Ah, but there is a subtle difference with the Monty Hall puzzle. There the door is open so the information is verified as true. Now, what if the guard lied about B having to be executed? Now we have two possible situations:
1) The guard said the truth (i.e. B is to be executed)
2) The guard lied (i.e. B is to live) = A is to die.
Assuming that they are equally possible we now have a completely different problem.
Originally posted by PawnCurryI already conceded a few posts above you, thanks to the link that you posted. π
To BLR - proof of the Monty Hall problem.
Assume there are three safes labelled A, B, C. There are only three possible outcomes each with P=1/3.
P(A) = P(B) = P(C) = 1/3 where P(A) is the probability of the prize being in safe A, etc
The three possible outcomes are shown below:-
A B C
1 O O X
2 O X O
3 X O ...[text shortened]... Phew! Do we get a prize for contributing to one of the longest threads on RHP?
π
Originally posted by ilywrinNow that's just silly.
Ah, but there is a subtle difference with the Monty Hall puzzle. There the door is open so the information is verified as [b]true. Now, what if the guard lied about B having to be executed? Now we have two possible situations:
1) The guard said the truth (i.e. B is to be executed)
2) The guard lied (i.e. B is to live) = A is to die.
Assuming that they are equally possible we now have a completely different problem.[/b]
Difficult for me to argue in English (I'm German) but let me put it
this way: with the MH-problem the question is not whether the price
is in A or B after the moderator has opend C. The question is whether
the candidate improves his chances by changing his first selection.
That's exactly the same situation as if the candidate had selected
door A and then the moderator asks "now, should I open that door
A for you or would you prefer me to open doors B AND C
You may empirically "prove" this problem by using a simple
EXCEL-sheet:
Let Col-A denote where the car is; let Col-B give the canditate' random
selection; so Cols A and B both should contain the formula
Randbetween(1;3)
Col-C is titeled "candidate does not switch" and we know that in this
case he will win if his first selection was correct; thus formula in Col-C
should read
if(A1=B1";"win";"lose"π
OK?
Now copy/paste this line 5000 times - and the let EXCel count the
numer of "win"s
Convinced???
But --- is the prisoner's problem REALLY the same as the MH-problem?
Not sure about this ...
Thankyou PawnCurry, very good explanation... the penny finally dropped when I realised that the guard specifying that one of the others will die does not exclude any of the three outcomes defined initially for C, since if B was going to die, the guard would have said B and if A would have died the guard would have said A. Hence the guards response to C's question would change depending on which of the three outcomes were chosen initially and hence would make no difference to C's probability of dying. However, if B is stated to die, from this point on A would know that the outcome where B lives is excluded from his future and hence his chances of living increases.
Sheesh, even after getting it, it still baffels the mind a little, since for A and B you can now exclude one of the outcomes for the future but for C you can't.
Originally posted by BLReidThe only condition that needs to be assumed is that all of the prisoners are equally likely to be executed.
As EdSchwartz stated above, the truth of the guard's statement is implied in the spirit of the puzzle. But you already know that don't you?π
Now, why would the problem imply anything like what you just stated?
Let's see the possibilities (Question to be answered: "Have the chances of survival for A improved"π:
1) Guard's statement is true. Then you may apply the logic of Monty Hall. Negative answer to the question asked.
2) Guard's statement is a lie. Then A is done for. Again negative answer to the question asked.
3) Guard does not know/remember. Again negative answer to the question asked.
So to make this short, there is no need to think of the guard's words as implicitly stated as truth, for in all cases the chances of A for survival do not improve.
In any case my point was the following, there's no need to assume unnecessary things when solving a problem. On the contrary, one should assume as least as possible, or otherwise one risks to narrow his search to a particular case of the original problem. Feel free to agree or disagree with me, we live in a free world (most of us anyway) π
I understand your point, but I think that a fair amount of assumption must be made as far as the guard's veracity. Without that, we would all be spinning our wheels here (not that we aren't just wasting time anyway π).
*edit: Important to note that I am the guy who was wrong for a fair portion of this thread, so what do I know anyway?!π
I think with a lot of these puzzles you need to make a certain amont of assumptions, such as being told true information etc, otherwise you could reduce things down to absurd proportions.
You could introduce plainly ridiculous possibilities such as being broken out of prison - how would you factor that in? What if one of the prisoner's knows he is going to die and hangs himself - does that mean the other 2 will be executed? I know the examples I've given are extreme (and a tad ridiculous!), but in order for their to be a problem worth taxing our brain-cells, you do need to make a few basic assumptions that aren't explicitly stated in the problem.
In this case, I personally would make the assumption that I was being told the truth. Monty Hall, is a no brainer, as he PROVES which safe is empty by opening it.
I think what all of this shows, is that whoever coined the phrase "lies, damned lies and statistics" was spot on! A lot of these probability questions require us to ditch our immediate presumption as to the obvious, and go right back to basics.
I remember being shown the equivalent of the "3 Prisoners" problem while at school (ooh, 20 years ago?), and struggled to come to terms with it then - when my brain functioned a lot more cleraly than it does now!
Even now, KNOWING the answer, AND being able to prove it (at least myself), the "natural" answer would seem to be 50/50. I still have to perform some mental gymnastics and the odd leap of faith to get the real answer.
Anyone got any more??!!
π
Originally posted by EdSchwartzor even better take two bombs π, the probanility of three bombs on one flight should be even smaller.
How to minimize the probability that there's a bomb on your flight?
Take a bomb with you yourself, because the probability for TWO
bombs on one flight is very small
π