Its inherent in the understanding of what "differential privacy" is. But please listen to her explain it herself, http://www.sciencefriday.com/segments/crowdsou

Its inherent in the understanding of what "differential privacy" is. But please listen to her explain it herself, http://www.sciencefriday.com/segments/crowdsourcing-data-while-keeping-yours-private/ http://www.sciencefriday.com/segments/crowdsourcing-data-whi... Basically, its either your statics approach meaninglessness and maintain the individuals privacy or "overly accurate estimates of too many statistics is blatantly non-private"