Monty Hall Problem

Monty Hall Problem

Pick one of three doors, then the host reveals a goat behind another. Should you switch? Yes. Switching doubles your odds from 1/3 to 2/3, though almost nobody believes it at first.

The Monty Hall Problem is a probability puzzle named after Monty Hall, host of the TV game show "Let's Make a Deal". Its answer feels wrong to almost everyone the first time they hear it.

You're on the show. In front of you are three doors: Door A, Door B and Door C. Behind one is a brand new car. Behind the other two are goats. You want the car.

Here's how it goes:

  1. You pick a door, say Door A.
  2. Monty, who knows where the car is, opens one of the other two doors to show a goat (say Door B).
  3. He then asks if you want to stick with Door A or switch to Door C.

Should you stick, switch, or does it not matter?

Your gut probably says it's 50-50 now, since there are two doors left. But you actually double your chances of winning by switching.

Here's why:

When you first pick, you have a 1 in 3 chance of choosing the car and a 2 in 3 chance of choosing a goat. If you stick, you keep that 1 in 3 chance. If you switch, you win whenever your first pick was a goat, which happens 2 times out of 3. So switching wins the car 2 in 3 times. Monty's reveal isn't random. He always shows you a goat, and that gives you information. In this game, switching is the smart move, and it's a neat reminder that our instincts about probability can be badly off, just as they are in the birthday paradox.

The Monty Hall Problem is a classic probability puzzle based on the TV game show "Let's Make a Deal", hosted by Monty Hall (Selvin, 1975). It is a showcase for conditional probability, and an even better showcase for how badly human intuition can misfire when probabilities are involved (Tversky & Kahneman, 1974).

The setup: a contestant picks one of three doors. Behind one is a car; behind the other two, goats. The host, who knows what is behind every door, then opens one of the two remaining doors to reveal a goat. The contestant can now stick with the original pick or switch to the other unopened door.

Does switching help, hurt, or make no difference? The counterintuitive answer is that switching doubles your chances, raising the probability of winning the car from 1/3 to 2/3.

The puzzle connects to several principles and fields:

  1. Conditional probability and Bayes' theoremThe problem can be worked through with conditional probability and Bayes' theorem (Bayes & Price, 1763), which tells us how to update probabilities in light of new information. Once Monty reveals a goat, applying Bayes' theorem to each door shows that the car is more likely to be behind the door you didn't pick, so switching raises your odds.

  2. Behavioral economics and decision-makingThe puzzle is a staple of the literature on human judgement and behavioral economics (Tversky & Kahneman, 1974). We lean on heuristics, or mental shortcuts, and these can lead us astray when probability is involved. Here the shortcut says "two doors left, so it's 50/50", while the actual probabilities say otherwise. It's the same family of error as the gambler's fallacy, where a feeling about how chance ought to behave overrides the actual odds.

  3. Randomized algorithms and the exploration-exploitation trade-offThe problem can be framed as a simple instance of a wider family of problems in artificial intelligence and reinforcement learning, where an agent must balance exploring new options against exploiting what it already knows (Sutton & Barto, 2018). In this framing, switching doors is exploration and sticking is exploitation. The lesson carries over: update your beliefs when new information arrives, and take calculated risks to maximise reward.

Few puzzles pack so much into three doors and a goat, though the birthday paradox gives it a run for its money. Monty Hall teaches conditional probability, exposes the limits of intuition, and links to decision-making, behavioral economics and artificial intelligence.

References

  • Bayes, T., & Price, R. (1763). An essay towards solving a problem in the doctrine of chances. Philosophical Transactions of the Royal Society of London, 53, 370-418.
  • Selvin, S. (1975). A problem in probability (letter to the editor). The American Statistician, 29(1), 67.
  • Sutton, R. S., & Barto, A. G. (2018). Reinforcement learning: An introduction. MIT Press.
  • Tversky, A., & Kahneman, D. (1974). Judgment under uncertainty: Heuristics and biases. Science, 185(4157), 1124-1131.