Assuming this is a one off and not a repeated iteration;
I’d take 1A because I’d be *really* upset if I lost out on $27k due to being greedy and not taking the sure $24k. That 1⁄34 is a small risk but to me it isn’t worth taking—the $24k is too important for me to lose out on.
I’d take 2B instead of 2A because the difference in odds is basically negligible so why not go for the extra $3k? I have ~2/3rds chance to walk away with nothing either way.
I don’t really see the paradox there. The point is to win, yes? If I play game 1 and pick B and hit that 1⁄34 chance of loss and walk away with nothing I’ll be feeling pretty stupid.
Let’s say you prefer 1A over 1B, and 2B over 2A, and you would pay a single penny to indulge each preference. The switch starts in state A. Before 12:00PM, you pay me a penny to throw the switch to B. The die comes up 12. After 12:00PM and before 12:05PM, you pay me a penny to throw the switch to A.
But why would I pay to switch it back to A when I’ve already won given the conditions of B? And as Doug_S. mentions, you can take my pennies if I’m getting paid out tens of thousands of dollars.
I do see the point in it being difficult to program this type of decision making, though.
Assuming this is a one off and not a repeated iteration;
I’d take 1A because I’d be *really* upset if I lost out on $27k due to being greedy and not taking the sure $24k. That 1⁄34 is a small risk but to me it isn’t worth taking—the $24k is too important for me to lose out on.
I’d take 2B instead of 2A because the difference in odds is basically negligible so why not go for the extra $3k? I have ~2/3rds chance to walk away with nothing either way.
I don’t really see the paradox there. The point is to win, yes? If I play game 1 and pick B and hit that 1⁄34 chance of loss and walk away with nothing I’ll be feeling pretty stupid.
But why would I pay to switch it back to A when I’ve already won given the conditions of B? And as Doug_S. mentions, you can take my pennies if I’m getting paid out tens of thousands of dollars.
I do see the point in it being difficult to program this type of decision making, though.