This post for a big part was inspired by Jon and Frontier’s work on MEV and sequencers, will be extending and diving into some of the ideas that would help with better block production and transaction flow in the future for rollup and alt L1 space.

• What are Rollups?

• Architecture of rollups

• What are Sequencers and MEV?

• Shared sequencers and different approaches.

Rollups have been regarded as the future of Ethereum scaling because they solve the congestion issues on the mainnet and partly because base layer Sharding is out of the picture for Ethereum but as the focus on Rollups increased they are still limited by the Data Availability and bandwidth of the base chain. This could be partially solved by the implementation of Proto-Danksharding which introduces blob-carrying transactions but that would still be limited to 2 X 125kb blobs per block so in crypto, blockspace is a commodity like real estate. Everyone wants to be at the hotspot where the activity/nightlife is good but as more people pour(Demand ∝Price) the price shots up so does the eth gas fee as more people are willing to pay for their transactions to be included in the latest block but one of the ways to solve this is built taller buildings and accommodate more people in the same footprint which is what rollups do. In simple terms, Rollups just compress a batch of transactions and post their proof of execution on the mainnet to prove their validity.

This way they can accommodate more transactions in the footprint of very few transactions saving blockspace. As rollups compress and do off-chain verifiability which means that the whole transaction data won't be on the mainnet but off-chain data stored in the machines that produced the proof of your transaction or the light clients.

• In this pipeline, Users send a transaction that first goes to a rpc/relayer which acts as the entry point and is then sent to sequencers that make blocks and submit them to proof producers. One machine could perform multiple operations, maybe the sequencer is also the proof producer depending on the case But why this architecture?

This is different from the relayer directly sending the transactions to the validator scheduled to produce blocks because of Proposer Builder separation(PBS). To make the system more censorship-resistant and save users, PBS was introduced in which the role of a block builder and proposer was separated. Now multiple block builders make blocks by packing transactions from users and only sharing the block headers with the amount of gas fee reward for Proposer but the transaction themselves are not shared. The proposer chooses the blocks with the highest rewards and includes them in the slot which still aligns the validator values to make more profits and user experience.

In older architecture, a block builder/validator would directly choose the transactions to include in the block which led to user exploitation(aka MEV) where Builders could arbitrage the user orders by front running them and taking profits from the slippage thus disincentivizing users like in traditional Finance where big HFT firms pay to get their supercomputers inside the exchange itself so they could read and write before the wider world and front run the opportunities which Defi is trying to solve by providing everyone the same data at the same time. MEV is basically the arbitrary reordering or insertion of transactions by block producers to increase the value extraction. Though there is nothing inherently bad about MEV because in an open permissionless system, agents work to maximize their output but it can hurt users.

These Block producers/sequencers are very important for the transaction pipeline to work efficiently as they have the power to include and drop any transaction but it becomes even more important in the rollup space as right now most of the big optimistic and zk rollups have centralized sequencers meaning that no one can run their own node outside of the team. This assumes a high degree of trust by the users and a security assumption of 1 of 2 as even if the sequencer drops the transaction, the user can still post their data directly to the rollup contract which would in the end defeat the purpose of using a rollup to save gas but it has some Big pros for its users and teams->

• Soft Conformation: As rollups post transaction proofs or batches where the size of the proof does not increase at the same rate as the number of transactions it's economical for them to batch as many transactions and find the optimal time to post on the chain. Though this approach is great for saving gas but comes at the cost of user experience as someone posting a transaction would have to wait for till the proof is posted, by using single sequencers the rollup can give a soft confirmation and giving the final hard confirmation later when its posted as there is no one to interfere the proof production other than the founders.

• Extractions: Being able to maximize the gas and reward from a block incentivizes the team to improve the liveliness and security of the rollup.

We can't forget at the end that the biggest ethos of being in crypto is Decentralisation and on that front rollups of today don't look quite good but teams are working + progressing on their roadmap of better decentralization but some cons are ->

• User experience: Recently after the Arbitrum airdrop, due to high load the Arbitrum sequencer hit the rate limit, and because of it being a centralized sequencer system the network was not able to process transactions for a few hours.

• Censorship: With most of the sequencers based out of the USA and the recent OFAC ban on Tornado Cash, the teams would have to be responsible to censor the transactions that tried to interact with the tornado cash smart contract. As a big part of the eth block producers were using the same MEV framework of Flashbots they were of complaint meaning censoring the tornado cash transactions.

One of the ways to handle this is open-sourcing the sequencer as Starknet did with Blockifier and Lambda team made another implementation of the sequencer but this has also led to long confirmation times for the starkex ecosystem and loss of MEV to the team which I think is a good enough tradeoff if you are looking for decentralization and security but wait what if there is a better approach? Glad you asked about it my fren

To push for decentralization and better block production than centralized sequencers, Shared sequencers where rollups don’t have specified relays for their block production but rely on a network of shared relayers. It aggregates the security and liveliness of many rollups into a single layer eg Espresso

Credits: DBA from Jons Post

There are a lot of approaches to better transaction sequencing but Shared sequencers get a little bit good part of both worlds(a bit bad too)

• Better Security & Trust: Shared sequencers inherit better security as it's a network of participants and you just need to trust that at least 1 of N sequencers will pick up your transaction rather than 1 of 1 or 1 of 2 security assumptions in general rollup systems.

• Cross Domain MEV: It would allow better cross-rollup communication and better arbitrage opportunities as the data from different state layers are readily available. This is good for rollup ecosystems like Arbitrum but App chains that see MEV as their biggest revenue stream wont have much value alignment with Shared Sequencers.Also, check out this paper going into Cross domain MEV

• Competition: The introduction of more participants introduces competition among agents to improve their services and quote fair prices.

• No Soft Conformation

L1 enriched aka based rollups share their sequencers with L1 and inherit the security and censorship-resistant as they follow the same PBS system as the L1. Scheduled Builders and searchers would also be able to produce blocks for the rollup taking away the authority of centralized systems. Some Ws

• Decentralization: It would allow L1 PBS to extract MEV for the rollups driving up their rewards and making it attractive for more folks to run Builder/proposer nodes for the base layer.

• Liveliness: Based rollups inherit their liveliness from L1 as the block production pipeline is the same.

Some Ls

• MEV: It's one of the biggest topics of consideration in any transaction supply chain discussion. If we get based rollups then the app chains will give up on one of their prime income sources which is MEV as fees on rollups as supposed to be less they would lose a huge chunk of their revenue. Having an application fee where users pay a fixed amount of fee for interaction with the dapp.

• Conformations: Having centralized sequencers give the rollups the ability to give soft conformations as blocks are built by the founding team which has some degree of trust that they won't do a reorg or produce an invalid state transition. Having shared PBS will lead to longer wait times as we would have to wait for the fraud-proof window to expire.

Cosmos ecosystem stands out as the biggest advocate for modularity and app chain thesis where every dapp has its sub-chain that inherits security from the shared protocol, in this case, IBC(Inter Blockchain communication) protocol. This opens up a huge world of possibilities for dapp developers giving them the flexibility that would not be possible with monolithic chains or would require whole network upgrades but it also comes with its challenges. Ethereum with its few rollups faces backlash because of its immature sequencer ecosystem but with the cosmos having thousands of chains the problem is amplified where each chain has to set up its own sequencer and builder. One of the ways to solve this is having a shared sequencer network kinda like what Astria is building where The sequencer network produces blocks containing transactions for multiple rollups and directly to the data availability layer and rollups retrieve transactions from there and produce new state roots for hard conformations.

With MEV and a few different factors to maximize rewards, the Ethereum block production started centralizing and a few entities producing more than 60% of the blocks to tackle this with a better system than having Shared sequencers, the flashbots team released SUAVE (Single Unifying Auction for Value Expression) which would act as a middleware for block production by building a credibly neutral encrypted mempool for transaction flow which will be resistant to bad MEV(to some degree) and being chain agnostic it would accelerate cross chain MEV extraction. With better MEV it would attract more agents to become a part of the network and better decentralization.

SUAVE is more like a shared builder rather than just a sequencer which works as a plug-and-play for any chain but one of the considerations is that it would have to report all the state transitions happening due to its transactions back to the builders.

In this post I explained how rollups work and what are some good ways to reach a better level of sequencers decentralization, I never looked on the MEV front before this but it was a fascinating ride learned a lot about how complex actors collude to just verify a transaction on-chain. There are pros and cons to every approach but having a neutral shared sequencer network is going to be important for credible rollups in the future.

How would having a shared mempool of transactions for all the rollups and side chains affect the rent paid by them to the base layer?Right now rollups pay huge amounts of eth to the base layer as their DA and conformation layer, how does that look like in the future with sovereign rollups and many using Celestia as their DA but ethereum mainnet as their conformation layer? Might explore this in the next post!

PS: Thanks if you read this till the end! I am still learning so if you see any mistakes or corrections feel free to reach out!

Resources used->

• Unity is Strength: A Formalization of Cross-Domain Maximal Extractable Value

• Rollups Aren’t Real

• Flash Boys 2.0 paper

• Beyond Extraction

• Market Friedly PBS

• Tiers of Finality

• Dune Dashboards

This post for a big part was inspired by Jon and Frontier’s work on MEV and sequencers, will be extending and diving into some of the ideas that would help with better block production and transaction flow in the future for rollup and alt L1 space.

• What are Rollups?

• Architecture of rollups

• What are Sequencers and MEV?

• Shared sequencers and different approaches.

Rollups have been regarded as the future of Ethereum scaling because they solve the congestion issues on the mainnet and partly because base layer Sharding is out of the picture for Ethereum but as the focus on Rollups increased they are still limited by the Data Availability and bandwidth of the base chain. This could be partially solved by the implementation of Proto-Danksharding which introduces blob-carrying transactions but that would still be limited to 2 X 125kb blobs per block so in crypto, blockspace is a commodity like real estate. Everyone wants to be at the hotspot where the activity/nightlife is good but as more people pour(Demand ∝Price) the price shots up so does the eth gas fee as more people are willing to pay for their transactions to be included in the latest block but one of the ways to solve this is built taller buildings and accommodate more people in the same footprint which is what rollups do. In simple terms, Rollups just compress a batch of transactions and post their proof of execution on the mainnet to prove their validity.

This way they can accommodate more transactions in the footprint of very few transactions saving blockspace. As rollups compress and do off-chain verifiability which means that the whole transaction data won't be on the mainnet but off-chain data stored in the machines that produced the proof of your transaction or the light clients.

• In this pipeline, Users send a transaction that first goes to a rpc/relayer which acts as the entry point and is then sent to sequencers that make blocks and submit them to proof producers. One machine could perform multiple operations, maybe the sequencer is also the proof producer depending on the case But why this architecture?

This is different from the relayer directly sending the transactions to the validator scheduled to produce blocks because of Proposer Builder separation(PBS). To make the system more censorship-resistant and save users, PBS was introduced in which the role of a block builder and proposer was separated. Now multiple block builders make blocks by packing transactions from users and only sharing the block headers with the amount of gas fee reward for Proposer but the transaction themselves are not shared. The proposer chooses the blocks with the highest rewards and includes them in the slot which still aligns the validator values to make more profits and user experience.

In older architecture, a block builder/validator would directly choose the transactions to include in the block which led to user exploitation(aka MEV) where Builders could arbitrage the user orders by front running them and taking profits from the slippage thus disincentivizing users like in traditional Finance where big HFT firms pay to get their supercomputers inside the exchange itself so they could read and write before the wider world and front run the opportunities which Defi is trying to solve by providing everyone the same data at the same time. MEV is basically the arbitrary reordering or insertion of transactions by block producers to increase the value extraction. Though there is nothing inherently bad about MEV because in an open permissionless system, agents work to maximize their output but it can hurt users.

These Block producers/sequencers are very important for the transaction pipeline to work efficiently as they have the power to include and drop any transaction but it becomes even more important in the rollup space as right now most of the big optimistic and zk rollups have centralized sequencers meaning that no one can run their own node outside of the team. This assumes a high degree of trust by the users and a security assumption of 1 of 2 as even if the sequencer drops the transaction, the user can still post their data directly to the rollup contract which would in the end defeat the purpose of using a rollup to save gas but it has some Big pros for its users and teams->

• Soft Conformation: As rollups post transaction proofs or batches where the size of the proof does not increase at the same rate as the number of transactions it's economical for them to batch as many transactions and find the optimal time to post on the chain. Though this approach is great for saving gas but comes at the cost of user experience as someone posting a transaction would have to wait for till the proof is posted, by using single sequencers the rollup can give a soft confirmation and giving the final hard confirmation later when its posted as there is no one to interfere the proof production other than the founders.

• Extractions: Being able to maximize the gas and reward from a block incentivizes the team to improve the liveliness and security of the rollup.

We can't forget at the end that the biggest ethos of being in crypto is Decentralisation and on that front rollups of today don't look quite good but teams are working + progressing on their roadmap of better decentralization but some cons are ->

• User experience: Recently after the Arbitrum airdrop, due to high load the Arbitrum sequencer hit the rate limit, and because of it being a centralized sequencer system the network was not able to process transactions for a few hours.

• Censorship: With most of the sequencers based out of the USA and the recent OFAC ban on Tornado Cash, the teams would have to be responsible to censor the transactions that tried to interact with the tornado cash smart contract. As a big part of the eth block producers were using the same MEV framework of Flashbots they were of complaint meaning censoring the tornado cash transactions.

One of the ways to handle this is open-sourcing the sequencer as Starknet did with Blockifier and Lambda team made another implementation of the sequencer but this has also led to long confirmation times for the starkex ecosystem and loss of MEV to the team which I think is a good enough tradeoff if you are looking for decentralization and security but wait what if there is a better approach? Glad you asked about it my fren

To push for decentralization and better block production than centralized sequencers, Shared sequencers where rollups don’t have specified relays for their block production but rely on a network of shared relayers. It aggregates the security and liveliness of many rollups into a single layer eg Espresso

Credits: DBA from Jons Post

There are a lot of approaches to better transaction sequencing but Shared sequencers get a little bit good part of both worlds(a bit bad too)

• Better Security & Trust: Shared sequencers inherit better security as it's a network of participants and you just need to trust that at least 1 of N sequencers will pick up your transaction rather than 1 of 1 or 1 of 2 security assumptions in general rollup systems.

• Cross Domain MEV: It would allow better cross-rollup communication and better arbitrage opportunities as the data from different state layers are readily available. This is good for rollup ecosystems like Arbitrum but App chains that see MEV as their biggest revenue stream wont have much value alignment with Shared Sequencers.Also, check out this paper going into Cross domain MEV

• Competition: The introduction of more participants introduces competition among agents to improve their services and quote fair prices.

• No Soft Conformation

L1 enriched aka based rollups share their sequencers with L1 and inherit the security and censorship-resistant as they follow the same PBS system as the L1. Scheduled Builders and searchers would also be able to produce blocks for the rollup taking away the authority of centralized systems. Some Ws

• Decentralization: It would allow L1 PBS to extract MEV for the rollups driving up their rewards and making it attractive for more folks to run Builder/proposer nodes for the base layer.

• Liveliness: Based rollups inherit their liveliness from L1 as the block production pipeline is the same.

Some Ls

• MEV: It's one of the biggest topics of consideration in any transaction supply chain discussion. If we get based rollups then the app chains will give up on one of their prime income sources which is MEV as fees on rollups as supposed to be less they would lose a huge chunk of their revenue. Having an application fee where users pay a fixed amount of fee for interaction with the dapp.

• Conformations: Having centralized sequencers give the rollups the ability to give soft conformations as blocks are built by the founding team which has some degree of trust that they won't do a reorg or produce an invalid state transition. Having shared PBS will lead to longer wait times as we would have to wait for the fraud-proof window to expire.

Cosmos ecosystem stands out as the biggest advocate for modularity and app chain thesis where every dapp has its sub-chain that inherits security from the shared protocol, in this case, IBC(Inter Blockchain communication) protocol. This opens up a huge world of possibilities for dapp developers giving them the flexibility that would not be possible with monolithic chains or would require whole network upgrades but it also comes with its challenges. Ethereum with its few rollups faces backlash because of its immature sequencer ecosystem but with the cosmos having thousands of chains the problem is amplified where each chain has to set up its own sequencer and builder. One of the ways to solve this is having a shared sequencer network kinda like what Astria is building where The sequencer network produces blocks containing transactions for multiple rollups and directly to the data availability layer and rollups retrieve transactions from there and produce new state roots for hard conformations.

With MEV and a few different factors to maximize rewards, the Ethereum block production started centralizing and a few entities producing more than 60% of the blocks to tackle this with a better system than having Shared sequencers, the flashbots team released SUAVE (Single Unifying Auction for Value Expression) which would act as a middleware for block production by building a credibly neutral encrypted mempool for transaction flow which will be resistant to bad MEV(to some degree) and being chain agnostic it would accelerate cross chain MEV extraction. With better MEV it would attract more agents to become a part of the network and better decentralization.

SUAVE is more like a shared builder rather than just a sequencer which works as a plug-and-play for any chain but one of the considerations is that it would have to report all the state transitions happening due to its transactions back to the builders.

In this post I explained how rollups work and what are some good ways to reach a better level of sequencers decentralization, I never looked on the MEV front before this but it was a fascinating ride learned a lot about how complex actors collude to just verify a transaction on-chain. There are pros and cons to every approach but having a neutral shared sequencer network is going to be important for credible rollups in the future.

How would having a shared mempool of transactions for all the rollups and side chains affect the rent paid by them to the base layer?Right now rollups pay huge amounts of eth to the base layer as their DA and conformation layer, how does that look like in the future with sovereign rollups and many using Celestia as their DA but ethereum mainnet as their conformation layer? Might explore this in the next post!

PS: Thanks if you read this till the end! I am still learning so if you see any mistakes or corrections feel free to reach out!

Resources used->

• Unity is Strength: A Formalization of Cross-Domain Maximal Extractable Value

• Rollups Aren’t Real

• Flash Boys 2.0 paper

• Beyond Extraction

• Market Friedly PBS

• Tiers of Finality

• Dune Dashboards