Scientifically Speaking, Nice Guys Do Not Actually Finish LastS

Think only jerks can catch a break in this cruel world? Nope. David Rand, a Harvard University researcher, studied the behavior of 800 individuals he recruited through Amazon's Mechanical Turk to prove it.

Rand had people play a game to find out whether they would be nice or mean in a social network.

We had them play a cooperation game embedded in a social network. At the beginning of the game, each person was randomly connected to set of other players (her 'friends'). In each round of the game, each player chose whether to be generous and pay a cost to give a benefit to each of their friends, or to be selfish and not. Then the players got to find out whether some random other people where generous or not, and based on that information could choose to form new friendships, or break existing friendships. The amount of money they got paid at the end of the study depended on how many points they earned, so being generous really cost you something (and was really materially beneficial for your 'friends').

One key difference between this study and others, for example the famous Stanford prison experiment, is that the subjects could choose their friends. Unlike the in the Stanford study, people in Rand's experiment could break their connection with someone who was a jerk and move on to someone more amiable.

The key insight of our paper was that allowing people to make and break friendships supports generosity. In the game worlds where friends were fixed, or where friendships were shuffled randomly, selfish people did better and selfishness spread. But in the games where people had control over their friendships, generosity was favored and being nice spread.

Since most of us can choose who to befriend, this makes me happy. All you mean girls and boys out there: know that scientifically you are at a disadvantage.

And the method in which Rand performed his study is almost as interesting as the study itself: He used Amazon's Mechanical Turk, the online hub that matches tech employers with workers to do things that humans are better at than computers—like tag photos or choose categories for products. Rand used it to recruit and pay folks for the study. Turns out, Mechanical Turk could transform social science big time. Rand told me about it over email:

…the main advantage MTurk offers over the physical lab is making it easy to recruit subjects who are more diverse in terms of age, education, income and geography.

It also makes studying so many people an "order of magnitude" less expensive than it would have been had Rand recruited subjects through solicitations at universities or other institutions and had them come to the lab. He also collected the data in just a few months—much faster than by traditional methods. "Even more importantly," Rand said, "it involved MANY fewer man-hours, because the experimenters didn't have to go down to the lab, set things up, wait for people to show up, etc."

It's exciting to imagine other insights Mechanical Turk might help uncover. In fact, Rand has lots more already in the works, including one asking whether people are intuitively cooperative or if we have to try hard to be. Personally I would like to know why everyone lies so much. [Proceedings of the National Academy of Science]

Image: Shutterstock/Yuri Arcurs


You can keep up with our Science Editor, Kristen Philipkoski, on Twitter, Facebook, and occasionally Google+