August 14, 2008The Game Show ProblemHow about a little Math, Scarecrow? The best ones look simple but are counter-intuitive. And I will admit, I didn't get this. Even after reading this good explanation (ain't the Internet grand?) I reluctantly came around. I saw this in the movie "21," a decent (3.675 starts) if imperfect flick. They gave it enough time that I felt they must have had good backing, but I couldn't get it. It's "The Game Show Problem" and our beloved protagonist, Ben Campbell, solves it correctly to ingratiate himself with his professor at MIT (played by Kevin Spacey) and eventually secure his spot on the school's unofficial intermural blackjack squad. Here it is, identical to its appearance in 21: Suppose you're on a game show, and you're given the choice of three doors. Behind one door is a car, behind the others, goats. You pick a door, say #1, and the host, who knows what's behind the doors, opens another door, say #3, which has a goat. He says to you, "Do you want to pick door #2?" Is it to your advantage to switch your choice of doors? I watched the movie twice and gave the screen the angled-head-quizzical-dog look both times. I majored in Math (probability was not my thing, and I left school to pursue a music career) but it seems like third grade rules apply, that once the door is opened, you have a 50-50 shot with either door. I err in good company; many of the PhD commenters make the same claim. Writing a little java program to test it empirically, it becomes obvious. Sticking with Door #1, you will win only if it is in there (33%). The host will not show you the car, so switching gives you the known best choice of the other two. Counterintuitive. Clever. I would not have made the Blackjack team. UPDATE: I just ran the computer program 1,000,000 times (boy, my finger is tired!) and sticking with door 1 wins 33.3134%, switching 66.6866% |
This really is fascinating. The Refugee likes to dabble in statistics and probability (albeit from a mundane marketing perspective). At a macro level, it demonstrates how human intervention changes probability. As one of the teachers commented, I'm not sure how one would write a formula for this - and that's the point. The Refugee needs to ponder the related implications for a while.
Posted by: Boulder Refugee at August 15, 2008 1:15 PMI refused to believe it until I was writing the code to test it. There are only two outcomes, that is 100% of the time, you win or you lose. If you choose #1 and stick, you win 33% of the time. Ergo, if you switch (the only other choice), you win (100-33)%
Posted by: jk at August 15, 2008 2:37 PMHere's another way to look at it: when picking the first door, you and 1/3 chance of being a winner and therefore a 2/3 chance of being a loser. After being allowed to change sides, you now have 2/3 chance of winning. Odd - counter-intuitive - but real.
Posted by: Boulder Refugee at August 15, 2008 3:40 PM | What do you think? [3]