Despite multiple photos and surveillance video images of two suspects involved in the Boston Marathon bombings last month, as well as state-of-the-art facial-recognition software and two government databases, investigators were unable to identify the two suspected perpetrators, even after releasing several of the images to the public.
It was only after one suspect was shot in a showdown with police and was fingerprinted on his way to the hospital that authorities finally had a name to go with a face — Tamerlan Tsarnaev, a Chechen youth who was already on two U.S. terrorist watch lists after Russian authorities warned the U.S. that he had become “radicalized.”
The watch lists were supposed to alert authorities if Tsarnaev attempted to travel overseas, but they failed as well, even after he returned last year from a six-month trip to Russia. The databases contained a misspelling of the suspect’s name — “Tsarnayev” instead of Tsarnaev — and two incorrect dates of birth.
The facial-recognition system failed because none of the images captured of the suspects at the bombing site were full-frontal shots that the system’s algorithms could recognize.
These are two of the technology failures discussed in a new documentary about the bombing manhunt produced by NOVA and airing on PBS stations tonight.
Viktor Mayer-Schönberger of the Oxford Internet Institute tells NOVA that the database failure shows how law enforcement’s use of its multi-million-dollar databases can’t even keep pace with the state-of-the-art in Google searches.
“Google actually has the ability to do messy, big-data analysis that can deal with misspelled words, and law enforcement can’t,” he says in the documentary. “If law enforcement had used a Google-style big-data analysis, chances are that they might have prevented the Boston bombing from happening.”
Whether that’s true or not, there were two other technologies that succeeded big in the sensational manhunt — the tracking of signals from a cellphone left in a car the suspects were driving, which led police to a showdown with the suspects on a residential street, and a high-tech thermal-imaging camera that captured the now-iconic images of one of the suspects hiding in a backyard boat.
Of all the technologies reviewed in the documentary, the thermal-imaging camera is the most remarkable and the most worrisome, due to its capabilities. A demonstration provided to NOVA by Flir, the maker of the equipment, says the camera has the ability to spot a dropped cigarette from a mile away due to heat the butt emits.
Helicopter footage obtained from the Massachusetts State Police by NOVA shows the department’s thermal-imaging footage of Tsarnaev’s younger brother Dzhokhar in the boat, beneath its shrink-wrapped cover, following the noise of the helicopter overhead as the pilots watch him through the thin plastic.
“They were actually looking at us as we flew over,” one of the helicopter officers tells NOVA. “I was pretty surprised that not only was there a person in the boat, but the definition and clarity that I got from the camera.”
Images from the thermal-imaging camera were released weeks ago following the incident, but NOVA got access to footage that appears to show an explosion going off inside the boat during a volley of gunfire that occurred while the suspect was in the boat. NOVA doesn’t address the explosion, but it could be the moment when authorities say Tsarnaev shot himself in the mouth.
In addition to the thermal-imaging camera, NOVA also got a look inside the command center for New York’s Domain Awareness system — a surveillance system that aggregates a constant stream of data from 4,000 closed-circuit surveillance cameras in Manhattan as well as environmental sensors and license-plate readers that record the plate of every vehicle in every lane on every bridge and tunnel entering lower Manhattan.
The Domain Awareness System is connected to New York’s 911 emergency phone system so that as calls come in, possible terrorist-related events can be flagged and augmented with data from other sources. A database and map display the precise location information of the call as well as the location of nearby surveillance cameras, which can be called up for real-time viewing or rewound for images stored going back 30 days.