Monty Hall problem (again)

Status
Not open for further replies.
game show problem
Forget it. You're lost in your own little world. This is a discussion forum, not a blog.

You put forth a thesis; it's been shown - in so many ways, by so many people - where it is flawed, and you have sunk so low as to contradict your own account of the rules. And, it has become quite plain that you have no idea how probabilities work anyway.

Time to lock this up.
 
Forget it. You're lost in your own little world. This is a discussion forum, not a blog.

You put forth a thesis; it's been shown - in so many ways, by so many people - where it is flawed, and you have sunk so low as to contradict your own account of the rules. And, it has become quite plain that you have no idea how probabilities work anyway.

Time to lock this up.
Agreed, 161 replies but feels a hell of a lot longer than that.
 
Please do not troll.
No point in posting any more evidence why Selvin and Savant were wrong (illegal). Responders here don't accept facts.
Enjoy your feel good moments.
 
win c = 1/2 for both.
Nope. Goes from 1/3 to 2/3.

The mistake you continually make is that you assume that once the door is opened, your odds instantly change from 1/3 to 1/2. They do not. They are still 1/3. Your original odds are still there.
 
We can look at this as cards on a table too.

1. 52 cards are arranged in a line on the table.
2. Player wins if they can turn up the Ace of Hearts.
3. Player points to card #27.
4. There is a 1-in-52 chance that the player's chosen card is the Ace of Hearts.
5. Monty turns 50 cards face up - all except card #27 and (for some mysterious reason) card #13.
6. Monty has eliminated 50 possibilities. There are now only two cards on the table. One is the card the player has chosen (card #27) and the other (card #13) is unknown (to the player).
7. As before, there is still a 1-in-52 chance that the player - having chosen card #27 - had correctly chosen the Ace of Hearts.
8. But, of 50 original possibilities, there are now only two.
9. The chances that card #13 is the Ace of Hearts is 1 in 2.

If you can find any logical flaw in the above steps - as per the rules you stated, feel free to point out the error.
This is all fine up to a point, but step 9 needs further examination/explanation to determine if there's a "logical flaw". So, let's unpack that.

Yes, it is true that once the number of unrevealed cards has been reduced from 52 to 2, the chances that either one of the 2 remaining unrevealed cards is the ace of hearts would be 1/2, if no other information is available to us, other than "There are two unrevealed cards on the table and one of them must be the Ace of Hearts". But in DaveC's scenario (quoted), the player does have more information than that.

To see the difference, consider the following variant on the above:

0. The player is out of the room and cannot yet see what is going on.
1. 52 cards are arranged in a line on the table.
2. Player wins if they can turn up the Ace of Hearts.
3. A random device selects one of the 52 cards at random. It is card #27. This card is designated as "the player's chosen card" (even though the player has no choice in the matter).
4. There is a 1-in-52 chance that the player's chosen card is the Ace of Hearts.
5. Monty turns 50 cards face up - all except card #27 and (for some mysterious reason) card #13.
6. Monty has eliminated 50 possibilities. There are now only two cards on the table. One is the card the player has chosen (card #27) and the other (card #13) is unknown (to the player).
6a. The player is then shown into the room and sees two cards unrevealed on the table, while 50 other cards are revealed as cards that are not the Ace of Hearts.
7. There was a 1-in-52 chance that the player's chosen card - card #27 - was the Ace of Hearts at the time that card was selected at random.
8. But, of 52 original possibilities, there are now only two.

Does this change things for the player, compared to the initial scenario? I think it does.

Now, the interesting thing here is what happens in each scenario when we add step 10:
10. The player is then given the option to "stay" with card #27 or to "switch" to card #13.

In the original scenario (at the top, above), the player is fully aware of everything that happened. That is, the player witnessed steps 1 through 6. In the variant scenario, however, the player is not aware of what happened because he wasn't in the room prior to step 6a.

Assuming the player in each case does not know the "rules" of the game, but has to infer them from what he has seen, what is the best strategy for the player to "win" by turning up the Ace of Hearts?

In the variant game:
The player, upon entering the room, can see 50 cards that are not the Ace of Hearts face-up on the table and two cards face-down. He can deduce (assuming a fair deck) that one of the two face-down cards must be the Ace of Hearts. Now here's the important bit: he has no reason to suppose that card #13 is any more likely than card #27 to be the Ace of Hearts. So, it's a 50-50 guess. "Stay" or "Switch" therefore makes no difference to him. He can only guess at which card the Ace of Hearts might be. So, in this case, the probability that the card #13 is the Ace of Hearts, based on the information the player has, is 1/2. There is no advantage of choosing "stay" over "switch", as far as the player can tell. (We know better, because we have advantage of knowing exactly how the game played out, but the player is not in our position.)

But what about the original game:
In that game, the player is aware of exactly how it came to be that 50 cards were revealed and 2 cards were left unrevealed. He is aware that Monty knew which of the 52 cards was the Ace of Hearts from the start, and that Monty deliberately turned over 50 cards that Monty knew were not the Ace of Hearts.
The player can (and should) reason as follows:
1. The probability that I chose the correct card (#27) in the first place was 1/52.
2. Therefore, the probability that the Ace of Hearts was one of the 51 cards I didn't choose was 51/52 at the time I chose card #27.
3. After Monty eliminated 50 of those 51 cards by revealing that they aren't the Ace of Hearts, I now know that there's a 51/52 chance that card #13 is the Ace of Hearts, while the probability that card #27 is the Ace of Hearts remains unchanged at 1/52.
4. Therefore, the sensible thing to do is to "switch" to card #13, because I'll have a 51 times better chance of winning if I switch, compared to if I "stay".
 
If the player chooses A, then the probability of A on the table =0.
If the player chooses a not-A card, then there are 51 cards on the table.
There is only 1 A card on the table, so the probability is 1/51.
You're still playing dumb, I see.

There are two groups of cards:
Group 1 consists of the single card that the player chose.
Group 2 consists of the 51 cards on the table, not chosen by the player.

If the player chose the A, then the A is in Group 1. The probability that the A is on the table is then zero as far as we are concerned. After all, we know the player chose the A. Things are different for the player, though. The player doesn't know if he chose the A. So, he estimates the probability of the A being on the table as 51/52. We know that the A isn't on the table, in fact, but the player doesn't.

If the player chooses not-A, then the A is in Group 2. The probability that A is on the table is then 1 (=51/51), and clearly not 1/51 as you claim it to be. If the player didn't choose the A, then the A must be on the table.

Again, though, things are different from the player's perspective. You and I might know that the player didn't choose the A, but the player doesn't know that. The rational player's perspective is the same as before. He reasons that he had a 1/52 chance of choosing the A, so he estimates the chance that the A is on the table at 51/52.

Are you going to continue to play the fool?
 
Last edited:
By the way, phyti, all your bullshit about "sessions" is just that - distracting rubbish.

Each specific player of the Monty Hall game (or the 52 card variants I just discussed) only ever plays one "session".

If we consider the game as it applies to a great many players, then we can start to ask questions like "On average, how often will a player play 'session 1' vs 'session 4'?" and such. We can also ask questions like "In the limit as the number of plays gets very large, what proportion of games will be 'won' by players who adopt the 'switch' strategy?"

In the Monty Hall game, the answer to the second question is that 2/3 of games will be won by players who adopt the "switch" strategy in the 3-door game. Similarly, 51/52 of games will be won by players who adopt the "switch" strategy in the 52-card variant in which the player is aware of the entire process.

I have already answered how often players will play each "session" many times in analysing the 3-door Monty Hall game. Consistently, phyti, you have failed to address the point that players will play each of the two "sessions" in which they happen to choose the door with car initially exactly half as often as they will play each of the two "sessions" in which they happen to choose a non-car door initially.

You continue to be dishonest interlocutor. Shame on you.
 
At this point, repeating your incorrect claim that the prior knowledge has no effect, is pointless.

So - just do it!

Get a pack of cards.
Shuffle them.
Pick one.
Then think about whether you'd be better off keeping that 1 or swapping to what's effectively all of the other 51.
He'll never do it. He knows what the outcome would be.
I would guess you refuse and stubbornly cling to your belief.
I no longer accept that he believes his own bullshit.

He ceased being honest in this discussion more than a year ago. Nothing has changed with him since then.

He's just trolling at this point.
 
Last edited:
And you are still making the same mistake regarding frequencies of those sessions.
Note, the player has a 1/6 chance of picking a door at the start - yet you have 10 sessions, with the player picking the car 5 times out of those 10. Do you not see how this interpretation of yours breaks the real probability of the game. The player will pick the car 1/6 of the time, yet you have them picking the car in 5 out of 10 "sessions", and you are weighting all "sessions" equally.
That is your mistake. You refuse to adjust it, despite having it pointed out to you repeatedly, and hence you are stuck regurgitating your same nonsense results.
phyti consistently runs away every time he is caught out making this error - which he continues to do despite endless correction.

He's either an idiot or a troll. Troll seems far more likely.
 
Last edited:
px = player chooses door x
hx = host opens door x, if not player choice nor door with the car

1 2 3 door ID
c 0 0 car location
sequence of events
stay________ switch

Session 1. p1 h2 p1___ p1 h2 p3
Session 2. p1 h3 p1___ p1 h3 p2
Session 3. p2 h3 p2___ p2 h3 p1
Session 4. p3 h2 p3___ p3 h2 p1
Session 1 is played once in every 6 games.
Session 2 is played once in every 6 games.
Session 3 is played twice in every 6 games.
Session 4 is played twice in every 6 games.

The player wins by switching when he plays sessions 3 or 4, but loses when he plays sessions 1 or 2.

Probability that player wins by switching is:
P(plays 1)P(wins 1) + P(plays 2)P(wins 2) + P(plays 3)P(wins 3) + P(plays 4)P(wins 4)
=(1/6)(0) + (1/6)(0) + (2/6)(1) + (2/6)(1)
=2/6 + 2/6
=2/3

Adopting "always switch" strategy doubles the chances of the player winning, compared to "always stay".

i.e. switching gives the player an advantage (doubles his chances of winning).

Comparison of differences in column 3 yields
win c = 1/2 for both.
There is no advantage by switching.

No questions about door probability.
Only had to count to 4 and divide by 4.
A simple game.
Dishonest troll.

Point out the fault in my analysis here, if you can, or else admit that it is correct.
 
You're still playing dumb, I see.

There are two groups of cards:
Group 1 consists of the single card that the player chose.
Group 2 consists of the 51 cards on the table, not chosen by the player.

If the player chose the A, then the A is in Group 1. The probability that the A is on the table is then zero as far as we are concerned. After all, we know the player chose the A. Things are different for the player, though. The player doesn't know if he chose the A. So, he estimates the probability of the A being on the table as 51/52. We know that the A isn't on the table, in fact, but the player doesn't.

If the player chooses not-A, then the A is in Group 2. The probability that A is on the table is then 1 (=51/51), and clearly not 1/51 as you claim it to be. If the player didn't choose the A, then the A must be on the table.

Again, though, things are different from the player's perspective. You and I might know that the player didn't choose the A, but the player doesn't know that. The rational player's perspective is the same as before. He reasons that he had a 1/52 chance of choosing the A, so he estimates the chance that the A is on the table at 51/52.

Are you going to continue to play the fool?
"The player doesn't know if he chose the A. So, he estimates the probability of the A being on the table as 51/52."

If the A has 1/52 chance of being in the deck, then the chance of being on the table, a set 1 card smaller, is 1/51.
You are comparing 2 different size sets, 1 vs 51.
 
...

If the A has 1/52 chance of being in the deck, then the chance of being on the table, a set 1 card smaller, is 1/51.
You are comparing 2 different size sets, 1 vs 51.

That's a bit garbled.

There are fifty two cards in a deck, and one of them is the Ace of Hearts.
Shuffle the deck, spread the cards on a table, and pick one card at random:

There's a 1 in 52 (1.9%) chance you have the Ace of Hearts in your hand, and
There's a 51 in 52 (98.1%) chance the Ace of Hearts is still on the table.

Agree so far? (If not, then I agree with post #161).
 
That's a bit garbled.

There are fifty two cards in a deck, and one of them is the Ace of Hearts.
Shuffle the deck, spread the cards on a table, and pick one card at random:

There's a 1 in 52 (1.9%) chance you have the Ace of Hearts in your hand, and
There's a 51 in 52 (98.1%) chance the Ace of Hearts is still on the table.

Agree so far? (If not, then I agree with post #161).


Dave C could write a simulation where 51 people each remove a different card from the table. Would there be 1 winner and 50 losers?
 
Last edited:
Not simultaneously. the A can only be in 1 place.

(Edit: the above seems to have been removed from the quoted post. I'll leave my reply in place.)

Yes, the Ace can only be in one place, the odds are about where it likely is.

In this case, it's most likely (98.1% vs 1.9%) to be among the 51 cards on the table, not the 1 card in your hand.

1 Card in your hand51 Cards on the Table
???????????
??????????
??????????
??????????
???????????
The Ace of Hearts might be in the left column or the right column. Yes - not both. But which is most likely?


(If you were betting on watching one person draw one card from a fair deck; would you put money on them getting the Ace of Hearts, or not getting the Ace of Hearts?)

Dave C could write a simulation where 51 people each remove a different card from the table. Would there be 1 winner and 50 losers?

If all 51 cards were removed from the table by 51 different people:
There's a 1.9% chance the Ace was in your hand and all of them are losers.
There's a 98.1% chance the Ace was on the table, so one of them is a winner.
 
Last edited:
A suggestion, seeing as an unfamiliar poster may enter this thread expecting MH but gets card games instead, is this an appropriate point to split the thread?

A good title would be, " Phyti doesn't understand card game probabilities either."
 
Dave C could write a simulation where 51 people each remove a different card from the table. Would there be 1 winner and 50 losers?
No, but I might consider writing a sim based on your rules as posted in #148.
If that's the rules then that's the rules.

But frankly I really don't see the point.
51/52 members in this thread already know how the odds work, and don't need yet another explanation.
The 52nd doesn't know, and doesn't care to learn.
So who exactly would yet another sim benefit?
 
(Edit: the above seems to have been removed from the quoted post. I'll leave my reply in place.)

Yes, the Ace can only be in one place, the odds are about where it likely is.

In this case, it's most likely (98.1% vs 1.9%) to be among the 51 cards on the table, not the 1 card in your hand.

1 Card in your hand51 Cards on the Table
???????????
??????????
??????????
??????????
???????????
The Ace of Hearts might be in the left column or the right column. Yes - not both. But which is most likely?


(If you were betting on watching one person draw one card from a fair deck; would you put money on them getting the Ace of Hearts, or not getting the Ace of Hearts?)



If all 51 cards were removed from the table by 51 different people:
There's a 1.9% chance the Ace was in your hand and all of them are losers.
There's a 98.1% chance the Ace was on the table, so one of them is a winner.

'most likely' is not a probablity.
The person with the card in hand does not have the A.
You are saying there are 51 possible cards that could be the A.
Probability needs a history of card choices which allows it to be calculated.
P=number of successful choices/all possible choices. P=1/51.
If 51 people remove a card from the table, 1 will be successful and 50 will not.
50/51=the probability of any 1 card on the table not being A. 1-50/51=1/51.
The solution depends on choices, not locations.
 
Status
Not open for further replies.
Back
Top