Do you feel that with IPFS using content hashes (CID v1, v2, etc) caching would be an easier problem to solve than fetching content that can change? I had thought IPFS would make this problem easier. And thanks for the detailed response, it has got me thinking about a lot.