Protecting Privacy with MATH (Collab with the Census)

The video discusses the challenges of maintaining privacy in large surveys like the US Census. It explains the inherent trade-off between data accuracy and privacy, highlighting the mathematical techniques used to protect individual data. The video also covers the advancements in privacy protection, particularly in the 2020 Census, and emphasizes the importance of balancing privacy with the need for accurate data.

What is the primary goal of surveys like the US Census?

• To promote cultural events

• To increase government revenue

• To collect personal opinions

• To gather demographic data for political and social insights

Why is it challenging to maintain privacy when publishing survey results?

• Because powerful computers can reconnect data to individuals

• Because surveys are not accurate

• Because removing names from data is sufficient

• Because surveys are too expensive

How can attackers use published statistics to deduce private information?

• By hacking into the Census Bureau

• By asking participants directly

• By using brute force computational power to find plausible data combinations

• By guessing randomly

What is a 'plausibility peak' in the context of privacy?

• A point where data is most accurate

• A point where data is completely random

• A point where some data combinations are more likely than others

• A point where all data is equally likely

What is the purpose of adding random noise to published data?

• To increase the size of the dataset

• To make the data more interesting

• To protect privacy by making data combinations less distinguishable

• To confuse the data analysts

How does adding noise to data affect its accuracy?

• It makes the data completely inaccurate

• It increases accuracy

• It has no effect on accuracy

• It reduces accuracy but protects privacy

What is the trade-off when adding noise to data for privacy?

• Less privacy means more data

• More privacy means more accuracy

• More privacy means less accuracy

• Less privacy means less data

What was a significant change in the 2020 Census regarding privacy?

• It stopped collecting data

• It used mathematically rigorous privacy protections for the first time

• It was conducted entirely online

• It included more personal questions

Why is it important to have a mathematically robust way of protecting privacy?

• To reduce the cost of surveys

• To make data collection faster

• To guarantee there are no peaks in plausibility

• To ensure data is colorful

What should individuals demand from organizations collecting their data?

• Faster data processing

• Free access to all data

• A guarantee of no plausibility peaks

• More personal questions