If it is possible that to make a simulation which matches our experience, then it is likely possible to make an unbounded number of such simulations. Thus, if such simulations are possible, it is vanishingly unlikely that we are executing directly on the underlying substrate.
If it is not possible, then, well, it's not.
So to a good approximation, the question "do you believe it is more likely than not that we are living in a simulation?" is equivalent to the question "do you believe that a simulation of the phenomenon you have observed is possible?"
And... well, sure, there's not a strong reason to think it's /impossible/, based on the evidence available to us. So, yeah, more likely than not.
Another way of phrasing this: Do you think it's more likely than not that there's some physical law, as yet discovered, that makes high fidelity simulation impossible? Such a law is certainly imaginable (limits on information density, magical-ness of souls, whatever); but if you don't have a reason to believe such a law is likely, then you probably believe we are more likely than not in a simulation.
There’s a counter argument from my physics undergrad brother that I found convincing.
We already do have a law of physics that is relevant here. We know that the information capacity of space is finite and fixed. A centimeter of space can only store so much information before it becomes a black hole. That means that to build a simulation in our universe you can only ever subdivide a fixed pie of information. That means the more relevant thing to ask is if a quantity of information is more likely to exist in the base reality or the simulated one. Because we have to assume that the base reality is not carpeted over with simulation super computers it seems safe to assume a random bit of information is more likely to be part of the base reality rather than a simulation all else being equal.
I think the idea of the universe being a simulation is just more fun
> That means the more relevant thing to ask is if a quantity of information is more likely to exist in the base reality or the simulated one.
Well, what quantity of information? It seems relatively unlikely that one would bother simulating an entire universe for billions of years at this level of fidelity; what's the point? On the other hand, the quantity of information needed to simulate your current experiences, including the experience of having memories, is probably in the megabytes; human bandwidth just isn't that high.
The same logic suggests that, even if you discount simulation for reasons of faith or whatever, Boltzmann brains are worth considering. The idea that the experience you're having right now of reading my precious prose is worth keeping a universe running, or even a large-scale simulation, is a bit self-centered of you, isn't it?
I think I'm very suspicious whenever infinities come into probability calculations, because I think human institution's break down when it comes to infinites. What I like about pointing out that the information content of the a cubed square foot has a ceiling is that it implies that there can't be an infinite number of simulations except in the way that there is an infinite amount of space (at least without throwing out that bit of physics) and an infinite amount of space is something I feel more capable of reasoning about.
Unless we think most information exists in simulations than if you take a random bit of information out of the universe it's more likely to be associated with the root existence's goinings on than it is with any kind of simulation. This works because it's an argument about that bit of information, not anything else. I think you could push back and say that really actually information isn't what's important and that actually even if we think that most information won't be part of a simulation most consciousnesses will be, but idk, that seems suspect, we don't have any reason to think that. Certainly most simulations we do now seem not to include consciousnesses.
I think the idea of Boltzmann brains fall apart because in that sea of possibility space it actually seems much more likely for the seeds of a universe to form than a complete brain full of consistent memories of writing 3/4th of a post on a randomly generated website called hacker news. I think it's just another illustration of the problems apply infinity to probability.
Also worth considering that simulating the universe for billions of years could run on the equivalent of a raspberry pi in some higher order universe. It just seems huge to us because we are living in it
Who ever said that the computer doing the simulation is working under laws of physics that are identical to our own? There are plenty of sets of physical laws that would support such a simulation, and a great many of them are entire computational complexity categories stronger than our own - consider, for example, a reality that is truly real-valued, rather than approximately real-valued with some semi-bounded probabilistic error. And then it's not "tiling some large fraction of reality" that's needed; if you go up a few alephs you end up with computers where our experienced reality falls out of people noodling and with stuff like "run all discrete programs with length less than N bytes" the same way we do things like compute the error of the fast inverse square root for all 32 bit floats.
It is a counter argument against that we can build a perfect simulator of our universe, but not a good one against that our universe is simulated.
In fact if I were to build a simulator, I most likely have to design a mechanism to prevent its residents from observing beyond a certain micro scale due to limited cpu/mem resources and laziness to implement all the details. Tiny black hole is a good mechanism to reduce resource consumption when simulating a fixed volume of this universe. Imagine living in the world of Minecraft, the minimal unit is a block. Trying to look inside of it yields nothing. All physically meaningful characteristics are described by its surface. In our universe this looks very much like a blackhole.
I you're really maybe missing the simulation business w.r.t. computing, etc. It's like cells in a dish debating the world outside them without any context of it.
It's pretty trivial to have this reality be a 'container', as it were, within some superset reality, regardless of the information density of said contained reality.
Think of it like a human body sort of being its own unique, low-entropic thing.
I personally like thinking that in one sense, we're all a bunch of bits of fleeting consciousness on the edge of some fractal of realities, and the 'substrate' is simply possibility itself.
After all, the mandelbrot fractal is drawn by which series terminate, and often how long it takes them to terminate.
Why not us? Why can't we be one infinitesimal reality in the entire sea of possible realities, held together by the fact that our reality happens to have coherent rules that allows it to exist as some possible state in the greater states of some amorphous soup of possibility?
Just a thought. I know it may be more out there for some.
I think it's fine to imagine that there might be a root universe that doesn't share this one's restrictions on information density, but we don't have any reason to believe it and without believing it the argument that there should be near infinite simulated realities starts to break down
I think the problem with that assumption is that a simulation would have to represent every bit of information under simulation. From the perspective of a human being, most of the universe could be simulated only at a very high level of abstraction - there’s no need to maintain the state of every subatomic particle in a star in a neighboring galaxy, for instance.
The argument presented above rules out a infinite chain of simulations inside simulations. It may still be possible to have a finite chain of simulations, where the fidelity successively decreases. At some point down the chain the simulation can't have enough capacity anymore to contain conscious beings, which wonder, whether they live in a simulation.
I think that maybe brings in a question of if you should assume a piece of information say the existence of this 't' is uncorrelated with being simulated or not. If not than I think it can be shown that it's probability of being part of a simulation is just equal to the fraction of information in the universe that's dedicated to simulations which I think most people would imagine is less than 50% given the knowledge that that space has a max information capacity. I do admit that you might think information associated with consciousness is more common in simulations than out, but idk, also maybe not. Personally I find the particular set of facts I'm presented with to not seem particularly simulated. My life isn't particularly interesting and I think most information in the kinda of limited resolution simulation you're imagining should have unusually high value of some kind.
This would only be true if the arrow of time only moves toward the future. If it also moved backwards, even in a greatly limited manner, you wouldn't need to track everything. You'd basically have infinite storage for a finite amount of data.
The second law of thermodynamics creates the arrow of time, only pointing to the future right? And aren't all other laws of physics time symmetrical? What explains these laws co-existing better than it being part of a simulation?
The simulations aren't nested, they're simultaneous and all under one hypervisor! Maybe one of my old Qubes laptops is running the second layer of the simulation - I should probably turn it back on.
I don't understand that but feel confident that if there were anything to understand I would understand it, and it makes me want to quote Billy Madison.
Edit: I don't understand the argument that if many good simulations exist, we must be in one, either. It seems bizarre to me. So having a bit of an odd suggestion about information density is as good of a response as any.
If we're in a simulation, wouldn't that simulation run in a universe that may have different capabilities than our simulated one, such as a different maximum density or even a completely different set of physical laws that governs it?
Sure, but then we're back to knowing nothing about it. I think what many people find compelling about the simulation argument is that it claims that we can just follow forward our laws of physics and see that we are more likely than not living in a simulation, but I think that's not true
> If it is possible that to make a simulation which matches our experience, then it is likely possible to make an unbounded number of such simulations.
Why? This seems to me to be the weakness in the argument.
Of all the universes in which it is possible for a technological species to evolve and create a simulation of our universe, what’s the probably of said simulations having a given incentive or conducive cost/benefit ratio for said species?
Theoretically this could range from “can only do one once before our budget runs out and we move on” to your “unbounded” claim. But with what distribution?
This question seems fundamental and so reduces the initial question to a more complex one than what you pose: is it possible and if so how plausible?
Unless I’m missing something, leaping over this factor, as seems to be the mainstream approach, indicates to me that some techno-utopic-transcendentalism bias is at play.
> Unless I’m missing something, leaping over this factor, as seems to be the mainstream approach, indicates to me that some techno-utopic-transcendentalism bias is at play.
Yeah, at some level. The universe that we observe doesn't seem to be set up, in the current epoch, to have very tight resource limits other than time. Energy is plentiful. The main cost of anything is opportunity cost. Sure, simulating a universe might cost /our/ civilization so much compute capacity that we have to choose between than and advertising Christmas sales, and that's clearly no choice at all -- but it only takes a small percentage of similar planets to hold civilizations that are just slightly more advanced to make this a reasonable freshman project, and at that point, plentitude creeps in again. Basically -- and I agree this can totally be interpreted as utopianism -- it's hard to imagine a line between "it is possible to build a computer capable of computing <X>" and "it is expensive, on the scale of reasonably-advanced civilizations, to build a computer capable of computing <X>."
> it's hard to imagine a line between "it is possible to build a computer capable of computing <X>" and "it is expensive, on the scale of reasonably-advanced civilizations, to build a computer capable of computing <X>."
Well it’s not so much about whether there is a line but what the probability distributions are and whether it continues to make sense to think that of all sentient beings the majority are likely simulated.
And while I personally get the argument you and the parent post make, I think it’s worthwhile highlighting that it’s likely not a simple matter of whether it’s possible and that the biases/utopianism that facilitate making that leap are also factors and worth making explicit.
Personally, I find it hard to conclude that a sufficiently advanced civilisation would necessarily be concerned with running so many simulations when there are probably a number of things they could spend time on that we can think of and many more we can’t because we’re not that advanced.
I think your summary of the question conflates a couple very different things. One of those things is the possibility of a simulation like the matrix, where you could attach computer IO to a human brain with the brain unable to tell the difference. But "we are living in a simulation" requires that technology be possible, and it also requires that at some point everyone did that, and they never left, and that something in the real world causes all their offspring to get wired in young enough that we don't notice it happening at any point in the process of pregnancy, birth, and infancy.
I don't see any reason to assume it's entirely impossible to make a computer system that provides brain IO indistinguishable from the real world. It'd obviously be very far beyond us, but it seems possible that a sufficiently advanced computer could manage it. But accepting that doesn't mean I have to accept that enough people did it to establish a population, they did so permanently, they forced it on their offspring (conceived both in the real world and simulation), and they never told anyone or left clear signs in the simulated world.
Alternatively it could assume that we are ourselves simulated, just programs unaware that we're programs. But that leaves many of the same questions (who did it, why keep it going forever, etc). We currently could dedicate all of humanity's exaflops of computing power to Monte Carlo simulations of Snakes and Ladders, but why? I don't think there's any reason to step from "theoretically it's possible" to assuming any amount of likelihood.
I always took the simulation to mean “society”, nation state norms, gossip about one another, elites… truisms of society being “facts” like GDP, not facts like speed of light.
Like in the “clearly people doing the work is why potatoes are on store shelves, not due to the shareholders of Ore-Ida, which is hallucination.”
There is always the option that humanity eradicates itself into oblivion before the technology for building a simulation has been developed.
Looking at how green/co2 certificates work, looking at politics and misinformation, looking at escalating wars out of stupidity, looking how many countries have been far-right-winged lately into Sharia law,...I think that's the far more likely option.
Humans are petty, humans are irrational, humans forget too quickly.
Always bet on humans acting like psychotic apes wanting more bananas even when their belly is so stuffed that it almost explodes.
Whether you want to admit that this is how the planet works or not doesn't matter. In the end, right wing populism always wins because they bet on stupidity and irrational beliefs, not on compromise and rationality.
If it is not possible, then, well, it's not.
So to a good approximation, the question "do you believe it is more likely than not that we are living in a simulation?" is equivalent to the question "do you believe that a simulation of the phenomenon you have observed is possible?"
And... well, sure, there's not a strong reason to think it's /impossible/, based on the evidence available to us. So, yeah, more likely than not.
Another way of phrasing this: Do you think it's more likely than not that there's some physical law, as yet discovered, that makes high fidelity simulation impossible? Such a law is certainly imaginable (limits on information density, magical-ness of souls, whatever); but if you don't have a reason to believe such a law is likely, then you probably believe we are more likely than not in a simulation.