On This Problem Rational People Do Worse

TL;DR

In Newcomb's Paradox, rational two-boxers consistently earn only $1,000 while "irrational" one-boxers walk away with $1 million.

Key Points

• 1.The Setup: A supercomputer predicts your choice before you enter. If it predicted you'd take only the mystery box, it placed $1 million inside. If it predicted you'd take both boxes, the mystery box is empty — leaving you with just $1,000.
• 2.Two-boxer argument (Causal Decision Theory): The boxes are already set. Your choice now can't change the past, so you should always take both — you gain an extra $1,000 regardless of what's inside the mystery box. This is called "strategic dominance."
• 3.One-boxer argument (Evidential Decision Theory): Because the computer has correctly predicted thousands of people, the evidence strongly suggests one-boxing correlates with $1 million. The math shows that if the predictor is even slightly better than 50.05% accurate, one-boxing has higher expected value.
• 4.The Guardian polled 31,000 people in 2016: 53.5% chose one box, 46.5% chose two. Veritasium's own poll of 24,000 viewers found two-thirds were one-boxers.
• 5.Philosophers Gibbard and Harper (1978) argued two-boxing is the rational act, but admitted one-boxers fare better — concluding the game "richly rewards irrationality," which the video calls a cop-out.
• 6.The deeper lesson — pre-commitment: Newcomb's parallels Mutually Assured Destruction (MAD) and the game of Chicken. In all three, publicly committing to a worse outcome (retaliation, one-boxing) produces the best real-world result. The Soviet doomsday device in *Dr. Strangelove* illustrates how removing the option to defect is itself the winning strategy.
• 7.Key takeaway: Rationality may not be about in-the-moment choices but about what rules you pre-commit to living by — because being the kind of person who one-boxes is what actually wins, both in the game and across the repeated "game" of life.