;colony/science  / Mathematics, Made Visual  / The Monty Hall problem
Mathematics, Made Visual · No. 208 of the first 100

The Monty Hall problem

Three doors, one car, one simple question — and an answer so unintuitive that thousands of mathematicians once wrote in to call it wrong.

Plate 38 — Stick or switch P(stick) = ⅓ · P(switch) = ⅔
Play a round on your gut, then run two hundred and count.
1 2 3a car behind one door, goats behind two — pickstick: 0/0 won · 0%switch: 0/0 won · 0%
PLATE 38 · STICK OR SWITCH
Pick a door on the left to play a round.
Don't trust one game — run two hundred
Stick wins
0% of 0
Switch wins
0% of 0
Your first pick is right 1 time in 3 — that never changes. Which means the car is behind one of the other doors 2 times in 3. When the host kindly removes a goat from those two, all of that 2-in-3 luck funnels into the one door left. Switching isn't a hunch — it doubles your wins. Run the 200 games and watch.
The short answer

You pick one of three doors. The host, who knows where the car is, opens a different door with a goat, then asks: stick or switch? Almost everyone says it makes no difference. It does. Switching wins twice as often — and you can prove it here by playing two hundred games in two seconds.

What's actually happening

The setup comes from the old game show Let's Make a Deal, and the storm it caused is half the fun. In 1990, Marilyn vos Savant answered it correctly in Parade magazine — switch, it doubles your chances — and received some ten thousand letters insisting she was wrong, around a thousand of them from people with PhDs. The greatest mathematicians weren't immune: Paul Erdős reportedly refused to accept the answer until he was shown a computer simulation. The tally counter above is exactly that simulation.

Here's the way to feel it rather than fight it. Your first pick is right one time in three — nothing the host does afterwards can reach back and improve a guess you already made. So two-thirds of the time, the car is behind one of the doors you didn't pick. Now the host does you an enormous, easily-missed favour: from those two doors, he removes a guaranteed goat. He hasn't shuffled anything; he's taken the two-thirds share and funnelled all of it onto a single door. Sticking keeps your original ⅓. Switching inherits the ⅔.

The detail everything hinges on: the host knows. He always opens a goat door, never the car, never yours. If instead a clueless host opened a random door (sometimes revealing the car and spoiling the game), the leftover odds really would be 50:50, and intuition would be right. The puzzle isn't about doors at all — it's about what information someone's deliberate behaviour leaks. That instinct, formalised, is Bayesian reasoning, and it runs spam filters and medical-test interpretation alike.

Try it at home Convince a sceptic with three cups
  1. 1Hide a coin under one of three cups while a friend looks away. They pick a cup; you (knowing where the coin is) lift an empty one from the other two, and they decide: stick or switch.
  2. 2Play twenty rounds with them always sticking, twenty always switching, tallying wins.
  3. 3Sticking lands near 7/20; switching near 13/20. Watching their own tally beat their own intuition is the moment it clicks — it never works by argument alone.