I have three questions. First, they RAPPOR is used when the user's data is repeatedly used (longitudinal data). What does it really mean? Does it mean asking the same user the same question over and over again, or asking the same user different questions? Second, how does the decoding work now that the data is hashed? If anybody wants to use the data ever, a decoding algorithm need to be in place. Third, where is the decoding algorithm implemented? Presumably in the server, correct?
1.) It is not the same question, but different "questions". You can think of a fitness tracker that measures your steps. These will vary between days. So same data type, different data. 2.) Decoding is a bit complicated that's why the video takes a bit longer, but it will come on this channel. 3.) Yes, decoding happens on the server.