The limits to Crowdsourcing

I am a big fan of crowdsourcing done right. GE's use of Kaggle's crowd to solve complex aviation challenges, or the use of Amazon Mechanical Turk for crowd scans of satellite imagery in the search for Jim Gray are two good examples.

This week, however, we have seen two controversial examples of crowdsourcing.

First,  the Masters in Augusta last weekend used viewer called-in information to assess Tiger Woods a 2 stroke penalty for a rule violation. Bubba Watson, last year's winner,  reacted for many:  "Nobody calls in during a basketball game or a football game. ... They're definitely not calling balls and strikes during a baseball game." Crowdsourced referees are suspect - they may not like Tiger or Bubba. That's why we have paid officiating.

Far more concerning was the crowdsourced sleuthing of the Boston bombing (far ahead of the FBI releasing the photographs of the  suspects on Wednesday). An example of the crowd identified "suspects" was on this album. One of the suspects, according  to the Associated Press was a "high school student, of Moroccan descent, said he is scared to go outside after he was portrayed on the Internet and on the front page of the New York Post as connected to the deadly Boston Marathon bombings."

My own reaction when I saw those photos was the law often issues "disinformation" early on in an investigation - so the crowd was primarily looking for backpacks with sags (due to a heavy bomb), hoodies, primarily on Boylston street,  brown folks.  For two whole days, the crowd tried to identify suspects based on a limited number of photos from amateur cameras used at the Marathon and the Boston Police scanner live stream, whereas the law enforcement had access to a far bigger evidentiary database of store, police and other videos and first-hand witness interviews in addition. Sure enough this crowd did not even know about the Tsarnaev brothers till the FBI released their photos.

The week in Boston will be analyzed from countless perspectives in the coming months. I hope one angle looks at the opportunities and limits of crowdsourcing.