Privacy and Anonymity in a World of Interconnected Data

> Arvind Narayanan
>
> speaking on
>
> Privacy and Anonymity in a World of Interconnected Data
>
>
>
> Arvind Narayanan is a post-doctoral computer science researcher at
> Stanford University and a junior affiliate scholar at the Center for
> Internet and Society, housed at Stanford Law School and a part of
> the Law, Science and Technology Program. Arvind completed his Ph.D,
> which exposed the problems with data anonymization, at the University
> of Texas, and now studies privacy from a multidisciplinary perspective
> focusing on the intersection between technology, law and policy.
>
> Arvind will describe the issues around the privacy and anonymity of
> online consumers in this age of massive data collection from internet
> sites and mobile applications. His doctoral thesis, in a sentence,
> is that the level of anonymity that consumers expect—and companies
> claim to provide—in published or outsourced databases is fundamentally
> unrealizable.

Arvind began by explaining that when fingerprints were discovered about 100 years ago it was a huge breakthrough in criminal justice. No longer was it possible for a criminal to escape because they couldn't be proven to have been there. The idea that everybody could be uniquely identified took on a life of its own in the public imagination. Since then the field of privacy law has developed a lot.

Arvind showed us pictures of many other things that also now are known to have unique fingerprints. Computers have unique ID's from manufacturing requirements. Old typewriters have unique ID's because life has worn their keys in individual ways. Digital cameras have unique ID's because every camera is just slightly less than perfect in a different way, and the differences can be spotted by sophisticated computers with the right test algorithms. Even a blank piece of paper can be uniquely identified by its grain.

Arvind explained that such information is important to know about. Consider a blogger that is providing information about an awful regime. If their identity isn't known they are relatively safe. If their camera gives them away they could be apprehended. Then he explained that pictures can be "derezed" to hide such information by reducing the detail level to less than 70% of that in the original shot.

Online there are also digital footprints to be watched out for. It turns out that 87% of us can be uniquely identified just by our zip codes, birthday, and sex. Any anonymous database that holds those clues can be at least partly broken into by comparing those characteristics with public information. Arvind talked about a couple of major data dumps that had simply released too much information about individual people for their privacy to be guaranteed by the companies that promised it.