Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

On whether we're living in a simulation

scottaaronson.blog

Post reply on HN
Print view View original

191–200 of 206 posts

Re: On whether we're living in a simulation

#191

Post by vidarh »

jncfhnb wrote: post #103
vidarh wrote: post #22

Earlier quoted context omitted.

That variant is simply Pascal's Wager. It's more interesting to ask whether, of we are in a simulation, it is possible to prove and/or find "escapes". All our experience with computational sandboxes so far is that preventing information leaks or escapes is extremely hard. Of course, we still have the problem that if we fi d nothing we don't know what that means. From my perspective, I'm on the side that it's a fun ph…

Pascal’s wager seems so… idk, unenlightened? If you start from a status quo of feeling comfortable with your obliteration, it sounds kind of ridiculous to ask god to grant you an eternal life of heaven. Have we not concluded that eternal life begets meaninglessness? Are we really imagining ourselves to be eternally pleasure seeking or revelries entities? Like what does that actually mean? It seems, frankly, more in l…

As an atheist, I don't know if I'd say I feel "comfortable" with obliteration. I just don't believe in any evidence suggesting an alternative. If I "woke up" after death and faced some god, you better believe I'd look for a way of avoiding misery.

The part of trying to "save" atheists by appealing to a chance for eternal life isn't the problem to me. If Pascal's Wager wasn't logically flawed, then it'd seem a small imposition to try to live accordingly for eternal life.

The fatal flaw is that this presumes that you know which deity to placate, and even worse that you know that acting a certain way won't actually piss off a deity instead.

And so while I wouldn't decline, I do think acting according to my personal integrity is the best choice I can make absent any evidence of a better course of action, on the basis that if I turn out to have been wrong and there is an afterlife, a deity I could respect would be one that values keeping your integrity over sucking up, and one who is cruel enough to condemn people to misery for not following arbitrary rules would be one I'd be miserable trying to serve anyway, and one worth spending eternity planning rebellion against.

Re: On whether we're living in a simulation

#192

Post by vidarh »

steve1977 wrote: post #137
vidarh wrote: post #47

Earlier quoted context omitted.

Depends what you mean by "go there", and whether they are in a simulation. Consider that if we're in a simulation, we're code, and an "escape" would involve getting out of a sandbox and perhaps being able to access their version of the Internet, and move around in it. In that sense we might well be able to "go there". Now imagine they're in a simulation, in which case their entire corporeal experience is code. There'…

> There'd be nothing inherently making it impossible for us to escape our sandbox and theirs and find a way to give ourselves bodies in "their" simulation in that case. But we would still be code and in a simulation. Just in a different simulation.

If we're code, and it for all practical purposes is indistinguishable from what we consider reality, what does it matter?

Though if we knew, and had a way out, we'd then have a choice to make if we're satisfied with that or want to try to see if there is another layer outside that again.

Of course, if we can't find an "escape" it isn't proof that there is nothing outside.

Re: On whether we're living in a simulation

#193

Post by vidarh »

Erem wrote: post #101
pavel_lishin wrote: post #83

Earlier quoted context omitted.

If we're just code being executed on some machine, we could be executed on a different machine - one that can move around in "their" world. Basically, getting uploaded into an android frame. (Assuming that their world would even be comprehensible to us; but again, as others have pointed out, so much of this is effectively theological...)

I mostly enjoy this hypothesis because I'm a developer. I like imagining The poor sap that has to deploy us. And what do their logs look like? What does their tech debt look like? "Hey yeah there's this emergent bug where bad things happen to good people. We would love to fix it but you know the intern wrote the implicated service in Haskell so we would have to rewrite it in Go and nobody has time for that this quart…

One of the most distressing parts of the simulation argument, is that I can't really get away from the idea that if we are in a simulation, the most likely simulation to be in would be some broken dev or QA build.

Re: On whether we're living in a simulation

#194

Post by vidarh »

steve1977 wrote: post #135
pavel_lishin wrote: post #83

Earlier quoted context omitted.

If we're just code being executed on some machine, we could be executed on a different machine - one that can move around in "their" world. Basically, getting uploaded into an android frame. (Assuming that their world would even be comprehensible to us; but again, as others have pointed out, so much of this is effectively theological...)

But we would still be code, we could not escape our "non-corporality" I guess. Or if you want to have a theological take, we could not become god(s).

If we are in a simulation there's nothing but code preventing us from being gods in every way that matters.

Who'd want a "real" existence any way, if you can be in a simulation where control of the simulation would mean you could change your entire physical reality?

If we are in a simulation it also weakens the presumption that there even is any such thing as a corporeal existence - maybe it's infinite layers of simulations.

Re: On whether we're living in a simulation

#195

Post by vidarh »

andrewflnr wrote: post #151
thwarted wrote: post #122

Earlier quoted context omitted.

Can entities inside the simulation get info about your CPU architecture? (From HAKMEM): Item 154 (Bill Gosper): The myth that any given programming language is machine independent is easily exploded by computing the sum of powers of 2. If the result loops with period = 1 with sign +, you are on a sign-magnitude machine. If the result loops with period = 1 at -1, you are on a twos-complement machine. If the result loo…

> path-dependent nature of our hardware architecture, not anything fundamental Anyway, there's no reason entities in a physics simulation would have such direct access to the primitives of the underlying computer, and trying to crash your host simulation does not seem like a reliable, healthy, or in any other way advisable approach to finding out if you're in one.

There's no reason to assume we will know how to prevent such direct access in a way that isn't sufficiently more costly that simulations without sufficient protections are more likely than those who are. It would presume simulations are limited to being run more often by entities willing to be exceedingly conservative and careful than by entities making mistakes that creates openings.

You don't need to try to crash your host simulation - trying to find a way to gain control to e.g. effectively become "gods" of the simulation, or find a way out, or just explore would all be other motivations.

If I thought we were definitely in a simulation and could find a way to gain information about the outside, you can bet I'd want to be part of an effort to do just that - if we crash it, who cares? We'd be gone and wouldn't know we messed up. Severe corruption would be far worse than a crash.

Re: On whether we're living in a simulation

#196

Post by PeterisP »

vidarh wrote: post #189
FartyMcFarter wrote: post #182

Earlier quoted context omitted.

That assumption is baked into the question being asked though. When we talk about X bits of storage it's generally accepted that this means being able to store any bit string of length X, not just compressible ones.

The question was: > What's the reason for the simulation to be smaller than the simulator? And the answer to that is that there's no reason. There's no reason to assume the state of a simulation isn't compressible, because have no knowledge about whether or not it is, and we do have plenty of knowledge of how to create simulations that appear complex but that reduces to easily computable sequences "from the outside".…

I'd (obviously? perhaps not) define the "size" of the simulation as the amount of information (i.e. uncompressible state) it contains, simply because for other measures of size many trivial simulations are infinitely large while having tiny incompressible information, so these other measures of size are not useful and not used.

So a more compressible i.e. more regular, simpler simulation is the smaller one, and the simulation still has to be smaller than the simulator, and we literally don't care about the compressibility or incompressibility or uncompressed size, because why would we, that's not information-theory information.

Re: On whether we're living in a simulation

#197

Post by steve1977 »

vidarh wrote: post #192
steve1977 wrote: post #137

Earlier quoted context omitted.

> There'd be nothing inherently making it impossible for us to escape our sandbox and theirs and find a way to give ourselves bodies in "their" simulation in that case. But we would still be code and in a simulation. Just in a different simulation.

If we're code, and it for all practical purposes is indistinguishable from what we consider reality, what does it matter? Though if we knew , and had a way out, we'd then have a choice to make if we're satisfied with that or want to try to see if there is another layer outside that again. Of course, if we can't find an "escape" it isn't proof that there is nothing outside.

If you knew there was a „higher dimension of existence“ but you also knew you can never be part of it, maybe you would rather not know it in the first place?

Re: On whether we're living in a simulation

#198

Post by staticman2 »

_yb2s wrote: post #183
staticman2 wrote: post #179

Earlier quoted context omitted.

In the sense an intelligent design advocate would also claim they can test "specific hypothesis" about god, perhaps.

Only if “god” is a computational or mathematical system that leads to clearly defined and testable predictions. That isn’t typically consistent with most religious concepts of God. I can surely imagine an intelligent design advocate trying to formulate a testable physical prediction inspired by theology, but if done rigorously, they are going to be disappointed in the result. And I’ll venture to guess that’s why you…

When you do your experiments and they don't show you are in a simulation, do you conclude that we are living in reality or do you conclude "Wow, this simulation has better code than I expected!" or "Wow, I guess the simulation creators invented better computers than we imagined!"

I compare it to religion because, well, I think you are claiming nonfalsifiable ideas are falsiable, and I don't think doing a bunch of math changes this or framing this in a way that sounds scientific changes this.

Anyway someone on substack said it better than me.

"any simulation can only be detected from the inside by performing a measurement and finding a discrepancy between the results and what would have resulted if the measurement had been performed outside the simulation. Without access to the outside of the simulation, there's nothing to compare against and no way to tell simulation from reality".

Re: On whether we're living in a simulation

#199

Post by _yb2s »

staticman2 wrote: post #198
_yb2s wrote: post #183

Earlier quoted context omitted.

Only if “god” is a computational or mathematical system that leads to clearly defined and testable predictions. That isn’t typically consistent with most religious concepts of God. I can surely imagine an intelligent design advocate trying to formulate a testable physical prediction inspired by theology, but if done rigorously, they are going to be disappointed in the result. And I’ll venture to guess that’s why you…

When you do your experiments and they don't show you are in a simulation, do you conclude that we are living in reality or do you conclude "Wow, this simulation has better code than I expected!" or "Wow, I guess the simulation creators invented better computers than we imagined!" I compare it to religion because, well, I think you are claiming nonfalsifiable ideas are falsiable, and I don't think doing a bunch of mat…

We can only experimentally test for very specific artifacts from specific types of simulation. If we don’t see them we can conclude that we’re not in that specific type of simulation, but that is all we can conclude, anything more is speculation.

I understand the point that you are trying to make, and the quote makes it clearly, but it does not apply in the situations I am talking about, where you can actually make and test a prediction. That’s why these specific testable hypotheses are written up by physicists and appear in arXiv or physics journals, and are not considered theology.

Indeed the general idea that we are in some type of simulation is mostly not falsifiable. At best, we could falsify only that we are not in a few specific types of resource constrained simulations that have predictable artifacts.

Re: On whether we're living in a simulation

#200

Post by andrewflnr »

vidarh wrote: post #190
andrewflnr wrote: post #87

Earlier quoted context omitted.

> All our experience with computational sandboxes so far is that preventing information leaks or escapes is extremely hard. I believe this is due to the path-dependent nature of our hardware architecture, not anything fundamental. x86 is not really designed with virtualization in mind. Theoretically, perfect airtight simulation is the primary ability of computers, and we're just too sloppy to exploit it properly. Ano…

> Can entities inside the simulation get info about your CPU architecture? Not really, not unless you really messed up (let's assume you don't make any gross concessions to performance over correctness). I would consider it magnitudes more likely to assume we would really mess it up than to assume there are no information leaks, and I think it's wildly unrealistic to assume we wouldn't make gross concessions to perfo…

The answer to both points: the universe exhibits, as far as we can tell, a high degree of correctness. We've already tested it in pretty diverse and extreme circumstances and found it consistent. Clearly if our universe is a simulation, the simulators are both willing and able to give a shit about correctness, enough that there's no reason they would use the shoddy engineering that plagues modern attempts at sandboxing. The cut-rate simulations humanity is or will be capable of anytime soon are irrelevant.
Post reply on HN