Robin
Penultimate Amazing
- Joined
- Apr 29, 2004
- Messages
- 14,971
No, can and did are your words. I don't care what the agent can do. What the agent did at t=30 does not even make sense in a statement that exists at t=20. It is like saying "I did that 30 minutes from now".My point is that there are two different statements. One is that John can choose between three different choices at t<30. The other is that John chooses between three different choices at t=30.
My statements referred to the probability that the agent will turn left at t=30.
Things you can't see can also be in your timeline, things that don't impact you can be in your timelineInserting the statement doesn't change anything. In fact, it's not even in John's timeline (it doesn't affect John's timeline in the least since he never sees it).
But that does not mean that the timeline is not changed.While it is true that the statement exists "in time," the programmer being unaffected by the agent's timeline is key. The programmer can peek at what the agent will do at t=30, and insert the statement into the world at t=20 before the agent has performed the action without disrupting the timeline.
So the chances of the prediction in the safe being wrong are then 1/2 at t<30Sure. The chances of John choosing to turn or stop are each 1/2 at t<30.
So what you are saying is that if I did a complete dump of the array that holds the events relating to t<30 before the call to the free will function, and did another dump of t<30 after the programmer predicts the agents choice, then they would be absolutely identical in every way?Huh? What multiple universes are you referring to here?
Are you saying that the 'LEFT' in a safe will appear in both dumps. No they will not, the programmer inserts the prediction after the call to the free will function. The free will function is called in the context of a different history to the one containing the description.
Let me put it this way, you need to do this:
Code:
God.isOmnipotent=true;
p = God.getPrediction(agent,t);
if ( p == null )
God.isOmnipotent = null; // cannot say one way or other
else if (p != agent.make_free_choice() )
God.isOmnipotent = false;
In the above example, unless you can avoid this in coding.And when exactly is God (or the programmer) ever wrong?
So a statement exists in the safe at time t=20 saying "At time t=30 agent turns left". There is a 50% probability that at time t=20 the agent turns right. But the statement "At time t=30 agent turns left" cannot possibly be wrong. And you can't see any conflict here?There is no conflict with the three statements.
From my original problem:See, that's where you got it wrong. The statement is actually that the programmer witnessed the agent having already turned at t=30. In fact, the programmer isn't predicting anything, he's observing it as it happens at t=30. He inserts his observation into the world at t=20. Even though the statement exists before the event occurs, it's not a prediction -- it's an observation of the actual event.
So you see it is not that I got it wrong, it is that you have changed the problem in order to give God and out.The book has predicted he will go right
But the problem says what it says, it is about predictions. If God chooses to peek into the future to make these predictions that is his own business, but he must phrase them as predictions, as testable statements about John's behaviour at a particular time. They will be right or they will not and if they are not then God will not be omnipotent. http://www.internationalskeptics.com/forums/showthread.php?postid=1239967#post1239967
But the prediction by God which is in existence at t=20. So there is a 1/2 probability at t=20 that this prediction will turn out to be wrong.Correct, at t=20 it is 1/2 as I said. At t=30 it goes to 1 since the agent has already taken the action.
Except for this:There is no falsification of history at all.
Which sounds like falsification to me. It would now require recalculation to be consistent.Now at time 10, we insert a new wall on which is written the word "LEFT" completely contained inside the safe.
Wait, did you really say "You, as the programmer, can predict with 100% accuracy the agent's future movement"? I thought you said it was not a prediction? But now when it supports the "God can be omnipotent" part it is a prediction.Furthermore, this is a completely programmable simulation. You can program it yourself in a few hours and try it. There is no trick at all. You, as the programmer, can predict with 100% accuracy the agent's future movement by observing it first hand even though you cannot predict the results of the random function.
Again, you can't have it both ways. You have inserted a 100% accurate prediction into t=10 to demonstrate omniscience. Then at t=20 there exists a 100% accurate prediction that the agent will turn left. And yet at t=20 there is a 50% probability that it will turn right.
There is a contradiction.