In 1906, statistician Francis Galton watched 800 people at a county fair guess the weight of an ox. He expected chaos. What he found stunned him: the average of all guesses was 1,197 pounds. The actual weight? 1,198 pounds. The crowd, somehow, was nearly perfect.

But before you start crowdsourcing your life decisions, consider this: the same species that nails ox weights also produces stock market bubbles, lynch mobs, and that one time everyone agreed parachute pants looked good. Groups can be eerily brilliant or spectacularly dumb. The difference isn't luck—it's structure. And once you understand the rules, you can spot which version of the crowd you're dealing with.

Here's the cruel irony of collective intelligence: the moment people start talking to each other, they often get dumber together. Psychologist Solomon Asch demonstrated this brutally in the 1950s. He showed participants three lines and asked which matched a reference line. Easy task. But when planted actors confidently gave wrong answers first, real participants conformed about a third of the time—agreeing the short line was actually the long one.

This is why brainstorming meetings often produce worse ideas than people thinking solo. The loudest voice anchors the room. The boss's opinion becomes everyone's opinion. The first idea poisons the well. Researchers call this information cascade—once a few people commit to a view, others follow not because they agree, but because they assume the early movers know something they don't.

The fix is almost embarrassingly simple. Before any discussion, have everyone write down their independent estimate, prediction, or vote. Then aggregate. Companies like Google use this technique for project forecasts. Juries deliberate better when initial votes are private. The crowd's wisdom lives in the diversity of independent thoughts—and dies the moment we start performing agreement.

Takeaway

A group's intelligence depends on protecting each member's independent thinking before mixing it together. Premature consensus is just confident ignorance with extra steps.

Imagine you need to solve a hard problem. You can hire either ten of the world's top experts or ten thoughtful people from wildly different backgrounds. Conventional wisdom says go with the experts. Mathematical modeling by Scott Page suggests otherwise: under many conditions, the diverse group wins.

Why? Because experts in the same field tend to make the same mistakes. They were trained on the same textbooks, learned the same heuristics, and developed the same blind spots. Ten cardiologists looking at a strange case may all miss the same unusual pattern. A team that includes a cardiologist, a nurse, a data analyst, and a patient advocate brings different errors—and different errors cancel each other out.

This is the diversity prediction theorem: a group's accuracy equals the average individual accuracy plus the diversity of their perspectives. Translation: adding someone who thinks differently can boost group performance more than adding someone who simply knows more. Which explains why the most innovative teams aren't always the most credentialed ones—they're the ones where someone is always asking the awkward question nobody else thought to ask.

Takeaway

Cognitive diversity isn't a moral nicety—it's mathematical horsepower. The person who sees the problem differently is often worth more than the person who knows it better.

Once you have independent, diverse opinions, how do you actually combine them? Spoiler: it matters more than you'd think. The wrong aggregation method can turn brilliant individuals into a confused mob.

For numerical estimates—how many jellybeans, when will the project finish, what's the sales forecast—the humble average is shockingly powerful. But not always the best. The geometric mean handles outliers more gracefully. The trimmed mean, which discards the most extreme answers, outperforms both when a few people are wildly off. For predictions, prediction markets (where people bet on outcomes) often beat polls because they weight confidence: people who are sure put more skin in the game.

For qualitative decisions, methods like the Delphi technique have groups vote anonymously, share the distribution, then vote again. This captures the wisdom of discussion without the contagion of conformity. Even simple ranked-choice voting outperforms straight majority for surfacing options people can live with. The lesson: collective intelligence isn't just about who's in the room—it's about the math that turns their answers into a decision.

Takeaway

How you combine opinions can matter as much as whose opinions you collected. Wisdom needs not just contributors, but a careful recipe.

Crowds aren't magic. They're machines—and like any machine, they produce garbage when you feed them garbage and gold when you set them up properly. Independence, diversity, and thoughtful aggregation are the three gears. Strip one out, and you don't get wisdom. You get a bubble, a panic, or a really bad meeting.

The next time you're in a group decision, ask yourself: did everyone think alone first? Are we actually different, or just nodding alike? And are we combining views, or just letting the loudest one win? Get those right, and the crowd really can be smarter than you.