You assume that a human cannot be described with something less powerful than a Turing machine. There is no reason to believe that knowing what a human will do in this instance is equal to decidability of computer programs.
Re:Newcomb's paradox
by Dylan16807 on Saturday April 04, 2015 @05:03AM
(#9320410)
I almost objected, but you're right. The human has finite time with which to calculate. The predictor can easily be assumed to have more time. There's no problem calculating if a program will halt in finite time.
I assume nothing of the sort. I merely assume that humans, like computers, are capable of evaluating the truth of something and then deliberately reporting the opposite. That is all that is required.
Re:Newcomb's paradox
by connerbrooks on Saturday April 04, 2015 @03:13AM
(#9320240)
I think this may be the proof.
Start by assuming that the predictor is a decider P for a computational model of a human M in a situation encoded by w. We can then construct a Turing Machine S that decides A_TM.
S <M, w> {
1. Run P on w.
2. If P halts, run M on w; if M accepts, accept, if it rejects reject.
}
Since S decides A_TM we must have made a mistake somewhere in our proof, our only mistake could be that the predictor P is a decider. Therefore the Predictor is not a decider.
The isomorphism between the program which defeats decidability and the two-box Newcomb problem is left as an exercise.