Long-Lived Verifiable Multi-Party Computation

Olivier Pereira

When running a distributed computation, the impact of a long-term loss of privacy can be a central problem, while the impact of a long-term loss of verifiability is often null: the computation is over, and becoming unable to verify the results after several years has no consequence.

Based on this observation, we develop efficient protocols for the evaluation of functions getting their secret inputs from multiple parties in a way that can be verified for a bounded amount of time, while making sure that the information disclosed for this verification will never leak any secret, even in the long-term.

Our protocols are based on a traditional worker-client setting, in which clients delegate a joint computation on secret data to one or more workers, who provide the computation result and a proof of correctness in a non-interactive way.

We implemented our protocols and show various applications in the context of voting, auctions and linear optimization. In many applications, the verification task of the clients can be made cheaper than the cost of performing the computation itself, with the additional confidentiality benefits.