How Can The Crowd Be Made Smarter?

The modern world is awash with peer feedback, with everything from employers to restaurants coming with a rating provided by fellow consumers.  This feedback has been shown to be hugely influential in our decision making, but is it as flawless as the ‘wisdom of crowds’ would suggest?  That was the question posed by a recent study from Washington University in St. Louis.

The researchers examined data from the financial platform Estimize, which provides a facility for professional and amateur financial analysts to estimate the quarterly earnings-per-share for traded companies.

The data revealed that the less each person knew about the guesses of the other Estimize users, the more accurate both their own guesses were, and also that of the crowd as a whole.  What’s more, this was quite a significant boost, as when all estimates were visible, the crowd beat the Wall Street consensus around 57% of the time, versus 64% of the time when guesses were kept hidden.

“The problem with seeing others’ information is that people tend to herd with others,” the researchers explain. “That makes individual forecasts more accurate, but… reduces the consensus accuracy.”

Following the herd

The apparent herd-like behavior of the crowd was evident from the data, with users tending to follow the pack whenever they had access to the estimates of the crowd.  This was found to make users smarter individually, but the crowd dumber collectively.  This herding behavior was especially evident when influential users made their forecasts early on.

Indeed, so stark were the results from the experiment that Estimize were moved to change their approach and prevent users from seeing the estimates of others before they’d posted themselves.

“We were also quite lucky to collaborate with Estimize to run experiments where we can randomize the information sets of users,” the authors explain.

They believe that their findings would be applicable to any website where crowds are encouraged to make predictions, whether voting platforms, crowdfunding sites or product reviews.  If the aim is to aggregate the wisdom of crowds, then a degree of ‘blindness’ would serve you well.

Facebooktwitterredditpinterestlinkedinmail