@alcinnz @baldur …the term Differential Privacy refers to a property of the way data can be transformed in aggregate. An individual submission isn't differentially private, which is why in my blog post I used Multi-Party Computation to get a larger anonymity group first. Most papers in this field assume you have a large non-anonymous database and you need to allow people to query that data without violating privacy. But see also the older https://en.wikipedia.org/wiki/Randomized_response method.