By Timothy B. Lee
The American Civil Liberties Union is challenging the National Security Agency’s dragnet surveillance of Americans’ phone calling records. On Monday, the ACLU asked the court to issue a preliminary injunction halting the program while its legality is litigated.
The program only collects metadata about Americans’ phone calls—who they call, when, and how long the calls last. In defending the program, the government has cited a controversial 1979 Supreme Court decision that held that phone records are not protected by the Fourth Amendment because consumers do not have a reasonable expectation of privacy in their calling records.
But Ed Felten, a professor of computer science at Princeton University (and, full disclosure, my former graduate school advisor) argues that this intuition is wrong. In a legal briefsupporting the ACLU’s request, Felten argues that the distinction between call “contents” and “metadata” isn’t always clear. Sometimes, the mere fact that someone called a particular number reveals extremely sensitive personal information.
Certain telephone numbers are used for a single purpose, such that any contact reveals basic and often sensitive information about the caller. Examples include support hotlines for victims of domestic violence and rape, including a specific hotline for rape victims in the armed services.
Similarly, numerous hotlines exist for people considering suicide, including specific services for first responders, veterans, and gay and lesbian teenagers. Hotlines exist for suffers of various forms of addiction, such as alcohol, drugs, and gambling.
Similarly, inspectors general at practically every federal agency—including the NSA—have hotlines through which misconduct, waste, and fraud can be reported, while numerous state tax agencies have dedicated hotlines for reporting tax fraud. Hotlines have also been established to report hate crimes, arson, illegal firearms and child abuse. In all these cases, the metadata alone conveys a great deal about the content of the call, even without any further information.
And, Felten argues, metadata becomes even more revealing when it’s collected in bulk:
Two people in an intimate relationship may regularly call each other, often late in the evening. If those calls become less frequent or end altogether, metadata will tell us that the relationship has likely ended as well—and it will tell us when a new relationship gets underway. More generally, someone you speak to once a year is less likely to be a close friend than someone you talk to once a week.
Consider the following hypothetical example: A young woman calls her gynecologist; then immediately calls her mother; then a man who, during the past few months, she had repeatedly spoken to on the telephone after 11p.m.; followed by a call to a family planning center that also offers abortions. A likely storyline emerges that would not be as evident by examining the record of a single telephone call.
Likewise, although metadata revealing a single telephone call to a bookie may suggest that a surveillance target is placing a bet, analysis of metadata over time could reveal that the target has a gambling problem, particularly if the call records also reveal a number of calls made to payday loan services.
If a government employee suddenly begins contacting phone numbers associated with a number of news organizations and then the ACLU and then, perhaps, a criminal defense lawyer, that person’s identity as a prospective whistleblower could be surmised.
In short, the distinction between call metadata and call contents is not as clear in practice as it might seem in theory. Sucking up everyone’s phone records gives the government access to a lot of highly sensitive information, like whether you’ve had an affair, gotten an abortion, or provided secret information to a reporter. Felten’s examples give ammunition to those who believe the Supreme Court should revise its interpretation of the Fourth Amendment.