Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

Here's been a bunch of questions on the Monty Hall problem, so I'll assume people know the basics.

This answer helped clarify a few things for me, but talking with some colleagues yesterday, someone brought up the idea that as you increase the number of doors, the probability of winning the car by switching approaches 1. Intuitively, this makes sense, but I also know that infinite sets aren't always intuitive.

Is it really the case that if the number of doors increases to infinity, the probability of winning approaches 1? If not, why not?

EDIT: To clarify, the infinite case would look as follows:

  1. There are an infinite number of doors. One has a car behind it, the others have goats.
  2. The host knows which door has the car behind it.
  3. The contestant selects a door
  4. All doors except the selected door and an additional one are opened revealing goats.
  5. The contestant is asked if they would like to switch doors.

The question is: in this case, does P(winning by switching doors) = 1? Or is this even well defined? What changes when the number of doors is large but finite?

share|cite|improve this question
Intuitively, it seems that $p(winning) \rightarrow \infty$ with the number of doors. – Newb Oct 30 '13 at 16:52
Will the host open all but one door? The main problem is that with infinitely many doors there is no such thing as s uniform distribution – Hagen von Eitzen Oct 30 '13 at 16:52
@RichardRast because the idea here is that you pick a door, which, with a large number of doors, is unlikely to be the door containing the car. Then the host opens all doors except for two, one of which contains the car... – Newb Oct 30 '13 at 16:55
Ah, so they open almost all the doors. I figured the host would only open one door. Thanks for the edit, it's much more clear now. – Richard Rast Oct 30 '13 at 16:57

I've always found the Monty Hall problem interesting in that it's used to illustrate conditional probability, but there is a much simpler approach that I've always used when discussing this problem and its variants:

In essence, the Monty Hall problem boils down to two events:

  1. The car is behind the door the contestant selected (S)
  2. The car is behind the door Monty Hall has left closed (M)

Note that S and M are exhaustive, disjoint events, so $P(S) + P(M) = 1$. If the contestant selects from among N doors at random, the probability of selecting the door with the car is $P(S_N)=\frac{1}{N}$. From the above observation that the events S and M are disjoint-exhaustive, we can conclude $P(M_N)=1-P(S_N) = \frac{N-1}{N}$.

Taking the limit as $N\rightarrow \infty$ we get $P(S_{\infty})=\lim\limits_{N\rightarrow \infty} \frac{1}{N} = 0$ hence, as the number of doors increases, the success probability of switching does indeed approach 1.

share|cite|improve this answer
I agree with your critique, but the question was about reaching the limit, I.e. the actual infinite case – Ryan Reich Oct 30 '13 at 17:59
@RyanReich: If you actually had an infinite number of doors, there are philosophical/conceptual issues, as some commenters pointed out. Putting a uniform distribution on an infinite number of doors gives you the degenerate/defective f(x)=0 distribution. You would need two priors on this problem for the actually infinite case: A distribution on what door you will select and a distribution of likely locations of the car. – user76844 Oct 30 '13 at 18:32
@RyanReich (cont'd): BTW, the OPs third line talks about limits. The "clarification" is what is drawing the criticism. – user76844 Oct 30 '13 at 18:34

You can only validly state this problem in a decent probabilistic way if you either consider only finite versions and take the limit for the probability and the finite number of doors tends to infinity, or when considering one problem with infinitely many doors you specify a probability distribution that gives the a priori probability for each door that the car is behind it. You cannot say that infinitely many doors all have the same probability, because (given there is no more than one car) that probability would have to be$~0$, and it would imply that the probability that there was any car at all would be$~0$ as well, which it is not supposed to be. There is no such thing as a uniform probability on a countably infinite set.

Given the problem statement, the chances you win by switching is exactly the probability that you did not choose the door with the car in the first place. So in the limit of finite problems, the success probability for switching, $1-\frac1n$ for $n$ doors, indeed tends to$~1$.

In an infinite version with a given probability distribution, you can make your chances of success as close to$~1$ as you like by choosing a very improbable door in the first place (one situated way beyond the furthest known galaxy might be fine, but it really depends on the given probability distribution), but quite possibly you cannot make it $1$ exactly (namely if there are no doors that have been assigned probability$~0$).

I cannot help feeling sorry for those infinitely many undesired goats.

share|cite|improve this answer

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.