Academics Get Personal Over Big Data

by Brett Winterford,  iTnews Australia

Princeton University computer science professors Arvind Narayanan and Edward Felten have published a rebuttal to a June paper by Information Technology and Innovation Foundation researcher Daniel Castro and Ontario privacy commissioner Ann Cavoukian, which concluded de-identification tools did a sufficient job of ensuring data privacy and that research to the contrary was based on flawed assumptions and incomplete research.

De-identification tools have been put forth as a way to protect the privacy of the wealth of personally identifiable data being used by everyone from governments to marketers, while still enabling the data to be economically useful. Castro and Cavoukian’s paper argued concerns about re-identification from de-identified data were “greatly exaggerated” and the skills involved in such work are not readily available in the “real world.”

However, Narayanan and Felten’s rebuttal critiques eight problems with the previous paper, lists many examples where motivated actors can combine aggregated data sets to re-identify de-identified data, and argues the only skills required to do so are basic programming and statistical knowledge.

Narayanan and Felten say efforts should be directed toward other techniques, such as differential privacy, and organizations should be willing to accept some loss of utility and convenience in order to ensure privacy interests are respected.  Report