The full toll of the marathon bombing is still being counted. The number of injured reported keeps ticking up and we can only hope that the death count remains at three. But in at least one way, Monday’s tragedy stands out from similar acts of terrorism. Simply by virtue of the technologies available today, it is likely the most widely recorded event of its kind in history.
Boston Police Commissioner Edward Davis described the Copley area as “the most complex crime scene” the department has ever seen. But never has it been more possible to reconstruct such a complicated scene from so many vantage points.
The FBI has said it is bringing up video analysis experts from its headquarters to begin sifting through the video submissions not only from cameras installed at local businesses, but from citizens’ cameras and smartphones.
As former U.S. counter-terrorism official Richard Clarke wrote on Facebook:
First, the FBI will stitch together hundreds of hours of video camera recordings from private and public surveillance and traffic cameras, as well as recordings made by private citizens attending the race. They will look for when the bombs might have been left behind and then examine the faces of everyone who was in the area around that time. They will try to put names to those faces, using facial recognition matching software, drawing on drivers license, passport, and visa databases.
“It is the all-encompassing eye witness,” said Rick Mislan, a professor at the College of Computing Security at RIT. “All of those different viewpoints from that scene can now be put together.”
But despite the existence of facial recognition technology, The Verge reports that much of this process is still relatively manual, with analysts watching video in order to painstakingly record what happens second by second. In 2011, video analysis was used to identify suspects in riots following the Stanley Cup in Vancouver. As The Verge reports:
After gathering those 5,000 hours of footage, Vancouver police then contacted LEVA through the Forensic Video Analysis Response Team to go through all that video. Fredericks says 52 analysts then spent 14 days processing that footage, identifying 15,000 criminal acts perpetrated by 300 rioters.
A similar article at Wired noted that the military is struggling with this same problem:
The military has found that its explosion of imagery data has stressed its ability to process it, to the point where its futurists are hunting for algorithms that can pre-select images a human analyst sees.
On the tech news site Hacker News, users yesterday debated the possibility of building a site to crowdsource analysis of video from the bombings, as well as the possibility that such an exercise would lead to profiling or would be ignored by authorities.
Mislan told me he did think there was room for technologists to contribute to authorities’ video analysis, if not in this instance then for the future.
“A 3-D video-rendered model…would definitely help,” he said. The Verge report concludes on a similar note, quoting video analyst Larry Compton saying:
“What needs to be developed is a solution for governments to accept public video and images for such situations, as there were likely several Zapruders who captured evidence,” he writes, referencing the famous film of the Kennedy assassination. “It needs to be far more comprehensive and useful than just random submissions to the 911 centers.”
In the near term, such independent efforts are unlikely to play a pivotal role in this investigation. And as one Hacker News user pointed out, “vigilantism can be dangerous.” But that doesn’t mean technologists can’t build a tool to improve authorities’ ability to deal with video evidence so that future investigations are strengthened. In the next few years, with smart glasses, drones, and other technologies becoming more ubiquitous, the volume of video recorded during emergencies will only increase.
Though he noted the legitimacy of the public’s concerns over privacy and emphasized the importance of trust to such a system, Mislan believes such an “eye witness” system should be technologically possible. “I don’t see why we couldn’t have that,” he said.