"The reason is that in a group, individual errors on either side of the true figure cancel each other out"
About this Quote
Michael Shermer points to the engine behind what is often called the wisdom of crowds. When many people independently estimate an unknown quantity, their over- and under-shoots tend to balance, and the average of their guesses gravitates toward the truth. Francis Galton famously discovered this in 1906 when a crowd at a country fair collectively estimated the weight of an ox; the median of their hundreds of guesses was uncannily close to the actual weight. Jelly-bean jars, polling averages, and even market prices often display the same pattern.
The mechanism is statistical, not mystical. Each person brings a different slice of information and a different pattern of error. If those errors are mostly random around the true value, then aggregating them reduces noise, much like how multiple noisy measurements refine a signal. As the number of independent estimates grows, the variance of the average shrinks; the crowd becomes a better instrument. Diversity of perspective and independence of judgment are crucial here. So is an honest aggregation rule, such as an average or median, that does not overreact to outliers.
There are limits. If errors are correlated by a shared bias, they will not cancel. A leading hint, a misleading headline, or a strong social cue can push everyone to err in the same direction. Herding, echo chambers, and financial bubbles are examples where dependence and conformity make groups dumber, not smarter. Even the choice of aggregator matters: the median can be sturdier than the mean when outliers or trolls lurk. The practical lesson is to design group judgments to preserve independence and diversity and to use transparent aggregation. Forecasting tournaments, prediction markets, and well-run polling averages embody these principles. When the errors are independent and the incentives aligned with accuracy, many imperfect minds together can outperform most single experts, not by magic but by mathematics.
The mechanism is statistical, not mystical. Each person brings a different slice of information and a different pattern of error. If those errors are mostly random around the true value, then aggregating them reduces noise, much like how multiple noisy measurements refine a signal. As the number of independent estimates grows, the variance of the average shrinks; the crowd becomes a better instrument. Diversity of perspective and independence of judgment are crucial here. So is an honest aggregation rule, such as an average or median, that does not overreact to outliers.
There are limits. If errors are correlated by a shared bias, they will not cancel. A leading hint, a misleading headline, or a strong social cue can push everyone to err in the same direction. Herding, echo chambers, and financial bubbles are examples where dependence and conformity make groups dumber, not smarter. Even the choice of aggregator matters: the median can be sturdier than the mean when outliers or trolls lurk. The practical lesson is to design group judgments to preserve independence and diversity and to use transparent aggregation. Forecasting tournaments, prediction markets, and well-run polling averages embody these principles. When the errors are independent and the incentives aligned with accuracy, many imperfect minds together can outperform most single experts, not by magic but by mathematics.
Quote Details
| Topic | Decision-Making |
|---|
More Quotes by Michael
Add to List
