Mathematical Solutions For Maintaining Privacy
from the works-in-theory... dept
I’ve heard of similar solutions before, but here’s a report about technical means to protect private information while still allowing companies to get accurate (aggregate) demographic information and researchers to get honest responses on online forms. The first method involves adding or subtracting a random sum to any answers the person provides, and then fixing it in the aggregate. I don’t quite understand why the system described needs to be so complicated. All you need are three tallies: (1) showing the total amount (2) showing the total adjustments and (3) the number of entries. With that you can get the accurate average without violating anyone’s privacy. The other method discussed is with yes or no questions on private information. Have a random coin flip. If the coin is heads, then answer honestly. If the coin is tails – then you are assigned an answer. This way no one knows if the people who answered in potentially embarassing ways are doing so because they’re being honest or because the coin flip told them to. Of course, both methods break down if the data is somehow connected – and someone can associate one entry with another. Also, I doubt most people would trust companies to actually follow through with such systems to keep their data private. Even if a company used such a system, I imagine many people either wouldn’t understand how it worked or would assume that the company was making it up, and would still be looking at their personal info.