it’s “writing back” to an EVM that’s hard
Let me see if I can relay my thinking here, by setting up a scene in Q2:
- RiscZero ZK proving with settleability with STARK to SNARK into Ethereum is fully open source and generally available
- We’re post Eigenlayer launch. The norm of co-processing with high (restaked) crypto economic security is normal, and we probably have Cartesi as a co-processor available, or a WASM one, or whichever (get a large sample or quorum of an honest majority). Cartesi as a co-processor with Eigenlayer (separate from Rollups product track)
- There’s reasonable DA infrastructure available such as Celestia or EigenDA or Espresso or Protodanksharding
- We’re talking about 100,000’s of appchains of all sizes and shapes, not a few.
So we can generally assume that some subset of problems are no longer, for ‘instant’ computation, limited to the capabilities of the EVM or lack of expressability - we can at bare minimum assume ability to write reasonable Rust programs and prove the results to Ethereum. We can assume data amount is no longer a problem (block size limits). This gives us the ability to look at this topic with fresh eyes and perhaps do more useful computation relating to being a client “on” EVM.
Appchains (or app specific rollups or whatever name) will exist in a multi-appchain world. It’s not just only about proving to Ethereum L1 or L2. It’s about proving to each other, too. Appchains are like islands with many types of bridges/ferries between them. Also, for good measure: Lambada won’t prove anything but resulting CID – not machine state hash, to others. Machine state hash is a technical detail and offers near-zero value to most developers.
See how things look over in Cosmos: https://twitter.com/0xPajke/status/1742744018553807181 (also, I will expect someone to write that for Cartesi)
Classical, full-blown Ethereum optimistic rollup security is the epoch, claims, fraud proofs (Dave), and 7 days delay with ideally L1 DA.
Do 90% of our appchain use cases need that for their assets and bridges? Would they be perfectly fine with 20% of that security from a set of co-processors? Would it be better to add the fraud proof on co-processor slashing instead of on every single appchain?
The best way of finding that out is to create the ability to choose and grow as an appchain and progressively become more secure.
The short answer is:
We can do more than just classical security. And it’s a new world to understand how everything fits in a multi-appchain multi-bridge world. We’ll see Eigenlayer based bridges for even other blockchains (Filecoin, etc).
For most of our initial Lambada customers, they’d be perfectly happy with a sampling from something like Witness Chain to represent their state. In my humble opinion.