Quote Originally Posted by Invader View Post
Misunderstanding. I didn't mean that we'd have the literal ability to replay the day as a means of testing, only that there'd be a difference if the day could be replayed with the extra person in order to illustrate that the wishful thinking would have some overall but non-specific effect.
I understood that you intended it as a purely hypothetical suggestion.

Quote Originally Posted by Invader View Post
Would it be possible then to have a computer that can model a situation according to our current understanding of physics and then have an observer attempt to effect the outcome, or is something fundamentally wrong with that?

I can think of an example of what I'm trying to suggest. If we didn't know about electromagnetism, for instance, and decided to build a machine that worked on other principles that could make a theoretical model of a situation based on that understanding of physics, the information it would produce would be in line with that model of physics. Say we have two conducting wires next to each other, one with alternating current flowing through it, and the other disconnected from any power source. The simulation would suggest that the second wire has no current moving through it. The real world would demonstrate the contrary, that the second wire has current induced in it from the first.

I'm still not sure that what I'm saying is clear, or at least that modern computers can simulate complex events and still be true to our understanding of the physical laws without being effected by any other unknown phenomenon that we aren't aware of yet. Does that make sense?
I think it makes some sense. There are a couple different points here.

To address the issue about what modern computers can and can't simulate: computer models often exhibit interesting and unexpected emergent behavior when only basic assumptions are programmed in, and these phenomena can be tested against the empirical facts. So your example about programming in the assumptions given by a certain theory of electromagnetism and then observing whether the emergent electrical behavior matches well with empirical data makes sense.

Whether or not a human observer could passively affect the outcome of the simulation is a somewhat different issue. It really comes down to the type of claims being made by the observer. To answer your other question, there is not any fundamental reason that I can see why they couldn't affect the simulation. The simulation is implemented as a part of the natural world just like everything else and thus should be open to this kind of manipulation. It just seems to me like it would be an unusual context in which to demonstrate these kinds of effects, and I could imagine some of the people who purport to be able to do this sort of thing to protest, perhaps reasonably, that it's an entirely different situation when dealing with computer simulations. And if they're not making this claim in the first place, we wouldn't really have a good reason for testing it