3 prisoner puzzle

This the easiest way I know of to explain the difference between choosing a card at random and choosing a card on purpose.

In each case, Tom and Dick are drawing cards from a deck of three, one of which is the Ace of Spades.

Case A: Both Tom and Dick choose a card at random.
Before anyone draws a card, there is a probability of 3/3 that the Ace is in the deck.
After Tom chooses a card, there is a probability of 1/3 that he has the Ace, and therefore 1/3 chance of removing it from the deck, leaving a probability of 2/3 the Ace is still in the deck. After Dick chooses his card (at random), there is a 1/3 probability that he has the Ace and 1/3 that it is still in the deck. When Dick turns over his card to show that it isn't, in fact, the Ace, the odds are 1/3:1/3 or 1:1 = 1/2 that the Ace is in Tom's hand.

Case B: Tom chooses a card at random, Dick chooses a non-Ace on purpose (assuming if they are both non-Aces he will choose one at random).
Before the draw, there is a 3/3 chance the Ace is still in the deck.
After Tom chooses a card, there is a 1/3 chance the Ace is in Tom's hand, and a 2/3 chance the Ace is still in the Deck.
When Dick looks at the cards and chooses a non-Ace, there is 0 chance of the Ace being and Dick's hand and therefore still 2/3 chance the Ace is in the deck.
The odds are then 1/3:2/3 or 1:2 = 1/3 that Tom holds the Ace.

The key is that, in Case A, Dick has a 1/3 chance of removing the Ace from the deck, while in Case B, he does not.
 
This the easiest way I know of to explain the difference between choosing a card at random and choosing a card on purpose.

In each case, Tom and Dick are drawing cards from a deck of three, one of which is the Ace of Spades.

Case A: Both Tom and Dick choose a card at random.
Before anyone draws a card, there is a probability of 3/3 that the Ace is in the deck.
After Tom chooses a card, there is a probability of 1/3 that he has the Ace, and therefore 1/3 chance of removing it from the deck, leaving a probability of 2/3 the Ace is still in the deck. After Dick chooses his card (at random), there is a 1/3 probability that he has the Ace and 1/3 that it is still in the deck. When Dick turns over his card to show that it isn't, in fact, the Ace, the odds are 1/3:1/3 or 1:1 = 1/2 that the Ace is in Tom's hand.

Case B: Tom chooses a card at random, Dick chooses a non-Ace on purpose (assuming if they are both non-Aces he will choose one at random).
Before the draw, there is a 3/3 chance the Ace is still in the deck.
After Tom chooses a card, there is a 1/3 chance the Ace is in Tom's hand, and a 2/3 chance the Ace is still in the Deck.
When Dick looks at the cards and chooses a non-Ace, there is 0 chance of the Ace being and Dick's hand and therefore still 2/3 chance the Ace is in the deck.
The odds are then 1/3:2/3 or 1:2 = 1/3 that Tom holds the Ace.

The key is that, in Case A, Dick has a 1/3 chance of removing the Ace from the deck, while in Case B, he does not.

This is probably the best explanation of this I've heard. And yet, I know I'm still going to have trouble translating it to the 2 goats/1 car problem that my brother and I argued over last Thanksgiving.
 
Dick has marked the deck of one ace of spades and two jokers. Tom chooses first among the cards dealt face down. His chances of drawing the ace are 1/3. Dick always chooses one joker or the other. His chance of drawing the ace is 0/2. Poor sucker Harry is always stuck with the remaining card which has a 2/3 chance of being the ace. The reason it is not 1/2 is that a third of the time, Tom has already chosen the ace, so in those rounds, Harry has a 1/1 chance.

In the first round, Tom choses the ace, Dick takes a joker leaving Harry with the other joker. In the second round, Tom choses a joker, Dick takes the other joker leaving Harry with the ace. In the third round, Tom choses the other joker, Dick takes the joker, leaving Harry with the ace, again.

In the ace-loses game: If Tom has the option of trading with Harry, he should not. In the ace-wins game, if Tom has the option of trading with Harry, he should as Harry will be holding the ace 2/3 times to his own 1/3 times.
 
But...

Tom didn't gain any new information from the warden.

He already knew that at least one of the other two guys would be set free. Knowing which guy changes nothing.

Right?

The probability stays exactly the same whether the Warden says:

"Dick will be set free."

"Harry will be set free."

"I'm not telling you who, but it will be two of you."

All three of those answers provides the same information on Tom's fate, which is none.

Right? If not, why not?
 
Last edited:
The probability stays exactly the same whether the Warden says:

"Dick will be set free."

"Harry will be set free."

I'm not telling you who, but it will be two of you."

All three of those answers provides the same information on Tom's fate, which is none.

Right? If not, why not?

Right - assuming the warden always tells the truth and would under no circumstances ever say Tom.

How Tom is supposed to know that, I don't know... which is why I think the cards and/or gameshow version (where the host must follow the rules) is a better-posed version.
 
Last edited:
So why does the probability change from Tom's point of view upon hearing the answer? He has received precisely no information as to whether or not he is part of the pardoned group.

EDIT: I guess this means I agree with the opening post, but I can't for the life of me understand what the ones who disagree are saying.
 
Last edited:
So why does the probability change from Tom's point of view upon hearing the answer?

It doesn't!

EDIT: I guess this means I agree with the opening post, but I can't for the life of me understand what the ones who disagree are saying.

It's evidently very easy to confuse oneself about this.

You might ask yourself a different question. Suppose Tom wants to predict which of the three is going to die, and (being morbid) he predicts he himself is the one. Then he gets nervous and asks the warden his question, and let's say the answer is Harry. He's now offered a chance to change his prediction. Should he stick with his original prediction (Tom), or should he switch it to Dick?
 
Last edited:
So why does the probability change from Tom's point of view upon hearing the answer? He has received precisely no information as to whether or not he is part of the pardoned group.

EDIT: I guess this means I agree with the opening post, but I can't for the life of me understand what the ones who disagree are saying.

That's precisely the point. Tom's probability started out at 1/3, and the warder's information has not changed that - therefore it is still 1/3. Before the warder gave his answer, Tom knew that he was going to say either Dick or Harry, and it makes no difference which, as far as Tom's probability is concerned. But it certainly does make a difference to Harry's probability (which has now become 2/3). And obviously to Dick's, too, which has now become 0.

I guess this is the point that some people find confusing. An answer that only mentioned Dick gave information about Dick (obviously), and also about Harry but none about Tom. Superficially this may seem wrong, because the reason for the asymmetry between Tom and Harry is not obvious (it is introduced because it was Tom that asked the question, thereby eliminating himself from the possible answers).
 
Here in Singapore, I have presented the Monty Hall puzzle to dozens of friends over the last year or so. No one got it right the first time but with most, it wasn't really difficult to make them see the light, so to speak. With some, however, and these tend to be the smart ones, it was really tough getting them to see where they went wrong. I think this is because smart people tend to be a bit too smug about themselves. They just don't believe they can be wrong, even if it's just occasionally, and especially not with a simple-looking puzzle like picking the right door from 3.

As an example, I bumped into a friend just last week. He's a member of Mensa and is known to have an IQ of over 150. A couple of years ago, I've seen some of his Mensa magazines and the puzzles inside there are really tough. They are always 4 with each issue. My friend has consistently been able to solve all of them. I know that for a fact because with every issue, they published the names of the members who were able to solve the previous issue's puzzles. My friend's name has appeared in every single one of them.

So I presented the Monty Hall puzzle to him. Guess what? He got it wrong. Which didn't surprised me actually. What really shocked me though, was that despite explaining the puzzle to him for almost an hour, he still couldn't see where he went wrong. He even said that although mathematically it would seem that switching doors gives the contestant a better chance of winning the car, in reality, the chances are the same.

So if it took a while for you to understand where you went wrong, don't feel too bad. Even very smart people can be quite stupid sometimes :D
 
I actually did, but with pieces of paper with "car" and "goat" written on them.
Good for you :) My question was to .13., but it seems they've seen their mistake now anyway -- good for them too :)
 
Here in Singapore, I have presented the Monty Hall puzzle to dozens of friends over the last year or so. No one got it right the first time but with most, it wasn't really difficult to make them see the light, so to speak. With some, however, and these tend to be the smart ones, it was really tough getting them to see where they went wrong. I think this is because smart people tend to be a bit too smug about themselves. They just don't believe they can be wrong, even if it's just occasionally, and especially not with a simple-looking puzzle like picking the right door from 3.

As an example, I bumped into a friend just last week. He's a member of Mensa and is known to have an IQ of over 150. A couple of years ago, I've seen some of his Mensa magazines and the puzzles inside there are really tough. They are always 4 with each issue. My friend has consistently been able to solve all of them. I know that for a fact because with every issue, they published the names of the members who were able to solve the previous issue's puzzles. My friend's name has appeared in every single one of them.

So I presented the Monty Hall puzzle to him. Guess what? He got it wrong. Which didn't surprised me actually. What really shocked me though, was that despite explaining the puzzle to him for almost an hour, he still couldn't see where he went wrong. He even said that although mathematically it would seem that switching doors gives the contestant a better chance of winning the car, in reality, the chances are the same.

Yes, it does seem strange, doesn't it?

Here's an even more bizarre example, concerning the mathematician Paul Erdős (of Erdős Number fame). It is told by his lifelong friend Andrew Vázsonyi in The Man Who Loved Only Numbers:

'I told Erdős that the answer was to switch, and fully expected to move to the next subject. But Erdős, to my surprise, said, "No, that is impossible. It should make no difference." At this point I was sorry I brought up the problem, because it was my experience that people get excited and emotional about the answer, and I end up with an unpleasant situation. But there was no way to bow out, so I showed him the decision tree solution I used in my undergraduate Quantitative Techniques of Management course.'

Vázsonyi wrote out a 'decision tree' not unlike the table of possible outcomes that vos Savant had written out, but this did not convince him.

'It was hopeless. I told this to Erdős and walked away. An hour later he came back to me really irritated. "You are not telling me why to switch," he said. "What is the matter with you?"'

I can't explain why some mathematicians and other clever people find the puzzle difficult. I don't think it's because "smart people tend to be a bit too smug about themselves", though. It's plain that there's a serious disconnect in some (otherwise) highly intelligent people between the talent for carrying out complex mathematical manipulations and the ability to relate these mathematical manipulations to the real (or even the logical) universe. It's a brain-wiring problem.
 
I can't explain why some mathematicians and other clever people find the puzzle difficult. I don't think it's because "smart people tend to be a bit too smug about themselves", though. It's plain that there's a serious disconnect in some (otherwise) highly intelligent people between the talent for carrying out complex mathematical manipulations and the ability to relate these mathematical manipulations to the real (or even the logical) universe. It's a brain-wiring problem.

It took a very long time for me to finally get it. My problem was (and still is) that I don't understand enough about the math of probability for the equations to be at all convincing. What the program I wrote finally revealed to me is that Monty opening a door without the car, the warden telling Tom that Dick will be spared, and someone deliberately choosing a non-ace from the deck is exactly the same as saying "yes, there is at least one goat/free man/non-ace left in the two you didn't choose" which the contestant/Tom/hypothetical somebody already knew.

If the contestant gets to choose which door Monty will open, if Tom gets to choose whose fate the warden will tell him, if .13. picks a card at random, it's a different problem.

The answer to the problem is counter-intuitive precisely because it is very similar to the case where the non-winner chosen randomly and revealed results in the odds going to 50-50. In fact, once I had convinced myself that the Monty Hall Problem is correct, I was also convinced that it didn't matter if Monty knew where the car was. In other words, revealing a door without the car didn't affect the odds that the contestant had chosen the car in any case. Another simulation (one I actually had to run this time) showed me where my mistake was. The fact that he could have chosen the car and didn't adds information that can be used to adjust the probability. Remove that possibility and there is no longer any basis on which to revise the probability.

Lately I've been getting bit in the ass by Monty Hall because I wrongly assume other probability problems are identical. "Deal or No Deal" was the last.
 
It's a brain-wiring problem.
Here in Singapore, we call it "don't know how to keng". Keng is a Hokkien word for "thinking street smart". For instance, in the old days, we used to have illegal gambling around. Gangsters put up make-shift tables where they invite passers-by to make a bet.

The game is actually quite simple. I think it's called 369 but I'm not sure. It involves a board with 6 numbers on it, 1 to 6. They have 3 die. The die are placed on a saucer. They cover it with a bowl and give it a good shake. Punters put their money down on the numbers of their choice. It's even money for each number. In other words, if you bet $1 on number 1 and there's a 1 among the 3 die when the die are revealed, they pay you a $1. If there are 2 number 1's among the 3 die, they pay you $2. There does not seem to be a house edge. They don't cheat. The die are not loaded but consistently, they win.

So, where's the edge? I've put this question to many highly-educated folks but often, the answer eludes them. For some strange reason though, the street smart gangsters who have hardly any education can figure it out.
 
Last edited:
They actually pay out 1:3, don't they? For my 1$ I am playing three games, so that's 1/3$ per game. I have a 1:6 chance of winning a game. So, for betting 1$ I'll get 4$ back once (including my own dollar) and I'll lose another 5$. The house comes out ahead with 2$.

How is that not an edge?

Or: What am I missing here?
 
They actually pay out 1:3, don't they? For my 1$ I am playing three games, so that's 1/3$ per game. I have a 1:6 chance of winning a game. So, for betting 1$ I'll get 4$ back once (including my own dollar) and I'll lose another 5$. The house comes out ahead with 2$.

How is that not an edge?

Or: What am I missing here?

I hope Rasmus wouldn't mind me quoting him here but I think I could have misled the readers here with my question. Please forgive me as English isn't my first language. I'll try again to explain how the game works.

There are 6 squares on the board. Within each square, there's a number. They are 1 to 6. They use 3 die. They put the 3 die on a saucer and cover it with a bowl. They give the bowl a good shake. Punters put down their bets. You can bet on a single number or several of them. But they don't accept bets that are between numbers, unlike the real casinos where you can place your chip to cover more than one number.

You bet on say, the number 1. They lift up the bowl and reveal the 3 die. Let's say the 3 die show a 1, a 4 and a 5. You win. If you bet a dollar, they pay you a dollar. But let's say the die show a 1, another 1 and a 4. They pay you 2 dollars because you hit the number 1 two times. Obviously, if there isn't a 1 among the 3 die, you lose.

Hope this is clear now.

Loh K L
 
Last edited:
I hope Rasmus wouldn't mind me quoting him here but I think I could have misled the readers here with my question. Please forgive me as English isn't my first language. I'll try again to explain how the game works.

There are 6 squares on the board. Within each square, there's a number. They are 1 to 6. They use 3 die. They put the 3 die on a saucer and cover it with a bowl. They give the bowl a good shake. Punters put down their bets. You can bet on a single number or several of them. But they don't accept bets that are between numbers, unlike the real casinos where you can place your chip to cover more than one number.

You bet on say, the number 1. They lift up the bowl and reveal the 3 die. Let's say the 3 die show a 1, a 4 and a 5. You win. If you bet a dollar, they pay you a dollar. But let's say the die show a 1, another 1 and a 4. They pay you 2 dollars because you hit the number 1 two times. Obviously, if there isn't a 1 among the 3 die, you lose.

Hope this is clear now.

Loh K L

Do I get 3 $ if my number shows three times?


I was assuming that. If my assumption was wrong then things would look different yet again.

If it was accurate, I stand by my original answer.
 
If you put $1 on the number 1 and all 3 die show this number, they will pay you $3. Because you hit the number 1 three times.
 
Last edited:
If you put $1 on the number 1 and all 3 die show this number, they will pay you $3. Because you hit the number 1 three times.

Then I will try and rephrase my answer:


If I bet 1$, I am going to play three games for 1/3$ each.

Let's assume the three die have the colour red, green and blue for ease of reference.

I bet 33 cents on the red dice. If I win, I will be paid out my 33 cents, plus a whole dollar in winnings.

I bet 33 cents on the green dice, also. If I win, I will be paid out my 33 cents, plus a whole dollar in winnings.

And I will bet 33 cents on the blue dice as well. Again, if I win, I will be paid out my 33 cents, plus a whole dollar in winnings.

So, the payout is 1:4 (or 1:3, I am never sure if you include the initial bet ...) whilst the chances of winning are only 1:6.

So the payout is lower than the chances of winnings - hence there's your house edge. And it's a lot bigger than the edge the house gets with roulette.
 
I'm not sure I understand your answer, Ramus but I think you have got it all wrong. I really don't know how I can make it clearer. Perhaps somebody who understands can help me out?
 

ISF - Join now!

Every member here is approved by hand. No bots, no spam, just people who care about evidence and honest debate.

Membership is free!

Create your free account

Back
Top Bottom