A being, who you believe to have superior predictive powers (say God or some super intelligent computer), makes you an offer. Before you are two boxes, identical in appearance. Within one box, let us call it 'Box A', there is either one million dollars or nothing. Within the other box, 'Box B', there is definitely one thousand dollars.
The being, whom in deference to tradition we will call the Newcomb Being, will allow you to either take both boxes, or just Box A.
Twenty-four hours ago, the Newcomb being made a prediction about what you would choose. At this point, it placed the money in the boxes. If the Newcomb Being predicted that you would take both boxes, it left nothing for you in Box A and the $1000 in Box B. If the Newcomb Being predicted that you would only take Box A, it would leave the million dollars in it, as well as the thousand dollars in Box B. If the being predicts you will base your choice on a random event (like flipping a coin), it will leave Box A empty.
Now you have grounds to believe that in such matters, the Newcomb Being is always 100% correct in predicting your actions.
Will you pick (a) both boxes or (b) Just Box A
This is the problem known as the Newcomb paradox.
Argument (I) - 'Two boxer' argument
Some people are dead sure that the best choice is to pick both boxes. According to them, no matter what choice you make now, it cannot change the contents of what is in the box A now (Remember The entity filled the contents of the boxes BEFORE you made the choice). Since you can't change what is in the box A, you might as well take both boxes to maximise your gains.
Argument (II) - 'One boxer' argument.
Others argue that the rational choice is to pick just Box A and win the million. They argue, that because the entity knows you will argue as in argument (I) above and pick both boxes, he will put nothing in the Box A because he can predict your reactions with 100% accuracy. So picking both boxes will get you only 1,000.
So you should just choose to take box A and win the million.
They also argue that if you carry out the experiment, those who choose only Box A always win the million and those who choose both, get only 1,000. So obviously the best choice is to pick box A only.
Who is right? What would you do?
PS : One variation of the problem, says that instead of the Newcomb Being being 100% accurate in prediction it is only 95% accurate. It does not really change the problem, since the expected gain of picking only Box A (=95%*1,000,000 + 5%*0) is still higher if you go for the 'one boxer' argument (=95%*1,000 + 5%*(1,000+1,000,000).
PSS : No one knows what the right answer is, both arguments seem to be equally rational, but they can't be both right! it is interesting to see what your intuitions tell you.
