Crowdsourcing, Sociopolitical Legitimacy, and the Boston Marathon Bombing
(Post originally written April 23, 2013).
I woke up this morning to read that Reddit had apologized for some of the blunders made by users searching for the identity of the Boston marathon bombers. Given the scolding given by the FBI to further detective work once the FBI had identified their targets, this proves that crowdsourcing was a mistake. Crowdsourcing has failed, and the public should never help the police again…which is the exact incorrect conclusion to draw from this case.
I pause to point out that yes, once the FBI knew the proper identities, the FBI was certainly right to tell the crowd to stop searching and take down any photos that did not include the correct suspects. And the crowd’s desire to search for personal information (such as in poor Sunil’s case), instead of limiting themselves to spotting suspicious patterns in photos, was a terrible mistake. But it’s important to note that several in the crowd pointed out such errors and did their best to limit these crowd instincts. Cases such as Richard Jewell were duly noted and cited. Rather than suggesting that crowdsourcing as a whole is to blame, let’s look more closely at when crowdsourcing is most effective, and whether crowdsourcing was given a fair trial.
First, effectiveness. Crowdsourcing was indeed a great idea early in this case. I stress the mention of early. In my opinion, crowdsourcing is best when (1) members of the crowd may have novel sources of information not available to the original team, (2) the original team can fully share information with the crowd, and (3) the amount of information to sort through is massive. The Boston Marathon case had (1) and (3), but not (2). In the first day or two after the bombing, crowdsourcing could uncover photos taken by private citizens that the FBI did not have access to, or find patterns that had not yet been uncovered. Once the FBI had some leads, leads that they understandably could not share with the public at first, (2) meant that the crowdsourcing could no longer be fully effective. But until that point, crowdsourcing still made a great deal of sense because of (1) and (3).
Second, innovation experiments cannot thrive under a high-pressure environment that only rewards a correct outcome and penalizes any errors. The Harvard Business Review articles on failure are highly recommended as background reading. Paradoxically, the media, while busy making their own wild guesses and speculations, seemed almost offended that a crowd would venture to do the same thing. Aldrich and Fiol, in their well-regarded 1994 paper (and also Rao’s 1994 paper on the automobile industry), describe sociopolitical legitimacy as endorsement by legal authorities, governmental bodies, and other powerful organizations. In this case, the media was influential in de-legitimizing crowdsourcing in the eyes of ordinary citizens. Unfortunately, Internet crowdsourcing efforts were immediately criticized by the Atlantic, Slate, and other influential media bodies. Unscrupulous media organizations such as the New York Post ripped off various ideas the crowd came up with and trumpeted their scoops, only to cheerfully blame “the Internets” when those ideas were incorrect. Crowdsourcing can’t be effective in a fishbowl environment, as it takes time to discard bad ideas and elevate good ones.
There are good reasons to beware of crowdsourcing in criminal cases. However, based on my paragraph about effectiveness, I do believe the Boston Marathon Bombing case was a rare situation where private citizens may initially have had access to visual data that was previously unknown by police, and where large amounts of information made more analyst eyeballs valuable. Crowdsourcing by Reddit and other sites faltered when some crowd members attempted to be both judge and jury. Or if you prefer, detective and arresting officer. However, that should not mean that crowdsourcing itself was a failure, and it does our national and personal security a grave injury if this means citizens are now more reluctant to help police efforts.