Odaily News Ethereum researcher Justin Drake published a post on the ethresearch forum discussing a new rollup design called “native rollups.” In short, it relies on the Ethereum L1 validator for proof, namely the state transition function and verification.
This contrasts with Optimism Rollup (e.g. Optimism, Arbitrum) or zk-Rollup (e.g. Starknet, ZKsync), which push the computational burden of execution to L2 and then rely on fraud or zk proof systems to generate state roots and proofs back to mainnet.
These proof systems are code-heavy and prone to bugs and other vulnerabilities, which is why Rollup sequencers (the entities that order transactions on L2) have historically been centralized. Concerns about sequencer centralization have in turn spurred “based” Rollup designs such as Taiko, which rely on Ethereum L1 validators to perform ordering.
Drake’s proposal suggests introducing “execute” precompiles (hardcoded functions in the EVM) that will validate the EVM state transitions of user transactions. Native rollups achieve the following two breakthroughs:
No longer need to invest in and maintain expensive miner-prover networks and specialized GPU hardware, as proofs will be processed and executed by L1 validators;
There is no longer a need to maintain a complex governance structure including a trusted security committee to approve contract upgrades to achieve EVM equivalence.
This actually makes native rollup "trustless" by inheriting the security of Ethereum L1.
Finally, like based rollups, native rollups will enjoy “synchronous composability” and will not be limited by the 12-second block time. Thanks to the “execution” precompilation, L1 validators only need to verify the zk proof without having to perform the calculations themselves.