Have your data and hide it too: An introduction to differential privacy

Many applications rely on user data to deliver useful features. For instance, browser telemetry can identify network errors or buggy websites by collecting and aggregating data from individuals. However, browsing history can be sensitive, and sharing this information opens the door to privacy risks. Interestingly, these applications are often not interested in individual data points (e.g. whether a particular user faced a network error while trying to access Wikipedia) but only care about aggregated data (e.g. the total number of users who had trouble connecting to Wikipedia).

The Distributed Aggregation Protocol (DAP) allows data to be aggregated without revealing any individual data point. It is useful for applications where a data collector is interested in general trends over a population without having access to sensitive data. There are many use cases for DAP, from COVID-19 exposure notification to telemetry in Firefox to personalizing photo albums in iOS. Cloudflare is helping to standardize DAP and its underlying primitives. We are working on an open-source implementation of DAP and building a service to run with current and future partners. Check out this blog post to learn more about how DAP works.

DAP takes a significant step in the right direction, Continue reading