gRR comments on What a reduction of "could" could look like - Less Wrong

53 Post author: cousin_it 12 August 2010 05:41PM

You are viewing a comment permalink. View the original post to see all comments and the full post content.

Comments (103)

You are viewing a single comment's thread. Show more comments above.

Comment author: gRR 18 February 2012 11:56:07PM 1 point [-]

I thought about a different axiomatization, which would not have the same consistency problems. Not sure whether this is the right place, but:

Let X be a variable ranging over all possible agents.

(World Axioms)
forall X [Decides(X)=1 => Receives(X)=1000000]
forall X [Decides(X)=2 => Receives(X)=1000]

(Agent Axioms - utility maximization)
BestX = argmax{X} Receives(X)
Decision = Decides(BestX)

Then Decision=1 is easily provable, regardless of whether any specific BestX is known.
Does not seem to lead to contradictions.