Independent errors cancel out, revealing an accurate collective estimate

Surowiecki opens with the famous story of Francis Galton observing a country fair contest where attendees guessed the weight of an ox; no individual guess was particularly accurate, but the average of all the guesses came remarkably close to the ox's true weight, closer than most individual experts managed on their own.

The underlying mechanism is statistical: if each person's guess contains some random error, some guessing too high and others too low for different individual reasons, those errors tend to average out across a large enough sample, leaving the systematic signal — the actual weight everyone was roughly perceiving — intact while the noise cancels.

This only works, crucially, if the errors are genuinely independent rather than correlated; if everyone is influenced by the same bias or the same piece of misleading information, their errors will all skew in the same direction and averaging won't help at all. The ox-weighing story is charming, but its real payload is this statistical mechanism, which recurs throughout every example in the book.