A perfect predictor already filled the box — do you still have a real choice?
A near-perfect predictor has already decided what to put in a hidden box, based on forecasting what you're about to do — and is right almost every time. In front of you sit that hidden box, containing either $1,000,000 or nothing, plus a second, transparent box holding a visible $1,000. You may take only the hidden box, or take both boxes. The contents were locked in before you chose. Which do you take?
Reveal the answer
There's no consensus answer — that's the paradox. Expected-utility reasoning favors taking only the hidden box, since an accurate predictor overwhelmingly rewards one-boxers with the million. Dominance reasoning favors taking both, since whatever is already inside the hidden box can't change based on a decision made after the fact, so grabbing the extra $1,000 can only help. Physicist William Newcomb devised the puzzle around 1960; philosopher Robert Nozick formalized it in 1969, and Martin Gardner introduced it widely in his March 1973 Scientific American 'Mathematical Games' column, where readers favored one-boxing roughly 3-to-2.