for @Litten to discuss philosophy because he’s odd
The Game
The game has the following form. You will be presented with two boxes:
-
Box A will contain $10,000.
-
Box B will contain $1,000,000 or nothing.
You will not be able to see inside the boxes, so you won’t know whether Box B contains $1,000,000 or nothing. You will then be given the choice between taking home both boxes or just Box B.
There are two crucial things to take into account when making your decision.
-
The amount of money in Box B will be determined by Perfect Predictions’s most accurate clairvoyant - approaching 100% accurate – on the following basis. If she predicts that you will take home both boxes, then no money will be put into Box B. If she predicts that you will take home only Box B, then $1,000,000 will be placed into the box .
-
The prediction will already have been made by the time the game starts, and the amount of money in Box B already fixed.
Here’s a table representing possible outcomes.
Assuming you want to win as much money as possible, should you take home both boxes or just Box B?
This is the main thing
if you want to win as much as possible you take both but gaslight the being into thinking you’re just picking b blah blah blah
So what are we doing
i think its optimal for me to take both boxes and its optimal for the clairvoyant to guess both boxes and i end up with 10k. the clairvoyant has no reason to predict only box B that’d be throwing
it’s trying to get at the whole game theory suggests you should take both thing
if they predict only box B they are are giving me a million for free. regardless of my choice
assume that this clairvoyant has infinite money and does not care about giving it away
Wow that was quick
so they must predict A B and looking at the chart i can win at most 10k
i’m two years early to the problem
well still assuming their goal is to not give me money (is it their goal?) they should predict A B
their goal is to be right
as often as possible, but obviously not exactly 100% because then the paradox vanishes
then we would both mutually benefit if i go for B and they predict B. i get a million and they get to be right
again, if someone has a greater than 50% chance of picking correctly (like, say, a game theory professor who’s been watching you for the better part of a semester and presumably has better than coinflip odds of knowing what you would choose) expected value suggests you should take just b. but you’ll always get 10,000 more by picking both no matter what
it’s a devilish problem
I’ve actually heard of this one before. Hmmm, maybe it could be worked into a misc or something like that.