Can the scientific method ever disprove Magic and Miracles (Simulated)?

Question

I have seen the idea behind this question before but I don’t think it was well formulated. Here is my attempt to ponder on reality based on Plato's “Myth of the Cave” and show how valid it still is.

According to remarkable thinkers such Buddha, Descartes, Einstein etc. reality is an illusion. What makes reality "real" it's persistence and consistency eg. "Day after day most people endure the same routine,they communicate approximately in the same way with roughly the same people, they look about the same as the day before and they engage in about the same routine and mundane situations and sensations day after day". This permanent coherence is what makes reality "real" for the human mind and different from for instance “lucid dreams”.

Thought experiment 1: Harry Potter’s Magic World

Let's presume that a bunch of “cruel” anthropologists have an advanced virtual reality gadget and they have raised a group of kids since babies 24x7 in and advanced version of the Harry Potter's magic world video game. They will be trained to use magic wands, fly on brooms, etc. (for them it would be as natural and ordinary that it would not be even cool). Since they have not known any other reality just like in the Myth of the Cave, that video game would be "their reality" besides their science cannot prove/disprove the true origin of the forces behind that "magic".

Would it be somehow possible for them to find out that they are living in an fantasy world? Would they even want to?

Thought experiment 2: Sid Meier’s AI Matrix Civilization

We are living in an 7000+ year strategy game that began before the invention of the wheel. This artificial machine has the ability to know future events (it records snapshots of the whole world including the memories of the users playing in it or it is perhaps a quantum computer therefore all our destiny is already written but we think we have “free will”). This machine is programmed to make people ignorant of its true nature as long as most of its users accept it is not fake. Once it reaches this state it will reset itself before bronze age as it has been doing it over and over again for billions of years getting better learning more and more with each iteration. It will allow magic, alchemy (turning lead into gold), miracles, holy figures, unicorns, mermaids, UFOs, paranormal activity as long as it does not give away its true nature with proof eg. “A remarkable psychic would never pass the scientific test” because it assumes humans would not want to live in a fake world. This AI knows how important for the human mind hope is and also knows how superstition and credibility work so it would try to find a balance.

Can the scientific method ever be valid enough to disprove the simulation hypothesis when that simulation can be intelligent? Is this even taken seriously? Is it safe to presume that the only thing that could make people suspect is a flaw in the matrix and even so it could be only a non-credible subjective experience no different from a paranoid delusion? Knowing that from a neuroscientific point of view the mind has difficulty to drastic changes eg “going to live in a different country to learn a very different culture and language at the age of 80” is it safe to assume that the desire to remain in the cave is in our nature and trying to be leave would be “crazy”?

References to authors asking those questions and personal opinions and above all new ideas would be appreciated.

“There are only two ways to live your life. One is as though nothing is a miracle. The other is as though everything is a miracle.” Albert Einstein

Answer 1

If you think of science more broadly as a collection of methods to point towards truth, using data and reason, then yes I think it theoretically can navigate these kinds of truth claims. (Prove or disprove seems to be rather strong wording, I think of science in terms of probabilities given what we currently know. And the scientific method is a rather quaint way to constrain science that most scientist's don't bother using anymore. Not to say there aren't good pieces and practices contained within it.)

The simulation thought experiment assumes there is another world from which the current world is being simulated from. If the simulated creatures science could ever become more insightful than the designers of the simulation's code prepared for then I suppose you could see imperfections in the simulated world. Once again this is basically the Matrix's premise. Notice it doesn't necessarily hinge on the simulated creatures processing more information than the simulation itself (which would be impossible anyways) but on insights into reality that the designers didn't prepare for. Granted this may be practically very difficult, because all data and experiments (necessary for science) would be mediated through the simulation machine.

Maybe a similar but more directed question than is: "Could a simulated creature outsmart the design the architect made of the simulation to accidentally discover the world of matter not just code?"

I frankly don't see why this would be impossible. Difficult, probably. Impossible, doubtful.

Edit:

This is definitely a difficult thought experiment. The main difficulties for me come from illusions of free will in the simulated creatures and not understanding enough information processing theory. So lets imagine a simpler experiment. As a thought experiment design a 2d box and set a 2d "ball" bouncing off the corners inside the box. Does it ever leave the box? What if your corners don't meet up enough? What if you make your box more complex. Say by making it 3D. Say a cube. Imagine the ball bouncing from wall to wall. Now add in a 3D maze with irregular outside boundaries. Now the there is much more surface error to make a mistake and let the ball out. You could even add a 4th dimension to this imaginary box. As you scan the box for imperfections that may let the ball out it becomes possible to miss a ball sized hole. Now in an almost infinitely more complex simulation it must be almost infinitely more difficult to seal all the corners. At what point of complexity of ball or box does this analogy break down and become misleading, I am not sure. But it seems to suggest the ball always has a chance of getting out. And we haven't even considered outside (from the 'real' world) influences drilling a hole.

Answer 2

One problem with these virtual worlds, whether or not they include our physical bodies, is whether such fantastic scenarios are in fact possible. For example there is a scientific hypothesis of "Boltzmann brains" spontaneously forming in outer space due to the statistical possibility in quantum physics being technically nonzero. But for this to happen the Universe must be infinite in either time or space, and a finitely small bubble of spacetime expanding at finite speed over a finite time, as science suggests ours to be, can never be infinite, so the scenario fails.

Your Harry Potter scenario is basically Neo or Morpheus in The Matrix, learning to reprogram themselves with amazing gymnastic feats, bullet time and so on. Human brains are highly plastic, especially when young. They reorganise themselves substantially through learning and during adolescence, ending up different from each other at a surprisingly macroscopic scale. A person blind from birth may re-use their otherwise useless visual cortex to develop spatial awareness. So if you plug a newborn babe into the Matrix, even in principle how is the thing going to know what inputs this particular brain needs to be fed? Is it really possible to map out all the subtleties of a given neuronal network simply by plugging in to its signalling? Artificial neural nets are notorious for learning stuff without the programmer's faintest idea about the algorithm it has built for itself, why should a human's biological neural net be more accessible to the Matrix? Right now, the evidence is that it just isn't, and probably can't be.

Create the whole personal entity in the machine and its capabilities must expand unimaginably. You are simulating not only the observable cosmos but every intelligence within it. What scale of Universe can supply the computing hardware for such an enterprise? Is such a vast but locally-connected Universe a genuinely plausible scientific hypothesis or is it as untenable as Boltzmann brains?

It is a particular feature of many logical systems that if you get one assumption wrong then, with a little ingenuity, you can obtain any outcome you want. Indeed, obtaining two contradictory outcomes is a standard way of demonstrating that some proposition is false. "I don't see why not" is just not good enough.

Answer 3

Is it safe to presume that the only thing that could make people suspect is a flaw in the matrix [...] ?

I'm going to take an easier version of the problem: is a flaw in the matrix sufficient to rationalize a simulation using the scientific method?

Taking Feynman's version of the scientific method:

1. Make a guess
2. Make predictions
3. Test predictions
4. Rinse and repeat until you can't prove the model wrong

Then 'proving' a simulated universe means invalidating any model that doesn't require a simulation. (I'm sticking to testable models, e.g. no thought experiment here - this is narrow, but it sounds reasonable to assume that the simulated universe is simpler than the universe the simulation runs in).

That in turns requires:

1. A simulation capable of modeling itself (counter example: a cell in Conway's game of life)
2. A decent model of the world in the absence of glitch (null hypothesis)
3. The ability to recognize a glitch as such (something that'd come out of a simulation inside the simulation)
4. The ability to predict glitches
5. The ability to validated those predictions

I believe at this point we need to assign intent to the simulation. Does it have easter eggs ("this is weird, the waveform of this proton spells 'coded by Bill' when you shine pink light on it")? Has the simulation a divine equivalent of a security manager (shut down the moment someone gets smart)?

Assuming a simulation allowing us to proceed, then we need to consider time: how frequent/dramatic are glitches that you can iterate your model before it's over? The scientific method doesn't do well attempting to prove miracles - or anything you can't really test.

Idea: rationalizing a simulation using the scientific method is only possible if it's buggy as hell?