I'm surrounded by people talking about this all the time. I feel it turns on what one considers "infallible predictor" to amount to.

At any rate, the problem as given above poses that the predictor is infallible, but we can quibble with what that amounts to and what sense it makes, and of course we can imagine related problems where we relax the infallibility. If the predictor really truly was infallible, then you would make 1.1 million taking one box but would only make 1000 taking one box, as tautological consequence of infallibility in this setup.

But some may feel the predictor being infallible also makes the problem nonsensical, by leaving no room for truly free choice even as we ask you to answer what choice you would make.

Er, *"but would only make 1000 taking two boxes", I meant to say. 11:25 AM Perhaps the core of the problem is this:

An infallible predictor predicts that you will decline a million dollars. They then offer you the million dollars. Do you accept it or decline it? One might say obviously you should and would accept it. One might also obviously you nonetheless won't accept it, by stipulation of the problem.

Or we might also postulate the problem in a time-travel setup.

Instead of being a predictor, they are a rememberer. Based on your decision, they will in the future send the money through a time machine into the past. Again, if you take just one box, they'll arrange for that one box to have a million bucks. If you take both boxes, they'll arrange for the two boxes together to have nowhere near a million bucks.

But perhaps saying they will do this AFTER you make their decision makes one more inclined to feel they can indeed be infallibly correct in their reaction to your decision. On the other hand, you may well react to this by feeling "Well, this just means you can't send money backwards in time". But certainly we could imagine a counterfactual universe with such time travel within it, could we not? But this illustrates again that one must grapple with what it amounts to to make choices freely or not. 11:31 AM I keep making typos. Sorry, I'm a terrible typer. In our lives, we do not have this kind of infallible prediction or time-travel, they are not a part of our ordinary experience, and so our ordinary intuitions do not speak to them. The very way we ask questions encodes presuppositions which may be incompatible with it, the very conceptual language of choice and so on. 11:35 AM Perhaps the truly most reduced form of this is like so:

On a table, there is a million dollars, free for the taking. As it happens, it is known with 100% certainty, an inevitable consequence of the laws of physics and the state of the universe up until this point, that you will not take the million dollars. Should you take the million dollars?