Newcomb's Paradox — The Oracle's Game

An Oracle with 95% accuracy has studied thousands of players and predicted your choice — before you even knew you'd play. Two boxes stand before you. The Oracle has already acted on its prediction.

Evidential · One-Box

Take only Box B

The Oracle foresaw you. If you're a one-boxer, Box B almost certainly holds $1,000,000. Correlation is your ally. Trust the evidence.

Causal · Two-Box

Take both boxes

The Oracle already decided. The money is fixed. Taking Box A always gives you $1,000 more. The past cannot be changed by your present action.

The Setup

Box A Transparent. Always contains $1,000.

Box B Opaque. Contains $1,000,000 if the Oracle predicted you take only Box B — otherwise empty.

Oracle Has played thousands of games. Correct 95% of the time. Already made its prediction.

Your choice Take only Box B — or take both boxes. What do you do?

⬥ The Oracle has made its prediction ⬥

🔮

Box B

??? — sealed

💰

🔮

Both Boxes

$1,000 + ???

The Unresolved Dispute

Evidential Decision Theory says one-box: your choice is evidence of the Oracle's prediction, so take Box B and expect ~$950,000. Causal Decision Theory says two-box: the past is fixed, taking Box A costs nothing extra, so always grab both.

Both positions are held by serious philosophers and mathematicians. William Newcomb posed the problem ~1960; Robert Nozick popularized it in 1969. No consensus exists. The paradox remains genuinely open.