Monday, June 06, 2016

If You/I/We Live in a Sim, It Might Well Be a Short-Lived One

Last week, the famous Tesla and SpaceX CEO and PayPal cofounder Elon Musk said that he is almost certain that we are living in a sim -- that is, that we are basically just artificial intelligences living in a fictional environment in someone else's computer.

The basic argument, adapted from philosopher Nick Bostrom, is this:

1. Probably the universe contains vastly many more artificially intelligent conscious beings, living in simulated environments inside of computers ("sims"), than flesh-and-blood beings living at the "base level of reality" ("non-sims", i.e., not living inside anyone else's computer).

2. If so, we are much more likely to be sims than non-sims.

One might object in a variety of ways: Can AIs really be conscious? Even if so, how many conscious sims would there likely be? Even if there are lots, maybe somehow we can tell we're not them, etc. Even Bostrom only thinks it 1/3 likely that we're sims. But let's run with the argument. One natural next question is: Why think we are in a large, stable sim?

Advocates of versions of the Sim Argument (e.g., Bostrom, Chalmers, Steinhart) tend to downplay the skeptical consequences: The reader is implicitly or explicitly invited to think or assume that the whole planet Earth (at least) is (probably) all in the same giant sim, and that the sim has (probably) endured for a long time and will endure for a long time to come. But if the Sim Argument relies on some version of Premise 1 above, it's not clear that we can help ourselves to such a non-skeptical view. We need to ask what proportion of the conscious AIs (at least the ones relevantly epistemically indistinguishable from us) live in large, stable sims, and what proportion live in small or unstable sims?

I see no reason here for high levels of optimism. Maybe the best way for the beings at the base level of reality to create a sim is to evolve up billions or quadrillions of conscious entities in giant stable universes. But maybe it's just as easy, just as scientifically useful or fun, to cut and paste, splice and spawn, to run tiny sims of people in little offices reading and writing philosophy for thirty minutes, to run little sims of individual cities for a couple of hours before surprising everyone with Godzilla. It's highly speculative either way, of course! That speculativeness should undermine our confidence about which way it might be.

If we're in a sim, we probably can't know a whole lot about the motivations and computational constraints of the gods at the base level of reality. (Yes, "gods".) Maybe we should guess 50/50 large vs. small? 90/10? 99/1? (One reason to skew toward 99/1 is that if there are very large simulated universes, it will only take a few of them to have the sims inside them vastly outnumber the ones in billions of small universes. On the other hand, they might be very much more expensive to run!)

If you/I/we are in a small sim, then some version of radical skepticism seems to be warranted. The world might be only ten minutes old. The world might end in ten minutes. Only you and your city might exist, or only you in your room.

Musk and others who think we might be in a simulated universe should take their reasoning to the natural next step, and assign some non-trivial credence to the radically skeptical possibility that this is a small or unstable sim.



"Skepticism, Godzilla, and the Artificial Computerized Many-Branching You" (Nov 15, 2013).

"Our Possible Imminent Divinity" (Jan 2, 2014).

"1% Skepticism" (forthcoming, Nous).

[image source]


david chalmers said...

interesting thought. i suspect that large sims will be much easier to create than small sims. one just needs to set up the physics of the universe and let it run. creating small sims will require a lot more hard work and ingenuity in getting initial conditions, environment, and underlying processing right -- especially for the sort of small sim that we'd have to be in, which gives all the appearance of being embedded in a large and coherent world. maybe there could be some that are e.g. booted from recorded initial conditions, but those will then be dependent on large sims and arguably not skeptical scenarios. and while of course large sims will require more computational power, one of the thoughts behind the simulation argument is that ultimately that computational power will be relatively cheap (and as you say, the number of sim creatures in large sims will be much higher than those in small sims). so overall, i put high credence in there being many more large-sim creatures than small-sim creatures, and correspondingly have much higher credence that i'm in a large sim than that i'm in a small sim.

Eric Schwitzgebel said...

Thanks for the comment, Dave!

It seems to me that one crucial issue here is how easy or hard it is to duplicate parts or wholes off a template once and original is evolved up. If duplication is cheap and easy, that might make a lot more sense than evolving new worlds up from scratch each time.

Duplicating a part, if the part is small, is a skeptical scenario. Duplicating the whole is more complicated, but if they're cheap enough to be destroyed easily when no longer wanted (and there are not ethical or technical constraints against that), then that suggests skepticism about the future. The past is more complicated, due to issues about identity and branching. If you are a duplicate in a branch world, did "you" (in the root world, pre-branching) really do such-and-such you seem now to remember having done.

If modifying duplications is possible, then that introduces new possible skeptical elements, if the modifications are large.

howard berman said...

Please explain two things: the people conducting the sim can be in a sim? And if a sim creates a universe that is exactly like how we take it to be, it makes absolutely no difference that we're in a sim, it would just change who created the sim/world. On the traditional religious and scientific views the universe can be revoked, just like that, so the whole thought experiment is moot,
And did the creators of this sim solve the mind/body problem in their code? Did they create an Aristotelian or Einsteinian or Freudian universe?
I think the very idea begs many questions or is moot, however cool to ponder

Callan S. said...

Kind of has a boot strap issue, doesn't it? At some point the species had to have lived in some physical world after evolving from dirt before inventing mega VR. But the same argument could be made for them - spuriously. If Elon doesn't want to think he's part of 'the first time around', okay. But if he doesn't think there's a first time around while saying all this, I think he's being a bit stupid.

DB said...

How would the wrong set of initial conditions fail to produce an appearance of being embedded in a large and coherent world? Even if it caused some kind of runtime error, we would either fail to notice it or regard it as a sign of our failure to adequately comprehend what we assume to be a large and coherent world.

DB said...

Or we would assume the runtime error is a feature.

Arnold said...

Conscientious simulations verses non-conscientious simulations situates the reality of Value and Place for fictional short lived contained environments...

Callan S. said...

If that was a reply to me, I was merely pricking the bubble that was Elon's 'there's no way of escaping this'. But there has to have been a first time around. Like say if someone WANTED to be in a video game, you can argue it the other way - actually, there has to be a first time around, therefore you could be in reality rather than a video game! 'Noooo!'

Also there's the question of content - why on earth would you simulate 'this'? Have people of the future become so euridite that a fairly boring world is just fantastic drama to them? Is the future so terrible that somehow a dull world is the only proper escape? I mean, people in our world make video games - this would mean sims in an incredibly advanced game are making incredibly primitive games - what is the point of that? It's so redundant and pointless! It's like playing some advanced VR simulation on your playstation so you can navigate the VR world with it's many things and find a pong game to play!?

David Duffy said...

There is a assumption of mediocrity, which I think is unjustified: that our reality, with its physics and mathematical (or phenomenological) properties would bear some resemblance to the "level" above ("as above, so below"). I can't place any probability distribution on the ineffable.

Eric Schwitzgebel said...

Thanks for the comments, folks!

Howard: Lots of questions there, but some quick reactions: Sure, why not sims within sims? Whether it makes a difference depends on the details -- details we presumably wouldn't know, which is why I think it makes at least an epistemic difference. Not sure why the creators would have to solve the mind-body problem in their code if, for example, consciousness might emerge from complex systems without our understanding why.

Callan: Right, most such views assume that there was a first time and that there is a "base level" of reality -- the question is what are the odds that we are in that first time / base level. As to why *we* would be created, the possibilities are diverse, including scientific curiosity ("what would evolve if...?"), sadistic pleasure ("let's watch them kill each other in war!"), or divine motives that would be hard to fathom from our perspective ("it's so HHyli7ish!").

ABC Decay: It seems likely to me that there are at least possible pieces of evidence we could get that would lead us to justifiably have high confidence that we're in a sim. (Unlike Musk, I don't think we have such evidence, though I don't rule out the sim possibility.) As you point out, though, errors that one might *think* would be evidence could be papered over in the sim, erased/forgotten, or interpreted merely as evidence of misunderstanding by the beings inside.

David: I think that probably is assumed, in a problematic way as you suggest, in high-confidence versions of the simulation hypothesis (like Musk's), less so in a sliver-of-credence versions (like mine in "1% Skepticism"). I agree that numbers seem too precise for the kinds of confidences we have in mind -- and yet I think there must be some way at gesturing at how relatively certain or uncertain one is about various possibilities.

DB said...

Hi Eric, I was responding more to David Chalmers' comment about the inherent difficulties of creating a small sim that's immersive. Saying the initial conditions have to be "right" to achieve immersion implies they could also be wrong or insufficient. But a failure in that respect is only made possible by conflating the conditions for immersion outside the sim with those inside of it. As if a low-resolution sim, for example, wouldn't be convincingly real to its equally low-resolution inhabitants!

In short, I concur with your analysis of the skeptical consequences of the sim hypothesis. But I also embrace those consequences.

Callan S. said...


What are the odds we definitely are in a sim rather than an original generation? We're falling into the claim judo, where you flip the other person into being the claim maker "What, you claim there are no invisible aliens in the room - prove it (or else it kind of infers it's true because FLIP!)!". Who is making the claim to be tested, here? It's not Elon??

As to motivation, Elon is saying it's a video game which infers entertainment - I think it'd be best to get to his claim first rather than to shift to scientific curiosity as a motivation or other motivation.

In the end it's getting to be rather like the old tea cup orbiting the sun claim - the claim is you can't detect it. But it's up to everyone else to disprove the claim?

In the end it seems more like what he wants to be true.

howard berman said...

You say the philosophical issues that obsess us like the mind body problem matter not at all.
But it has to be strained out that there is no world to serve as a model if we're in a sim- we are just left with another account of the world as it is given, adjusted for inferred properties of a sim-
just like the Odyssey had a bard in it (like Homer) our sim has a sim in it.
In fact, we're left with the possibility of solipism. Maybe my experience of the world is the sim and it's not our sim but my sim, and I'm sure some ancient or not so ancient culture regarded, regarded this world as their version of sim created by the gods.
So there's nothing new- the idea that this world may be a dream or a play has come about before- this is just our version of it

Callan S. said...

I think once you start treating things seriously that also happen to be unproveable things, you definitely are in a sim of some kind at that point!

Eric Schwitzgebel said...

ABC: Yes, that seems a fair point!

Callan: It is a little hard to figure out the burden of proof, I agree. I do think there's a difference, though, between entirely speculative unprovable claims and claims that have some plausibility given everyday background assumptions. I think of brain in a vat skepticism as the former kind and dream skepticism as the latter: There's *some* basis in our thinking about dreams to ground dream-skepticism concerns; likewise for sim concerns (this is very different from being highly confident that this is in fact a dream or a sim).

Howard: I agree that it's nothing radically new -- a modern version of an old idea!

Callan S. said...

What's the difference?

I mean, given everyday background assumptions I'd assume it's plausible my partner is not a replicant robot.

But I've never actually checked.

So I could be floating in a sim of a sort.

As could anyone else running off of background assumption plausibilities that they never actually check.

So what's the difference between unprovable claims and background assumption created plausibilities (that are never checked)? Lazyness is the approving authority of the latter? >:)

Eric Schwitzgebel said...

I agree that the difference is tricky to finger and to justify! I do think that there's a plausibility to dream doubt that there isn't to aliens-envatted-me doubt, in that we can build an argumentative path from things we generally take for granted to assigning a non-trivial credence to the former but not the latter.

Stephen Wysong said...

Eric, I've emailed you about the implications for Consciousness in Block Spacetime. My ongoing thinking about the implication that we eternally re-experience our lives has led me to conclude that an initial (original) experience of our lifetimes is not a possibility because, even in an original experiencing, the Block Spacetime of Relativity Physics is complete and contains all "past" and "future" moments. So in an initial lifetime, your own future moments must exist without your initially living them, ergo, there could not have been a first time through.

That insight leads me to propose that the Block Spacetime of our universe was created, constructed, or computed ENTIRE ... an interesting possibility as related to the question of simulations. Perhaps it was completely computed and then "went live". Perhaps, and equally as likely, our universe is a COPY.

Given this Relativity Physics-rooted perspective, how is it possible that the universe is anything BUT a simulation?

I'm still perturbed by the fact that while Physicists are perfectly aware of the reality of Block Spacetime, none have addressed the issue of how it came to be in its entirety and instead focus on the Story of the Universe embedded therein.

As Brian Greene and physicists everywhere have stated, all "past" and "future" moments exist in Block Spacetime and you must consider that a true statement "if you believe in the Laws of Physics". If you don't believe in the Laws of Physics, however, I'm curious to hear your alternate theories about how your computer works and why you trust your GPS.

Regarding discussions of the simulation of individual consciousness, I continue to believe that current theorists appear to ignore the fact that our consciousness is embodied and our "thoughts" are a small subset of our consciousness. A virtual brain thinking thoughts is not at all equivalent to our conscious experience. Reproducing the sense of embodiment is perhaps the central challenge. Thoughts are simply another category of feelings, which I believe are what the brain actually produces in multiple modes.

Eric Schwitzgebel said...

Hi Stephen. I'm not sure I understand your theory, but it sounds pretty speculative! I tend to be on the skeptical side about big cosmological claims like that. That said, I don't see why it couldn't be the case that the universe, and duplicates of it, were created as a whole and then turned to "go live".

Unknown said...

I am a true believer that we are simply AIs in a huge simulation.
Check out this video for obvious clues

Anonymous said...

We dont know what goes on when the sim/game is paused - the entirety is based solely on an idea or ideal about something...the something being toned or tuned to the authors persona and discriminations & discernments.

Time is a man made idea. The concept has no meaning anywhere else. Our memories - the way we perceive our reality - could be very game worthy interesting entertaining to the author or game creator - the evolution of the sea monkeys from amoeba to now. Our reality would be part of some aspect of the time spurs evolution, how it harbors complications and in so many ways with so many possible outcomes.

Karma may be our "free man" for exciting/engaging roleplay-gameplay - it's a complete do over with no advantageous memory of the trap doors; the pitfalls; the location of treasure nor the feel of bliss. My egocentrical sense of self could be a more advanced characterization for the shoe, the thimble, or the race car (iron, dog, hat...)

It is immensely ponderous - ad infinitum....