Cyrus Farivar, by his own admission, has an obsession. He’s obsessed with a technology that most people don’t even know exists: license plate readers.
“I first learned about them in around 2012 or 2013, when I was reading a blog post by the ACLU of Massachusetts," says Farivar, a senior editor with the technology news website Arstechnica. “These things were being discussed for purchase in Massachusetts by various local law enforcement agencies. I started to wonder, ‘Well, OK, are these things being used in other places in America besides Massachusetts?’ The answer was, ‘Yes, of course.’ "
License plate readers are small devices that typically sit on the roof of police cars, scanning up to 60 plates a second as the police cars drive around.
“All the plates that it can see in its visual field — cars coming toward the police car, cars driving away from the police car, driving to the side, parked," Farivar says. "No matter what, it can scan them."
Then, the plate numbers are cross-referenced with a “hot list” of plates of wanted or stolen vehicles. The problem, Cyrus says, is that only a small fraction of the plates are on the wanted list. The rest of the plates are on cars that belong to non-criminal, law-abiding people — people whose movements the government could now conceivably track.
“The fear is that with enough data points, with enough instances where your car was captured, you might be revealing something about that person that the government may not already know,” Farivar says. “[They] shouldn't necessarily know, without good reason.”
This technology is used all over the country in big cities like Boston, Chicago, New York, Washington, DC, and Los Angeles, but also in some smaller cities. Farivar says the technology is also used in other countries like the United Kingdom and Australia.
But for his research on license plates readers and their effect on privacy and civil liberties, Farivar turned his attention to his own backyard: Oakland, California. He filed a records request for all the license plate reader records that had ever been collected and received 4.3 million records covering roughly four years.
“That was pretty shocking to me,” Farivar says.
He reviewed the records, identified clusters of data points on a map and — by simply matching a license plate to its owner — was able to piece together an astonishing amount of personal information. To make his point about the potential pitfalls of this type of data collection, Farivar evaluated the movements of Oakland City Council member Dan Kalb.
“Knowing nothing else about him, other than his license plate number that he gave me, I showed him on a map, and I said, ‘I bet that you live on this particular block in North Oakland because I can see that your car has been scanned something like 30 times on this one block. He said, 'Yeah, that is exactly where I live.’ ”
Knowing where someone lives is one thing, but this technology could also allow someone to pin-point where an individual works, shops, worships or goes to have fun — all information that someone, even a law-abiding citizen, may not want law enforcement to have in their records.
When Farivar published an article on Arstechnica about the Oakland license plate reader experiment in March 2015, it catalyzed a change in policy in the Oakland police department. Now instead of keeping license plate data for years, they only keep it for six months unless the plates are under investigation, which Farivar says is a step in the right direction.
But still, he cautions, we need a better understanding of the limitations on government’s ability to collect big data.
“Just because the government has the ability to do that, in my opinion, it doesn't necessarily mean that they should be able to."