# Monty Hall problem

1. Aug 16, 2011

### lowerlowerhk

http://en.wikipedia.org/wiki/Monty_Hall_problem

This is a game about probability:
Say there is three doors. Two goats and a car are hidden behind.
Wanting to pick the car, you randomly picked a door.
Then the game host open one of the unchoosen door which happened to contain a goat.
And the host offer you a chance to switch door - to switch to the remaining unchoosen door.
The question (not my question though) is, should you switch door?

The correct answer is, switching is always better because the only way you could lose the game by switching is to pick the correct door at the first place, which is unlikely.

But I don't understand how the probabilities transfer during the host's action. Let's say the host is going to open a door which he knows a goat is inside. To me, before the door is opened it contain some probabilities to be a car. After the door is opened, that probability is transferred away. But why the car's probability only goes to the unchoosen door instead of distributed evenly between all unopened doors?

2. Aug 16, 2011

### BloodyFrozen

Did you look at the diagram at the bottom?

You pick goat 1, Host shows Goat two, switch->you win
You pick goat 2, Host shows Goat one, switch->you win
You pick car, Host shows either goats, switch-> you lose

2 wins out of 3

In the beginning you have a 2/3 chance of picking a goat

Last edited: Aug 16, 2011
3. Aug 16, 2011

### lowerlowerhk

Doing an exhaustive list does yield the result, but I am not asking for that. I am asking why the probability transfers to one door only.

4. Aug 16, 2011

### BloodyFrozen

5. Aug 16, 2011

### lowerlowerhk

It said the probability of the unchoosen door increased, but it didn't tell why doesn't the probability of the choosen door increase accordingly. May be I missed something?

Last edited: Aug 16, 2011
6. Aug 16, 2011

### mathman

The probability of the original chosen door is 1/3. Opening a door when the host knows is not the car (and at least one of the two remaining has a goat) does not give you any more information about the door you chose. However, the fact that the host opened one of the other doors revealing a goat does give you more information about the remaining door.

7. Aug 16, 2011

### Staff: Mentor

8. Aug 16, 2011

### disregardthat

The crucial thing is (which makes the probability "change") is that the host knows which door the goats are behind, and will only open a door with a goat behind. So you get more information about the game after he opens a door. He would never open a door with a car behind. If we randomly chose a door, it wouldn't matter if you chose the other closed one.

9. Aug 16, 2011

### Anonymous217

Exactly. If the host had chosen a door arbitrarily, where there could have been a car, then switching would not make a difference.

Note the considerable difference between this and Deal or No Deal, where the case is chosen at random, without the "secret knowledge" seen in the Monty Hall problem. So switching cases does not make a difference.

10. Aug 16, 2011

### disregardthat

Another thing which is crucial is knowing that the host would open a door and ask you whether you wanted to switch regardless of what door you chose. If he did that only when you chose the car, you would always lose.

11. Aug 17, 2011

### daveyp225

One thing I never understood about the phrasing of this problem is that it seems necessary to make the choice to switch before the offer is even made to obtain the "edge" in the solution. During the second round of the game he is effectively asking you to make a choice between two doors, changing the problem.

12. Aug 17, 2011

### Dr. Seafood

How can you switch before the offer is made? Suppose you choose the first door -- you can't really "switch" since both the other two doors are both still closed. The advantage of switching comes from that you know that Monty knows which door hides which item, and so he was deliberate in his choice of which doors contents to reveal when making the offer to switch.

The second round of the game contains more information than (often) considered at first. So it's not exactly the same as randomly choosing a door from two in hopes that you win. You also know that (1) if you picked a goat, he would reveal the (only other) remaining goat; (2) if you picked the car, he would arbitrarily choose (either of) the remaining goats.

13. Aug 17, 2011

### Jocko Homo

I don't understand this...

Suppose Monty chose a door random. Before he opens the door, it may or may not have the car behind it. However, once he opens the door, we will know whether the car is behind it or not. If there isn't a car behind it, doesn't the situation revert back to the classic Monty Hall problem? How could it be different merely because of the method Monty used to choose a door? How does that express itself in the problem?

14. Aug 17, 2011

### daveyp225

I said that one must choose to switch before the offer is made, not switching before it is made. I understand the probability of making the choice to switch before he opens the door and being correct is 2/3. However, if I had suddenly incurred amnesia after the first step of the problem, and was presented with either keeping door 1 versus choosing door 2 over 1, how is the probability not now 1/2? Also, how is the Monty Hall problem any different than one with 4 doors, but one false door already open?

15. Aug 17, 2011

### HallsofIvy

Staff Emeritus
Because in this scenariio, you have completely dropped the third door, reducing the problem to one out of two equally likely possibliities: probability of either, 1/2.

Not at all different. That "fourth door" does not matter.

16. Aug 17, 2011

### Anonymous217

Not at all. The fact that there isn't a car behind it doesn't translate to Monty having the "secret knowledge" necessary to affect the probability. Given you chose a goat, it was merely 50% chance that he picked out the other goat. However, with the secret knowledge, it's 100% chance he picks out the other goat.
It's hard to elaborate conceptually since it's more or less intuitive. If all else fails, try an exhaustive list again. In terms of this list, note that the 50% causes more possibilities to arise, and these extra possibilities tie in to make the 50% overall mark of switching.

17. Aug 21, 2011

### DaveC426913

I still don't get it. Monty opening one of the doors has changed the probability of both the door you didn't pick and the door you did pick.

So how does switching increase the likelihood of success?

Bah, I'm just going to brute force this problem and prove one way or 'tother once and for all...

18. Aug 21, 2011

### DaveC426913

Monty Hall
A: goat B: goat C: car

1 You choose A
1.1 Monty opens B
1.1.1 You stay with A, You lose
1.1.2 You switch to C, You win

2 You choose B
2.1 Monty opens A
2.1.1 You stay with B, You lose
2.1.2 You switch to C, You win

3 You choose C
3.1 Monty opens A
3.1.1 You stay with C, You win
3.1.2 You switch to B, You lose
or
3.2 Monty opens B
3.2.1 You stay with C, You win
3.2.2 You switch to A, You lose

Conclusion: It makes no difference whether you switch or not. If you switch, you have a 50/50 chance of winning the car. If you do not switch, you have a 50/50 chance of winning the car. The fact that your first choice had a 1/3 chance of winning is a red herring, and is what keeps tripping people up.

19. Aug 21, 2011

### DaveC426913

And yet...

...it's wrong.

Here's a simulator (only works in IE)
http://www.grand-illusions.com/simulator/montysim.htm

It shows manifestly that switching increases your wins from 1-in-3 to 2-in-3.

Logic has let me down. I am going to go join a church. Goodbye PF.

20. Aug 21, 2011

### zhentil

Ok, suppose there are ten billion doors. You open one, and then monty reveals 9,999,999,998 goats. Still fifty-fifty? If so, please come to my new casino.