The Monty Hall Problem: When Intuition Fails

A reader wrote to a newspaper columnist in 1990 and accidentally started one of the most heated math debates in American history. Thousands of PhD holders wrote angry letters insisting she was wrong. She wasn't.

The columnist was Marilyn vos Savant, who at the time held the Guinness World Record for the highest recorded IQ. Her column, "Ask Marilyn," ran in Parade magazine, and the question she received was about a game show. The answer she gave was correct, simple, and so counterintuitive that it made professional mathematicians furious.

The Setup: Three Doors

The problem is named after Monty Hall, who hosted the TV game show Let's Make a Deal from 1963 to 1986. Here's how it works. You're a contestant on a game show. In front of you are three closed doors. Behind one door is a car. Behind the other two are goats. You pick a door. The host, who knows what's behind every door, opens a different door to reveal a goat. Then he asks you: do you want to stick with your original choice, or switch to the remaining unopened door?

Most people say it doesn't matter. They're wrong. You should always switch. Switching wins the car two-thirds of the time. Staying wins only one-third of the time.

If that strikes you as absurd, you're in excellent company.

Why Your Gut Says "It Doesn't Matter"

The most common reaction goes something like this: "After the host opens a door, there are two doors left. The car is behind one of them. So it's 50/50." This reasoning feels airtight. It's also wrong, and the mistake has a name: equiprobability bias. When people see two options, they instinctively assume each is equally likely. Two doors, two outcomes, must be 50/50.

But the host's action isn't random. Monty Hall doesn't open a door at random and happen to reveal a goat. He always opens a door with a goat behind it, and he always opens a door you didn't pick. That constraint is doing enormous work. It means his choice carries information about where the car is, and that information changes the odds.

Think of it this way: when you first picked a door, you had a 1/3 chance of being right. That means there was a 2/3 chance the car was behind one of the other two doors. When the host opens one of those two doors and shows you a goat, the entire 2/3 probability collapses onto the single remaining door. Your original pick is still sitting at 1/3.

The Evidence: 10,000 Simulated Games

Words can only do so much. Let's run the experiment. Here's a JavaScript simulation you can paste into your browser console or any code editor. It plays the Monty Hall game 10,000 times and counts how often each strategy wins:

function montyHallSimulation(trials = 10000) {
  let stayWins = 0;
  let switchWins = 0;

  for (let i = 0; i < trials; i++) {
    // Place car behind a random door (0, 1, or 2)
    const carDoor = Math.floor(Math.random() * 3);
    // Contestant picks a random door
    const pick = Math.floor(Math.random() * 3);

    // Host opens a door that isn't the pick
    // and isn't the car
    let hostOpens;
    do {
      hostOpens = Math.floor(Math.random() * 3);
    } while (hostOpens === pick || hostOpens === carDoor);

    // Staying wins if original pick has the car
    if (pick === carDoor) stayWins++;

    // Switching wins if original pick was wrong
    // (the remaining door has the car)
    if (pick !== carDoor) switchWins++;
  }

  return { stayWins, switchWins, trials };
}

const result = montyHallSimulation();
console.log(`Stay wins:   ${result.stayWins} / ${result.trials}`);
console.log(`Switch wins: ${result.switchWins} / ${result.trials}`);

Run it yourself, or trust the math. Across 10,000 trials, the results consistently land near these numbers:

The gap isn't subtle. Switching doubles your chances. You can try generating your own random numbers between 1 and 3 to simulate door placements by hand, but 10,000 automated runs make the pattern undeniable.

Why Switching Wins: A Bayesian Walk-Through

The simulation shows that switching works. Here's why, broken down as a probability tree. Suppose you pick Door 1.

Scenario A: Car is behind Door 1 (probability 1/3). You picked right. The host opens Door 2 or Door 3 (doesn't matter which). If you switch, you lose. If you stay, you win.

Scenario B: Car is behind Door 2 (probability 1/3). You picked wrong. The host must open Door 3 (the only goat he can reveal). If you switch to Door 2, you win. If you stay, you lose.

Scenario C: Car is behind Door 3 (probability 1/3). You picked wrong again. The host must open Door 2. If you switch to Door 3, you win. If you stay, you lose.

Count the outcomes. Staying wins in 1 scenario out of 3. Switching wins in 2 scenarios out of 3. That's it. The host's forced behavior is what creates the asymmetry. In Scenarios B and C, the host has no choice about which door to open, and that lack of choice is a signal pointing you toward the car.

In Bayesian terms, the host's action updates your prior. Your initial belief was 1/3 for each door. After the host reveals a goat behind, say, Door 3, the posterior probability for Door 1 stays at 1/3, while the posterior for Door 2 jumps to 2/3. The evidence (the host's forced move) concentrates probability onto the door you didn't pick.

The Controversy That Shook Mathematicians

When Marilyn vos Savant published her answer in September 1990, the backlash was immediate and intense. She received an estimated 10,000 letters, roughly 1,000 of them from people with PhDs. Many were not polite. A professor from Georgetown wrote: "You are the goat!" Another from the University of Florida said: "You made a mistake, but look at the positive side. If all those PhDs were wrong, the country would be in serious trouble."

The mathematician Paul Erdos, one of the most prolific in history with over 1,500 published papers, reportedly refused to believe the answer until a colleague showed him a computer simulation. Even then, according to accounts from people who knew him, he wasn't fully comfortable with it until he worked through the formal proof himself.

What happened here wasn't just a math mistake. It was authority bias in action. Thousands of credentialed experts were so confident in their wrong intuition that they assumed the person disagreeing must be the one making an error. The fact that vos Savant was a woman writing a popular magazine column (not an academic journal) made it easier for them to dismiss her. Several letters explicitly referenced her gender or her venue as reasons she couldn't be right. The episode became a case study in how credentials can actually hinder clear thinking when they make people too certain of their first instinct.

You can read vos Savant's original explanation and the follow-up responses in her archived Parade columns. The clarity of her reasoning holds up remarkably well three decades later.

Where This Applies Beyond Game Shows

The Monty Hall problem isn't just a party trick. The underlying principle shows up in real decisions where new information should change your strategy but your gut tells you nothing has changed.

Medical testing. A screening test comes back positive. Your doctor says the test has a 95% accuracy rate. Most patients assume they're 95% likely to have the condition. But if the condition affects 1 in 1,000 people, the actual probability (even with that accurate test) might be below 2%, because false positives vastly outnumber true positives in a large population. The math is structurally identical to the Monty Hall problem: new information (the test result) should update your prior, but people anchor on the wrong number.

A/B testing in software. You run two variants of a landing page. Variant B outperforms Variant A for the first week. Your instinct says B is better. But if the sample size is small and you haven't accounted for multiple comparisons, that early signal might be noise. Experienced product teams know to "switch doors" when the data tells them to, even when the original variant felt more promising. Using a dice roller or random number generator to mentally simulate small-sample variance can build intuition for why early results mislead.

Investment decisions. You pick a stock. New information arrives that changes the landscape. Do you stick with your original pick because you already committed? That's the "stay" strategy, and in many contexts it's wrong for the same reason staying is wrong in the Monty Hall problem. The arrival of new information should genuinely change your assessment, not just confirm your existing position. Behavioral economists call this the endowment effect when applied to decisions: people overvalue things simply because they already hold them.

The practical rule: Whenever new evidence eliminates some possibilities but not others, the remaining alternatives you didn't originally choose become more likely. Your first pick doesn't get stronger just because you made it first. Updating your beliefs when the situation changes isn't indecisiveness. It's rationality.

The Monty Hall problem endures because it sits at the exact boundary where human intuition breaks down and mathematical reasoning takes over. We want the world to be 50/50 when there are two options left. It often isn't. And the people who recognize that, whether they're game show contestants, doctors, or investors, tend to make better decisions over time.