Challenging epistemology: Interactive proofs and zero knowledge (Q959048)

The essay under review is a contribution to both the epistemology of mathematics, with a focus on mathematical practice, and to the philosophy of computer science. Much of it is a philosophical case study on interactive zero-knowledge proof systems, a particular class of interactive probabilistic protocols which was originally designed to clarify the relationship between cryptographic proofs and mathematical proofs [cf. \textit{S.~Goldwasser, S.~Micali} and \textit{C.~Rackoff}, Proc. 17th ACM Symp. Theory of Computing, 291--304 (1985; Zbl 0900.94025)]. Interactive zero-knowledge proofs are defined as dynamic communications between a prover and a verifier where the prover does not provide the verifier with any additional information beyond the truth of his claim. The author explores what research on interactive zero-knowledge proofs has to teach philosophers about the epistemology of mathematics and of theoretical computer science. He concludes that interactive zero-knowledge proofs are not mathematical proofs and argues that any lessons from the work on interactive zero-knowledge proofs for the epistemology of mathematics (such as the need to account for interaction or for probabilistic evidence -- or even proposals to characterize mathematical proofs solely through their persuasive power) are either vintage or misleading. Finally, the author investigates, based on the contrast between mathematical proofs and interactive zero-knowledge proofs, the considerable differences between concepts, perspectives, and epistemic principles adopted by some theoretical computer scientists (complexity theorists, in particular) and those adopted by mathematicians. However, he ultimately remains skeptical about the need to develop a separate epistemology of theoretical computer science outside the epistemology of mathematics.