Thread by @norswap

I think the mempool is going to be one of the next blockchain module where we will see a lot of innovation.

Encrypted mempools, side-pools, app-specific mempools, custom appchain mempools, ... read all about it below!

First, at the base layer, with @project_shutter and Flashbots SUAVE building encrypted mempools.

But we'll also see the rise of new "side pools", for instance to support account abstraction via EIP-4337.

The 4337 mempool broadcasts "user operations" instead of traditional transactions, which are later bundled and posted on-chain by bundlers.

See the EIP (eips.ethereum.org/EIPS/eip-4337) and this thread by @VitalikButerin which explains the possibilities and remaining problems:

https://twitter.com/VitalikButerin/status/1576199517434949634

EIP-4337 needs a "general" mempool.

I think we will also see the emergence of app-specific mempools.

In fact, this already exists, for instance in @CoWSwap.

These mempools allow doing some pre-processing, matching, and aggregation in the mempool directly. For instance, CoWSwap matches buy and sell orders off-chain in order to minimize slippage.

Quick sidebar: CoWSwap is really cool because this matching is itself decentralized, a competition between "batch builders" to get the best price for users. See this great talk by @AnnaMSGeorge for more details:

Often, the possibilities offered by these app-specific mempools hinges of the notion of "intent": the users sign-off on *what* they want accomplished, not *how* they want it accomplished.

Specialized actors then monitor the mempool and try to fullfill these intents as well as possible.

For CoWSwap, this is done through order matching.

But you could go a lot more generic! You could imagine a system where an intent would grant specific permissions (e.g. unlock assets), but the transaction would revert if the specified conditions are not fullfilled by the end of the transaction.

If I understand correctly, @anomanetwork is trying to build such a system. But you don't necessarily need a whole new system to do this. There is a TON of innovation that can be driven via a new mempool module + smart contracts on EVM chains.

For instance, I could sign an intent that says: on a budget of 1 $ETH, I want to purchase any three #motorheadz NFTs, to be carried out within the next 3 hours.

If a "builder" finds a way to carry the trade, he can use my signed intent to unlock the ETH, purchase the NFTs, and transfer them to me!

Another example: optimistic game resolution. Player trade a series of intents, representing game moves. If everybody respects the rules, it's only necessary to post a single transaction on-chain, representing the agreed-upon result of the game.

But if a player tries to cheat, you could use the intents (representing the game history) to build a proof of cheating on-chain.

If the intents are traded privately, this is a solution reminiscent of state channels.

If the intents are broadcasted in the mempool, mempool watchers can act as referees, and mount a proof of cheating on behalf of the player that has been wronged.

These app-specific mempools can of course be integrated in app-specific chains!

And of course, it's my dearly held opinion that all app-chains should be app-specific *rollups* with Ethereum as their settlement layer!

This is of course where I plug the #OPStack, our initiative at OP Labs to build a modular rollup stack where it is easy to mix and match parts!

Bedrock (next mainnet release) is the foundation, but we want to push the easy of customization to the max, and offer a wide selection of open-source modules!

But we'll need help to do this. If you want to build on the OP Stack, work on new modules, and help push forward the future of blockchain tech and modularity, don't hesitate to be in touch!

DMs are open!

smh I forgot to number this tweet thread :D