We say that we can use the very basic standard transaction and include the encrypted md bundle encrypted with the public key of the target block builder inside the call data. So we still pay for the delivery of this transaction carrier transaction, which means that a peer-to-peer network will be delivering this uh transactions everything else, because it's simply not looking any different from any other transaction.

B

So it shouldn't there's no reason to censor it. There's no reason to reject it in the memphis. It's a transaction that pays for itself and delivers some information to the uh to the miners.

B

Now as part of the solution, because we have the registry that talks about miners, selecting so validators selecting which blog builders they accept, uh the blog builders can tell how they build blocks, so they can say these are the sections that we accept. For example, in the front of the block. We include the partner, a which might be one of the layer, two solutions providing batch transactions in the front.

B

Then, as the second level, you can have flashbots mev and then public people and then maybe flashpoints privates private transactions. When we talk about flashbacks private transactions, this would be transactions that they're more like protected transactions, not the ones that extract the mov but protect themselves from mev. So if you trade on uni swap- and you don't want to be sandwiched, uh you want to use the end points from flashbacks that allow you to simply deliver the transaction without any more complex, more complex bundle. So this is not available. Now, but should be available soon.

B

I think it's part of the deliveries of flashbacks.

B

um The other big thing that comes with ethereum 2 is the multi-block mev and why it's important, because not many d5 protocols, analyze this multi-block mev solutions and because of lack of analysis. They may miss some particularly risky scenarios. uh Some threat to the stability of their uh d5 protocols, uh especially around governance around some, like multi-block execution of series of trades that can prevent other strong from, for example, liquidating the pull at the time when it's under collateralized severely undercollateralized.

B

Now we can do like some series of blocks of execution white comes only in theorem too, because now, with the miners, you have a bit like less of a chance, less probability that you'll mine, multiple blocks in a in a row, but in ethereum two, the validator can know in advance that they'll be mining multiple blocks in a row that if they have a big chunk of the network, they can know it about it.

B

Like a few minutes before that they will have two three or four blocks and they can prepare some attacks and ac attacks. uh Nasty attacks on the on the network on the d5 really so something to explore and something uh to to research- and there are some great articles already about it, but much more to to uncover uh what happens to me: vs gx after ethereum 2.. uh So since we've seen this block a proposal, builder separation, where you only uncover the bundles after proposer, selects the header.

B

So this was like a built-in solution for the privacy of any of mev bundles. If you don't have that solution, if you don't have block proposer separation as block builder proposer separation, then then you can use the current mechanism. Like the current. The merge.

B

Specification with the knowledge that, if you start signing things inside the enclave and share the signed block with the blog builder before publishing it to the network, then you no longer have ability to sign another block because you're risking splashing. So this is something fresh that we that we explore now. Looking how sgx looks in ethereum too, um there might be some new ideas here, but yeah.

B

This all should work that should be researched more, and this is uh everything that I prepared for you today, and uh anyone who is interested in joining flashbots uh would be great if you send an email to jobsetflashbox.net and if you want to follow what I talk about when I talk about mev and flashbots, then follow me on twitter or anybody else from the flashbots crew.

B

Much and I'll be happy to answer any questions.

A

First of all, thank you so much for this wonderful presentation. Now I'd like to open the floor for uh any technical question, william or anita.

C

I guess the first question I'd want to ask it's just kind of like the general question where it all starts a lot of the mev, though it does seem like uh from from your classifications. I think this is an all ndv, uh but a lot of them if he seems to come from the fact that transaction sequencing ultimately is up to the miner.

C

Do we do we think that's a bug or a feature like would in a perfect blockchain? If such thing would ever exist, would there be arbitrary transaction ordering the miner can affect, or do you think that it's better if we keep it this way with the mdv.

B

um So it's a good question because there are already many challenges to uh to the status quo. So first of all we often would ask question: can we can we make sure that on the protocol level, we enforce some ordering that cannot be modified by miners and there are a few problems with that. So if you say, if you define the order based on some hash, like some randomization of what the transaction hash is, then whoever wants to order transaction can modify some of the transaction call data so long until they find the desired hash.

B

That will place the ordering that they want. So then, when we propose this solution we can ask. Can we introduce some kind of required delay, so verified verifiable delay function into this reordering, the hashing of the transaction?

B

It causes the problem of latency of the transactions like so suddenly, there is some delay before the block can be constructed and still the powerful players that can parallelize a lot of a lot of simulations of various block. Orderings can take advantage of it, but some solutions are explored by layer, 2 providers. So you can look at the video solution from startnet for their sequencer.

B

They do explore the space, so in a perfect blockchain, uh probably yes, you would have um randomization within a block and then the healthy arbitrage based med on d5 happening in the next block, based on the previous block modifications, instead of something that can front run you when you execute defy properties.

B

The other challenge to this and is also related to to this perfect world sequencing is also a bit more risky when you think about legislation legislation. Now there are suggestions that the miners or validators should be responsible for for reporting and um and some kind of fair or approved way of sequencing transactions.

B

uh I think the problem and challenge with this one and threat big threat to the network is that a miner that's no longer thinking only about their own benefit, which, by the protocol design most often almost entirely, is aligned with the network benefit.

B

uh The deminer may start following some kind of um government enforced rules on what transactions to allow or not, and then this sensoring that we were talking about that. It's not and not provided within the tooling, that is ecosystem uh aligned. The censoring can be introduced by the governments if they, if they have the ability to to enforce the behavior of minors.

B

So I would say that sometimes this desired perfect world might be a risky world that is far from perfect. um Sometimes it might be simply challenging from technical perspective.

C

That's very interesting: isn't the game theory would be that if um the miner needs to be receiving enough profit to resist any censorship right, but if we take away the profitability of the mev, then they're susceptible to channels just influencing them in other directions, because their bottom line is much much lower. Is that correct.

B

uh This probably as well, although what I was thinking about is just just when we try to enforce um the sequencing not by the rules of the protocol, but us by some external off-chain powers.

B

uh Then this becomes risky because it opens the way, a path to some other behaviors like censoring of transactions and and shutting down the miners.

D

Does the like distribution of the validators affect this as in you know, if you have, however, many in different countries or the spread or just in terms of uh yeah, I mean is as far as the legislation goes like. Is that wouldn't that kind of couldn't you compensate for some of that with a distribution of people in different areas of the world.

B

So I believe that, yes, you can keep running away from the governments on the on the miners or validators side like from regulations into like some more crypto friendly environments, but this opens a lot of questions about how how long and how often you can run and keep moving your operations, and this might be different for validators- might be different for miners.

B

The miners will want to be in the places where uh cheap power is available and the validators will be in the places where most likely, where the infrastructure cost is the lowest, but also it might be related to the to the entities that have a lot of capital to stake right, so so they're different type of players and they may have different ability to decentralize their operations geographically. So the very design of ethereum ii supports a decentralization of validation, because the validators should not fail together when they fail together, then they risk slashing.

B

So you, you are encouraged to make sure that your geographically distributed, that your infrastructure is not concentrated on a single provider of infrastructure and so on, like single cloud provider. um So this should all help. Yes, I mean, if you're asking, does it help? Is it does it mitigate the problem of the the government's trying to uh censor the transaction? Probably yes, and because of the network, design now encourages this uh we're in a slightly better place.

C

um I also wanted to ask how often speed is a factor in bundles, as in a lot of the setups that seem to be seemed to be discussed here and also just the the kind of classic eth1 setup as it is now. um How often are we seeing bundles that need to be executed? Really asap?

C

Does that constitute like a significant percentage of bundles.

B

I cannot tell you the exact number, because I was not looking at a statistic like that: it's just my intuition is that um the big percentage of bundles, probably is the one. That's rushed right, so the one that reacts immediately to the misalignment of prices on taxes and so on, and the liquidation opportunities uh appearing in the very block.

B

So sometimes they may have miv bundles that are prepared for a few blocks ahead. If you know that some events are unlocked on some block number. So if the protocol says that on block x, something will be available, like think of maybe some kind of token distribution events when when the distribution starts on a specific block, this is when your mev bundle is prepared to happen at specific block in the future.

B

So in a way you still care about being fast and when that block happens to be as faster as possible, and that's why you care about the front of the block, um but maybe that's that fast has to happen from some point in the future.

C

White hat sweeps would be an example of that. If someone has their tokens locked in a compromised account.

B

Yes, although because white hat uh behaviors would like some some of it will be manual operation like delivering set of transactions that protect the um the account which will take time so sure we care here about the speed, but because it's a manual process, it's actually very slow.

B

If you compare it to the kind of speed that you want to achieve with the bots that are trading mev, um if these already automated uh sweeps that try to remove the eve from from allowing the attacker, they remove the eve to prevent attacker from moving the funds from the account. So so this has to be fast, like reacting very fast to any new transactions that are incoming to the account.

B

So you want to reorder them, put them in the front and and execute so again, you care about being to execute your bundle immediately when the transaction appears in the member. So most of the mev and flashbots will will be the this word flash there is. Is there for a reason right? You want that to happen immediately and in the transactional.

C

Matter yeah. So when we talk about the setups like the one, that's sticking out in my mind, is the one where bundles are encrypted for a particular particular validator. So in in those cases, um how is how is speed assured? Because, wouldn't you have to wait for that validator to show up now? I think actually, that you probably answered half the question already, because in cases where you're waiting for block n, at least in the e2 setup, you should know in advance who the validator for black end is going to be.

B

uh Yes, exactly so so this is the situation when, especially on ethereum 2, you know in advance that specific validator will be creating block x, so you can send a bundle prepared for that block.

C

But with that, um but for the cases where it really needs to be kind of like a as soon as possible thing um setups like that probably would cause bundles to be published a little bit later. Is that correct? Yes,.

B

So I think the latency of the peer-to-peer delivery, it's something that would really not be um solution of choice for for bundle, delivery. It just provides some kind of protection and privacy, but it doesn't provide the speed of execution.

B

Is it introduces a lot of latency, but only on the like distribution side, so they appear to peer delivery instead of delivering directly over rpc by a relay with the simulations pre-simulations decryption is not a problem, it should be very fast, but the fact that you have to deliver transaction in the period of your manner is a problem.

B

A

All right yeah, I just have one last question for you. I mean like I'm just trying to understand. If, if we put mav in like simple terms, very simple terms, it's basically the miners are trying to make profit uh by front running or back running and sandwiching and generally exploiting transaction.

A

So if I can get some explanation for some terms that I came across like fees-based, forking attack or time-based attack or what kind of exploiting we are generally referring to.

B

Yeah, so I mean mev is not only that's not exactly exploiting. I mean like some of it. Is this bad me that we're talking about, but lots of it is also good mev, which is just aligning the decentralized exchanges and making sure that everything happens smoothly like when you, when you're designing the protocol to enable liquidations on a protocol. You really want those liquidations to happen because they are they're critical for the healthy operations of your of your protocol.

B

Like a maker dao wants the liquidations to happen immediately, especially at the times when the network gets clocked. You want to have the access to solutions like uh get close to miners immediately, so it's really important not to think of med as something bad. It has some negative externalities here and there and those we should uh try to remove, and we should research actively the solutions to remove them, but maybe by itself are simply a very, very positive impact on the ecosystem and the gas prices like if it's organized properly.

B

As for uh the time bandit attacks and the fee based forking, so like the feedback's working, I'm not entirely sure. If, if I answer properly here, because uh it's like I've heard this uh concept many times but time bandit, it's it's all about using the reorganization of the um of the blocks, recent blocks uh to extract more mev, and this is very, very dangerous for the network. So we consider this as one of the biggest threats to the network.

B

Luckily that's very hard to execute and long term misaligned with they with the miners incentive so executing this kind of reorganization attacks like a ankle band into like long long-range organization.

B

um This leads to the loss of trust from the community, uh the disconnection from the relay network, in probably in most of the cases, so it's simply long-term, not profitable for miners and also, if interesting result of that, is that if everyone starts to do that, then actually leads to a very political system that leads to everyone losing.

B

So when you start this game, it leads to everyone losing and like from flashbacks perspective, just to be aligned with the safety long-term safety of the network and aligned with the ecosystem needs and community um flashbots very, very clearly works on uh ensuring that the uh this rear attacks do not happen are not enabled by the protocol.

B

uh So there's the active research- and you can- you can see uh like a few hours of recordings of top researchers analyzing that and telling how to avoid it, how to protect against it. uh What how the markets behave in the cases of those rears happening as the fee-based forking, so it's similar like trying to bribe.

B

I believe it's trying to bribe a miner to fork the network to um to move different way than the miner, otherwise would follow just to be aligned with community so again, something that very often will be probably always will be negative for the network, because suddenly you have a miner that is choosing the fork not based on the long-term health of the protocol, but based on the payment to the miner, um but nowadays ethereum slowly becomes not very forkable because of the dfi ecosystem.

B

So we probably don't have to think that much about it as as we think about rearx.

A

Well, thank you for explaining that.

E

I think alita had one last question.

D

For you yeah, uh I just thought of so this whole thing is pretty new to me and I think it would be really helpful to like the concept of mvvs um and I think it'd be really helpful to if it's possible, give like um one sentence kind of summarize like what uh what it adds like what's being added here and um like what's what's the purpose and then like. What's the difference from how things are done or like the main drive to change, if.

B

Yeah, so just to ask a question: in most simple words, what I mean is uh what flashbacks is so mev exists. It simply exists. I just whether it's named or not it's there uh and mav is ability of a miner to use their position of being a block sequencer to extract value from the network, whether it's good or bad. The thing is because they are powerful in the way of being sequencer in the given moment. They can extract all that value for themselves. So that's what mev is.

B

It can be good or bad, and they do that by reordering, censoring or injecting transactions of their own.

B

What it adds to the network is in the bad mev scenario. It takes some sum of money from anyone who trades on the protocols with a slippage allowed and the good mev is. It makes sure that all the liquidations happen timely and that all the uh all the prices are aligned in all the d5 protocols. What flashbots adds on top of it?

B

It allows everyone to participate in value extraction, not only the miners, so it allows everyone to provide the service for the miners of providing this expertise of knowing how to extract mev and taking some of the payments to to the searchers. Instead of all the payment to the miners and without flashbacks. You also have less efficient mep, because the miners don't know how to execute all the liquidations and so on.

A

Well, well, I can't say if the whole mav thing is like overblown or not, but it clearly is an interesting concept and I'm glad to know about flash parts like. uh I think it's important for people to be aware of, especially when you are into mining material.

A

So with that we have hit the hour mark. If you would like to say any final comment for our viewers.

B

um Thank you for listening. um Keep exploring mvv keep researching participate in research, contribute to tooling make ethereum better together with flashbacks, together with other researchers, build your own tools. I think this is most important thing. This is where ethereum drives and has been driving for a long time in the thousands of contributors, thousands of developers, builders, uh building together the future of ethereum and putting a lot of effort into the research. So it's beautiful enemyv will be there. It will stay, but we'll make it just just a good mvp in the long term.

A

Very well, I appreciate you taking out time to manage and talking about mav and flashpot. It was awesome uh with the conversation today. uh I expect to see both maybe and obviously the flash part to be in action in coming years. Even uh you showed some diagrams how it will look like post merge, so I I hope that these presentation will help us understand more than the basics and more contributors will come up mostly, which is required in the area of research.

A

So thank you for that and on this note, for all our youtube viewers, thanks for watching and if you like it, please hit the like button, share it with people of your similar interests. We are leaving a bunch of resources for your references uh in the description section, including the slides and uh guess twitter handle in the description section as usual I'll be looking forward to your feedback. You may leave your comment.

A

Let us know which proposal you would like to get explained.

A

All suggestions are welcome, reach us on twitter or at dtncattle.com.

A

Next week we will be joined by philip kastungay to talk about eip 1271 standard signature, validation, method for contracts. So stay tuned subscribe to our youtube channel. If you haven't done it already, keep watching keep sharing your love with your dream, catalyst see you all in the next episode.