Nine years ago, federal agents stuck a location tracker on a nightclub owner’s car without a warrant. The agents thought their suspect might be dealing drugs, and four weeks of GPS data ultimately proved them right. He countered that the prolonged tracking had violated his privacy, and therefore his constitutional rights. The case fell apart in the Supreme Court, where justices debated the length of time that police could tail citizens before routine investigation became all-out invasion. Was four weeks too long? What about three days, or four hours?

Last week, a high-tech solution was proposed: let algorithms set the guidelines for us. Since computers are able to unearth subtle patterns in data, the thinking goes, they could help lawmakers quantify how much or how little surveillance is fair.

“Some justices think four weeks is too much and they’ve never been able to explain why,” says Steven Bellovin at Columbia University in New York City. “I saw there was a natural way to answer some of the questions by using these techniques.”

Bellovin, along with specialists in computer science and law, analysed previous research on tracking to learn what such data could uncover about an individual’s characteristics and daily habits. They concluded that one week of data would reveal enough information to constitute a threat to privacy, and so would be a reasonable place for the law to draw the line (NYU Journal of Law & Liberty, doi.org/s5d). (1)

We have started trusting computers to determine whether someone should go to jail.
We have started trusting computers for matters of justice.
We have started trusting computers for human matters.

We have stopped being human.
We have started thinking like computers.

So illogical.
So paradox.
So… human.

Give THAT to computer to analyze…