Big Data, Meet Big Brother

If computers can now predict our behavior, should governments watch our every move? “One who breaks an unjust law must do so openly, lovingly and with a willingness to accept the penalty.” That was Martin Luther King Jr.’s definition of civil disobedience. It does not appear to be Edward Snowden’s. He has tried by every method possible to escape any judgment or punishment for his actions.

Snowden has been compared to Daniel Ellsberg, the man who leaked the Pentagon Papers to the New York Times. But Ellsberg did not hop on a plane to Hong Kong or Moscow once he had unloaded his cache of documents. He stood trial and faced the possibility of more than 100 years in prison before the court dismissed the case against him because of the prosecution’s mistakes and abuses of justice. Gandhi and Jawaharlal Nehru spent years in prison in India for defying colonial British rule in their native land.

But while Snowden is no hero, his revelations have focused attention on a brave new world of total information.

We are living with the consequences of two powerful, interrelated trends. The first is digital life. Your life today has a digital signature. Where you eat, shop and travel; whom you call, e-mail and text; every website, café and museum you visit even once is all stored in the great digital cloud. And you can’t delete anything, ever. “This will be the first generation of humans to have an indelible record,” write Eric Schmidt and Jared Cohen in their book The New Digital Age.

The second is Big Data. Americans were probably most shocked by the revelation that the U.S. government is collecting massive quantities of their digital ­signatures—billions of phone calls and e-mails and Internet searches. The feds aren’t monitoring every last one. But they easily could, and this is the essence of the age of Big Data.

In ancient times—by which I mean a decade ago—computers would sort through random samples of data or try to create an algorithm to search for a criminal. But today, data is so readily available and computers are so fast and powerful that experts can analyze entire data sets, every last piece of information, to find needles in haystacks. As a result, they have stopped trying to figure out why something—say, crime—happens. Instead they look at crimes and notice what events or behaviors seem to precede them. In other words, the tricky work of ­turning information into knowledge has shifted from causation to correlation.

In their excellent book Big Data, Viktor ­Mayer-
Schönberger and Kenneth Cukier write about the police in Richmond, Va., who track criminal incidents against a variety of events: corporate paydays, sports events, concerts, gun shows and dozens of other possible triggers. The computer identifies patterns. Two weeks after a gun show, for example, there is always a jump in violent crime. Multiply this example by thousands, and you understand what the NSA computers are doing. The key to getting robust results, however, is that the computers must be able to sort through lots of information—Big Data.

As Mayer-Schönberger and Cukier point out, if the computers can make predictions based on data analysis, should we prevent bad actions by arresting people before they act? This is not as far-fetched as it sounds. The NSA program Prism aims to identify suspicious patterns to allow the government to prevent terrorism (i.e., to act before an attack takes place). A research project at the Department of Homeland Security that tried to predict terrorist behavior based on people’s vital signs—physiological ­patterns—was 70% accurate, according to the authors.

As far as we know, the U.S. government has broken no laws and has followed all established procedures, and Congress approved this program, though it did so in secret, writing laws that aren’t public. Obama Administration officials, echoing their (slightly less transparent) predecessors in the Bush era, insist that any fishing expeditions undertaken through terabytes of collected data are highly targeted and do not involve innocent Americans.

Maybe so, but over the past 33 years, the Executive Branch has made 33,900 requests for surveillance to a special court created to make sure there are solid grounds to grant these surveillance powers. The court has approved all but 11 of them. Is that genuine oversight? It is hard to say, for the court itself is secret. Shouldn’t we know more?

The larger question Big Data raises is, Should any government be permitted to use computer ­analysis—even if highly accurate—to observe, inform on, quarantine or even arrest people simply because they are likely to do something bad? That seems like a scenario from a horrifying sci-fi thriller. Yet here we are, very close to a real-world version. Is that compatible with life in a free society?