Tuesday, June 28, 2011

How Would You Prove That We ARE in a Simulation?

There's no Matrix you dummies! This guy is just a really clever Martian who can do good special effects. Is there a way to prove otherwise?

Most discussion of simulation arguments, from Descartes to Bostrom, has focused on trying to prove that we're NOT in a simulation, concerned that we don't see the world (enough) as it is. There is some profit in pointing out that "simulation" is ill-defined, and because of the way our senses filter and bind and create our experience, the distinction between a simulated world and the "real world" is somewhat arbitrary. Therefore, perhaps a more interesting question is what it would take for us to be convinced that we ARE in a simulation, if one of the simulation programmers appeared in our midst, and prove to us that this IS a simulation. What kinds of experiences or information would the revelations of the simulation programmers have to contain to make us think that we're in a simulation?

1) First trick would be for the simulators to demonstrate violations of physical laws at their will. Then again, an apparent violation of physical law could just be the result of advanced technology that you don't understand. "Look! Gravity reversed itself locally! Look, we changed it so you're surviving in vacuum just fine, and now I just ran time backwards and separate causes and effects, but only on your street! Obviously you must be in a simulation!" Right? Certainly such behavior of the world would be inconsistent with our current understanding of how things are supposed to work, but does that automatically mean it's a simulation? Maybe it's just aliens with Clarke's Third Law-level technology, and they're screwing with you. And for that matter, shouldn't we expect that there are limits on what even a simulator can do without the simulation no longer functioning? The source code of World of Warcraft is certainly not robust to any possible manipulation you could dream up.

2) They could show you (a la Thirteenth Floor) where the simulation breaks down at some boundary (scale, time, space, cognition). But would that prove it? We know already that there are such boundaries (uncertainty principle, the Big Bang, cognitive closure at least in the trivial sense of limited working memory), yet no one has seriously argued that this means we're in a simulation.

3) They could tell you what you're thinking, or tell you a 20 digit number you wrote on a piece of paper in your desk (and didn't tell anyone), if they really have access to the entire code of the simulation (and they could just stop it and do a CTRL-F). Then again, how do you know they're not aliens with awesome fMRIs or paper-reading X-ray machines?

4) They could do something fatal to you, and then have you re-awake. Again, maybe they have great technology! Even if they do something that would physically destroy your nervous tissue like drop you into a volcano and you wake up, maybe you're just waking up for the first time with memory implants (of a life up to and including falling into a volcano), or maybe there's a way to recover consciousness even after such total tissue destruction that we don't yet understand.

5) They could demonstrate continuity of experience; that is, one second you're walking around what you thought was 21st century Earth, the next you're being unplugged from the computer in your vat. How do you know the vat isn't the simulation? Even if there's no simulation technology being used here, a lot can be done to manipulate perception of time; and for that matter, maybe you're a clone with memory implants, and you're really waking up in that vat for the first time, with false memories.

So: are there tests that could even in principle separate simulators who create our universe, from mere clever aliens who co-inhabit it with us?

[Added later: Michael Shermer makes an interestingly analogous argument when he states, in a variant on Clarke's Third Law, that a sufficiently advanced alien would be indistinguishable from a supernatural being. In a predictably miffed answer, a counter-arguer misses some points, saying that the absence of recognizable modern telecommunications devices from the Bible shows that we can distinguish gods from aliens. Apparently the counterarguer insists that aliens use cell phones.]


TGP said...

Assuming mind-reading, time-stopping alien overlords with inscrutable reasons for creating an elaborate simulation isn't too different from believing in that a deity put humanity on the earth to test our moral fortitude, is it?

Having an adversary with the ability to rejigger any experimental outcome to conform with expectations pretty much much scuttles the scientific approach to gaining knowledge.

Also, unless you're central to the simulation, another human conceit, what's to keep the overlords from just blipping you out of reality if you start learning too much? "Did you see this blog post? Bloop! It's deleted and so is the blogger."

I think the better question to ask is why would an overlord bother with a simulation?
If they're mind-readers, they don't need it to gain information from us.
Conquerors with such abilities would really be taking the slowest route to defeating humanity.
Scientists studying our behavior would be doing it in a experimental setting with way too many uncontrolled variables.
If they're sadistic thrill-seekers messing with us for their own entertainment, we're just plain screwed.

Of course the real clever aliens overlord invaders would just put some silly special effects into the real world and watch the humans commit mass-suicide to 'escape the simulation.' "Hey, Flizz-Nar, look: free planet."

Michael Caton said...

You bring up exactly the two questions here that should make us ask about the relevance of this whole thought experiment, and that's exactly what I'm trying to demonstrate by reversing it. The first point (which you've made previously) is: why would an entity capable of doing things like this CARE? Any hand-waving explanation we can come up with also begs the question of why they wouldn't just wipe us out; incidentally, also the silliest plot hole in the Matrix (batteries? Come on. Algae wouldn't be better? Or if you need vertebrates for some reason, then whales or brachiosaurs?) Second is the question of wrecking all epistemology - whether their superpower is putting people in a fantasy world or just destroying all possible cause-effect expectations with magic technology, the effect is the same - the nature of the world and oneself in it is unknowable, except for the fact of one's existence.

To me this is a reductio ad absurdum of the whole enterprise. It's not a question of whether we manufacture our experience (we absolutely do, by for example knitting together visual and tactile input as coming from a single object but excluding others from lack of input instrument or just poor pattern recognition). What people consistently seem to be interested in is whether there are elves changing the set pieces to trick us, and that goes back to why they would care.

Anonymous said...

I think we both agree that the motivation to create an entire world simulation perfectly is silly.

Your original question about creating a TEST (an experiment to falsify the hypothesis that world is a simulation) is still unanswered.

I think that to avoid adolescent "the aliens are laser-proof so your lasers don't work" arguments, you must assume no 'supernatural powers' on the part of the overlords.

They can't read minds (but they could be very intuitive and perceptive.)

They can't time travel and dick with causality (but they could have much faster reaction time than a human.)

I think you have to start your experiment assuming no gods, else you have to prove that the devil didn't hide dinosaur bones. We know that following that path pitches the scientific method under a truck brimming with bullshit.

I think that you can use certain test results to make educated assumptions about potential overlord motivations. e.g. Assuming that mind-reading overlords defeat the test also implies that the simulation is not intended to interrogate us. A test that determines that the simulation would be net endothermic implies that we're not being used as batteries.

Now, the even cleverer alien overlords might have put us into a simulation to see how creatures figure out if they're in a simulation because they themselves are looking for better tests to determine if they are in a simulation...

Jonathan Colvin said...

i have argued that the incompatibility of general relativity and qm is evidence that we are living in a simulation. this is exactly what we might expect to see if the simulators are taking short cuts. for example in video games different objects frequently have incompatible physics models. so there you go: direct evidence that we are living in a simulation. now we have discovered the Truth. oh crap, the stars are going out....

Michael Caton said...

Jonathan, very Arthur C. Clarke of you but the Tibetan monks are unfortunately busy with more than computer outputs at the moment. But more seriously - your observation (the apparent incompatibility of QM and relativity) could be questioned as follows: why does inconsistency imply epistemological segregation? In fact doesn't Goedel say that an inconsistent system is what you should expect if it's complete? And why aren't the simulators smart enough to hide such an inconsistency, if they care at all; why isn't it something more basic, like a breakdown in experience? Eventually I got to the point where the simulation argument was more of a word game, but I think really the problem is not that we can't ascribe motives to the simulators, but rather no one rigorously defines what a simulation is. (Whether people agree, and the remaining question(s) are intersting, is another matter.)