Newcomb's paradox is a thought experiment that divides participants roughly evenly on whether to choose one or two boxes [2].
The dilemma matters because it probes the fundamental tension between the concept of free will and the possibility of a near-perfect predictor. If a supercomputer can accurately anticipate human choices, it challenges the notion that individuals possess independent agency over their decisions.
In the experiment, a participant is presented with two boxes. One box may contain $1,000,000 [1], while the other contains a smaller, guaranteed amount. The catch involves a supercomputer that has already predicted whether the participant will take only the large box or both boxes.
If the predictor anticipates the person will take only the large box, it places the money inside. If it predicts the person will take both, it leaves the large box empty. The participant must decide their action without knowing the predictor's current move, though the predictor's track record is nearly perfect [1].
This scenario creates a split in logic. One group argues for the "one-box" strategy, suggesting that since the predictor is reliable, taking only the large box is the only way to secure the $1,000,000 [1]. They believe the prediction is effectively a predetermined outcome based on their character.
Another group argues for the "two-box" strategy. This logic suggests that because the money is already either in the box or not, the act of choosing cannot retroactively change the contents. Therefore, taking both boxes provides the most value regardless of the prediction [1].
Data indicates that this logical divide is stark, with approximately 50% of participants choosing each option [2]. The experiment serves as a tool for philosophers and scientists to explore how humans perceive causality, and the limits of predictability.
“Newcomb's paradox is a thought experiment that divides participants roughly evenly.”
The persistence of a 50-50 split among participants suggests that the conflict between evident decision theory and expected utility is not merely a mathematical error, but a fundamental disagreement on how causality works. If a predictor can reliably forecast a choice, it implies a deterministic universe where 'choice' is an illusion, whereas the two-box strategy insists on a world where the past cannot be altered by present actions.
