The Wisdom of Crowds
The “Wisdom of Crowds” is a topic that was briefly mentioned in lecture, but certainly caught my attention as I just ran into it in another context as well.
The previous Sunday I was attending an ECO conference and upon walking in, they had each attendee guess the number of candies in a large jar. When at the end of the event, it was revealed that the average of the guesses of all attendees was 4 away from the true amount (with the true amount being around 650), with my guess was nearly 200 below the true amount, and the guess of my tablemate next to me was off by more than that as well… I was astounded. Of course, I understand how averages make sense and it would be perfectly logical for the average guess to be closer to the true value, it astounded my just how close the average guess was… prompting me to wonder just how smart crowds are…
When looking into this subject further, I was interested to see that studies from Scott Page and Lu Hong demonstrated that diverse groups made better average guesses than the average guesses of experts. Perhaps that is why my colleagues at the ECO conference (an undoubtedly diverse group), creative such an accurate average.
However, could this “guessing” only be relevant in this scenario?
Well, the “Wisdom of Crowds” is said to be effective in three main areas according to James Surowiecki. These areas are cognition problems, coordination problems, and cooperation problems. Surowiecki, in his 4 conditions of “collective wisdom”, too, notes that there must be “true diversity of opinions”.
You, like I was, may understand the merit of diverse thought, but not understand the benefit to random, diverse people over experts. However, the benefit is inherent to the diversity. The diverse thought is what experts lack, as they tend to think alike, thus their misconceptions often are due to group-think or overconfidence. While a diverse group of participants has no reason to necessarily think similarly to another participant.
A fun, “real life” example of this is the popular gameshow “Who Wants to Be A Millionaire”.
In the game, you must answer questions correctly consecutively to continue raising the pool of money you will win. If you get stumped, you have 3 options: remove half of the options, phone a friend, or ask the audience.
Omitting the first option for sake of this discussing, and assuming that a rationale contestant would phone a friend that has relative expertise on the question at hand, we see, perhaps shockingly, that the third option of polling the audience is actually the most accurate over time.
Historically, the “phone a friend” provides the correct answer about 65% of the time, while the “ask the audience” (random crowds), converged to the correct answer 91% of the time.
This short example is a fun way to see how the wisdom of crowds goes relatively unseen in daily life, but is incredibly powerful when employed. Of course, there are very few scenarios in which the wisdom of crowds can work to its true accuracy, but in the case of Who Wants to Be A Millionaire, you can count of the crowds (91% of the time that is).