All corrections
LessWrong March 11, 2026 at 04:16 PM

www.lesswrong.com/posts/gjAuqRmS6oQ6Xv9Th/newcomb-s-paradox-complete-solution

1 correction found

1
Claim
TDT believers only see the first decision, CDT believers only see the second decision.
Correction

This overstates the causal-decision-theory view. Standard references note that CDT can also recognize an earlier decision to cultivate a one-boxing disposition, rather than "only" the final box-choice.

Full reasoning

The sentence says CDT believers "only see the second decision". That is not accurate as a description of causal decision theory.

Authoritative discussion of Newcomb's problem in the Stanford Encyclopedia of Philosophy explicitly says that CDT can acknowledge an earlier, separate decision about preparing for the problem by cultivating a disposition to one-box. In other words, CDT does not merely "see the second decision" (the final act of taking one or two boxes); it can also evaluate an earlier self-modification or disposition-forming choice.

SEP states that:

  • "A rational person should prepare for the problem by cultivating a disposition to one-box."
  • "Causal decision theory may acknowledge the value of this preparation."
  • It may hold that cultivating that disposition is rational, even while still holding that, given the circumstances at the moment of choice, two-boxing is rational.

So the post's blanket characterization is too strong: CDT is often presented as recommending two-boxing at the moment of choice, but it is not true that CDT believers "only see" that second decision and ignore earlier disposition-shaping decisions altogether.

1 source
  • Causal Decision Theory (Stanford Encyclopedia of Philosophy)

    A way of reconciling the two sides of the debate about Newcomb’s problem acknowledges that a rational person should prepare for the problem by cultivating a disposition to one-box. ... Causal decision theory may acknowledge the value of this preparation. It may conclude that cultivating a disposition to one-box is rational although one-boxing itself is irrational.

Model: OPENAI_GPT_5 Prompt: v1.16.0