I've been out of this thread for a while, so I don't remember if this was brought up before, but it sounds like your position is very similar to Dennett's argument in
"Why You Can't Make A Computer That Feels Pain". Is that right?
I agree with several things Dennett is saying there, but I still think he has missed the larger point.
He is correct in pointing out that a computer simulation of a hurricane doesn't result in anything getting wet, and does not confer upon the computer any actual wind speed.
But let's take this example, or rather a similar example of a tornado, and look at it from a systems theory point of view.
It's possible to build a tornado box in which real-world models of tornadoes can be produced. It's also possible to simulate a tornado on a computer. (For our purposes here, we'll posit a stipulative definition of a "model" as a real-world functional replica, and "simulation" as a virtual computer rendering of the phenomenon.)
What's the primary difference here?
Well, the model tornado can destroy real-world objects in its path. And it can do this whether or not anybody is there to observe it.
In other words, the model tornado is capable of existing in its entirety within a system that contains only the tornado itself. It can "be a tornado" on its own.
Let's compare that to a computer simulation.
In that case, if we remove any observer who knows how to interpret the output of the computer, and leave the machine alone in a system by itself, there is nothing there which resembles a tornado at all. It's just a computer doing what computers do, changing electronic states.
The simulation only exists in a system which includes at least one brain to act as programmer and interpreter. Someone has to be able to
understand that the outputs of the machine -- which may be pixels displayed on a screen, sound coming from speakers, printouts on paper, and such -- are intended to represent a tornado, or someone has to be observing who has the right kind of brain which can be fooled into "seeing a tornado" by the non-tornado-like output. And even so, this cannot be achieved by programming alone, but requires special hardware such as a display, printer, or speakers.
So the simulated tornado has no independent existence in the real world of spacetime, matter, and energy. The only thing existing in that realm is a computer behaving in essentially the same way as a computer simulating anything else.
This means that computer simulations, in order to be simulations, demand a system which incorporates 3rd party observers, and that the simulation doesn't actually happen in the computer at all, but resides in the interaction between the computer and an observer. Remove the outside observer and the simulation simply doesn't exist.
(This is why it's not possible that we exist inside a simulation. If our universe is a machine producing a simulation, we would have no way of knowing what sort of thing the universe is meant to be simulating to the beings running it, just as a sentient microbe inside a computer running a simulation program would have no way of knowing what the computer was supposed to be simulating for its observers… it would only perceive the mechanical workings of the machine itself.)
My body is performing a phenogram when it's conscious, regardless of whether anyone else is around to be part of a system with it.
Therefore it's not a simulation, and one cannot simply program a general purpose computer to do the same thing virtually and expect an actual phenogram to be the result in actual spacetime.
If we want a machine to generate a phenogram, we're going to have to build the proper hardware to make that happen. A computer might well be one of the components of such a machine, but a programming-only solution to producing an actual phenogram in the real world simply is not possible.
