Unprecedented quantities of data that could help social scientists understand and ameliorate the challenges of human society are presently locked away inside companies, governments, and other organizations, in part because of worries about privacy violations. Differential Privacy is a technique that might help change this—a technological solution to the political problem of data sharing. Differential privacy provides mathematical guarantees to protect the privacy of individuals who may be in the data while simultaneously making it possible for social scientists to gain insights into societal level patterns and relationships. However, many differentially private solutions are statistically invalid—giving biased estimates or no measures of uncertainty. We aim to fix this. Just as differential privacy can protect individuals, we are developing statistically valid differentially private systems that will also protect researchers and society so that we can be confident of scholarly conclusions, policy advice, and personal recommendations.
As part of this effort, we are partnering with Microsoft and the Privacy Tools Project (at SEAS) to incubate OpenDP, a set of open source software tools for privacy-protective statistically valid analyses of sensitive personal data, and the first end-to-end system we've built is WhiteNoise. We are also working together to build a broader OpenDP community with stakeholders and contributors from across academia, industry, and government. Together, we will design, implement, and govern an “OpenDP Commons” that includes a library of differentially private algorithms and other general-purpose tools for use in end-to-end differential privacy systems. We are also integrating our tools into IQSS' Dataverse project.
For generous contributions to the Privacy Insights Project, we thank the Alfred P. Sloan Foundation, Microsoft, Facebook, and private donors.