ZK-gossip for lightning channel announcements

The discussion highlights the performance of cryptographic proofs, specifically focusing on the creation time for STARK and Groth16 proofs on different hardware setups. The STARK proof creation on an M1 Max laptop with 32GB of RAM takes approximately 80 seconds. However, creating a Groth16 (wrapped) version of the proof on non-x86 hardware is not currently feasible. To address this limitation, the use of a proof server provided by RISC0 allows for the creation of fully wrapped proofs in less than 30 seconds, suggesting the server utilizes powerful GPUs to achieve these speeds. This setup hints at the potential for outsourcing proof wrapping to non-trusted servers without compromising privacy since wrapping the proof does not expose private inputs.

Further exploration into optimization is considered crucial for reducing proof creation times. Very little effort has been devoted to profiling the Zero-Knowledge (ZK) application thus far. RiscZero offers a comprehensive guide on optimizing ZK Virtual Machines (ZKVM), available at RiscZero Optimization Guide. Additionally, alternative implementations of zkVM, such as SP1, claim to offer better performance, which warrants further investigation.

The conversation also touches upon the efficiency and size constraints of proofs, especially in the context of their application in networks like Lightning Network (LN). While current proof sizes might be acceptable, there's a suggestion that verification times need to significantly improve to become practical, aiming for sub-second or preferably sub-100ms on standard hardware. Although longer proof times are deemed acceptable due to the inherent delays in blockchain operations, the size of proofs is a concern for efficient distribution within the LN, hinting at a possible need for proofs to be compact enough to fit within a maximum message size of 65kB.