Any given research protocol entails a trade-off between privacy and accuracy. At one extreme, locking up data so no one can use it gives privacy but no accuracy or utility. At the other, fully open data provides plenty of accuracy and utility, but no privacy. In between are other protocols—like ones using fully homomorphic encryption, multiparty secure computation, or differential privacy—that provide differing combinations of accuracy and privacy. Together, one can imagine all these protocols forming a production possibility set. This grant supports a project by Cornell economist John Abowd to characterize the “efficient frontier” of such protocols. These are ones with the property that no other conceivable protocol could deliver more accuracy without sacrificing some privacy, or more privacy without sacrificing some accuracy. After assembling a library of such protocols, Abowd and his team will explore and measure public attitudes among these protocols and the tradeoffs, helping us understand public preferences toward the tradeoffs between accuracy and privacy.