Traditional cryptographic hash functions such as SHA-256 are well known in the Ethereum community and are well regarded for security and performance. However, they are not quite suited for SNARKs and STARKs as their circuits are too complex and slow in SNARK/STARK-friendly fields. That's why Zcash proofs originally took more than 40 seconds to create. In this talk we present and call for new designs that will make all the SNARK/STARK crypto much faster and thus bring better privacy to Ethereum. In particular, we cover: - existing algebraic hash functions such as MIMC. - Merkle tree enhancements: wide functions and trees with feedback; - improved algebraic function design using ideas from AES and SASAS schemes. - performance overview; - how new designs will affect EVM and smart contract programming. We will also present other directions and potential research bounties that can be offered to motivate the hash research, crucial for SNARKs, STARKs, and Ethereum.