I don’t understand where people get this misconception that ZK proofs are faster or can somehow make a system faster. It’s completely wrong. They can lower some of the costs for replication, but they can’t improve performance if replication cost isn’t a blocker. Let’s assume that zk proofs are completely free and instant. If sizeof(tx) < sizeof(state change) then it’s nearly always faster to download the tx and execute it locally because local memory bandwidth and latency is better than network bandwidth and latency. It would take more time to download the state result from the transaction than downloading the transaction itself and using local copy of the state to execute it and produce the result. So the only place where zk proofs actually help is if replication is expensive. Where is replication insanely expensive? On ethereum L1. It’s cheaper to post a state root and a zk proof so the L1 can skip executing all the state.

I don’t understand where people get this misconception that ZK proofs are faster or can somehow make a system faster. It’s completely wrong. They can lower some of the costs for replication, but they can’t improve performance if replication cost isn’t a blocker. Let’s assume that zk proofs are completely free and instant. If sizeof(tx) < sizeof(state change) then it’s nearly always faster to download the tx and execute it locally because local memory bandwidth and latency is better than network bandwidth and latency. It would take more time to download the state result from the transaction than downloading the transaction itself and using local copy of the state to execute it and produce the result. So the only place where zk proofs actually help is if replication is expensive. Where is replication insanely expensive? On ethereum L1. It’s cheaper to post a state root and a zk proof so the L1 can skip executing all the state.

the official solana account when a chain other than solana is accelerating