So there we go, that’s an introduction to the power of erasure coding – arguably the single most underhyped set of algorithms (except perhaps SCIP) in computer science or cryptography. The ideas here essentially are to file storage what multisig is to smart contracts, allowing you to get the absolutely maximum possible amount of security and redundancy out of whatever ratio of storage overhead you are willing to accept.
I’m not sure it’s under-hyped (major storage vendors who have products based on it are hyping them and those products are successful and widely known), but it’s an interesting read for those who aren’t familiar with it.
Edit: since this is my post it’s probably OK for me to add that I thought about erasure coding as a potentially better way to protect data around the network, but I thought it wouldn’t work well because it’d need a lot more communication among the nodes.
Storage products (storage arrays) based on erasure coding have been on the market for around for 10+ years, but I’m not aware of decentralized storage that uses such algos (I imagine it’s because it must be difficult to do over network).