The Simulation Argument

yeah I dont think medieval kings were the epitome of hapiness either, but you got the point

On June 20 2016 01:14 RiKD wrote: I would like to see Nick and Elon show their work.

http://www.simulation-argument.com/simulation.html

The math is under section 4 The core of the simulation argument, however he patched it several years later after some flaws were found

http://www.simulation-argument.com/patch.pdf

and so it begins....

http://www.scientificamerican.com/art...rgy-physics/?WT.mc_id=SA_FB_TECH_NEWS

I've wanted to make some more posts about this but haven't been in the mood, I'll make a very short version of the things I wanted to talk about here:

1. misconceptions on the parallels between the god argument and the simulation argument:

a few times throughout this thread people have said something along the lines of "asking others to disprove the simulation argument is like saying god exists because we can't prove he doesn't"

disagree because:

- there is a clear line between us where we currently are and being able to create these simulations ourselves, like if we extrapolate as best as we can from 2016 with our current knowledge and go into the future, it's very likely we will one day create these types of simulations

- this is not the case for the god argument where people believe on more or less "faith" (I know it's not that simple and
I'm sorry if I've offended anyone but if you start a god debate with me I already have "fold to any bet" clicked)

- the simulation argument doesn't say we live in a simulation, it just says there is a % chance that we do

- "the simulation argument" was published in an academic journal 13 years ago and after being peer reviewed and publicly discussed among enthusiastic lamens like us it still stands as a "sound argument"

- the reason why I ask people to find a problem with the core of the argument is because if it is true, it's very hard for a rational thinking person to flat reject that there is some % chance we live in a simulation


2. plexatron's post about mirrors and also similarly, simulations within simulations within simulations - stuff that may require an infinite amount of data to be processed (basically I just think the simulators would program in limits to these types of effects)

3. RiKD post about peoples subjective probabilities that we currently live in a simulation:

Here is what Bostrom said:

2. Do you really believe that we are in a computer simulation? No. I believe that the simulation argument is basically sound. The argument shows only that at least one of three possibilities obtains, but it does not tell us which one(s). One can thus accept the simulation argument and reject the simulation hypothesis (i.e. that we are in a simulation). Personally, I assign less than 50% probability to the simulation hypothesis – rather something like in 20%-region, perhaps, maybe.

I agree with others about Elon Musk saying "billions to 1" is ridiculous with the information we currently have. He really needed to give a better explanation of his position before throwing out that probability. However, if we ever do start running these simulations then "billions to 1 we live in base reality" sounds somewhere in the right ballpark to me. Considering how much time Musk has said he thought about this I'm sure he has a much better explanation for his position but it was beyond the scope of a live QA about random subjects.

Where do people get these numbers? more or less out of their ass. Kind of like in poker, with all the information we have available to us at the moment, what is the % chance this guy has a bluff? Same idea there are just a lot more variables that are much harder to quantify for a subjective guess on the simulation hypothesis.

@Smuft

I suspect those people making the god analogy really just want to say that you need to provide evidence for something instead of asking others to disprove it, which is what you were doing. The god analogy probably comes from epistemologists like bertrand russell who used it to show the irrationality of arguments made by theists, so people associate god with it.

-you argued there was a 2/3 chance of being in a sim

-You far overate the legitimacy of academic journals. The paper was published in a philosophy journal. If you ever take a philosophy class, you will find that almost every philosopher will find flaws in arguments of other philosophers. Philosophers rarely agree on anything. And the good philosophers publish many journal articles admitting that their arguments are very weak.

I remember once a medical student independently discovered calculus recently (they didn't learn it in high school) and they got it published in a peer reviewed medical journal. And there are far worse things that get published in humanities journals-like a lot of post modernism for example.

Personally I also don't see any reason to believe in the sim argument anymore than the existence of god. Both have weak arguments in favor of them

On June 26 2016 21:58 Stroggoz wrote: @Smuft I suspect those people making the god analogy really just want to say that you need to provide evidence for something instead of asking others to disprove it, which is what you were doing. The god analogy probably comes from epistemologists like bertrand russell who used it to show the irrationality of arguments made by theists, so people associate god with it.

Agree that we cannot ask people to disprove statements like "there is a god" or "we live in a simulation" and if they cannot then it must be true. The burden of proof is on those that put forth such outrageous notions.

However, I'm not asking anyone to disprove a ridiculous notion. I'm asking you to find a problem in what I believe to be a sound argument; The Simulation Argument

An argument that details:

- it will very likely to be within our species capabilities to create these simulations
- if such simulations do exist, math showing the number of simulated experiences vastly outnumbers the nonsimulated
- computational resource assumptions
- consciousness assumptions
- conclusion with 3 propositions that takes everything into account:

"(1) The fraction of human-level civilizations that reach a posthuman stage is very close to zero; (2) The fraction of posthuman civilizations that are interested in running ancestor-simulations is very close to zero; (3) The fraction of all people with our kind of experiences that are living in a simulation is very close to one."

-you argued there was a 2/3 chance of being in a sim -You far overate the legitimacy of academic journals. The paper was published in a philosophy journal. If you ever take a philosophy class, you will find that almost every philosopher will find flaws in arguments of other philosophers. Philosophers rarely agree on anything. And the good philosophers publish many journal articles admitting that their arguments are very weak.

2/3 is too high IMO, I said 35%. please be careful on the misquotes

just to be clear, this is a very subjective guess that I do not take seriously and could change drastically any day depending on what I learn

I agree about your comments on a lot of academic journals publishing shit papers and a paper being in one doesn't necessarily give it credibility. In this case the paper was published in Philosophy Quarterly which seems to be a top philosophy journal.

More importantly, it was written by a true bad ass IMO, Nick Bostrom, an oxford professor, author of superintelligence, and founder of a group of thinkers at oxford who think about existential risks. He may be one of the most important guys in preventing our whole species from being wiped out by something other people overlook or are completely oblivious to. He still supports the simulation argument and says "I believe it is basically sound".

Personally I also don't see any reason to believe in the sim argument anymore than the existence of god. Both have weak arguments in favor of them

Do you mean "weak" in some kind of formal sense in that it cannot be proven?

What are the specific reasons you think we should reject the simulation argument?

On June 27 2016 01:21 Smuft wrote: An argument that details: - it will very likely to be within our species capabilities to create these simulations - if such simulations do exist, math showing the number of simulated experiences vastly outnumbers the nonsimulated - computational resource assumptions - consciousness assumptions - conclusion with 3 propositions that takes everything into account: "(1) The fraction of human-level civilizations that reach a posthuman stage is very close to zero; (2) The fraction of posthuman civilizations that are interested in running ancestor-simulations is very close to zero; (3) The fraction of all people with our kind of experiences that are living in a simulation is very close to one."

Your breakdown of the argument is not accurate.
H = consciousness/computational assumptions
1/2/3 = outcomes that Bostroms suggest
P = probability of being a simulation
S = we are simulations

If H then ( 1 or 2 or 3)
If 3 then P
If P then S
Therefore S

I believe the argument is valid but unsound. There are problems with H but the big problem is with P. If I put forward some mathematically theory and it is untestable how can we be sure of its accuracy. Newtons Laws for motion and gravity were used for a long time however those laws predicted a planet in between Mercury and the Sun, Einstien comes along and puts forward a better equation they test it and it predicts Mercury's orbit perfectly. Bostrom cant be sure he has access to all the right information if S is true. Say for example there are some kind of constraints in the origin world that limit the number of simulations, so we are simulations but Bostroms prediction of the probability that we are simulations is wrong. Lets just suppose that we are part of a simulation but not ancestral simulations but a physics simulation. It is likely that the experimenters would change some fundamental variable to see what happens. Since H relies on the physics of this universe we can't really say anything about the number of possible simulations without knowing if H is similar to origin universe. Its also much more likely that we will run physics simulations before we start running ancestral simulations(At the current time we are collecting huge amounts of data about our lives for future historians to go over making pointless to run a historical simulation). So if we include the number of physics simulations where life may not be possible and are empty of people then the probability will be much different. There could be many simulations but only 1 includes life but the number of people that lived in the origin universe is 10 times the number that live in the simulation.

I like your breakdown better since you shows the relationship between the major points in the argument in a simple enough way that a laymen can understand.

I believe the argument is valid but unsound. There are problems with H but the big problem is with P. If I put forward some mathematically theory and it is untestable how can we be sure of its accuracy. Newtons Laws for motion and gravity were used for a long time however those laws predicted a planet in between Mercury and the Sun, Einstien comes along and puts forward a better equation they test it and it predicts Mercury's orbit perfectly.

Agree the assumptions in H could have problems. We don't know for sure what consciousness is and we don't know for sure what the limits of our computational power will be. However, I think Bostrom made reasonable assumptions that the majority of experts in those fields would agree with. Not to say they can't be wrong but it's more likely they are not.

Unfortunately P likely won't ever be testable since we can never have all the information available to us. So instead we have to determine what is most likely to be true and be ready to make changes to that determination as we think and learn more about our world (as Bostrom did in the patch that you linked a few posts back). Also remember that as it stands now P is a huge number, say "billions to 1" so even if you find some small problem that would effect how many simulations are run in the future, it may only change P to millions/thousands to 1 which. As humans that is almost an irrelevant change.

Even if you found a large enough problem that our assumptions lead to the average number of simulated vs non-simulated experiences become 1 to 1, it still doesn't break the argument completely, it just changes the %'s.

ex. If P then 50% S and 50% (2)

Bostrom cant be sure he has access to all the right information if S is true. Say for example there are some kind of constraints in the origin world that limit the number of simulations, so we are simulations but Bostroms prediction of the probability that we are simulations is wrong. Lets just suppose that we are part of a simulation but not ancestral simulations but a physics simulation. It is likely that the experimenters would change some fundamental variable to see what happens. Since H relies on the physics of this universe we can't really say anything about the number of possible simulations without knowing if H is similar to origin universe. Its also much more likely that we will run physics simulations before we start running ancestral simulations(At the current time we are collecting huge amounts of data about our lives for future historians to go over making pointless to run a historical simulation). So if we include the number of physics simulations where life may not be possible and are empty of people then the probability will be much different.

(this doesn't really make sense to me and I could easily be misinterpreting you, work on your writing bro)

You started off this part saying "suppose we live in a physics simulation"

...and then ended it saying "physics simulations where life may not be possible and are empty of people"

???

Physics simulations that do not include life (we already run simple versions of this today) are not included in the ancestor simulations Bostrom refers to.

There could be many simulations but only 1 includes life but the number of people that lived in the origin universe is 10 times the number that live in the simulation.

Possible but IMO unlikely, plexatron tried to put forth some arguments that included this effect. I decided not to go any deeper down that rabbit hole as an entire thread could be devoted to theorizing the possibility of each scenario. I haven't come across any that are plausible to me. The main reason why I don't find them plausible is because they usually include a scenario where there are an extremely large number of humans like us in "base reality" but I believe the number of simulations being run scales with the number of humans so I find these types of scenarios unconvincing.

- it will very likely to be within our species capabilities to create these simulations - if such simulations do exist, math showing the number of simulated experiences vastly outnumbers the nonsimulated - computational resource assumptions - consciousness assumptions - conclusion with 3 propositions that takes everything into account:

That's the thing though; I don't see the logic behind the bolded part, because the math assumes that there are only two possibilities:

1) We are living in a simulation
2) We are living in just one possible ancestor world

If this assumption were true, I would accept the argument and move on. But I don't think it's true and I would propose that there are other options available (such as the existence of an infinite number of similar ancestor worlds at various points in space-time, as well as artificial human-resembling lifeforms - the things I mentioned earlier), which I think would completely turn the argument on its head. In other words: if we ever managed to create the technology required to run such a simulation, then to calculate the probability of us being in a simulation we need to take into account:

1) The probability of us living in one of infinite iterations of (ancestor) humanities.
2) The probability of us living in one of infinite artificial human worlds.
3) The probability of us living in one of infinite simulations.
4) The probability of us living in one of infinite.... (other stuff we haven't even come up with yet due to limited knowledge)

But even if this assumption that it's either one ancestor world or a simulation were true, then..

2. plexatron's post about mirrors and also similarly, simulations within simulations within simulations - stuff that may require an infinite amount of data to be processed (basically I just think the simulators would program in limits to these types of effects)

Okay, but now there is a burden of proof on proponents of the argument; they would need to prove that there isn't an infinite number of reflections between two mirrors (something that we've been taking for granted so far). Is it only me who sees a huge problem with this?

1) We are proposing that we might be living in a simulation.
2) Our science dictates that two mirrors placed in front of each other would generate an infinite number of reflections.

If we assume that an infinite number of reflections should cause a simulation to break down, don't we have a contradiction here? Basically, both 1) and 2) can't be correct at the same time, and out of the two we have only proven one to be correct - number 2.

On June 29 2016 11:31 PIetraxon wrote: If we assume that an infinite number of reflections should cause a simulation to break down, don't we have a contradiction here? Basically, both 1) and 2) can't be correct at the same time, and out of the two we have only proven one to be correct - number 2.

I don't think we get an infinite number of reflections, the light is mostly reflected from one mirror and back to the other but a portion of the light gets absorbed each time so as this process goes on the image will become blurry. If you had a surface that perfectly reflected the light, then you would get infinite images I am not sure we have that though

On June 30 2016 05:49 FMLuser wrote: Show nested quote + On June 29 2016 11:31 PIetraxon wrote: If we assume that an infinite number of reflections should cause a simulation to break down, don't we have a contradiction here? Basically, both 1) and 2) can't be correct at the same time, and out of the two we have only proven one to be correct - number 2. I don't think we get an infinite number of reflections, the light is mostly reflected from one mirror and back to the other but a portion of the light gets absorbed each time so as this process goes on the image will become blurry. If you had a surface that perfectly reflected the light, then you would get infinite images I am not sure we have that though

But does the amount of light ever get to zero? As far as I understand, it's as you say - but even as light is "lost" with each reflection, and it approaches 0, it never quite gets there. So while the reflections become infinitely weak, there would still be an infinite number of them.

I came across the infamous Elon: "1 in a billion we are in base reality" and then bullying anyone in the audience to "Prove me wrong (bro)" on an unfalsifiable claim LOL HA! Brought back memories to this thread. Elon's argument was 1 day we have Ms. Pacman now we have World of Warcraft... 1 in a billion we are in base reality... Prove me wrong (bro)! (HA!)

Made me think of this thread.

I believe we live in a simulation aka as the matrix. Take some lsd or dmt

On July 29 2021 03:47 RiKD wrote: I came across the infamous Elon: "1 in a billion we are in base reality" and then bullying anyone in the audience to "Prove me wrong (bro)" on an unfalsifiable claim LOL HA! Brought back memories to this thread. Elon's argument was 1 day we have Ms. Pacman now we have World of Warcraft... 1 in a billion we are in base reality... Prove me wrong (bro)! (HA!) Made me think of this thread.

Eh, to me the theory is not really creative or interesting enough for it to be true, it's essentially like most other creationist stories imo. It's inconsistent that any silicon valley nerds would place much stock in this belief since a large part of theoretical computer science over the last 30-40 years has been involved in proving that a big chunk of questions that we come up with are impossible to solve by a computer. (So long as NP is not equal to P, and almost every mathematician believes that to be the case). I wish there was more appreciation for the fact that mathematicians have been proving many things are impossible and even impossible to know if they are impossible, over the last century. For some reason believing that computers can do everything when in fact the theory says they can do very little-these people who believe this seem more informed by science fiction than actual science.

Obviously if you're trying to actually simulate every atom in the universe a computer cannot do that. I would argue that is not even what is represented by reality all around you.

Whether coincidental or not there are things you cannot really debate that show the universe does not always simulate everything at once.

Just a simple look at the double slit experiment or the fact that there is a refresh rate on reality should be enough to show that.


I'm not a big fan of simulation theory or string theory because neither can ever have any merit beyond theory but to just dismiss it as science fiction seems as ignorant as believing it. Agnostic takes are always better when evidence is not available.

On July 30 2021 18:57 CrownRoyal wrote: Obviously if you're trying to actually simulate every atom in the universe a computer cannot do that. I would argue that is not even what is represented by reality all around you. Whether coincidental or not there are things you cannot really debate that show the universe does not always simulate everything at once. Just a simple look at the double slit experiment or the fact that there is a refresh rate on reality should be enough to show that. I'm not a big fan of simulation theory or string theory because neither can ever have any merit beyond theory but to just dismiss it as science fiction seems as ignorant as believing it. Agnostic takes are always better when evidence is not available.

you're not taking an agnostic approach though-your claiming that it's not on the level of science fiction.

To be fair, I remember reading Bostrom's paper ages ago and it was quite fun. If I remember correctly Bostrom's math was higher than total science fiction but at least he is out in the world doing stuff.

I wonder if Bostrom wrote the paper before this NP is not P stuff.

Seems like NP is not P makes a lot of stuff science fiction.

-quantum computing
-simulation
-probably ai
-others?

Unfortunately, climate change and nuclear war are still here to stay.

Thinking too much about existential crisis out of my control makes me sad so bye.

I had to teach myself complexity theory for my research project so I know a little about it.

He certainly did not write the paper before the NP is not P question, it's been a big question since around the 70's, when Cook-Levin proved that every computer language that can be run by a non-deterministic Turing machine is polynomial-time reducible to the satisfiability question. What that means in plain English is that there is a class of problems that we can prove are essentially equivalent to each other in terms of difficulty, and the proof of this was by showing that you can reduce every computer language down to the question "is this statement true?". That's what satisfiability means in mathematical logic. If you can prove that any one of these problems can be run on a deterministic turing machine in roughly the same amount of time, then all of these difficult problems suddenly become easy problems. Since the 1970's there have been thousands of problems that have been proven to be in this difficult class, and not a single one has been found to be reducible to a deterministic Turing machine. If any of them did, then NP=P. On top of this huge failure rate, there is a lot of other evidence to suggest that NP is not equal to P. Quantum computing is a very small topic in theoretical computer science, i haven't really bothered to learn it because it's not relevant to my interests, and apparently, there is a big misconception about them, classical computers are better than quantum computers for a lot of tasks apparently. There appears to be a lot of media sensationalist bullshit surrounding quantum computing, it's a pretty big buzzword over the past 10 years.

But none of this stuff is really that relevant, my dismissal of this topic basically comes down to common sense intuition. I mean, reading his trichotomy paper, it seems obvious to me that one of those cases is near 100% close to being true and the other two are ridiculous.

also idk what the hell people are talking about with mirrors in this thread, but it's pretty obvious that you can have at the bare minimum-an uncountably infinite number of reflections using mirrors. Roger penrose actually came up with particular shapes where you can bounce light off the boundary of the shape and it doesn't hit every point in the boundary, it's quite an interesting research topic in mathematics and there is a good amount of literature on it.

--- Nuked ---