The Rate of Cryptography

Our society increasingly relies on computations over the Internet. Typically, these follow a centralised paradigm, where clients delegate the computation to a trusted authority, giving it access to the inputs. This inevitably raises security concerns: what happens if the authority’s server were to be taken over by an attacker?

Secure multiparty computation (MPC) studies protocols that avoid this type of single points of failure, ensuring the privacy of the inputs and the correctness of the outputs even if a subset of participants were to be corrupted.
In my research, I investigate the cost of these solutions: given any notion of complexity, what is the maximum ratio between the cost of the insecure protocol and that of the secure solution?

In this talk, I describe how I have tackled this question by studying simpler primitives: distributed samplers, homomorphic secret-sharing, trapdoor hashing and laconic function evaluation. I give a brief discussion of these problems and an overview of open questions I would like to explore.