Wisdom of Crowds Explained

Why a large group of ordinary guesses can outperform a single expert — and the specific conditions that have to hold for that to be true.

The basic idea

The classic illustration is a county-fair guessing game: ask a crowd to estimate the weight of an ox, and the average of everyone's guesses tends to land close to the true weight — often closer than most of the individual experts in the crowd. No single person needs to be right. What matters is that people's errors point in different directions, so when you average across enough guesses, the errors largely cancel out and what's left is closer to the truth than any one opinion.

Why aggregation works

This isn't magic, it's statistics. If each person's guess is the true value plus some random personal error, and those errors aren't all correlated with each other, averaging shrinks the total error the same way averaging noisy measurements in any experiment does. The crowd doesn't need to be smart on average — it needs to be diverse, in the sense that people are drawing on different information, different mental shortcuts, and different blind spots, so no single bias dominates the pool.

From guesses to markets

A prediction market is a more disciplined version of the same mechanism. Instead of asking people to state a guess for free, it asks them to back a probability estimate with money. That changes the incentive structure: people with weak information or low conviction bet small or not at all, while people who are confident and well-informed bet larger. The resulting price is a crowd estimate, but weighted by how much each participant is willing to risk on being right — which tends to make it more informative than a simple poll or show of hands.

The conditions that have to hold

The wisdom of crowds is not automatic. Research and observation on collective judgment generally point to a few conditions that need to hold for the effect to show up:

• Independence. People need to form their views separately, not by copying the person next to them.
• Diversity of information. If everyone is reading the same headline and nothing else, there's no diversity of error left to cancel out.
• Decentralization. No single actor should be able to dictate the group's answer.
• A working aggregation mechanism. There has to be some honest way to combine opinions — an average, a vote, or a market price — rather than letting the loudest voice win.

Where it breaks down

Break any of those conditions and the crowd stops being wise. If traders start watching the price itself instead of their own information and pile in because it's moving, independence collapses into herding. If a market is thin, one large trader can effectively become the whole crowd. These failure modes are worth understanding on their own terms — a related guide on prediction market biases covers them in detail. The wisdom of crowds is a real and useful effect, but it describes a mechanism with preconditions, not a guarantee that groups are always right.