I personally think that HTTP should be gradually supplanted by things like IPFS, all documents would be content-addressable and both accessing an archiving would be done in a decentralized manner.