In April 1961, some of the most experienced people in the American government sat in one room and agreed on a plan. They would send a CIA-trained brigade of Cuban exiles to invade Cuba at the Bay of Pigs. Almost nobody in the room objected. The idea had been drafted under the previous administration and inherited wholesale, and the new team accepted it without ever really stress-testing it. A few people did have doubts. Arthur Schlesinger Jr. and Senator J. William Fulbright raised objections, and the group waved them away, then went on believing in its own cause. The invasion collapsed within days. What makes the story worth retelling is not that clever men failed. It is that they failed because the room was so agreeable.
🧠 The Room That Said Yes
A year and a half later, the Yale psychologist Irving Janis published Victims of Groupthink. The word itself was older; William H. Whyte Jr. had borrowed it from Orwell’s Nineteen Eighty-Four in a 1952 Fortune piece, but Janis gave it teeth. He studied the Bay of Pigs and the failure to anticipate Pearl Harbor, later adding the Vietnam escalation and Watergate, and the same shape kept showing up.
The decisions were not bad because the people were ignorant. They were bad because the people were close. Janis turned it into something like a law: “The more amiability and esprit de corps there is among the members of a policy-making ingroup, the greater the danger that independent critical thinking will be replaced by groupthink, which is likely to result in irrational and dehumanizing actions directed against outgroups.”
He sorted the failure modes into eight symptoms across three families. First the group overrates itself: an illusion of invulnerability, and an unquestioned belief in its own morality. Then it closes its mind, rationalizing warnings away and stereotyping outsiders as weak or stupid. Then comes the part that actually does the damage, the pressure toward uniformity. Members censor their own doubts. Silence gets read as agreement. Anyone who objects is accused of disloyalty. And self-appointed “mindguards” quietly shield the group from information that doesn’t fit.
At the Bay of Pigs you can watch it run in order. The exiles’ chances were never seriously questioned. Castro’s air force was assumed to be useless. When Schlesinger raised doubts, he ended up talking himself out of them, the quiet kind of self-censorship. The objections were heard, then filed away.
🤔 It’s the Friendliness, Not the Stupidity
Here is the part that feels wrong at first: the dangerous ingredient isn’t conflict. It’s the absence of it. A team that argues is, in this frame, safer than a team that gets along.
But the twist that keeps the story honest is this. The claim everyone remembers, that group cohesion is the main cause, turns out to be the shakiest one. A 1990 review by Park counted only sixteen empirical studies and found they gave Janis “only partial support.” His summary is blunt: “no research has shown a significant main effect of cohesiveness on groupthink.” Later studies pointed in opposite directions. One found that leaders hungry for power made groupthink likelier. Another found the reverse, that groups with dominant members actually made higher-quality decisions. What survives every review is the structural stuff: a homogeneous group, an insulated group, and a leader who announces the answer before anyone else speaks.
So the honest version is narrower, and more useful. Groupthink isn’t “being around people makes you stupid.” It’s “a room with no norm of dissent, no contact with the outside, and a boss who speaks first is a room built to fail.”
🔗 The Cozy Room Is the Risky One
This isn’t only a story about presidents. Cass Sunstein has pointed out that investment groups full of internal socializing and happy talk are more prone to bad decisions, while groups of relative strangers, who are willing to argue, tend to do better. The corporate record rhymes: when Marks & Spencer and British Airways ran their 1990s globalization drives, a content analysis of their press releases found all eight symptoms present, the loudest being the illusion of invulnerability. Within a couple of years both stocks had roughly halved.
The one-on-one conversation is the opposite shape. There is no audience to perform for, no loyalty to signal, no silence mistaken for agreement. A confidant who isn’t trying to keep the peace, who says the inconvenient thing precisely because no one else is watching, is small and unglamorous and structurally immune to this particular failure. That turns out to matter more than any dashboard.
Janis’s fixes are almost boringly practical. Give every member the explicit job of being a critical evaluator. Have leaders state no opinion while they hand out the task. Keep the leader out of some of the meetings entirely. Split the group into independent teams attacking the same problem. Invite outside experts and let members grill them. And name a devil’s advocate, rotating the role each time.
🎲 The Cure, and Its Catch
The best proof the diagnosis was right is what happened next. One year after the Bay of Pigs, the same administration faced the Cuban Missile Crisis. Same men, same stakes, but the process had been rebuilt. Kennedy invited outside experts, broke the group into smaller ones, encouraged people to argue, and deliberately stayed out of many meetings. They still came close to catastrophe, but they landed. Janis had a name for it: vigilant appraisal. Same people. Different room.
One last wrinkle: a devil’s advocate who argues a side he doesn’t believe is much weaker than someone who genuinely disagrees. If you want the benefit, you need a real dissenting voice, a person or a team actually committed to the other path. Fake dissent convinces nobody, the group included.
And if you have ever been part of a group that took an action none of you actually wanted, each person assuming the others were on board, that has its own name too. The Abilene Paradox. But that’s another candy.