3 prisoner puzzle

Could you quote the part of the OP that you believe I am not following?
If you plan to simply pick Tom as the prisoner to be executed every time, or even a majority of the time, then your opponent in this game gets to pick which person will be the one to ask the warden the question.
 
If you plan to simply pick Tom as the prisoner to be executed every time, or even a majority of the time, then your opponent in this game gets to pick which person will be the one to ask the warden the question.

If Dan the warden chooses Tom to be executed 90% of the time he will win 90% of the games. If he chooses Tom to be executed 50% of the time he will win 50% of the games. If he chooses Tom to be executed 33% of the time, he will win 33% of the games. In each case the fact of revealing the name of one prisoner who will be spared makes no difference to the final outcome, which is precisely the point of the original answer to the problem.
 
This is probably a lost cause, but here goes.

Dan O., you have mis-read the problem. You understood it to mean that the warden selects the unlucky prisoner in response to Tom's questions (in which case, who can say what the odds are? – nothing to say he's choosing randomly).

However, the scenario is very clearly that the selection has been made before Tom questions the warder - and therefore before the warder has any reason to favour (or disfavour) Tom.

"Please tell me if I'm the one who will be executed tomorrow?". The warden refused to tell him the answer.
Note: refused to tell him the answer. The answer already existed at that point.

It's not even ambiguous (leaving aside that it would be a completely pointless puzzle if it's not interpreted that way).


Warden: "Ah, I am prohibited from telling you wether you are the one to be executed. But one of the other two is going to be set free". (the warden flips a shinny new dollar coin in the air) "Tomorrow, Dick's life will be spared".
If the warden really did need to flip a coin then Tom's probability of being executed would be neither 1/3 nor 1/2; it would be a certainty.


I'm calling shenanigans.
Don't think so. I reckon he really does think we're all wrong.
 
If the warden really did need to flip a coin then Tom's probability of being executed would be neither 1/3 nor 1/2; it would be a certainty.

My bet is the coin flip is there to mislead. He doesn't state that the warden based his decision on it, or even that it was necessary for him to do it.

Don't think so. I reckon he really does think we're all wrong.

I think he's trying to be pedantic about the initial assumption that they are all equally likely to be executed and failed to read the problem carefully.
 
Warden: "Ah, I am prohibited from telling you wether you are the one to be executed. But one of the other two is going to be set free". (the warden flips a shinny new dollar coin in the air) "Tomorrow, Dick's life will be spared".

OK, so do I have to wait until tomorrow morning to find out my fate?

Incidentally, if you really want to do this we obviously need some kind of system that verifies you are choosing who's going to die randomly with 1/3 odds for each (before the question). For example we can pick a trusted third forum member who generates a list of numbers, each being 1, 2, or 3. Or we can find a pseudorandom number generator online and mutually agree on a seed.

Once we do that I don't care in the slightest whether the coin flip is really random (since it's totally irrelevant).
 
Last edited:
It seems? to be understood why the odds are 1/3 for repeated trials. Unfortunately, it remains unclear when the stated trial is broken down time wise.



Say we have the variables x, y, and z.

Two will be removed and one will remain. The odds of a particular one remaining are 1/3.

Let it be known to z that y will be removed.

Now z knows the remaining variables are x and z.

One will be removed and one will remain. The odds of a particular one remaining are 1/2.

Great, now I'm back to not even seeing why repeated trials would change this. Does answering this require having a time machine or, alternately, no concept of time?

Can’t the other two prisoners be variably labeled from the start as “the other that is set free” and “the other who’s fate is still unknown”. Wouldn’t the odds of dying between the main prisoner and “the other who’s fate is still unknown” be ½ whether or not he knows specifically?

Pointedly (I posted this previously): Does the question ask:

What are the odds at the start of the game of the main prisoner dying in repeated trials?

Or does it ask:

What are the odds of the main prisoner dying vs. “the other who’s fate is still unknown” in one specific trial?

It seems to be open to interpretation, with one answer being 1/3 and the other I still don't see being anything other than 1/2.
 
It seems? to be understood why the odds are 1/3 for repeated trials. Unfortunately, it remains unclear when the stated trial is broken down time wise.



Say we have the variables x, y, and z.

Two will be removed and one will remain. The odds of a particular one remaining are 1/3.

Let it be known to z that y will be removed.

Now z knows the remaining variables are x and z.

One will be removed and one will remain. The odds of a particular one remaining are 1/2.

Yes. If you make spontaneous decisions at every stage. But that is not the game that is being played here.

In thze puzzle, a decision of who will die and live is made up front. It is final.

Pointedly
(I posted this previously): Does the question ask:

What are the odds at the start of the game of the main prisoner dying in repeated trials?
No. Those odds are 1/3. Trivial, really.

Or does it ask:

What are the odds of the main prisoner dying vs. “the other who’s fate is still unknown” in one specific trial?
Yes. But their respective fates are known - from the beginning - to the warden.

It seems to be open to interpretation, with one answer being 1/3 and the other I still don't see being anything other than 1/2.
Run a couple of trials and tell us what your outcome is.

Pick one prisoner randomly who shall die. (Throw a dice with 1/2 for Tom, 3/4 for Harry and 5/6 for Dick.) The prisoner you selected will die.

Now tell Tom one of the other two prisoners that will live. (Pick randomly if you can chose.)

Count, how many times Tom will die.

It's really, really simple: The decision is made first. Nothing anyone says in-between changes the outcome. There is no way that Tom will not die in one third of the cases.

The entire, silly Q/A-game doesn't tell him anything new and it doesn't change anything.
 
OK, so do I have to wait until tomorrow morning to find out my fate?

Incidentally, if you really want to do this we obviously need some kind of system that verifies you are choosing who's going to die randomly with 1/3 odds for each (before the question). For example we can pick a trusted third forum member who generates a list of numbers, each being 1, 2, or 3. Or we can find a pseudorandom number generator online and mutually agree on a seed.

Once we do that I don't care in the slightest whether the coin flip is really random (since it's totally irrelevant).

I am following the rule set forth in the OP and what I said when I created the game. If you can show that I have materially not followed the OP or my own rules I will remit $1 to the forum coffers.

The OP says Tom will learn his fate tomorrow morning.
 
Pointedly (I posted this previously): Does the question ask:

What are the odds at the start of the game of the main prisoner dying in repeated trials?

Or does it ask:

What are the odds of the main prisoner dying vs. “the other who’s fate is still unknown” in one specific trial?

It seems to be open to interpretation, with one answer being 1/3 and the other I still don't see being anything other than 1/2.


OnlyTellsTruths:
It has nothing whatever to do with repeated trials vs single trial – that's a complete red herring. How on earth do you think the probabilities could change by repeating the exact same scenario multiple times? That's a very basic misunderstanding.

Plainly our explanations aren't working, but I do think the penny will drop sooner or later if we continue the discussion. Perhaps it would help if you go back to the single trial with Tom, Dick and Harry, and explain to us whether you think the 1/3 and 1/2 answers can both be justified and, if so, how.
 
How can the odds of dying between the main prisoner and "the other remaining prisoner who's fate is unknown" be anything other than 1/2.

You could have started with any number and it would increase ...... 1/n, 1/4, 1/3, 1/2, 1.
 
It seems pretty simple to me. The concept of probability is all to do with ignorance of causal relationships.

This ignorance is due to a lack of information. Sometimes it is possible to increase the amount of information. Sometimes it is not.

From a perspective of total ignorance the inferred chance of any particular person dying is 1/3.

From a perspective of knowledge of one person surviving his inferred chance of dying is zero and the remaining is 1/2.

From a perspective of knowledge of two people surviving their inferred chance of dying is 0 and the remaining unlucky fellow has a inferred chance of dying of 1.

That the actual causal relationship was that one person was always going to die and the other's won't doesn't matter: the question is what probability can you infer from the information you have?

When you know one person survives you can infer a probability of 1/2 for the other two.
 
If the scenario were finished and the main prisoner was chosen to die, do we all at least agree that at that point his odds of dying have increased to 1?
 
How can the odds of dying between the main prisoner and "the other remaining prisoner who's fate is unknown" be anything other than 1/2.

You could have started with any number and it would increase ...... 1/n, 1/4, 1/3, 1/2, 1.
Change the sceanrio so there are 10,000 prisoners. If one of the 10,000 talks to the warder the warder who says it is not 9,998. Does that make the prisoner's odds 1 in 2 ? what if it was 200 billion and the warder rules out everyone except the prisoner and one other are the odds still 1 in 2 ?

Can it be right that it doesn't matter how many prisoners there are the odds of the person asking are always 1 in 2 ?
 
I am following the rule set forth in the OP and what I said when I created the game. If you can show that I have materially not followed the OP or my own rules I will remit $1 to the forum coffers.

The OP says Tom will learn his fate tomorrow morning.

I suppose your point is that the OP didn't specify clearly that initially, each prisoner has an equal chance of being the one?

That's pretty lame.

As I said several times in this thread, the gameshow version of this puzzle is much better posed (to avoid silly possibilities like the one I guess you're exploiting). So, my $1 is at your mercy.
 
Last edited:
This is probably a lost cause, but here goes.

Dan O., you have mis-read the problem. You understood it to mean that the warden selects the unlucky prisoner in response to Tom's questions (in which case, who can say what the odds are? – nothing to say he's choosing randomly).

However, the scenario is very clearly that the selection has been made before Tom questions the warder - and therefore before the warder has any reason to favour (or disfavour) Tom.


Note: refused to tell him the answer. The answer already existed at that point.

You are mostly correct. But I have to run to work right now.
 
Last edited:
Change the sceanrio so there are 10,000 prisoners. If one of the 10,000 talks to the warder the warder who says it is not 9,998. Does that make the prisoner's odds 1 in 2?

Yes. If he's allowed to ask 9998 times (being told N-2 releases). Which would be the analogy. As in the .... 1/4, 1/3, 1/2, 1 part.
 
Last edited:
When you know one person survives you can infer a probability of 1/2 for the other two.

Not when we know that all three were equally likely to be executed and one had no chance of being named as going free.
 

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