News for hackers. Stuff that matters.
Re:Monty Hall, Erdos, and Our Limited Minds by simonh on Wednesday November 26, 2014 @03:48PM (#8664498)
The Monty Hall problem is a real classic, I remember puzzling over it for hours and discussing it with family and friends for days after hearing of it.

It amazes me how many people believe that human beings are primarily logical beings.

The best way I've come up to explain the Monty Hall problem is that in the first place you have a 2/3 chance of choosing a Goat and in that case switching guarantees you will win the car. Alternatively you have a 1/3 chance you chose the car to start with and switching guarantees you will win a goat. Therefore a strategy of always switching gives you a 2/3 chance of winning the car.

Re:Monty Hall, Erdos, and Our Limited Minds by hamburglar on Wednesday November 26, 2014 @04:14PM (#8664618)
An interesting thing to realize about the problem is that there actually is a way to make your odds the "obvious" 50/50, and that's to flip a coin to decide whether to switch doors at the end. The probability of a win on the two doors isn't equal, but the probability on the first door is 1/3 and the probability on the door you can switch to is 2/3, so using a coin toss to decide whether to switch would result in a 50% win rate.

This is because the way the problem is structured gives you six choices, not three: [A, A], [A, switch], [B, B], [B, switch], [C, C], [C, switch]. 3 of these are wins and three of these are losses. If you choose from these six options randomly you will win 50% of the time, but if you only choose from the "switch" options you will win 2/3 of the time.

Re:Monty Hall, Erdos, and Our Limited Minds by oneeyedpigeon on Wednesday November 26, 2014 @04:02PM (#8664564)
Wow - I think that's the most straightforward explanation I've ever heard. I think, to rephrase it, you're saying:

a) because a goat is revealed, swapping will always give you the opposite of what you had

b) you're more likely to pick a goat in the first place, so you're more likely to switch to a car

Re:Monty Hall, Erdos, and Our Limited Minds by hamburglar on Wednesday November 26, 2014 @04:46PM (#8664796)
The notion of revealing the goat before you make your second choice is a bit of a distraction. It becomes a lot more clear if you realize that in your first choice, you're dividing the doors into two groups: the group you choose has a 1/N chance of having a car in it, and the group you didn't choose has a (N-1)/N chance of having a car in it. Now, imagine that without opening any doors, Monty said: if you're willing to switch to the N-1 group, you win if there's a car in any of them. Do you take it? Of course.
Re:Monty Hall, Erdos, and Our Limited Minds by lukasb on Wednesday November 26, 2014 @05:17PM (#8664915)
Yeah this is it I think. If I can restate more simply for the 3-door case:

You pick. The doors are now in two groups: your door (1/3 chance) and the other two doors (2/3 chance).

The host picks. The second group is now reduced to a single door, still with 2/3 chance.

Now you decide - stick with the door with 1/3 chance, or switch to the door with 2/3 chance?

Re:Monty Hall, Erdos, and Our Limited Minds by jonsen on Wednesday November 26, 2014 @04:31PM (#8664713)
The most straightforward explanation I've heard is this:

The chance is 1/3 the car is behind the door chosen at first. The chance is 0 for the opened door. As the three chances must sum to 1, the last door has a chance of 2/3.

Re:Monty Hall, Erdos, and Our Limited Minds by dllthomas on Wednesday November 26, 2014 @05:17PM (#8664916)
That's great for defusing the "there are two options and it's random so it must be 50/50" initial of some, but it doesn't quite work for the "Monty chooses randomly, and happens to have revealed a goat" situation. I think the next point is that, when the host picks randomly, it is then correspondingly less likely that you've found yourself in that situation.
Re:Monty Hall, Erdos, and Our Limited Minds by jonsen on Wednesday November 26, 2014 @06:18PM (#8665140)
If the host picks randomly, the chances must sum to 2/3, not 1.
Re:Monty Hall, Erdos, and Our Limited Minds by dllthomas on Wednesday November 26, 2014 @06:43PM (#8665221)
Right.
Re:Monty Hall, Erdos, and Our Limited Minds by kybernetikos on Thursday November 27, 2014 @03:16AM (#8666424)
The best way of explaining it that I've found is to describe it as a gameshow where there are 100 doors, and you choose one door, then the host (who knows what's behind each of the doors) opens 98 doors with goats behind them. Do you stay with the door you started with, or switch to the remaining one?
Re:Monty Hall, Erdos, and Our Limited Minds by simonh on Thursday November 27, 2014 @10:01AM (#8667492)
That sort of hints at the answer, but it's not watertight. Wouldn't a more close analogy be if the host opened one of the other 99 doors? In which case it's still not obvious. After all he only opened one in the orriginal version of the puzzle. Not everyone will necessarily agree that your new version is similar enough to be directly applicable to the original.
Re:Monty Hall, Erdos, and Our Limited Minds by kybernetikos on Thursday November 27, 2014 @05:48PM (#8669005)
The analogy is not chosen merely for being similar, but to fulfill a purpose. It's based on the idea that in the original puzzle what you are faced with is a choice between your original door, or all the other doors. That's why you have a 1/3 or 2/3 chance of being right if you stick or switch respectively - because you are either choosing one door, or effectively two doors. The host showing you whats behind a door is just an implementation detail that causes you to have a stick or switch decision that has that effect. That's why a nice analogy is one that gives you a choice between 1 door or 99 doors, i.e. a chance of 1/100 or 99/100 of winning.

Your variant is a valid analogy too, but the probabilities in your scenario between sticking and switching are so close to each other, it's not going to trigger any new intuition in someone who is already confused.

Don't think of the 100 door puzzle as an analogy with the 3 door puzzle, think of it as a class of puzzle where there are N doors, you can pick one, and then the host (who knows what's behind the doors) reveals N-2 fails. It's clear that the 3 door puzzle is a specific instance of that, and if you have understood something intuitively about that puzzle class through considering N being a large number, that will still hold for N being a small number.

The class of puzzles that you describe is M doors, where you pick one and then the host (who knows what's behind the doors) reveals one fail. The three door puzzle is also a specific instance of that puzzle, but I doubt that you've learned anything from that, since examining M at other numbers probably hasn't given you any insight (still, if it did, those insights are indeed applicable to 3 door Monty Hall).

I've tried a number of ways of explaining this problem to people, and empirically the 100 doors one is the way that seems to work best. Interested to hear any versions you tell that seem to help people understand whats going on.

Re:Monty Hall, Erdos, and Our Limited Minds by plikan13 on Thursday November 27, 2014 @06:18PM (#8669083)
I've been wondering about another somewhat intuitive way to see the problem: suppose a third person enters the room and he is allowed to finish the game, knowing nothing of what went on before. In particular, he doesn't know your initial choice. Now it is obvious that his chances of winning are 1/2. It is also obvious that if you stick to your first choice, your chance of winning is still 1/3. Now it is odd that somebody who has LESS information about the problem has a better chance of winning than you. Meaning that you must have some way of getting at least the odds of the other guy. But there's only one thing you can do other than sticking to your first choice: taking the other door. So that seems the correct thing to do.