A machine that’s never been wrong predicts your choice.
You see two boxes:
Box A — open, has $1,000 Box B — 0 or $1M
If the machine predicted you’d take only Box B, there’s $1,000,000 inside.
If it predicted you’d take both, Box B is empty.
When I first read about this thought experiment, my gut reaction was simple: take one box. All or nothing. It felt right, spontaneous, exciting.
Then I started thinking. And the more I thought, the more “rational” it seemed to take both boxes. The machine has already made its prediction. The money is either in Box B or it isn’t. My choice now can’t change what’s already inside. So why not take the guaranteed $1,000 on top of whatever’s there?
This argument feels bulletproof. Until you remember that the machine predicts decisions with near perfect accuracy. And it has never been observed to be wrong.
This is Newcomb’s Paradox, one of the most famous problems in decision theory. It splits people almost perfectly in half, and both sides are convinced the other is missing something obvious.
Two boxers argue that your current choice cannot affect what’s already in the box. That’s basic causality. And regardless of what’s inside Box B, taking both boxes always yields $1,000 more than taking just one. This is strategic dominance. If one option is better in every possible scenario, why not choose it?
One boxers see it differently. The machine predicted your reasoning before you even began. If you are the kind of person who thinks “I will just take both,” the machine has already accounted for that and left Box B empty. The people who take one box walk away with a million. The “rational” ones walk away with a thousand.
So what does rational actually mean here?
At this point, it stops being about boxes and starts being about free will.
If a machine can predict your choice with near perfect accuracy, was the choice ever really yours? You sit there, weighing options, feeling like you are deciding freely. But the outcome may have been determined before you even entered the room.
What I find most interesting is this. The people who “win” this game are not the ones trying to outsmart the system in the moment. They are the ones who had already made a decision about who they are before the choice even appeared.
Not because they calculated better, but because they committed to a principle. A quiet decision to be a certain kind of person. And somehow that pre commitment, that decision to be who you are, turns out to be a surprisingly effective strategy.
It reminds me of the prisoner’s dilemma. In a single round, betrayal is rational. Over time, cooperation wins. Trust outperforms calculation in the long run.
Another way to see it is to follow the moment as my friend said. Choose based on what matters to you right now. Maybe one box is enough. Maybe taking both feels right.
Sometimes the most rational thing you can do is act irrationally. Not as a mistake, but as a strategy. Like eating a pastel de nata late in the evening, fully aware that sugar is not good for you, simply because allowing yourself that small, irrational pleasure is its own quiet act of freedom.
Because in the end, it is less about making the perfect choice in the moment, and more about deciding who you are before the choice appears.