paper and tools at dataprivacylab.org

test demographics at aboutmyride.org

The blogger mentioned our use of hidden names, so let me clarify. We consider profiles having hidden names as re-identifications because the names appear irregularly and are not explicit or obvious. Detecting them is one re-identification strategy we report in the paper. Here is how it happens. Some participants previously uploaded auxiliary files that contained genetic information, as received from other sources. The public appearance of an auxiliary file is a compressed file that does not reveal the participant’s name. Only after the file is downloaded and uncompressed might you see a new filename that includes the name of the participant. This begs the question, did the participant know his name was there? If the public profiles had a name field where a participant could enter his name, then we would know the participant intended for his name to be present. But the profiles have no such name field. In ad hoc discussions, participants voiced that these hidden names resulted from how they received the compressed file, implying a lack of action on their part to insert their names and no technical option to remove them (if they saw them). So, detecting hidden names is one of the re-identification strategies reported in the work.

Here are a few more points. Regardless of whether the profile had a hidden name, we also put names to profiles based on demographics (date of birth, gender, and ZIP), when present. The PGP staff scored results for correctness. To help participants, we offer a service for participants to learn how unique their demographics may be and we offer technical solutions for participants to change their profiles; see website above. Anyone can test how unique she may be from her demographics using our server at .

In terms of how this work relates to other misconceptions commonly voiced about re-identifications, this work demonstrates that a significant number of people can be re-identified by basic demographics (date of birth, gender, ZIP). Two undergraduate students working on term projects in two different courses did some of the original re-identifications, so it is not a matter of advanced science to do these kinds of re-identifications. Adam Tanner, the noted reporter, was not a collaborator, but instead, only had access to our paper after we publicly released it (same as the blogger). He reports having independently re-identified and interviewed some participants himself. There are many lessons learned from the experiment. Feel free to follow @LatanyaSweeney for updates.

It is my understanding that each joint author has the ability to license the joint work nonexclusively without the permission or consent of the other joint authors (though the law differs throughout most of Europe). That being the case, in terms of copyright, the wronged ex’s recourse would not be against the website, but against her ex-boyfriend (the poster) under some theory of the waste doctrine, which would be unproductive in terms of taking down the photo.

http://lawreview.uchicago.edu/sites/lawreview.uchicago.edu/files/uploads/79_3/01%20Bambauer%20ART.pdf

But if copyright is going to be deployed for evil, why not use it for good as well? Plus, it’s worth remembering that the whole point of copyright, originally, was to allow the British crown to censor more effectively.