Jump to content

The Simulation hypothesis


Free Northman Reborn

Recommended Posts

So given there is a fair amount of intellect floating around this forum, I thought I’d garner some of your views on the Simulation argument - as articulated by Nick Bostrom and supported by people as influential as Elon Musk, amongst others.

I was struck by Prof Bostrom’s appearance on Joe Rogan’s show this week, and in particular Rogan’s frustrating inability to grasp the basic principles of the Simulation thought experiment for the last half hour of the interview, resulting in Bostrom having to endlessly repeat the basic probabilities involved, to no avail.

I am intrigued by it because the logic seems so compelling. Bostrom reduces the argument to 3 possibilities, one of which is highly likely to be true. To paraphrase the three options:

1. All civilizations go extinct before they achieve the technological level to create simulations indistinguishable from reality.

2. All civilizations decide for some reason not to create simulations indistinguishable from reality.

3. We are living in a simulation.

Number 3 is then the simulation hypothesis, while all three together constitute the simulation argument. And as counterintuitive as number 3 is, the more one thinks about it, the more logical it appears to be.

So, are there any obvious refutations of this  argument?

 

The crux of the debate unfolds from 1 hour 56 minutes onwards.

Link to comment
Share on other sites

Misses two obvious alternatives...

4. We are the first civilisation to reach the level where we can even consider (much less create) a simulation indistinguishable from reality. No prior civilisation has reached this level.

5. A simulation indistinguishable from reality is impossible to achieve with limited resources.

Frankly, I consider #5 the most likely.

Link to comment
Share on other sites

Just now, The Marquis de Leech said:

Misses two obvious alternatives...

4. We are the first civilisation to reach the level where we can even consider (much less create) a simulation indistinguishable from reality. No prior civilisation has reached this level.

5. A simulation indistinguishable from reality is impossible to achieve with limited resources.

Frankly, I consider #5 the most likely.

5 is incorporated in 1, as it is consistent with all civilizations going extinct before being able to create a simulation indistinguishable from reality. After all, if they can never achieve it, then by default they will go extinct before doing so.

And 4 is the point that Rogan got stuck on, and which is refuted by the probabilities involved. Because, the probability that we are in the one and only base reality, rather than in one of the millions or billions of simulations to follow is, well, millions or billions to one. 

Link to comment
Share on other sites

The simulation hypothesis is philosophically not particularly new. It's just Bishop Berkeley with a fancy lick of paint. The refutation is the same. Sure, we could all be living in a simulation, but what difference does that make? You will continue to get hungry, and you will continue to eat. You will breathe and your lungs will work. You will flinch from extreme heat and cold. You will continue, in other words, exactly as if everything around you is real. So the entire thing is a pointless mental game you're playing with yourself. 

Link to comment
Share on other sites

Oh! This reminds me of thoughts I had some years ago! Not exactly about whether we live in a situation, but rather about the merits and challenges of actually making a perfect simulation of all of reality:
So let's assume we reach a point where we have gathered enough knowledge about the physics of the universe that we can put all of its rules into a simulation that starts at the big bang and allows us to observe an exact copy of all of history. It would basically make all time travel unnecessary and may allow us to completely rewrite history books when we can just take a peek at what actually happened through our little simulated Earth. We could also use it to seek other intelligent life in the universe without even venturing out ourselves.

But what happens once we move the simulation up to the point where we have created this simulation? Its processing requirements would exponentially grow as it now has to recursively simulate simulators simulating realities with simulators simulating realities.

And looking into the future with it might be entirely impossible because it has to infinitely recalculate its own impact on decision-making.

Link to comment
Share on other sites

12 minutes ago, Toth said:

Oh! This reminds me of thoughts I had some years ago! Not exactly about whether we live in a situation, but rather about the merits and challenges of actually making such a simulation:
So let's assume we reach a point where we have gathered enough knowledge about the physics of the universe that we can put all of its rules into a simulation that starts at the big bang and allows us to observe an exact copy of all of history. It would basically make all time travel unnecessary and may allow us to completely rewrite history books when we can just take a peak at what actually happened through our little simulated Earth. We could also use it to seek other intelligent life in the universe without even venturing out ourselves.

But what happens once we move the simulation up to the point where we have created this simulation? Its processing requirements would exponentially grow as it now has to recursively simulate simulators simulating realities with simulators simulating realities.

And looking into the future with it might be entirely impossible because it has to infinitely recalculate its own impact on decision-making.

Bostrom points out that the simulation need not simulate full reality down to the quantum level. For example, the vast majority of the universe can just be superficially rendered, with no underlying substance. The table you are sitting at could just superficially appear to obey the laws of physics, but have nothing actually going on at the atomic level. If an observer then decides to examine it under an electron microscope, the simulation can just create the necessary atomic structures for the duration of the observation. Thereby vastly reducing the computational power required to maintain the simulation. This is how video games already work, rendering only what is necessary for the observer at a point in time.

As for the energy requirements of the simulation, we should not limit ourselves to what we could do in the next centuries, but in the next million or billion years. Kardashev 3 level civilizations, with planet or solar system sized computers.

Link to comment
Share on other sites

I find this generally quite silly, so I don't want to spend too much time with it. But very smart people seem to take it seriously. Now here is my counterargument which is not a perfect refutation but good enough for me:

The premisses for figuring out the probabilities involved, i.e. physical and technical possibility of such simulation, # of cvilizations in the universe, # of civs reaching simulation tech etc. are all using our current knowledge of cosmology, physics, maths, computer science, evolution of life, sociology of civs etc. But if we really live in a simulation (whatever this exactly means, are we all still inviduals, something like brains in the vat or am I a single brain in the vat or should I rather say that I am a simulation rather than live in one?) almost all this is bunk.

What we seem to know is only the simulated pseudophysics, only the apparent laws and "natural constants" etc. When I conclude that I live in or am a simulation I have absolutely no reason to assume that this apparent physics (that which I based my probabilities for the argument on) is anything close to the real physics. To the contrary, the realization that it could be a simulation should make it very likely that the real physics is (totally) different than the simulated physics because there are far more ways to have some kind of simulated physics but maybe only one way for the real physics to be. So the probabilities turn out not to be based on any scientific knowledge (because we probably have no scientific knowledge besides maths and logics, everything is like the laws of a computer game).

This does not kill the argument because of course it is possible that the simulators choose to stick close to the real physics (but why should they? it's not that we who are bound to the simulation could compare, like we can between one of our flight simulators or computer games and our apparent non-simulated reality) and therefore our apparent physics is close, so this shows that if the apparent physics is close to the real physics, simulations would be possible and than the argument will go through. I still think that it casts sufficient doubt upon the pseudo-quantitative character (probabilities using SCIENCE) of the argument.

But for me this is enough to not bother much more with that stuff. I also have a bunch of further quibbles, e.g. that I would expect much simpler, less messy simulated physics (so the messiness of actual science (as opposed to the classic simplex sigillum veri) is a sign of grasping bits of the real world, not being fed something readymade). And there are of course "deeper" arguments concerning the nature of the mental (i.e. that it cannot be simulated because it is irreducibly non-physical or irreducibly bound to biological stuff), semantics (what does simulation even mean, if there is no non-simulated reality to contrast it with?)

NB the "classic" solipsist/skeptical scenario by Descartes, i.e. that all my apparent sensual knowledge is fed into my system by an evil demon does NOT suffer from the same problem. Because it does not presuppose any physics (there is no material world in this scenario) and does not use probabilities.  I am not following them closely but Bostrom and other transhumanists are IMO fooling themselves into an incoherent mix of idealism (the mental is sufficiently independent to be simulated) and materialist reductionism that I cannot take seriously.

Link to comment
Share on other sites

9 minutes ago, Jo498 said:

I find this generally quite silly, so I don't want to spend too much time with it. But very smart people seem to take it seriously. Now here is my counterargument which is not a perfect refutation but good enough for me:

The premisses for figuring out the probabilities involved, i.e. physical and technical possibility of such simulation, # of cvilizations in the universe, # of civs reaching simulation tech etc. are all using our current knowledge of cosmology, physics, maths, computer science, evolution of life, sociology of civs etc. But if we really live in a simulation (whatever this exactly means, are we all still inviduals, something like brains in the vat or am I a single brain in the vat or should I rather say that I am a simulation rather than live in one?) almost all this is bunk.

What we seem to know is only the simulated pseudophysics, only the apparent laws and "natural constants" etc. When I conclude that I live in or am a simulation I have absolutely no reason to assume that this apparent physics (that which I based my probabilities for the argument on) is anything close to the real physics. To the contrary, the realization that it could be a simulation should make it very likely that the real physics is (totally) different than the simulated physics because there are far more ways to have some kind of simulated physics but maybe only one way for the real physics to be. So the probabilities turn out not to be based on any scientific knowledge (because we probably have no scientific knowledge besides maths and logics, everything is like the laws of a computer game).

This does not kill the argument because of course it is possible that the simulators choose to stick close to the real physics (but why should they? it's not that we who are bound to the simulation could compare, like we can between one of our flight simulators or computer games and our apparent non-simulated reality) and therefore our apparent physics is close, so this shows that if the apparent physics is close to the real physics, simulations would be possible and than the argument will go through. I still think that it casts sufficient doubt upon the pseudo-quantitative character (probabilities using SCIENCE) of the argument.

But for me this is enough to not bother much more with that stuff. I also have a bunch of further quibbles, e.g. that I would expect much simpler, less messy simulated physics (so the messiness of actual science (as opposed to the classic simplex sigillum veri) is a sign of grasping bits of the real world, not being fed something readymade). And there are of course "deeper" arguments concerning the nature of the mental (i.e. that it cannot be simulated because it is irreducibly non-physical or irreducibly bound to biological stuff), semantics (what does simulation even mean, if there is no non-simulated reality to contrast it with?)

NB the "classic" solipsist/skeptical scenario by Descartes, i.e. that all my apparent sensual knowledge is fed into my system by an evil demon does NOT suffer from the same problem. Because it does not presuppose any physics (there is no material world in this scenario) and does not use probabilities.  I am not following them closely but Bostrom and other transhumanists are IMO fooling themselves into an incoherent mix of idealism (the mental is sufficiently independent to be simulated) and materialist reductionism that I cannot take seriously.

Thought provoking comments, even if you didn't want to spend much time on it.

To touch on just a couple of your points, no other civilizations beyond earth need be considered for this argument.  Even if we are the only intelligent civilization ever to arise in the universe, if we eventually create a fully realistic ancestor simulation and let it run, then the simulated civilization inside the simulation will eventually do the same, and so on and so on until a million sub simulations are created.

Also consider that due to the immense computational power that will be available, these simulations might run for billions of simulated years inside the simulation, but due to the processing speed of the computers might only last a few minutes in real time. Meaning that the base reality civilization can run millions of repeat simulations over a fairly short span of years. Not to mention that they could run millions of distinct simulations simultaneously, each covering billions of years of subjective time, while only lasting minutes on the outside.

And all of this just requires our single civilization ever to have arisen. No aliens needed whatsoever.

Secondly, the brain in the vat as in the Matrix is totally unnecessary. With sufficient computational power, it is assumed that conscious entities can be digitally created within the simulation program. Call it 7 billion artificially intelligent entities (programs) created within the simulation, each conscious and believing that it is a you or me or one of the other 7 billion people on this earth. But in reality only existing as code in the simulation. No brains in vats plugged into the computer.

As for the nature of physics in base reality - that is Elon Musk's ultimate question he would ask the programmers if he had the chance: What lies outside the simulation?

 

Link to comment
Share on other sites

The simulation hypothesis has also been on my mind...

1 hour ago, Free Northman Reborn said:

Bostrom points out that the simulation need not simulate full reality down to the quantum level. For example, the vast majority of the universe can just be superficially rendered, with no underlying substance.

And weirdly enough, it seems the universe has has been observed has characteristics that correpond to this...

One thing that's been troubling me is that as we come closer to the possibility (in terms of technology, especially computing power) of creating a credible simulation ourselves there seems to be two main possibilities:
- We (humans) create simulations for fun or entertainment, i.e. they provide something that "our" world doesn't. So if we created this simulation for ourselves it supposedly provides something the "real" world doesn't, i.e. different sensations or emotions. The obvious theory would be that our souls do no have access to physical sensations in the real world.
- OTOH, if the simulation is managed by a superior being, then it is meant to observe how we react/adapt to it. In other words it has a goal. Not in the sense that it can end (in this scenario we'd probably be unable to exist outside the simulation), but in the sense that humanity is meant to evolve in a specific direction and that the parameters of this simulation guide us toward our goal. The spiritual implications are rather far-reaching.

2 hours ago, mormont said:

Sure, we could all be living in a simulation, but what difference does that make? You will continue to get hungry, and you will continue to eat. You will breathe and your lungs will work. You will flinch from extreme heat and cold. You will continue, in other words, exactly as if everything around you is real. So the entire thing is a pointless mental game you're playing with yourself. 

The last two sentences are actually wrong. Of course no one would change their day-to-day habits, but the philosophical implications are likely to change the way one defines their life-goals. Also, if taken seriously the hypothesis would obviously affect the way our physicists conceive their theories.
In the long-run if the simulation theory came to be believed it would potentially affect civilization itself.
It's a bit like religion: on the face of it, having faith or not doesn't change anything. And yet depending on the way you approach the religion you choose to believe in and depending on how many other people adopt it, it ends up changing everything.

Link to comment
Share on other sites

My personal view, based on preference rather than logic, is that option 1 is true. Which is that simulations of this nature never become possible during the lifetime of our civilization. However, given our current trajectory, that seems like wishful thinking.

Option 2, that we choose never to run such a simulation even though we eventually will be able to, seems unlikely. So option 1 leaves us the best chance of not being in a simulation, in my view.

Link to comment
Share on other sites

I think you are missing my problem with the argument. If we live in a simulation these cool impressive numbers and probabilities are as good as pulled out of thin air. Because if we live in a simulation we have no reason to assume that we know anything about the real physics these numbers are supposedly based on.

The argument presupposes that our current knowledge is reliable and pretty close to the true physics of the universe (that's where it gets the numbers about what is computationally possible etc. from) and then proceeds to tell us that our naive realist view of the universe is totally wrong because all our "physics" stems just from a simulation. So it undermines itself, albeit not in a strictly logical way because of course it could still be true that the real physics is close to the apparent "physics" we based the argument on, but this doesn't seem likely, if one goes with the gist of the argument.

(Because probabilities are so easily abused, I'll do it again: There are zillions more ways to simulate a universe (just remember unrealistic physics from some computer games) than to actually have a physically consistent working universe, so the probability that a randomly picked simulated universe is closely similar to the one and only real one is vanishingly small!) I really think they shouldn't get public funding for this, and the more pitiable point is that very smart people are wasting their time and brainpower with such stuff.

So adding more impressive numbers about what could in principle be computationally possible is moot because we don't know what actually IS computationally possible or probable, we only "know" the bunk pseudoreality the simulators or simulation has fed us.

I also don't see why I should assume that there are 7 billion programs all believing that they are humans. (This would be the analogue to Berkeley: Human Minds are real but there is no physical reality and we get the sense impressions, the simulated reality, from God, the Supreme Mind (that's why we agree on apparent reality, the feeds have the same source which does not want to confuse us with different streams of pseudoreality.) This is again having the cake of apparent reality (7 billion people out there) and eating it (it's just 7 billion pograms).

One program (myself) and one simulated environment are enough. (This would be closer to Descartes evil spirit who feeds me the false impressions that I have a material body, encounter material things and other people etc.

I think a good test for the argument would be if one replaces the framing of modern computation with Berkeley's God or Descartes' Demon and if it suddenly becomes very implausible, we should get suspicious. As we should. We have been seduced by large numbers from SCIENCE and impressive pseudoquantitative probabilism: They can simulate billions of years in microseconds! God could have created the universe 10 min. ago exactly as it appeared 10min. ago with all our memories and traces etc. making it appear billions of years old - the same style of argument with a different framing. And these older arguments are actually (marginally) better because they do not presuppose (pseudo)physics as taken for granted at first but getting destroyed in the course of the argument.

Link to comment
Share on other sites

10 minutes ago, Jo498 said:

I think you are missing my problem with the argument. If we live in a simulation these cool impressive numbers and probabilities are as good as pulled out of thin air. Because if we live in a simulation we have no reason to assume that we know anything about the real physics these numbers are supposedly based on.

The argument presupposes that our current knowledge is reliable and pretty close to the true physics of the universe (that's where it gets the numbers about what is computationally possible etc. from) and then proceeds to tell us that our naive realist view of the universe is totally wrong because all our "physics" stems just from a simulation. So it undermines itself, albeit not in a strictly logical way because of course it could still be true that the real physics is close to the apparent "physics" we based the argument on, but this doesn't seem likely, if one goes with the gist of the argument.

(Because probabilities are so easily abused, I'll do it again: There are zillions more ways to simulate a universe (just remember unrealistic physics from some computer games) than to actually have a physically consistent working universe, so the probability that a randomly picked simulated universe is closely similar to the one and only real one is vanishingly small!) I really think they shouldn't get public funding for this, and the more pitiable point is that very smart people are wasting their time and brainpower with such stuff.

So adding more impressive numbers about what could in principle be computationally possible is moot because we don't know what actually IS computationally possible or probable, we only "know" the bunk pseudoreality the simulators or simulation has fed us.

I also don't see why I should assume that there are 7 billion programs all believing that they are humans. (This would be the analogue to Berkeley: Human Minds are real but there is no physical reality and we get the sense impressions, the simulated reality, from God, the Supreme Mind (that's why we agree on apparent reality, the feeds have the same source which does not want to confuse us with different streams of pseudoreality.) This is again having the cake of apparent reality (7 billion people out there) and eating it (it's just 7 billion pograms).

One program (myself) and one simulated environment are enough. (This would be closer to Descartes evil spirit who feeds me the false impressions that I have a material body, encounter material things and other people etc.

I think a good test for the argument would be if one replaces the framing of modern computation with Berkeley's God or Descartes' Demon and if it suddenly becomes very implausible, we should get suspicious. As we should. We have been seduced by large numbers from SCIENCE and impressive pseudoquantitative probabilism: They can simulate billions of years in microseconds! God could have created the universe 10 min. ago exactly as it appeared 10min. ago with all our memories and traces etc. making it appear billions of years old - the same style of argument with a different framing. And these older arguments are actually (marginally) better because they do not presuppose (pseudo)physics as taken for granted at first but getting destroyed in the course of the argument.

If the real universe is based on the same physics as ours, then the predictions of the Simulation argument are based on sound principles, meaning we might live in a simulation as proposed by Bostrom.

If, on the other hand, as you suggest, the physics of the real universe is utterly different from that in our simulated universe, then that is because we definitely live in a simulation, so the simulation hypothesis has a probability of 100%.

Link to comment
Share on other sites

@Rippounet: Pick some spiritual tradition, say Buddhism or Christianity that has some vague similarities to the Simulation hypothesis, be it karmic retribution (go to simulated hell because of some behavior in simulated live), guidance and goal-setting by a superior being or whatever. You actually get the idea that the apparent world is only a poor shadow of the real (spiritual) world in some of these traditions. (PK Dick apparently believed or made us believe that he entertained the belief that the devil creates the impression that 2000 years have passed since Christ walked the earth that we should despair and not believe in the Second Coming whereas we are actually still in the 1st cent. AD....) If you don't believe any of these traditions and their spiritual implications for leading one's live (probably because you find it a bunch of superstitions), what makes you take Bostromism any more seriously? The pseudoscientific guise?

Link to comment
Share on other sites

I don't see how there could be an arbitrarily large number of simulated universes. Surely there must be a hard limit? You can't use the physical properties of a universe to perfectly simulate that same universe, for the same reason that the only perfect map of a territory is the territory itself. So any simulated universe must be smaller and/or simpler than it's parent universe. If each layer is simpler, then eventually you reach a point where there isn't sufficient complexity to model consciousness or self-awareness.

Link to comment
Share on other sites

Quote

The last two sentences are actually wrong. Of course no one would change their day-to-day habits, but the philosophical implications are likely to change the way one defines their life-goals. Also, if taken seriously the hypothesis would obviously affect the way our physicists conceive their theories.
 In the long-run if the simulation theory came to be believed it would potentially affect civilization itself.
It's a bit like religion: on the face of it, having faith or not doesn't change anything. And yet depending on the way you approach the religion you choose to believe in and depending on how many other people adopt it, it ends up changing everything.

It's just another regurgitation of religion, though. Consider that the universe is an artificial creation, created by a God who stands outside that creation. To all intents and purposes that is the same as the universe being a simulation created by a programmer (or programming team, or even a super AI), even down to the moral argument: if you have the power to create artificial life with consciousness, even inside a computer programme, then why the fuck would you do so to inflict the most miserable horrors imaginable on so many of them?

The postulation is also almost philosophically meaningless: if the simulation of reality is 100% accurate then the differences between the simulation and reality became non-existent, then the simulation is all to intents and purposes reality. You might say, well someone could just turn off the simulation (presumably killing all of us instantly), but then if we are in reality, then we could also all be killed at any second by a radiation pulse from a nearby supernova, or even a false vacuum collapse, with absolutely no warning whatsoever. So again the difference becomes irrelevant.

Link to comment
Share on other sites

5 hours ago, Free Northman Reborn said:

5 is incorporated in 1, as it is consistent with all civilizations going extinct before being able to create a simulation indistinguishable from reality. After all, if they can never achieve it, then by default they will go extinct before doing so.

And 4 is the point that Rogan got stuck on, and which is refuted by the probabilities involved. Because, the probability that we are in the one and only base reality, rather than in one of the millions or billions of simulations to follow is, well, millions or billions to one. 

That probability argument is very soft though. It does seem to ignore the odds that our universe was created 1 day ago, 1 week, any random time period ago at as fine a scale as one wants to.

Generally I don't put much time in the simulation argument, it is a nice SF trope, but given the sheer amount of modelling involved seems to make it near impossible at any reasonable time ratio.

Link to comment
Share on other sites

58 minutes ago, Werthead said:

It's just another regurgitation of religion, though. Consider that the universe is an artificial creation, created by a God who stands outside that creation. To all intents and purposes that is the same as the universe being a simulation created by a programmer (or programming team, or even a super AI), even down to the moral argument: if you have the power to create artificial life with consciousness, even inside a computer programme, then why the fuck would you do so to inflict the most miserable horrors imaginable on so many of them?

The postulation is also almost philosophically meaningless: if the simulation of reality is 100% accurate then the differences between the simulation and reality became non-existent, then the simulation is all to intents and purposes reality. You might say, well someone could just turn off the simulation (presumably killing all of us instantly), but then if we are in reality, then we could also all be killed at any second by a radiation pulse from a nearby supernova, or even a false vacuum collapse, with absolutely no warning whatsoever. So again the difference becomes irrelevant.

Yes, yes, yes.  It's yet another way that we humans try to employ our (limited) neural power to understand, categorize, and explain chaos.  So all the thought is, as someone said, interesting from an SF perspective (and maybe as bar talk), but beyond that?  Bah humbug.

Link to comment
Share on other sites

1 hour ago, Werthead said:

The postulation is also almost philosophically meaningless: if the simulation of reality is 100% accurate then the differences between the simulation and reality became non-existent, then the simulation is all to intents and purposes reality.

If I have a Star Trek style replicator and replicate a 100 Euro bill, what I get is not legal tender, despite physical indistinguishability. In a similar fashion there still would be a slight difference between the real reality and the 100% accurately replicated reality.

But of course they don't mean a duplicated universe. They mean that there still is a "mother universe" (which is actually like the universe of our physical theories) and simulations are parts of it but not in the sence of physical  duplications but as data fed into some data-receiving "minds" (with the possibility that minds are also simulations). It's pseudomaterialist reconstruction of some traditional idealist or religious ideas, just with computers, that's why people who think Berkeley or certain buddhist philosophies are bs, fall for it.

Link to comment
Share on other sites

Archived

This topic is now archived and is closed to further replies.

×
×
  • Create New...