Sign In

Communications of the ACM

ACM TechNews

Differential Privacy the Correct Choice for 2020 U.S. Census

View as: Print Mobile App Share:

Said Sarah Radway, now a Ph.D. at Tufts University, “This is a serious controversy—there was even a lawsuit in a federal court. We showed that the Census Bureau’s judgments were correct, and that they made the right decision.”

Credit: MIT Technology Review

A study of the mathematical concept of differential privacy (DP) by Columbia University computer scientists concluded the U.S. Census Bureau's move to DP as a de-identification mechanism for the 2020 Census was appropriate.

DP maintains the privacy of an individual's personal data by injecting random changes into the data.

There had been concerns that such "noise" would result in artificial deflation of reported minority populations, and a subsequent loss in funding.

The researchers found DP offers a stronger privacy guarantee, while swapping puts a disproportionate privacy burden on minority groups.

Even when implemented with sufficient privacy, the researchers found swapping generally less accurate than DP.

From Columbia Engineering News
View Full Article


Abstracts Copyright © 2022 SmithBucklin, Washington, DC, USA


No entries found

Sign In for Full Access
» Forgot Password? » Create an ACM Web Account