How coding contests can be better at solving problems than Harvard

Harvard Business School and TopCoder recently performed a study where they took a big genomics problem being worked on by Harvard Medical School, broke it into discreet abstract parts, and the threw the problem’s parts to the crowdsourced coding community to solve. What they gleaned was an interesting insight: the smartest guy in the room isn’t always your best problem solver.
Speaking at GigaOM’s Structure:Data conference Wednesday, Harvard Associate Professor Karim Lakhani said crowdsourcing the genomics project did several things. First and foremost it generated dozens of different approaches to tackling the same problem.
Before the TopCoder contest was created, researchers were considering two different paths of investigation. The contest revealed 89 differing approaches to the problem, 20 of which were extreme values — possibilities Harvard and its National Institutes of Health counterparts had never considered.
Second, the contest was able to create motivation to solve problems that you wouldn’t necessarily find in a group of researchers. An institution like Harvard may have brainpower in spades, Lakhani said, but throwing a bunch of geniuses at a problem doesn’t necessarily lead to result if they’re unmotivated to solve it.
“When you go into a self-selection model you don’t have to worry about motivation,” Lakhani said. There are monetary rewards for winning a TopCoder contest, of course, but taking home a prize is not guaranteed. Everyone has their own motivation for participating, whether it’s cash, experience, scoring reputation points or even the free T-shirts given to each participant. “Because there are large numbers of people participating, there’s a greater chance you’ll find the right skills and the right motivation,” Lakhani said.
Ultimately crowdsourcing your science isn’t a replacement for having smart people of your own, Lakhani said, but it certainly helps, something Harvard’s counterparts at NASA have discovered.
CHeck out the rest of our Structure Data 2013 coverage here, and a video embed of the session follows below:
[protected-iframe id=”bdd301a1ce21b581fe4ed5f90e936fac-14960843-25766478″ info=”″ width=”640″ height=”360″ frameborder=”0″ scrolling=”no”]
A transcription of the video follows on the next page