What is the wisdom of crowds and when does it work?

The wisdom of crowds, popularized by James Surowiecki, is the finding that under the right conditions, the aggregated judgments of a large, diverse group consistently outperform even expert individual predictions. It works when four conditions are met: diversity of opinion (each person has private information), independence (people's opinions are not determined by the group), decentralization (people can specialize), and aggregation (a mechanism exists to turn judgments into a collective decision).

What is the most famous example of the wisdom of crowds?

The classic example is Francis Galton's 1907 ox weight contest. At a county fair, 800 people guessed the weight of an ox. No individual guess was correct, but the median of all guesses was 1,207 pounds — within 1% of the actual weight of 1,198 pounds. This predated modern research but demonstrated the core principle: independent estimates, when aggregated, cancel each other's errors and converge on accuracy.

When does the wisdom of crowds fail?

The wisdom of crowds breaks down when its four conditions are violated. The most common failure: herding and cascade effects, where people stop using private information and start copying others. This produces correlated errors that amplify rather than cancel. Financial bubbles are the classic example — crowd behavior converges on a shared (and wrong) narrative rather than aggregating diverse independent judgments. The crowd becomes a mob, not a wisdom engine.

The Wisdom of Crowds: When Groups Outperform Experts

Galton's Ox: Where It Began

In 1907, statistician Francis Galton attended a county fair in Plymouth, England, where 800 people were invited to guess the weight of an ox. Galton, a committed believer in elite expertise, expected the crowd to perform poorly — most participants were not livestock specialists. He collected the tickets to analyze the data after the contest.

What he found astonished him. No individual guess was correct. But the median of all 800 guesses was 1,207 pounds. The actual weight of the ox was 1,198 pounds. The crowd had been accurate to within 1%.

James Surowiecki built on this insight in his 2004 book The Wisdom of Crowds, demonstrating that this wasn't luck. Under the right conditions, large groups consistently outperform individual experts — not because any member of the crowd is brilliant, but because the errors of independent estimators cancel each other out while the signal accumulates.

The Four Conditions for Crowd Wisdom

Surowiecki identified four conditions that must be present for collective intelligence to emerge. When they are present, crowds are remarkably accurate. When they are absent, crowds can be catastrophically wrong.

1. Diversity of opinion

Each person in the group should have some private information — their own interpretation of facts, their own background knowledge, their own perspective. Homogeneous groups that share the same assumptions don't aggregate diverse signals; they amplify shared errors.

2. Independence

People's opinions should not be determined by the opinions of those around them. When individuals start copying each other's views — due to social pressure, authority deference, or herding behavior — the crowd loses its diversity and its error-cancelling property. Each opinion becomes correlated with others, producing collective errors that reinforce rather than cancel.

3. Decentralization

People should be able to specialize and draw on local knowledge. The person closest to a specific problem often has the most accurate information about it. Decentralized knowledge systems — like markets — aggregate these distributed insights better than centralized ones.

4. Aggregation

There must be a mechanism for converting individual judgments into a collective decision. Markets aggregate through prices. Polls aggregate through averages. Prediction markets aggregate through betting. Without effective aggregation, individual wisdom stays individual.

Why Aggregation Beats Individual Expertise

The mathematical intuition is clear: if errors are independent and unbiased, they cancel. If 100 people each make a guess with random error, the average of their guesses will be far more accurate than any individual guess — because the errors of overestimates cancel the errors of underestimates.

The deeper insight is that expertise is narrower than it appears. A single expert, however knowledgeable, has a limited perspective and particular blind spots. A diverse group of moderately informed people collectively covers more of the relevant information space. Their combined knowledge typically exceeds what any individual has access to.

The Diversity Prediction Theorem

Statistician Scott Page formalized this as the Diversity Prediction Theorem: the collective error of a group equals the average individual error minus the group's diversity (variance of predictions). The implication: adding a diverse thinker to a group improves collective accuracy even if that thinker is less accurate than anyone already in the group. Diversity beats expertise in aggregate performance.

Prediction markets

Prediction markets — where people bet on outcomes — consistently outperform traditional forecasting methods. The Iowa Electronic Markets have historically outperformed professional pollsters in election forecasting. Internal corporate prediction markets have outperformed official company forecasts. The aggregation mechanism of a market forces people to put money behind their beliefs, producing more honest and calibrated estimates than surveys.

When the Wisdom of Crowds Fails

The conditions for crowd wisdom are fragile. When they break down, crowds become dangerous.

Herding and information cascades

When people can observe what others are doing, they tend to copy behavior regardless of their own private information. This produces information cascades: early decisions, even if random, get amplified as subsequent people ignore their own signals and follow the crowd. The cascade can lock in a wrong answer with enormous confidence — a crowd that looks like it knows something but is actually just copying itself.

Financial bubbles

Markets are wisdom-of-crowds systems, but they fail when independence breaks down. During speculative bubbles, investors stop estimating intrinsic value independently and start extrapolating from what others are paying. Correlated errors replace diverse independent judgments. The crowd produces catastrophically wrong prices with high conviction.

Shared biases

If a crowd shares common biases — because they share demographics, media consumption, or cultural assumptions — their errors are correlated rather than random. The aggregation mechanism amplifies the shared error rather than canceling it. A demographically homogeneous group is not a wise crowd; it's a systematically biased one.

How to Apply Crowd Intelligence

Using Collective Intelligence in Practice

Seek genuinely diverse input before major decisions. "Diverse" means cognitive diversity — people with different backgrounds, training, and frameworks — not just demographic diversity. The goal is to aggregate genuinely different perspectives, not to create the appearance of consultation while getting predictably similar views.
Protect independence in group processes. Don't reveal your own opinion before asking others for theirs. Don't let the highest-status person speak first. Use structured techniques like blind polling, written pre-commitments, or anonymous input collection to preserve the independence condition that crowd wisdom requires.
Use prediction markets or betting mechanisms for important forecasts. When you need accurate probability estimates on important questions, look for prediction market data rather than expert opinion alone. For internal organizational forecasts, consider running anonymous prediction polls with small stakes to aggregate honest estimates.
Watch for herding signals. When everyone seems to agree with unusual speed and confidence, and especially when agreement emerged after public discussion rather than before it, treat the consensus skeptically. High-conviction rapid agreement is often a sign of information cascade, not independent convergence on truth.
Aggregate widely before filtering. When gathering opinions, get as many independent views as possible before applying any filter. Filtering first (asking only certain experts) reduces diversity and may systematically exclude the perspectives most likely to be right about the specific question you're asking.
Distinguish question types. Crowds work best for estimation questions with factual answers (how much does this ox weigh? what will the election result be?). They work less well for complex judgment calls requiring deep expertise, or for ethical decisions. Apply crowd intelligence selectively based on whether the question has a verifiable answer that diverse perspectives can genuinely inform.

Common Misconceptions

❌ "The wisdom of crowds means democracy is always right"

The wisdom of crowds is a statistical phenomenon that requires specific conditions — diversity, independence, decentralization, aggregation. Democratic voting meets some of these conditions and not others. Crowds are wise about factual estimation questions with verifiable answers; they are not reliably wise about complex value tradeoffs, technical questions beyond most voters' knowledge, or questions where shared biases dominate independent judgment.

❌ "More people always means better collective judgment"

Scale alone does not produce wisdom. A crowd of a million people sharing the same bias produces a million times more bias, not more accuracy. What matters is diversity and independence of judgment, not headcount. A small diverse group with genuinely independent views will outperform a large homogeneous one consistently.

❌ "Expert consensus is just the wisdom of crowds"

Expert consensus often fails the independence condition — experts in the same field share training, read the same journals, and are subject to the same peer pressures and paradigm constraints. This produces correlated errors rather than independent ones. Expert consensus is valuable for many reasons, but it is not the same as diverse crowd wisdom and should not be confused with it.

Conclusion

The wisdom of crowds is one of the most counterintuitive findings in social science: that diverse groups of non-experts, under the right conditions, consistently outperform individual experts. The conditions — diversity, independence, decentralization, aggregation — are specific and fragile. When they hold, collective intelligence is a powerful tool. When they break down, crowds become mobs.

For practical decision-making, this means learning to harness crowd intelligence when the conditions are met and learning to recognize the telltale signs of herding and cascade when they aren't. Seek diverse independent views before revealing your own. Use aggregation mechanisms that protect honesty. And treat high-conviction rapid consensus as a warning sign, not a comfort.

Apply This in Your Next Group Decision

Before your next team discussion, poll participants anonymously in writing before any group conversation begins. Compare the distribution of pre-discussion views with the post-discussion consensus. A convergence toward one view isn't necessarily wisdom — it may be social conformity. The pre-discussion diversity is often your best signal.

About Success Odyssey Hub

Success Odyssey Hub creates evidence-based content on decision making, mental models, and the psychology of high performance. Every article draws from peer-reviewed research and the documented practices of exceptional performers across business, science, and leadership.

Recommended Books

Thinking, Fast and Slow — Daniel Kahneman — Essential companion for understanding when individual and collective judgment succeeds or fails.
The Great Mental Models Vol. 1 — Shane Parrish — Frameworks for thinking more clearly about complex problems.