Making on chain ZK proofs sustainable with a refined L1 gas pricing model

Gas is for balancing the EVM’s usage against runtime execution and funding network maintenance.

This is a wild suggestion, but consider this:

If we take lower gas fees for a selected few pure functions by refunding gas spent on them post execution, the original EOA invoker still has to commit for gas without assuming any price reduction. That way, everyone can agree on a “standard library” for zero knowledge proofs and it would be much more reasonable to compute ZK proofs on chain. By agreeing on protocol rules set for gas pricing on standard library functions that implement the gist above, gas will still govern balanced execution, UX does not change, stakers will still be incentivized to maintain the network, and it will be more affordable to leverage ZK proofs.

What do you think? Is that a path for incorporating ZK proofs on chain better?