Proof of Knowledge

cryptography
pull zero-knowledge-proof
we know something if we can compute it
if we want to demonstrate knowledge of something, demonstrate either
- we can compute it efficiently
- we already have it
imagine the same ZKP protocols where the prover is modelled as a black box
- we allow rewinding this black box
- "show me edge A-B" and "show me edge A-C" answered by the black box independently for e.g. 3-color problem
- this allows computing things that aren't possible to compute in the simulation model, because we can change the distribution of transcripts we see to a subset (transcripts prefixed by up to our rewinding point)