The puzzle feels like a 50–50 choice after one losing door is opened, but the host’s action is not random: the host knows where the prize is and deliberately reveals a losing option. Your original choice keeps its 1/3 probability, while the 2/3 probability attached to the alternatives becomes concentrated on the remaining unopened door.
The Monty Hall puzzle is uncomfortable because the final picture shows two closed doors. Two doors look like two equal possibilities, so “one-half each” feels automatic. The missing detail is how the third door was removed.
Your first choice starts at one-third
With three doors and one prize, a first pick has a 1/3 chance of being correct. That also means there is a 2/3 chance the prize is behind one of the other two doors.
Now add a host who knows the prize location, always opens a door you did not choose, always reveals a losing door, and always offers the chance to switch. Those rules matter. The host cannot accidentally reveal the prize, so the opened door carries information.
The host does not reset the probabilities
Opening a losing door does not retroactively make your original 1/3 choice better. Its probability remains 1/3. The combined 2/3 probability that originally belonged to the two unchosen doors is now forced onto the only unchosen door the host was able to leave closed. Under the standard rules, switching therefore succeeds 2/3 of the time.
A larger version makes the effect easier to feel. Imagine 100 doors. You choose one, giving yourself a 1/100 chance. A knowledgeable host then opens 98 losing doors among the ones you did not choose. It is much easier to see why the single remaining alternative carries most of the probability.
Why intuition protests
People often treat the host’s reveal as if a random door simply disappeared. If an uninformed person opened a random door and happened not to reveal the prize, the probability calculation would be different. The classic puzzle depends on the host’s knowledge and constrained behavior.
That is the real lesson of Monty Hall. Conditional probability depends not only on what you observe, but also on the process that produced the observation. Two closed doors can look symmetric while having very different histories.