Pick the Door with the Prize!

What is the basis for reason? And mathematics?

Moderators: AMod, iMod

Mike Strand
Posts: 406
Joined: Wed Jan 06, 2010 6:54 am
Location: USA

Pick the Door with the Prize!

Post by Mike Strand »

One of my favorite applications of probability theory is the Monty Hall Prize Behind the Door game.

In the most basic version of the game, there are three doors, behind one of which is a prize. You, the player, don’t know the prize door, but your host, Monty, does. He asks you to pick one of the three doors as the prize door. At least one of the two remaining doors has no prize behind it. Monty opens one of these doors with no prize behind it. This leaves two unopened doors: The one you picked, and the one Monty did not open. Then Monty asks you if you want to switch your choice to the unopened door you did not pick in the first place.

Finally, Monty opens the door of your choice, and if the prize is behind it, you win the prize.

Should you make this switch? Justify your answer: For example, does your chance of winning the prize improve or not by making the switch, and why?

If there is enough interest, I’ll cover some generalizations of this game to more than three doors, with Monty opening varying numbers of non-prize doors and you deciding whether to change your initial choice.
surreptitious57
Posts: 4257
Joined: Fri Oct 25, 2013 6:09 am

Re: Pick the Door with the Prize!

Post by surreptitious57 »

I keep to my original choice simply because I have no reason to change my mind
Of course it might actually be the wrong door but I have no way of knowing it is
User avatar
Lacewing
Posts: 6604
Joined: Wed Jul 29, 2015 2:25 am

Re: Pick the Door with the Prize!

Post by Lacewing »

With 3 doors, it's hard to call. Although your odds are only 1 in 3 of picking the right door in the first place, your intuition may guide you. :) The more doors, the greater the probability that switching to Monty's remaining switch-door would be the right way to go, since odds are against you for picking the ONE correct door in the first place, while Monty can eliminate any number of doors to bring you to a choice between 2 doors, giving the illusion that you have a 50/50 chance. Right?
User avatar
HexHammer
Posts: 3354
Joined: Sat May 14, 2011 8:19 pm
Location: Denmark

Re: Pick the Door with the Prize!

Post by HexHammer »

I have yet to see actual proof, instead of the usual theoretical proof.
User avatar
vegetariantaxidermy
Posts: 13983
Joined: Thu Aug 09, 2012 6:45 am
Location: Narniabiznus

Re: Pick the Door with the Prize!

Post by vegetariantaxidermy »

HexHammer wrote: Tue Jul 03, 2018 5:54 am I have yet to see actual proof, instead of the usual theoretical proof.
Absolutely right. As no one ever seems to change their choice then there is no way of proving whether the theory translates into the practical. Logic would tell you that it makes no difference to change, since the prize is still behind the same door that it was in the first place.
It was done on 'Mythbusters' but one artificially changed his 'choice' over and over. That's not the same.
Mike Strand
Posts: 406
Joined: Wed Jan 06, 2010 6:54 am
Location: USA

Re: Pick the Door with the Prize!

Post by Mike Strand »

Thanks to surreptitious57, Lacewing, HexHammer, and vegetariantaxidermy, for your interest! I'll try to make your gracious attention to my post both interesting and rewarding.

Practical applications of probability -- games of chance such as card games and other casino games of chance, for example, are amenable to exact calculation of odds, so the gambling house will know its long-run profit. Luckily for the non-mathematical types, such applications are also amenable to simulation, to observe the long-run odds.

In this case (the three doors), if you don't want (or can't stand) to go through the math calculation, take three cards from a deck, one red suit and two black. The red suit represents the prize door. Have a friend keep track of which card is red. Mix the cards face down, spread them out, and point to one you guess is red. Have your friend turn up a black card. Two cards remain face down.

1. To simulate not switching: Stay with your first choice and flip the cards up. After, say, 50 trials of this, you'll find that you guessed red only about a third of the time. Not so surprising.
2. To simulate switching: Run another 50 trials, but this time, for each trial, switch to the other card that is still face down. What do you see happening?

I hope these simulations may be more enjoyable, at least for awhile, than, say, playing solitaire. If your friend isn't available, you can do the simulations by yourself and pretend you don't know which card is red.
surreptitious57 wrote: Tue Jul 03, 2018 4:00 am I keep to my original choice simply because I have no reason to change my mind
Of course it might actually be the wrong door but I have no way of knowing it is
surreptitious57, you have one chance in three of guessing the prize door (two chances in three of having the wrong door) at the outset. You're right, no way of knowing it, but please think in terms of odds, or chances. It's still one in three to win, if you don't switch. You'll find out that you do have a reason to switch! Not that you'll know for sure, in any case, until the game is over, but you'll actually improve your chances by switching.
Lacewing wrote: Tue Jul 03, 2018 4:11 am The more doors, the greater the probability that switching to Monty's remaining switch-door would be the right way to go, since odds are against you for picking the ONE correct door in the first place, while Monty can eliminate any number of doors to bring you to a choice between 2 doors, giving the illusion that you have a 50/50 chance. Right?
Good thought, Lacewing, increasing the number of doors -- but better (or worse) than 50/50, if you switch (don't switch). Say there are ten doors. If Monty opened eight non-prize doors, leaving two doors -- your first pick and one other door -- maybe you can see that switching is a good idea. There is a 9 in 10 chance that your first pick is wrong, because you only have a 1 in 10 chance of guessing correctly the first time. This means there is a 9 in 10 chance that if you switch, you'll get the prize! Switching is good even with the minimum three doors.
HexHammer wrote: Tue Jul 03, 2018 5:54 am I have yet to see actual proof, instead of the usual theoretical proof.
HexHammer, I hope you try the simulation described above. But my reply to Lacewing should give you a good clue to the proof in the three-door case.
vegetariantaxidermy wrote: Tue Jul 03, 2018 7:08 am Absolutely right. As no one ever seems to change their choice then there is no way of proving whether the theory translates into the practical. Logic would tell you that it makes no difference to change, since the prize is still behind the same door that it was in the first place.
vegetarian ..., there is a proof that changing (switching your choice) is the way to go, even with just three doors. Check out the ten-door situation above in my reply to HexHammer, which dramatizes the virtues of switching. The mathematical theory of probability provides such proof, and casinos depend on such proofs and the corresponding calculations, for their ill-gotten gains.

Now, I hope we'll all be able to calculate, or observe in a simulation, the odds of winning by switching in the three-door game!

Future fun(?) -- Ten doors, Monty opens less than 8 non-prize doors. Do you want to switch your choice to one of the other doors remaining unopened? How? You can use a random device. For example, if there are 7 doors left, yours and six others, throw a die to select one of the other six doors. Does this improve your odds of winning?
User avatar
vegetariantaxidermy
Posts: 13983
Joined: Thu Aug 09, 2012 6:45 am
Location: Narniabiznus

Re: Pick the Door with the Prize!

Post by vegetariantaxidermy »

Strange. My posts keep disappearing. I mentioned on my deleted post the Mythbusters experiment. They found the maths to be correct, and the one who changed his choice won the predicted number of times more than the one who didn't, but as I pointed out, in the real situation no one changes (or extremely rarely. That experiment has been done too, and they couldn't get a useful result because of it).
Mike Strand
Posts: 406
Joined: Wed Jan 06, 2010 6:54 am
Location: USA

Re: Pick the Door with the Prize!

Post by Mike Strand »

vegetariantaxidermy wrote: Tue Jul 03, 2018 9:33 am Strange. My posts keep disappearing. I mentioned on my deleted post the Mythbusters experiment. They found the maths to be correct, and the one who changed his choice won the predicted number of times more than the one who didn't, but as I pointed out, in the real situation almost no one changes (that experiment has been done too. They couldn't get a useful result because of it).
Thanks for hanging in there, vegetariantaxidermy! I hope you get a chance to see my reply to you and three other brave souls. You can do a simulation with three playing cards -- maybe better than solitaire or Go Fish.
User avatar
vegetariantaxidermy
Posts: 13983
Joined: Thu Aug 09, 2012 6:45 am
Location: Narniabiznus

Re: Pick the Door with the Prize!

Post by vegetariantaxidermy »

Mike Strand wrote: Tue Jul 03, 2018 9:37 am
vegetariantaxidermy wrote: Tue Jul 03, 2018 9:33 am Strange. My posts keep disappearing. I mentioned on my deleted post the Mythbusters experiment. They found the maths to be correct, and the one who changed his choice won the predicted number of times more than the one who didn't, but as I pointed out, in the real situation almost no one changes (that experiment has been done too. They couldn't get a useful result because of it).
Thanks for hanging in there, vegetariantaxidermy! I hope you get a chance to see my reply to you and three other brave souls. You can do a simulation with three playing cards -- maybe better than solitaire or Go Fish.
I'm sorry that you don't appear to understand what I mean. You could always ask.
Mike Strand
Posts: 406
Joined: Wed Jan 06, 2010 6:54 am
Location: USA

Re: Pick the Door with the Prize!

Post by Mike Strand »

vegetariantaxidermy wrote: Tue Jul 03, 2018 9:33 am I mentioned on my deleted post the Mythbusters experiment. They found the maths to be correct, and the one who changed his choice won the predicted number of times more than the one who didn't, but as I pointed out, in the real situation no one changes (or extremely rarely. That experiment has been done too, and they couldn't get a useful result because of it).
I apologize, vegetariantaxidermy, for not addressing your main point here, regarding human behavior, I think. In practice, the probabilities hold only for consistent behavior -- always switching or always staying with the original choice. If a person sometimes switches, sometimes not, the probability of winning falls somewhere between the two extremes. Only if the person's behavior is predictable (say, switch 30% of the time, not switch 70%) can the new long-run probability of winning be calculated.

We can be thankful, I guess, for the surprises that come from humans being "spur of the moment" decision-makers. Or can we?
Mike Strand
Posts: 406
Joined: Wed Jan 06, 2010 6:54 am
Location: USA

Re: Pick the Door with the Prize!

Post by Mike Strand »

Related problem: If a gambling casino lets customers play the three-door problem, what should they charge the customer to play each game? (The casino pays for the prize, but wants to make money). Keep in mind human whimsy -- sometimes switch, sometimes don't switch, and unpredictably.
User avatar
vegetariantaxidermy
Posts: 13983
Joined: Thu Aug 09, 2012 6:45 am
Location: Narniabiznus

Re: Pick the Door with the Prize!

Post by vegetariantaxidermy »

Mike Strand wrote: Tue Jul 03, 2018 12:57 pm
vegetariantaxidermy wrote: Tue Jul 03, 2018 9:33 am I mentioned on my deleted post the Mythbusters experiment. They found the maths to be correct, and the one who changed his choice won the predicted number of times more than the one who didn't, but as I pointed out, in the real situation no one changes (or extremely rarely. That experiment has been done too, and they couldn't get a useful result because of it).
I apologize, vegetariantaxidermy, for not addressing your main point here, regarding human behavior, I think. In practice, the probabilities hold only for consistent behavior -- always switching or always staying with the original choice. If a person sometimes switches, sometimes not, the probability of winning falls somewhere between the two extremes. Only if the person's behavior is predictable (say, switch 30% of the time, not switch 70%) can the new long-run probability of winning be calculated.

We can be thankful, I guess, for the surprises that come from humans being "spur of the moment" decision-makers. Or can we?
I see what they are getting at now. When you choose from the three doors, you have a 2 out of 3 chance of choosing a goat. Not particularly great odds. So after one of the goats has gone you now have 50/50 chance so mathematically you have a better chance of winning if you 'start again' and switch doors. It goes against the grain and our sense of what is logically the case though, hence the fact that pretty much everyone sticks with their original choice. It's a very interesting problem and lesson in what we consider to be reality despite the fact that it can't be proven in a real, practical sense.
I still get the sense that something is missing here though.
Mike Strand
Posts: 406
Joined: Wed Jan 06, 2010 6:54 am
Location: USA

Re: Pick the Door with the Prize!

Post by Mike Strand »

vegetariantaxidermy wrote: Tue Jul 03, 2018 1:15 pm When you choose from the three doors, you have a 2 out of 3 chance of choosing a goat. Not particularly great odds. So after one of the goats has gone you now have 50/50 chance so mathematically you have a better chance of winning if you 'start again' and switch doors. It goes against the grain and our sense of what is logically the case though, hence the fact that pretty much everyone sticks with their original choice. It's a very interesting problem and lesson in what we consider to be reality despite the fact that it can't be proven in a real, practical sense.
I still get the sense that something is missing here though.
Hi again, vegetariantaxidermy, and thanks for your continuing interest! You're on the right path, seeing that it looks better to switch doors. Try the simulation with cards, where you always switch. You'll see the odds of winning are better than 50/50. It's interesting, though, that if you flip a coin to decide whether to switch or not, your odds are indeed 50/50. The key is: Always switch!

Another good thing about doing the simulations: They help train your intuition to understand the actual odds. You start seeing that one of the two cards that are left - one your first choice, the other not - has got to be the red one (cover the prize). But your card only has a 1/3 chance of being the red one. Thus the other card has a chance of ..... ? You may be sorry you switched of course, but we're talking odds and long-run average outcomes here, not the outcome of any single trial of the game.
Last edited by Mike Strand on Tue Jul 03, 2018 4:52 pm, edited 2 times in total.
Mike Strand
Posts: 406
Joined: Wed Jan 06, 2010 6:54 am
Location: USA

Re: Pick the Door with the Prize!

Post by Mike Strand »

Back to the question of how much a gambling casino, which buys the prize, should charge a player of the three-door problem to play each game:

vegetariantaxidermy pointed out the inconsistency of human players (if I'm not mistaken), which makes the odds hard to predict.

The casino accounts for human whimsy, however, by assuming the worst case for the casino (best case for the player): This is that player will always choose to switch choice of door. Now you can figure out how much the casino should charge the player in order to make a profit. Obviously, the casino still wants to attract players, so can't charge a huge amount -- enough to cover the cost of the prize (on the average, since sometimes the player wins), and some amount of overhead. The casino is only interested in making money over many, many games.

By the way, is anyone ready and willing to state the odds (in the 3-door game) of winning, if the player always switches doors?
wtf
Posts: 1178
Joined: Tue Sep 08, 2015 11:36 pm

Re: Pick the Door with the Prize!

Post by wtf »

The problem becomes clear once we explicitly state the assumptions.

* Monty knows what's behind each door.

* Monty must reveal exactly one door.

* Monty must reveal a door concealing a goat.

Now the puzzle is clear.

The player makes a choice.

1/3 of the time the player chooses correctly and the other two doors conceal goats. Monty reveals one of the goats. Switching picks the other goat and the player loses.

The other 2/3 of the time, the player has initially chosen a goat. There are two doors left, one concealing the car and the other concealing a goat. Monty must reveal the door concealing the goat. The remaining door contains the car. Switching wins the car.

So 1/3 of the time switching loses; and 2/3 of the time switching wins.

Let me repeat this. 2/3 of the time the player's initial guess is a door with a goat. That leaves one door with a car and one with a goat. Monty is required by the rules of the game to reveal the door with the goat. That means that switching guarantees we get the car. This case happens 2 times out of 3.

1 time out of 3 the player's initial guess is the door with the car. Switching loses.

So switching is the winning strategy in 2 out of 3 equally likely cases.
Post Reply