Why Groups Often Think Worse Than the People in Them

The Puzzle of the Smart Group That Fails

In April 1961, a small team of senior American officials approved a plan to land 1,400 Cuban exiles on a beach at the Bay of Pigs, expecting the invasion to trigger a popular uprising against Fidel Castro. The men in that room were not fools. They included some of the most capable minds in Washington, several of them Harvard-trained economists and lawyers who had built reputations on rigorous analysis. Yet the plan collapsed within three days, the exile force was captured or killed, and the decision is still taught in university courses as one of the clearest examples of a group making a choice that almost none of its members would have made alone.

That gap is the puzzle. Individually, the planners understood the risks. Several later admitted they had private doubts about the operation’s chances. But those doubts rarely surfaced in the meetings where the decision took shape. Something about the group setting itself narrowed the discussion, suppressed dissent, and produced a false sense of confidence that no single participant actually held.

This pattern is not confined to Cold War history. It shows up in corporate boardrooms, hospital committees, city councils, and juries. The question worth asking is not simply “why do groups make mistakes,” since individuals make mistakes too. The real question is why a collection of capable people, working together with more combined information than any one of them possesses alone, can still produce worse judgment than its individual members would reach on their own.

What Psychologists Mean by Groupthink

The psychologist Irving Janis studied the Bay of Pigs decision and several similar failures, including the escalation of the Vietnam War, and in 1972 proposed a concept he called groupthink. Janis defined it as a mode of thinking that group members fall into when the desire for agreement overrides their motivation to realistically evaluate alternatives.

Janis identified several conditions that tend to produce this pattern. Group members begin to feel an unwarranted sense of invulnerability, believing that their group’s judgment cannot go seriously wrong. Individuals who have private doubts engage in self-censorship rather than voice them, often because they assume they must be the only one who disagrees. This creates what Janis called an illusion of unanimity: because no one speaks up, everyone assumes the silence reflects genuine agreement, when it may reflect widespread private doubt.

Groupthink also tends to appear alongside strong group cohesion, insulation from outside opinions, and a directive leader who signals a preferred outcome before the group has fully debated it. Under those conditions, the group starts filtering out information that would complicate the emerging consensus rather than seeking it out.

It is worth being honest about the limits of this theory. Janis built groupthink from historical case studies rather than controlled experiments, and later laboratory research testing his model has produced mixed results. Some of the specific mechanisms he proposed hold up more consistently than others. What survives across decades of follow-up research is the broader insight: cohesive groups under pressure to agree quickly can suppress the very disagreement that would have improved their decision.

Why People Conform Even When They Know Better

Groupthink explains what happens inside a decision-making body, but a more basic psychological force is often operating underneath it: the pressure to conform.

In the 1950s, the psychologist Solomon Asch ran a now-famous series of experiments in which a research subject was asked to judge the length of lines while seated among several other participants who were secretly instructed to give an obviously wrong answer. About three-quarters of the actual subjects went along with the incorrect group answer at least once, even though the correct answer was visually unambiguous.

Asch’s work distinguished between two different reasons people conform. Informational conformity happens when someone genuinely believes the group must know something they don’t, so they update their own judgment to match. Normative conformity happens when someone privately disagrees but goes along anyway to avoid standing out or facing social disapproval. In the Bay of Pigs meetings, the second kind of pressure appears to have mattered more than the first. Several planners were not persuaded that the operation would succeed. They simply found it easier to stay quiet than to challenge a plan the president and his closest advisers already favored.

This helps explain why adding more people to a discussion does not automatically add more independent judgment. If each additional voice is influenced by the voices that came before it, the group is not really combining separate opinions. It is amplifying whichever opinion happened to be expressed first and most confidently.

The Diffusion of Responsibility

A second force works alongside conformity: when responsibility is shared, individuals often feel less personally accountable for the outcome.

Social psychologists have documented this effect across many settings, from decisions about whether to intervene in an emergency to decisions about how much risk to accept in a group project. When a choice belongs to everyone, it can end up belonging to no one in particular, and that diffusion changes how carefully any single person scrutinizes it.

This same dynamic shapes risk-taking. Early research on group discussion found that groups sometimes shift toward riskier decisions than any individual member would choose alone, a pattern researchers originally called the “risky shift.” Later work, led by the social psychologist Serge Moscovici, showed the effect was broader than risk alone: groups tend to move toward a more extreme version of whatever position their members already lean toward before the discussion begins, a phenomenon now called group polarization. A team that starts out cautiously optimistic about a plan often becomes more confident after discussing it together, not because new evidence emerged, but because the discussion reinforced the shared lean and made dissent feel increasingly out of place.

When More Voices Make Things Worse, Not Better

A related but distinct problem occurs when people make decisions in sequence rather than simultaneously, each one observing what those before them chose. Economists call this an information cascade.

Imagine a group evaluating a proposal one member at a time. The first person gives an honest assessment based on their own information. The second person has some independent information too, but also sees what the first person said, and if the two conflict, may reasonably conclude the first person’s information was stronger and go along with it. By the time the discussion reaches the fifth or sixth participant, most people are no longer contributing their own independent signal. They are simply relaying the emerging consensus, even though that consensus may have been built on very little actual evidence.

This is a structural problem, not a personal failing. Rational individuals, behaving sensibly given what they can see, can collectively produce a herd that has abandoned the independent judgment that made the group valuable in the first place. Financial bubbles follow a version of this logic: each investor observes rising prices and reasonably infers that others must know something, even when the rise reflects nothing more than the same inference being made in a loop.

What Popular Belief Gets Wrong About Crowds

None of this means that groups are inherently worse than individuals. In some conditions, they are dramatically better.

In 1906, the statistician Francis Galton examined the results of a contest at a country fair in which several hundred people guessed the weight of an ox. No individual guess came close to the correct weight, but the median of all the guesses combined was within one percent of the ox’s actual weight. This became one of the founding examples of what is now called the wisdom of crowds: aggregating many independent estimates can produce a result more accurate than any single expert’s guess.

The key word is independent. Galton’s fairground guessers did not confer with each other before writing down their estimates. Each person’s error was essentially random, and when many random errors are averaged together, they tend to cancel out. That is a fundamentally different situation from a boardroom discussion, where people hear each other’s opinions before forming their own and where social pressure can push everyone toward the same answer rather than letting errors cancel.

This distinction resolves the apparent contradiction. Crowds are not smart or dumb by nature. They are smart when they aggregate independent judgments and dumb when they replace independent judgment with imitation. The Bay of Pigs planners were not lacking in expertise. They were lacking in independence, because the format of their discussions rewarded agreement over dissent.

The Conditions That Turn Deliberation Around

If suppressed independence is the underlying problem, the research also points to what restores it.

Janis’s own prescriptions for avoiding groupthink centered on deliberately reintroducing dissent: assigning someone the explicit role of devil’s advocate, inviting outside experts who have no stake in the group’s cohesion, and having leaders withhold their own preferences until after the group has debated freely. NASA adopted several of these practices after the 1986 Challenger disaster, another case in which engineers with real safety concerns felt unable to make those concerns prevail in a launch-decision meeting under organizational and schedule pressure.

Structured techniques that force people to record judgments privately before any group discussion, sometimes called the Delphi method, try to capture something closer to Galton’s ox-weighing crowd: independent estimates gathered before social pressure can homogenize them. Anonymous voting, written pre-commitments, and rotating who speaks first are simpler versions of the same idea. None of these techniques eliminate the risk of poor group judgment entirely, but they intervene at the actual point of failure, which is the moment independent information gets converted into a single, prematurely unified opinion.

Why This Still Matters

The instinct to explain a bad group decision by pointing to the individuals involved, calling them incompetent or reckless, misses what the research actually shows. Groups do not usually fail because their members are unqualified. They fail because ordinary social pressures, ones that operate on capable and well-intentioned people, can quietly convert a room full of independent judgments into a single, unexamined consensus.

The Bay of Pigs planners were not short on intelligence or information. They were short on a structure that let their private doubts reach the table. That distinction matters because it is fixable. Unlike a shortage of expertise, a shortage of independence can be corrected through deliberate design: through dissent that is invited rather than merely tolerated, and through judgments that are formed before they are shared rather than after.

The most capable group is not the one with the most talented members. It is the one that has figured out how to keep those members thinking independently even while they work together.