Author: Simson Garfinkel, Kyle Polich
Publisher: Data Skeptic
Publication Year: 2020
Summary: The following podcast episode features an interview with Simson Garfinkel who talks about his work with the U.S. Census and how they have changed the ways they are protecting data privacy for the 2020 Census. This new “Differential Privacy” method is more open to the public (with the code being published for anyone to look at) than previous privacy methods based on secrecy. This new method involves destroying the “encryption key” (the random numbers that are applied to the raw data to form the published data). Without this key, there is no way to back-calculate what the raw data is. This allows the code that applies the data transformation to be published so there is transparency and the public knows that their privacy is secured. This new privacy method, however, does present some potential problems that data scientists that have used census data in the past will need to be aware of