The unwisdom of crowds

James Surowiecki has disputed the wisdom of crowds, even though he literally wrote the book on the subject.


be more accurate, Surowiecki criticizes people who misapply the term,

or who simply invoke “the wisdom of the crowd” to justify doing stupid

things based on popularity. Under certain, explicit circumstances,

crowds can be scarily intelligent. But what are those circumstances?


of all, a crowd can only be wise when there is one, true answer. Crowds

are good at probabilities, math problems, and the like. For example, crowds are

great at setting odds at a race track. The crowd-chosen favorites win out over time.

But in those cases, the crowds were working on a problem which had an

objective answer. A horse was going to win the race. The crowd was

great at estimating which horse it would be, and over time, the crowd

would always prove smarter than any single expert gambler (assuming

there is such a thing).

The wisdom of crowds breaks down,

however, in instances of information cascade. If “the crowd” is so

smart, then how did everybody get taken in by the dot-com stock bubble?

Simple, a crowd is also a herd, and as with any herd, most people are

going to follow the group.

Surowiecki uses the example of two identical

empty restaurants next door to each other. The first customer arrives

in front of the eateries, sees them both empty and, with no data to

distinguish them, arbitrarily chooses one over the other. The next

customer arrives and sees that one restaurant has a customer and one

doesn’t, and chooses the occupied restaurant under the assumption that

the first customer knew something he didn’t, implying one restaurant

was better. The process repeats itself, until restaurant A overfills

while restaruant B sits empty, even though the two restaurants are

otherwise indistinguishable. All due to the random choice of the first


The dot-com bubble worked the same way. All the

investors assumed the other investors knew what they were doing–-twenty

million stock speculators can’t be wrong, supposedly-–and they all

collectively suffered when the bubble burst. No wisdom in that crowd.


how is the race track different from the stock market? Both are

essentially speculative markets with wagers placed against expected

performance of a certain entity, either a company or a thoroughbred.

Well, by and large, the crowd does not collaborate at the race track.

Yes, odds are posted, but bettors generally don’t confer with each

other about whom to bet on. The opposite is true of the stock market,

where collective analysis is the order of the day, and a few key

influence-peddlers coax the general investment crowd one way or the


Thus, Surowiecki implicates, crowds are wise only when

each member of the crowd is working on the same problem somewhat

independently. Crowds are wise in aggregation, not collaboration.


race track crowds are far more diverse intellectually than the stock

market crowd, and they rely on a much wider array of indicators than

stock analysts. That is to say, most active investors are educated,

with a certain reliance on statistics and business acumen that within

itself is quite varied, but when compared against the whole of human

society is actually pretty niche.

Bettors at a race track,

however, run the gamut from math savants to the barely literate, with

bets placed as often on the name or color of a horse as the

statistictical record of its past success. How often does an investor

pick a stock based on the color of the company’s logo or the catchiness

of its product jingle? Collectively, the race track’s more diverse

array of bettor insights and preferences is more accurate over time

than the comparatively homogenous set of expert stock investors,

because the varied biases for prediction factor out against each other.


diversity, even to an absurd degree, is a healthy thing for crowds.

Darwin taught us that, too. You don’t know how the environment is going

to change, so a diversity of species and genetic stock is the best

defense against the unknowable future. Apparently, its also the best

way of guessing the unknowable via an aggregate of the crowd.