This is a promising approach. I believe what the author is suggesting is that we construct an unbroken chain of custody binding various real-world proofs of ide

This is a promising approach. I believe what the author is suggesting is that we construct an unbroken chain of custody binding various real-world proofs of identity to an arbitrary address on the blockchain. For instance, If there is a secure and unforgeable way to take periodic photographs of ourselves, which doesn't sound infeasible given the existence of modern facial recognition systems, we could certify the evidence in a mutable fashion using the timestamping data. The general idea is to provide a way to document some uniquely identifiable set of details that only we could have access to at a given time. It doesn't necessarily have to be a photograph. Doctored personal IDs, attestations from third parties, and biometric devices could serve a similar function. All we actually need to do to prove uniqueness is to be the person who maintains this continuously updating record of entries. The main draw back has more to do with the fact that you could still create more than one identity by re-using the uniquely identifying information, in which case we would need some kind of pattern recognition system to analyze the number of times the information has been invoked to create multiple identities in some larger public dataset.