Should you switch or stay? Why almost everyone gets it wrong on the first try.
Researchers such as Nobel Prize winner Daniel Kahneman have shown that human common sense often fails to make sense — especially when we need to devise strategies based on statistics.
Humans are very good at reasoning about individual events. But statistics is not about a single event — it is about repeated events and choosing strategies that improve our chances over time. And this is exactly where our intuition tends to break.
The Monty Hall Problem
A famous example of this is a simple game involving three cups and a ball, known as the Monty Hall problem.
The Rules
- The host hides a ball under one of three cups: A, B, or C
- You, the player, choose one of the cups — let's say cup A
- The host — who knows where the ball is — opens one of the remaining cups to reveal it's empty (say, cup C)
- Now only two cups remain: A and B
- The host asks: would you like to stay with A, or switch to B?
The Common Intuition
Most people say it doesn't matter. The reasoning sounds convincing:
- "The ball was already placed before I made my choice"
- "My decision cannot influence where the ball is"
- "So switching shouldn't change the outcome"
And in a sense, that reasoning is correct — for one specific round, switching cannot change where the ball is. Either the ball is under A, or it is under B.
But here is the key point: this reasoning focuses on one sample. Probability is not about a single sample — it is about choosing a strategy that works better over many repetitions.
The Statistical Reality
When we repeat the game many times, something interesting happens:
- Always stay with your original choice → you win about 1/3 of the time
- Always switch → you win about 2/3 of the time
Why should the host opening a cup change the probabilities? The answer lies in one important detail: the host knows where the ball is. When he chooses which cup to open, his action is not random — he will always open a cup that does not contain the ball. And that fact quietly changes the meaning of what we observe.
A Different Perspective
To see this more clearly, imagine a slightly different version of the game. You choose cup A, but instead of opening a cup, the host asks:
Would you like to switch from A to the other two cups together?
In other words — would you prefer A, or the pair B and C?
Now the choice is obvious:
- Probability the ball is under A: 1/3
- Probability the ball is under B or C: 2/3
Switching to the pair of cups is clearly the better choice.
Now imagine the host reveals more information — after you conceptually switch to B and C, he lifts one of those cups. Because he knows where the ball is, he always lifts the cup without the ball first. Suppose he lifts cup C, and it is empty. Now only cup B remains from the pair you chose.
The two-thirds probability that belonged to B and C together now belongs entirely to B.
The host is not changing the probabilities. He is simply revealing information about the option that already had two thirds of the probability.
Why This Matters
Switching is the better strategy — not because it changes where the ball is, but because it takes advantage of the information revealed by someone who already knows the answer.
The Monty Hall problem reminds us that:
- Human intuition works well for single events
- But when reasoning about statistics and strategies, our common sense can easily lead us astray
- Decisions based on probability need to account for strategies that work over many repetitions, not just one-off events
- Information revealed by a knowledgeable party can fundamentally change the probabilities, even when it doesn't seem like it should