I'm expecting that the auditors will eventually make the same recommendation.

But I don't think we should go to 1.0 with RIPEMD160 as our hash function.

RIPEMD160 is a bad choice for a few reasons.

1. The birthday bound on RIPEMD160 is 2^80 for a second preimage. A 2nd preimage would at very least allow for state corruption in our database and possibly allow for forgeries

2. RIPEMD160 is roughly the same speed/throughput as SHA256 on current hardware. Future generations of Intel and ARM server hardware have will have native SHA256 instructions that allow massive speed ups. Here is how fast SHA156 is on Ryzen: https://crypto.stackexchange.com/questions/50618/how-fast-can-a-sha-256-implementation-go/50620

3. RIPEMD160 is 10x more expensive to verify on the EVM than sha256.
   https://github.com/ethereum/go-ethereum/blob/master/params/protocol_params.go#L69-L72

Bikeshed

It's a reasonable choice to ask if we should adopt a modern hash function. The consensus among hash function experts is that SHA2 will never be broken and will likely be the most widely supported hash function for the rest of human existence.

BLAKE2b is almost 5x faster in software than RIPE160MD but doesn't currently have ethereum ecosystem compatibility and is unlikely ever to be implemented in widely available hardware. Tho hardware sha256 is only 2x faster blake2b.

Shake128 is kinda slow. Farfalle is very fast but too new.

Ethereum KECCAK is indefensible.