Nearly four years ago the U.S. Commission on Evidence-Based Policymaking unanimously agreed that the country needed new strategies and supports for advancing privacy-preserving technologies. These are the innovative capabilities that enable sharing of information in an increasingly-protected manner – techniques such as multi-party computation, homomorphic encryption, and synthetic data.
Multiple demonstration projects and pilot projects successfully displayed the value of these approaches in recent years, as well as the importance of clear communication with public administrators about the role, need, and potential for privacy-preserving technologies in practice. For example, a project in partnership with Allegheny County, Pennsylvania applied multi-party computation to human services data, successfully demonstrating that disparate systems could share and link records that are cryptographically protected while also producing relevant and meaningful summary insights.
During a dialogue hosted by the Data Foundation with experts in privacy-preserving technologies, panelists shared perspectives from their own experience in deployment, use, and future potential. Chris Sadler with New America’s Open Technology Institute, Mayank Vaira with Boston University, Ninu Khazanie with Google, and Amy O’Hara with Georgetown University collectively focused not just on how to explore and research new technologies, but also practical applications and strategies for adoption and use of the existing technologies in appropriate contexts.
Adopting the approaches in practice is realistic, possible, and potentially cost-effective. The former Boston Mayor Marty Walsh, now the US Labor Secretary, previously supported a wage study led by a team at Boston University that was a low-cost strategy for generating new knowledge about wage gaps while protecting confidential information. Communication and translation of the role these efforts play in supporting and enhancing data analysis capabilities was a well-agreed upon point in the expert discussion.
As the privacy-preserving technologies are increasingly applied in governmental contexts, a clear need for accountability and transparency will also be vital and operationalized through explainability and replicability. In other words, these technologies can actually enable replicability in new ways, while also publishing and sharing approaches using open standards across a data enterprise implementing the techniques.
Looking forward to future adoption of these approaches at scale will require additional pilot projects in government as well as investments by domestic, non-intelligence agencies. Through additional testing the possibilities of production uses can be further assessed and the value proposition more clearly explained to recognize and realize the value of privacy-preserving technologies.