The Berkeley Marina, a stretch of park and
an enclosed yacht harbor, juts into the San Francisco Bay. Visitors
regularly drive there to admire the dramatic view that includes Alcatraz
Island and the Golden Gate Bridge. When I lived in the area, I often
parked my car, surveyed the vista for many minutes, then drove away if
the weather did not prove ideal for windsurfing.
When fertilizer salesman Scott Peterson made his last visits to the
Berkeley Marina from his hometown of Modesto, Calif., he had a practical
goal in mind. He wanted to see whether his pregnant wife whom he had
murdered had washed up along the shoreline. In the Peterson case a
decade ago, police placed a GPS tracker onto his vehicles when they
began to suspect that he was lying. Peterson had driven a few times to
the marina homicide site, even though he told police he had only learned
his wife had gone missing after returning from a fishing expedition to
the San Francisco Bay. That GPS evidence eventually helped convict him
of murder; he is now on death row.I thought of this story, revealed at Peterson’s murder trial which I occasionally attended, when a prominent data broker announced two weeks ago that it had begun selling locational information on license plates that have been filmed and identified. In recent years, police have also widely embraced license plate recognition to track suspected criminals. Repo men use the technology to recover vehicles; casinos in Las Vegas employ it to monitor cars in their parking lots. And now data broker TLO has begun selling information about the time and location at which cars have been sighted.
“With a massive database of one BILLION vehicle sightings and the addition of up to 50 million new sightings each month, Vehicle Sightings provide valuable information for both locating subjects and investigating the historical whereabouts of both individuals and vehicles,” advertises TLO, a data broker that caters to lawyers, private investigators, law enforcement and insurance firms, among others.
The service charges $10 per category of each license plate look up, divided into current, recent and historical. Cars are photographed or filmed and then matched with license plate recognition software. Initially I imagined a database that knew almost as much as a GPS locator: that you drove out of state three weekends ago, stopped off at the pharmacy on the way, spent the afternoon at a baseball game, then had dinner at a specific restaurant.
In reality, the feature is quite far from the all-knowing eye in the sky, although it can still reveal intimate clues. I searched for my own car, as well that those of two relatives with their permission. Of five cars that I looked up, three cars turned up nothing, but I found data on the other two.
One car had a single sighting: it was parked on Manhattan’s Upper West Side at 12:40 in the morning last December. The report included a picture of the car and license plate. A link showed exactly where the car was on Google GOOG +0.3% Maps. For another car, the search turned up data from August last year which showed it parked in Austin, Texas, a few minutes after noon. The lot was in front of a building of doctors’ offices, potentially revealing intimate information about that person’s activity that day.
Simple math suggests it may be a while before such license plate recognition systems can regularly spot specific vehicles. TLO advertises it has a billion vehicle sightings, but according to the U.S. Department of Transportation, there are more than a quarter of billion registered vehicles in the country. That means TLO would hold an average of four sightings per vehicle.
“While the coverage is nationwide, certainly there will be areas with more expansive coverage than others,” said James Reilly, TLO’s senior vice president of sales and business development. “Variables such as the amount of time the vehicles are stationed in inaccessible areas (i.e. secured lots at places of employment, gated communities, etc.) could certainly affect the number of opportunities for ‘sighting.’”
Law enforcement agencies are among the biggest users of automated license plate recognition. Drivers’ information is one of the few sectors of personal data granted strong legal protections, along with areas such as medical, financial and job hiring data. TLO’s core clients such as law enforcement, private investigators and others can obtain the information if they have a legally permissible use.
Still, anyone can photograph and catalog license plates in public. Some private companies such as MVTRAC, which says it has spent many millions of dollars collecting hundreds of millions of vehicle sightings, supplies repo men, law enforcement and others access to its database of recorded plate numbers.
“People are afraid of the government collecting phone records from Verizon and the government’s response was you don’t understand we’re not really spying on you, it’s metadata,” said Scott Jackson, MVTRAC’s founder and CEO. “People don’t understand what metadata is. It’s abstract data that is just sitting there and it has no meaning whatsoever until one day there is a Scott Peterson who all of a sudden now he is a suspect.”
“Now you can take that plate and you can reach into the metadata that up to then had no meaning whatsoever, and the next thing you know you can interpolate various patterns of movement, what his patterns were in terms of habits.”
Yet Jackson said his industry was years and years away from possessing enough data to reconstruct a random person’s driving on a given day. Obviously GPS or cell phone data would far more accurately record what any individual car does than photos captured externally. In the Peterson murder case, the GPS evidence proved important to the prosecution because the victim’s body washed ashore months later in the San Francisco Bay near where he had often visited. Yet GPS data is harder to obtain. Last year the U.S. Supreme Court ruled that police needed a warrant to attach a tracker on a suspect’s car.
One possible longer term issue around license plate recognition is that new firms in the field seeking to gain market share could gather specific data such as who was visiting what churches or mosques, underground clubs, or medical clinics and perhaps distribute that information more freely than companies now do.
“Because legislation and our Congress is not going to be able to keep up with the dramatic growth of big data, people who are aggregating data do have at least a reasonable social responsibility to make sure that they handle themselves with reasonableness,” says Jackson of MVTRAC. “I don’t think it would be reasonable for me to put out all the data and say these are all the cars that drove by at this intersection — although lawful.”
No comments:
Post a Comment