Crowdsourced Sleuthing Offers Extra Eyes and Ears, Some Wrong Turns

Sophisticated technology and crowdsourcing have helped police and the public work together in identifying the suspects in the Boston bombing. But some of the theories posited online have targeted innocent people. Ray Suarez interviews former deputy homeland security adviser Richard Falkenrath and Will Oremus of Slate.

Read the Full Transcript


    We now turn to the sophisticated tools law enforcement used to find the suspects, and how technology allowed crowdsourcing to become part of this major investigation.

    We are joined by Richard Falkenrath of the Chertoff Group. He's a former deputy homeland security adviser and special assistant to President Bush. He is also a contributing editor at Bloomberg TV. And Will Oremus, a staff writer for Slate and lead blogger for Future Tense, where he reports on emerging technologies, tech policy and digital culture.

    Will Oremus, what is Reddit? And how in this case did it become a tool for the use of law enforcement?

  • WILL OREMUS, Slate:

    Reddit is an online message board that has just rocketed to popularity in the past two years.

    It's a place where anybody can go if they just sign up and post a message about pretty much anything they want. These messages are divided into a bunch of different topic threads called subreddits. And in the course of the Boston bombing case, a couple of Reddit users have developed a thread called Find Boston Bombers dedicated to trying to help the authorities do their job.

    First, they were trying to find suspects. Then they were trying to help locate where the suspect might be once the FBI had circulated photos of them. And they have gotten a few things right and a lot of things wrong in the process. And it's stirring up a lot of controversy about their role.


    Richard Falkenrath, you can understand the appeal of something like force extension, extra eyes and ears, in effect enlisting hundreds of people to help the police.

    But do you also lose a little control, in effect, unleash peer-to-peer sleuthing, with people talking to each other, rather than to the cops?

  • RICHARD FALKENRATH, Chertoff Group:

    You lose more than a little bit of control.

    And, frankly, I'm skeptical that this crowd-sourced information was that useful to the investigators on the inside. They have tools themselves that are pretty sophisticated. They have done this before. And they are working, frankly, with a far greater amount of data than the crowd-source analysts were working with.

    And I think the statement yesterday from the FBI special agent in charge of the investigation was really quite telling, that he found this activity troublesome and led to an enormous amount of unhelpful speculation and misidentifications.


    Well, in this case, two young fellows, Salah Eddin and his friend Yassine, were identified on the covers of some nationally published newspapers as major suspects here.

    Will Oremus, walk us through some of the — well, the blind alleys that this kind of investigative technique walked the investigation down.


    So, over the course of the past few days on Reddit, there has been one potential suspect after another who rises to prominence on the site, and everyone starts delving into a certain photo to see if that could be the person.

    It's interesting. The format of Reddit is such that anyone can post anything, but everyone also gets to vote on which posts rise to the top of the page. The positive outcome of that is that if there's something that's posted that really valid, that's really interesting, that will go to the top of the page. Everybody will see it and everybody will start investigating that lead.

    Now, the downside is sometimes it's the most sensational thing that rises to the top of the page, and it's not always right. So one of the first people that they honed in on, on this Reddit was — they called him blue robe guy. It was a guy who was actually wearing a blue fleece.

    I guess robe sounded more sinister. They were dissecting his appearance, his location at the marathon finish line. Of course, he wasn't involved at all. Two people that were investigated by the crowd on Reddit ended up on the front page of The New York Post. They, too, were innocent.

    I don't think that's Reddit's fault. To me, that's The New York Post's fault. But it does emphasize the way in which these public crowd-sourced investigations can result in harm for people who happen to be mistakenly identified.


    But, Richard Falkenrath, once officials released the photos of their suspects, we had cases of people examining their own photo record of the day, and able to fill in their movements and whereabouts once they knew who they were looking for, in one notable case, a very sharp, high-quality, H.D. photo of Dzhokhar Tsarnaev moving from the scene. As other people are running from it in terror, he's walking to the corner and turning the corner. And we see him in very high fidelity, which I'm sure helps investigators in some sense figure out where he went after the explosions.


    Well, Ray, recall that well before the authorities released the photographs of the two individuals, they had requested that people in the vicinity of the explosion submit all their digital imagery to the FBI for analysis.

    So one would certainly hope that the person who had that high-definition photograph had supplied it to the FBI well before he found it and decided to publicize the fact.


    Are there new techniques and new machines that save police from having to go frame by frame through videos, leaf through photographs, that can look for a suspect very quickly through enormous data files?



    Yes, there are, something I have a fair bit of experience with in my time at the NYPD.

    But the new techniques to do analysis are having trouble keeping up with the massive increase in the amount of data that comes in the door in an investigation like this. So there is truly a phenomenal amount of digital imagery, unstructured digital imagery that comes in as a result of the crowd-source collection.

    So the analysis technologies are struggling to keep up with the collection technologies. And that's one of the basic problems they have. There will still be an investigator going frame by frame through the key feeds for the purpose of identifying which one is the best and who really to zero in on.


    After investigations, Will Oremus, there are often what are called after-action assessments.

    Should the tech world be doing one now about its role in helping uncover — or not — the two suspects in the marathon bombing case?


    Oh, they are. They absolutely are, and the media world as well.

    There's a lot of debate about whether this type of crowd-sourced sleuthing by armchair investigators does more harm than good. I think there certainly the potential for it to do good in some cases. There was a case a year or two ago where Reddit users were able to help solve a hit-and-run accident. Someone with great knowledge of cars was able to identify the taillight, identify the make and model of the car, helped police solve the crime.

    But this is a very different scale of crime. And I think the Reddit users are finding out that investigations are a lot harder than they might have thought, and that the harm along the way can be severe. So, there's been a lot of introspection, both in the media, the tech world and on Reddit itself. To their credit, a lot of the Reddit users are saying, hey, are we doing any good here? Are we really helping?

    But it's a diverse community of people. There are tons of people on the site. Some people are saying, hey, let's back off. Let's get out of this business and leave it to professionals. Others are saying, hey, no, look, we may have gotten it wrong here, but maybe we will get it right next time.


    To be continued.

    Will Oremus of Slate and Richard Falkenrath, thanks, both of you, for joining us.


    Thanks for having me.