• Arkady
    768
    Yes, evidently. I just wonder what he would posit as the reason for accepting such an assumption. Perhaps given that it's supposed to be an "ancestor" simulation specifically, he would say that such a simulation would by definition closely (if not necessarily exactly) resemble the ancestral state of the civilization doing the simulating.
  • RogueAI
    2.6k
    Yes, evidently. I just wonder what he would posit as the reason for accepting such an assumption.Arkady

    Maybe because the real world might have a lot in common with a simulated world? It seems a bit ad hoc to me.
  • Arkady
    768
    Yeah, it definitely seems like something's missing. However, as I said I haven't read up on this in detail, so Bostrom or other proponents of the argument may well have handled this objection somehow. It seems pretty obvious, so I doubt I'm the first to bring it up.
  • noAxioms
    1.4k
    But if this is indeed a simulation, then anything we purport to know about our present levels of technology (and thus any extrapolation therefrom) is illusory, because we don't actually possess that technology: such technology is simulated.Arkady
    Take airplanes. If the simulation initial state was set in the 20th century, then it includes airplane technology. It is 'given' so to speak. If the initial state is started before that, then airplanes are our own invention.. Either way, we possess the technology. It isn't illusory. We actually can make airplanes that fly. If you crash in one, you really die, as opposed to say a video game where if you 'die', you simply exit the game. Getting shot in a video game is indeed an illusion.


    Perhaps given that it's supposed to be an "ancestor" simulation specifically, he would say that such a simulation would by definition closely (if not necessarily exactly) resemble the ancestral state of the civilization doing the simulating.Arkady
    @RogueAI correctly pointed out that only somebody who knows about humans would want to simulate them, so it is presumably our decedents, be they human anymore or not.

    The initial state would presumably resemble some factual state in the past of 'reality', as best they can estimate it, but it would subsequently evolve in a totally different path, regardless of what you define the term 'ancestor simulation' to mean. You're indeed not the first to bring this up.

    So why would they want to run such a simulation? It won't reproduce 'what really happened', so there must be some other reason to simulate an entire planet at the level of full consciousness. I can't think of one. Not with those requirements.




    How does Searle say [the arm] goes up in the TED transcript?fishfry
    "we know the basic part of the answer — and that is, there are sequences of neuron firings and they terminate where the acetylcholine is secreted at the axon end-plates of the motor neurons, sorry to use philosophical terminology here. But when it is secreted at the axon end-plates of the motor neurons, a whole lot of wonderful things happen in the ion channels and the damned arm goes up."
    That's a wordy version of what I said, which is "there's wires connecting the parts where the will is implemented, to the parts where the motor control is implemented". Under Chalmers, there isn't such a wire, hence the magic.

    Because I can't believe that a computer program of any complexity, running at any speed, could ever be conscious.
    Nobody ever said the program was conscious. It's dumb as rocks, implementing a fairly small program that simply knows how to move the particles around. It implements physics and is no more conscious than is physical law. It has no external input, so right there it doesn't qualify as being conscious. Some programs do have such input, but not most simulations.

    Anyway, you don't believe a simulated person could be conscious, so you make up an arbitrary rule that forbids it. I think that's what you're saying, but personal belief isn't evidence against somebody's hypothesis. It's only an irrational reason that you don't accept the hypothesis.


    Programs play chess and drive cars, and I'm duly impressed. Not same as being conscious.
    I say the car wouldn't be able to do its thing if it wasn't conscious of what's going on around it. Not the same as human consciousness, sure, but it's still a form of consciousness. A car stays conscious even when it's off, a sort of security feature that has caught vandals and thiefs.
    So maybe you have more of a Searle definition of 'conscious' which is 'only if a genuine human is doing it'. He actually defines the word early in his talk, but it's just 'awake' as opposed to 'asleep', something that regularly comes and goes with a human.

    What would constitute evidence of what might be possible in the future?
    Mathematics. Known physical limits. Psychology. Fermi paradox. All vague things, I admit, but at least not empty.

    The ultimate argument against my position is that some configurations of atoms are self-aware, and someday we may figure out what those configurations are.
    The computer doesn't need to know which configurations. It only has to simulate physical law. It means that if they successfully simulate a conscious being, they still won't know how consciousness works.

    This referred to the claim that everything physical is computational. If you agree with me that you don't assert this, then we're in complete agreement. In fact I think we might be in a lot of agreement in general.
    Both the physics community and I are in general agreement in that our physics does not appear to be computational. Bell's theorem even 'proves' this, but it is based on empirical evidence, and one has to accept empirical evidence for the proof to hold.

    Programs don't have souls, don't have life energy, aren't alive.
    OK, but naturalism is in contrast with concepts like souls, life energy, vitalism, etc. None of these things is necessary to be alive, and indeed, a running program is no more alive than is your brain processes.
    They do have a hard time defining 'life'. I mean, given one example of Earth life, it's pretty easy: Anything that trances its ancestry to the earliest life form. But that definition fails as a metric to decide if something alien is alive or not.

    Our theory of gravity works, but we know it's not quite right.
    It's 'right' enough to know where the moon will be 17 years from now, but the physics is chaotic enough that we don't know where it will be 17 millennia from now.

    Oh no, that's chaos theory. Even if we had all the details of the initial state, we can't necessarily predict the future.
    Indeed, but we can for a limited time. For the rolling lumpy rock, yes, that's a chaotic function, but with sufficient precision, we can predict its brief path until it stops, with arbitrary precision. Same with the weather. Our current precision gets us maybe 6 days of what that storm will do, and much of that error is due to lack of perfect model, and lack of detailed initial state.

    Tiny rounding errors add up to great differences in output. Nearby points in the initial state space lead to vastly different outcomes. We know this.
    Which is exactly why there's no point in doing an ancestor simulation. It will show an alternate history that bears little resemblance to what the books say. If started far enough back, it will not evolve humans.

    He says that in the future, computations will instantiate consciousness.
    That's very different than us being a program.
    It is "I am a human" vs "I am that on which the laws of physics supervene". The program can't be conscious because it has zero sensory input. It has nothing to be conscious of.

    Very distinct. The universe, or God, instantiates all the stuff around us. It is the stuff around us. It's the exact ultimate laws of the universe. The execution of a model is just that. It lets us predict, to sufficient accuracy, how the galaxies will move. It doesn't move the galaxies and it's not exact.
    This is Searle's language game again. Instantiation if an anthropomorphic god does it, and 'execution of a model' if anything else does the exact same thing. The model may be a map, but the execution of it is territory.

    Gravity simulations do not attract nearby bowling balls. They do not instantiate gravity.
    Nonsense. If they didn't instantiate gravity, then the simulated moon would not orbit the simulated Earth. That's what you defined instantiation to be. Are we changing the definition now of 'instantiation' to be 'not simulated'?


    My reply is half the size of your post, in an effort to stem the tendency to growth.
  • fishfry
    3k
    "we know the basic part of the answer — and that is, there are sequences of neuron firings and they terminate where the acetylcholine is secreted at the axon end-plates of the motor neurons, sorry to use philosophical terminology here. But when it is secreted at the axon end-plates of the motor neurons, a whole lot of wonderful things happen in the ion channels and the damned arm goes up."
    That's a wordy version of what I said, which is "there's wires connecting the parts where the will is implemented, to the parts where the motor control is implemented". Under Chalmers, there isn't such a wire, hence the magic.
    noAxioms

    Where is the will that initiates the process?

    Nobody ever said the program was conscious. It's dumb as rocks, implementing a fairly small program that simply knows how to move the particles around. It implements physics and is no more conscious than is physical law. It has no external input, so right there it doesn't qualify as being conscious. Some programs do have such input, but not most simulations.noAxioms

    You are agreeing with me again?

    Anyway, you don't believe a simulated person could be conscious, so you make up an arbitrary rule that forbids it.noAxioms

    I made no rules. I expressed an opinion.

    I think that's what you're saying, but personal belief isn't evidence against somebody's hypothesis. It's only an irrational reason that you don't accept the hypothesis.noAxioms

    Ok. My reasons are irrational.

    You sound like I said something that annoyed you.

    Thanks for keeping this brief, anyway. I didn't understand much of this particular post. The rest of this convo has been interesting.
  • Arkady
    768
    Take airplanes. If the simulation initial state was set in the 20th century, then it includes airplane technology. It is 'given' so to speak. If the initial state is started before that, then airplanes are our own invention.. Either way, we possess the technology. It isn't illusory. We actually can make airplanes that fly. If you crash in one, you really die, as opposed to say a video game where if you 'die', you simply exit the game. Getting shot in a video game is indeed an illusion.noAxioms
    If we're in a simulation, and we make airplanes within the confines of this simulation, then it seems to me that we don't actually possess the technology. We at most possess a simulation of that technology. If we're in a simulation, what does "actually" flying mean? We're merely simulating the flying experience, making it simply a hyper-advanced flight sim. Pilots in flight sims aren't actually flying, after all.

    I think that's the whole point of a simulation: nothing is actual, since it's by hypothesis simulated. If we're to posit that "simulated" = "real", then what work is the "simulated" descriptor doing?

    @RogueAI correctly pointed out that only somebody who knows about humans would want to simulate them, so it is presumably our decedents, be they human anymore or not.noAxioms
    Well, it's a truism that only beings who know about humans would want to simulate them, as in order to simulate something you must have knowledge of it, else how do you construct a verisimilitudinous simulation of it? However, that truism needn't limit the simulators to our descendants: perhaps they're advanced aliens which at some point in cosmic history made contact with humans, perhaps they're advanced AI like in the Matrix, and so forth.
  • noAxioms
    1.4k
    If we're in a simulation, what does "actually" flying mean? We're merely simulating the flying experience, making it simply a hyper-advanced flight sim. Pilots in flight sims aren't actually flying, after all.Arkady
    You seem to be referring to a virtual reality. The simulation hypothesis is not a virtual reality. The people (us) are simulated. In a VR, we would be real, and only our experiential feed is artificial.

    So in the simulation hypothesis, everything in our universe is as real as we are, and therefore it is meaningful for them to say that they actually fly. From the point of view of those running the simulation (if they're paying any attention to it at all), they might say that the simulation is simulating the flying of some of the simulated people, but that seems a needlessly wordy way to put it.

    perhaps they're advanced aliens which at some point in cosmic history made contact with humans, perhaps they're advanced AI like in the Matrix, and so forth.
    The Matrix is also an example of a VR, not an example of the simulation hypothesis.
    Aliens (or our robot successors) might indeed be running the simulation, but the simulated history (especially an initial state) would then likely not bear much resemblance to actual history.
    The robots would have perhaps some DNA evidence of mythical humans, and to demonstrate that humans might have been responsible for the genesis of the earliest machines, they run simulations of human evolution from primitive state to eventually creating their successors. They'd probably have to run it thousands of times to get one where it works before we go extinct.
    I have no idea how something robot/alien could create an initial state if they don't have a real human to copy. Perhaps they grow one from the DNA, and then populate their sim with what they learn from that.


    Where is the will that initiates the process?fishfry
    I can't answer for your view, but for the naturalists, it comes from different places, depending on what sort of thing is wanted.
    Most will comes from subconscious places (Limbic system), such as choices as to which way to swerve around the tree or to cheat on your spouse. But the will to choose option C in a multiple choice test comes from higher up (Cerebrum for instance).

    Ok. My reasons are irrational.
    I said that because the reasons seem backwards: Conclusion first, then selection of premises to support that conclusion. This is rationalization, something humans are very good at. I don't consider humans (myself included) to be very rational creatures.

    You sound like I said something that annoyed you.
    Not at all, but I apologize if my words annoyed you. The effect was not intentional.
  • Patterner
    665
    The Matrix is also an example of a VR, not an example of the simulation hypothesis.noAxioms
    I guess the Matrix is a simulation to many sentient programs, and VR to many other sentient programs (Smith and Oracle, for example) and humans.
  • fishfry
    3k
    Where is the will that initiates the process?
    — fishfry
    I can't answer for your view, but for the naturalists, it comes from different places, depending on what sort of thing is wanted.
    Most will comes from subconscious places (Limbic system), such as choices as to which way to swerve around the tree or to cheat on your spouse. But the will to choose option C in a multiple choice test comes from higher up (Cerebrum for instance).
    noAxioms

    I have no limbic system. Only a simulation of a limbic system in a computer, if I understand you correctly (clearly I don't, right?) A computer simulation of a limbic system cannot create emotions any more than a simulation of gravity attracts nearby bowling balls.

    I said that because the reasons seem backwards: Conclusion first, then selection of premises to support that conclusion. This is rationalization, something humans are very good at. I don't consider humans (myself included) to be very rational creatures.noAxioms

    I agree with you there. We all have prejudices that make us gravitate towards one pole or the other of unanswerable questions.

    Not at all, but I apologize if my words annoyed you. The effect was not intentional.noAxioms

    Well I confess that I have no proof for my opinions or biases.
  • noAxioms
    1.4k
    I have no limbic system. Only a simulation of a limbic system in a computer,fishfry
    You sound like Arkady, but no, that statement is misleading. It makes it sound like the limbic system is simulated but you are not. So either "I have a limbic system", or "The simulated 'I' has a simulated limbic system". Either of those wordings is at least consistent. Your opinion (and mine, but for very different reasons) of course is that neither you nor your limbic system are the product of a simulation.
    Nobody is claiming that a simulation of X creates an X in the simulating world, which is the strawman you seem to use in your gravity example every time where you deny an equivalent straw claim that simulation of gravity would create gravity in the GS world. That you persist in this suggestion means that yes, you're not getting it right, perhaps deliberately so.
    So no, a simulation in the GS world of a limbic system does not create emotion in the GS world. I agree with that. It is exactly for that reason that the program running the simulation isn't conscious.
  • fishfry
    3k
    You sound like Arkady,noAxioms

    I don't read this thread, I only respond to my mentions. So I have no idea what @Arkady may have said. If I sound like him he must be an individual of deep insight and wisdom :-)

    but no, that statement is misleading. It makes it sound like the limbic system is simulated but you are not.noAxioms

    I asked where will comes from. Intensionality. Caring. Feelings. We know from biology that feelings come from the limbic system. But if I'm simulated, so is my limbic system. Raising the question once again of how a computer program can have feeeelings, nothing more than feeeeelings.

    So either "I have a limbic system", or "The simulated 'I' has a simulated limbic system". Either of those wordings is at least consistent. Your opinion (and mine, but for very different reasons) of course is that neither you nor your limbic system are the product of a simulation.noAxioms

    Well I agree with that.

    Nobody is claiming that a simulation of X creates an X in the simulating worldnoAxioms

    That's exactly what's claimed.

    , which is the strawman you seem to use in your gravity example every time where you deny an equivalent straw claim that simulation of gravity would create gravity in the GS world.noAxioms

    Just pointing out that computer simulations of gravity don't attract bowling balls (clearly true) and that therefore simulations of brains do not necessarily implement minds.

    That you persist in this suggestion means that yes, you're not getting it right,noAxioms

    No hope for me, clearly, after all this time.

    perhaps deliberately so.noAxioms

    Oh, I have bad will. But even so, you admit I have will! Therefore I am NOT likely to be a computer simulation. I will, therefore I am. Or as the song goes ... if it weren't for bad will, I wouldn't have no will at all.

    But really. After all this you have to accuse me of bad will? How am I supposed to take that?

    So no, a simulation in the GS world of a limbic system does not create emotion in the GS world. I agree with that. It is exactly for that reason that the program running the simulation isn't conscious.noAxioms

    So you agree with me after all. Or at least, I agree with what you wrote here. A program isn't conscious, it does not implement or instantiate consciousness, and it does not "simulate" consciousness. Consciousness is not the kind of thing that can be simulated, unless you think chatbots simulate consciousness. Many people believe that these days.

    Deliberately not getting it right. No. False. I'm not trolling you to annoy you. Why did you say that?
  • noAxioms
    1.4k
    Nobody is claiming that a simulation of X creates an X in the simulating world
    — noAxioms

    That's exactly what's claimed.
    fishfry
    Who makes that claim? Quote it please. If you can't do that, then you're making a strawman assertion.

    simulations of brains do not necessarily implement minds
    Not minds/people in the GS world, no. The claim is that we (the simulated people with yes, simulated minds) are in this simulated universe, and not in the universe running the simulation.

    you admit I have will! Therefore I am NOT likely to be a computer simulation.
    A simulation of a person without will would be a simulation of a body in a vegitative state.

    After all this you have to accuse me of bad will?
    What, my saying 'deliberate'? You seem to be putting words in people's mouths that they didn't say, and I don't find you to be an ignorant person.

    A program isn't conscious,
    Not the simulation being discussed here, correct. A running computer process forever without inputs by definition cannot be conscious any more than you would be without inputs ever.

    unless you think chatbots simulate consciousness. Many people believe that these days.
    I have a very loose definition that you would not like, but my opinion there is irrelevant. The chatbots (which perhaps imitate, but not simulate anything) at least have input, but so does a thermostat. The simulation in question does not.
  • fishfry
    3k
    Who makes that claim? Quote it please. If you can't do that, then you're making a strawman assertion.noAxioms

    The voices in my head. Put there by our Simulator who art in heaven.

    Not minds/people in the GS world, no. The claim is that we (the simulated people with yes, simulated minds) are in this simulated universe, and not in the universe running the simulation.noAxioms

    Yes, we're characters in a video game, with the assumption that Ms Pac-Man has an inner life. I believe I rejected that assumption a while back.

    A simulation of a person without will would be a simulation of a body in a vegitative state.noAxioms

    I don't see that. Isn't a simulation of a person without a will exactly what they call a philosophical zombie? It would literally be a terrific chatbot operating inside a highly realistic flesh and bone bot. Your neighbor, for instance. What makes you think they have a will?

    What, my saying 'deliberate'? You seem to be putting words in people's mouths that they didn't say, and I don't find you to be an ignorant person.noAxioms

    Perhaps I over reacted.

    Not the simulation being discussed here, correct. A running computer process forever without inputs by definition cannot be conscious any more than you would be without inputs ever.noAxioms

    Hmm. That raises some questions. The simulation program has no input. You write the code, then you execute the code and it does what it does.

    What is its output? How exactly do the Simulators examine its inner life? In other words, they run the program, and inside the program I come into existence. Me with my subjective experience. (How does that happen? Remind me please). Clearly they are interested in what I'm thinking and experiencing ... or are they only interested in my actions? So two questions:

    1) Do the simulators have access to my internal mental states, and if so, how? Copious log files of everything I'm thinking? and

    2) How do I perform actions for the Simulators to watch? They're running ancestor simulations, so they must want to see what I'm going to do next. How do they "watch" me? What are the outputs?



    I have a very loose definition that you would not like, but my opinion there is irrelevant. The chatbots (which perhaps imitate, but not simulate anything) at least have input, but so does a thermostat. The simulation in question does not.noAxioms

    You are avoiding the question of whether the sims are self-aware? I didn't understand this remark.
  • noAxioms
    1.4k
    I don't see that. Isn't a simulation of a person without a will exactly what they call a philosophical zombie? It would literally be a terrific chatbot operating inside a highly realistic flesh and bone bot. Your neighbor, for instance. What makes you think they have a will?fishfry
    You seem to have a dualistic definition of 'will'. All of your examples (pacman, p-zombies) are dualist/VR references. Bostrom's hypothesis is not. He's not proposing we're in a video game. All this has been said before.

    The simulation program has no input. You write the code, then you execute the code and it does what it does.
    That's what a simulation is, yes. It has an initial state conveyed to it, and that is input of sorts, but once the simulation begins, there is no further input of any kind. If there was, it ceases to be a simulation. I've run plenty of these myself. It was my job for a while. The sims would run without any I/O at all for perhaps a week, and I don't think results were available until the end, but they could be reported as they happen.

    What is its output?
    Output (state of system at any given time) can be had any time, often at the end, but it doesn't have to be. A weather sim is a single simulation of a storm, and it could output the stats of the storm at regular intervals, or it could wait until the end and output the whole thing in a lump. It has to complete in hours, not days, to be useful. My chip sims were a little difference since each chip was run through a series of discreet tests, mostly designed to see how fast you could clock it before it started misbehaving, but also to check the design for bugs. Those sims still output everything at the end, but they didn't have to.

    How exactly do the Simulators examine its inner life?
    They don't. It makes no more sense than asking what it is like for a human to be a bat.

    In other words, they run the program, and inside the program I come into existence. Me with my subjective experience. (How does that happen? Remind me please).
    Same way it happens in the real (materialist) world: Particles interact and do their thing. Your experience is a function of matter interactions (not so according to someone like Chalmers, whom you referenced with the p-zombie mention above).

    Clearly they are interested in what I'm thinking and experiencing
    The simulation itself cares about what you're thinking, but only because it needs to change physics due to it. The runners of the simulation may or may not care. Certainly they don't have enough people to care about every single individual. It's an ancestor simulation of the whole human race. They perhaps want to see what history unfolds, and they care no more about what anybody is thinking than you do about what anybody is thinking. You only care about what they say to you, what they do. You may wonder what goes on inside, but that's a motive for a single-person simulation, not a planetary scale one.

    1) Do the simulators have access to my internal mental states, and if so, how? Copious log files of everything I'm thinking? and
    If 'the simulators' are those that put together the simulation, who want the ancestor sim, then they have perhaps access to the same data as we do with a pimped-out MRI scan: A picture of where the matter is. You're not getting thoughts from that. To log thoughts, something needs to interpret that matter state and render it into language for readable by the simulators. I suppose such log files are possible, but much of thoughts are not in language form.
    And per above, if this is the sort of detail one wants, it makes far more sense to simulate one or a very small number of people. So the motives are probably different for the ancestor sim.

    2) How do I perform actions for the Simulators to watch? They're running ancestor simulations, so they must want to see what I'm going to do next. How do they "watch" me? What are the outputs?
    Up to them to design a way to do it that is useful for their purposes. I suppose one could insert a sort of point of view interface that lets one look from any event anywhere (much like the little guy you can steer around in google maps), and lets it move at the observers control. The sim would need to save all state (and not just current state) for this to work since it probably wouldn't be useful if it was 'live', displaying only what constitutes the current state of the sim.

    You are avoiding the question of whether the sims are self-aware?
    I presume that 'the sims' are the humans in the simulation.
    The hypothesis is that the sims are us, so tautologically they're as self-aware as you are.

    If 'the sims' is a reference to the simulation software, program, or process, well that's a different answer since people are not hypothesized to be any of those things.
  • Ludwig V
    1k
    I presume that 'the sims' are the humans in the simulation.
    The hypothesis is that the sims are us, so tautologically they're as self-aware as you are.
    If 'the sims' is a reference to the simulation software, program, or process, well that's a different answer since people are not hypothesized to be any of those things.
    noAxioms
    So the humans are entities created by the software? Then how are they not real people and not simulations of anything?
    Particles interact and do their thing. Your experience is a function of matter interactionsnoAxioms
    Quite so. But my experience is real experience, not a simulation of experience. So the people "inside" your software are real people.
  • noAxioms
    1.4k
    So the humans are entities created by the software?Ludwig V
    I would say the humans are entities created by rearrangement of matter, and that the matter in this case happens to be simulated by the running process in the supervening world. It's a choice of how to word things is all.

    Then how are they not real people and not simulations of anything?
    They are (hypothesized as being) you, and you are real, per your definition:
    If I'm experiencing fear, the fear is real.Ludwig V

    But my experience is real experience, not a simulation of experience.
    You seem to be inconsistent with your usage of 'real'. Have you switched to a different definition?

    So the people "inside" your software are real people.
    It's not my software. It's the software of the entities running the simulation, which isn't me. I am hypothesized to be the product of that simulation, not hypothesized to be creating or running one.
  • Bylaw
    550
    First, if the world is simulated, why don't its 'designers' simply 'pop out' at times and leave us with some trace of their existence? Guidance through such a virtual world might be helpful, and yet there is no trace of anyone 'programming' or 'guiding' us anywhere.jasonm
    Some possibilities:
    they don't want us to know
    it's merely entertainment for them
    they don't realize we're conscious, they think of it as more like a 3D film
    the do give us guidance, but not a lot - perhaps the voices prophets hear, perhaps insights people get regarding morals or science or whatever.
    Perhaps it's a work of art and the whole idea is to let it run itself.
    Perhaps it is in an experiment and there are experiments where they interfere and where they don't
    Then:
    how the heck would we know the motives of creatures other than us and that advanced

    Similarly, why don't we sometimes notice violations of the laws of physics? If it's just a simulation, does it matter if the laws of physics are perfectly consistent? This applies to any law of this simulated world, including propositional logic. Again, if you are there, leave us with some trace of your existence through 'miracles' and other types of anomalies that our world does not seem to have. And yet there seems to be no instances of this kind.jasonm
    No instances of anomalies? There are often anomalies. Perhaps in the end they will be explained, perhaps not. In any case, we now explain away anomalies even if we really don't know.

    But again, you're making assumptions. Perhaps they don't want anomalies in their experiment, entertainment, artwork, whatever this is to them. It matters to them. And since they're making it.....
    Third: what type of computing power would be required to 'house' this virtual universe? Are we talking about computers that are bigger than the universe itself? Is this possible even in principle?jasonm
    We wouldn't know how big the universe is. We only know what we know about our universe, which would be simulated. Whatever is outside it in which it is running would be beyond our ken. I'm sure educated, medieval people would dismiss descriptions of things we can do now as being impossible. But what did they know about humans would later be able to do? What do we know?

    As far as Occam's Razor....
    Probability and Indifference: Bostrom's simulation argument doesn't posit that the simulation hypothesis is necessarily simpler or more straightforward than the idea that we live in a base reality. Instead, it suggests that given certain plausible assumptions about the future capabilities of civilizations, the probability that we are in a simulation might be high. The argument hinges on three propositions:

    The fraction of human-level civilizations that reach a posthuman stage (capable of running simulations) is very close to zero.
    The fraction of posthuman civilizations that are interested in running ancestor simulations is very close to zero.
    The fraction of all people with our kind of experiences that are living in a simulation is very close to one.
    If the first two propositions are false, then the third proposition must be true, meaning we are almost certainly in a simulation.

    Reframing Occam’s Razor: Bostrom might argue that Occam's Razor should be applied to the assumptions underpinning each hypothesis. The simulation hypothesis, when considered in the context of his argument, doesn’t necessarily introduce more assumptions than the assumption that we live in the one base reality, especially given the potential vastness of simulated realities versus a single base reality.

    Technological Plausibility: Bostrom might point out that the simulation hypothesis stems from an extrapolation of known technological trends. Given the rapid advancement in computing and virtual reality, the assumption that future civilizations will have the capability and possibly the desire to run detailed simulations is not implausible. Thus, it is not an extraordinary leap in assumption.

    The Simulation Argument’s Structure: Bostrom’s argument is structured to show that at least one of the three propositions must be true, making it a probabilistic argument rather than one based solely on the principle of simplicity. The argument demonstrates that if advanced civilizations are likely and interested in running simulations, it becomes statistically more probable that we are in a simulation.

    Not Claiming Proof: Bostrom doesn’t claim that the simulation hypothesis is definitively true; rather, he argues that it is a hypothesis that should be taken seriously given the logical structure of his argument. He acknowledges that the base reality hypothesis is simpler in some ways but insists that the simulation hypothesis has significant probabilistic support under certain assumptions.
  • fishfry
    3k
    You seem to have a dualistic definition of 'will'. All of your examples (pacman, p-zombies) are dualist/VR references. Bostrom's hypothesis is not. He's not proposing we're in a video game. All this has been said before.noAxioms

    I'm coming to the end here. My interest in this topic is far exceeded by my word count at this point. At one point I thought I understood the sim/VR distinction. But once I found out that Bostrom explicitly assumes that the simulation implements consciousness, the distinction becomes moot.

    That's what a simulation is, yes. It has an initial state conveyed to it, and that is input of sorts, but once the simulation begins, there is no further input of any kind. If there was, it ceases to be a simulation. I've run plenty of these myself. It was my job for a while. The sims would run without any I/O at all for perhaps a week, and I don't think results were available until the end, but they could be reported as they happen.noAxioms

    Ok, no input.

    Output (state of system at any given time) can be had any time, often at the end, but it doesn't have to be. A weather sim is a single simulation of a storm, and it could output the stats of the storm at regular intervals, or it could wait until the end and output the whole thing in a lump. It has to complete in hours, not days, to be useful. My chip sims were a little difference since each chip was run through a series of discreet tests, mostly designed to see how fast you could clock it before it started misbehaving, but also to check the design for bugs. Those sims still output everything at the end, but they didn't have to.noAxioms

    You did not answer the question. Can you see that?

    What is the output? I did not ask WHEN is the output. I asked WHAT is the output.

    They don't. It makes no more sense than asking what it is like for a human to be a bat.noAxioms

    So the sims have an inner life (one of Bostrom's hidden assumptions) but the simuilators have no knowledge of it? You are really out on a limb. They care out their ancestor sims act but not what they think and feel, even though (somehow) they managed to make them think and feel?


    Same way it happens in the real (materialist) world: Particles interact and do their thing. Your experience is a function of matter interactions (not so according to someone like Chalmers, whom you referenced with the p-zombie mention above).noAxioms

    So YOU know how consciousness works. Why do you bother even trying to communicated with one so ignorant as me, who doesn't think ANYONE knows that?

    The simulation itself cares about what you're thinking, but only because it needs to change physics due to it. The runners of the simulation may or may not care. Certainly they don't have enough people to care about every single individual. It's an ancestor simulation of the whole human race. They perhaps want to see what history unfolds, and they care no more about what anybody is thinking than you do about what anybody is thinking. You only care about what they say to you, what they do. You may wonder what goes on inside, but that's a motive for a single-person simulation, not a planetary scale one.noAxioms

    So step one, they figure out how to implement consciousness using computers; and step two, they entirely ignore that and focus on behavior.

    And again, how is that behavior communicated to them? What is the output? What is the output? Ask yourself if anything you're saying makes sense?

    If 'the simulators' are those that put together the simulation, who want the ancestor sim, then they have perhaps access to the same data as we do with a pimped-out MRI scan: A picture of where the matter is.noAxioms

    An MRI does not provide access to internal mental states. You know that.

    You're not getting thoughts from that. To log thoughts, something needs to interpret that matter state and render it into language for readable by the simulators. I suppose such log files are possible, but much of thoughts are not in language form.
    And per above, if this is the sort of detail one wants, it makes far more sense to simulate one or a very small number of people. So the motives are probably different for the ancestor sim.
    noAxioms

    You're just speculating about your own confused ideas. You are not making sense. Perhaps we're at a point of putting this convo to rest. Bostrom says the computers implement consciousness. And I am asking you, what are the outputs of the simulation?

    Up to them to design a way to do it that is useful for their purposes. I suppose one could insert a sort of point of view interface that lets one look from any event anywhere (much like the little guy you can steer around in google maps), and lets it move at the observers control. The sim would need to save all state (and not just current state) for this to work since it probably wouldn't be useful if it was 'live', displaying only what constitutes the current state of the sim.noAxioms

    Ok. I'm asking you question you can't answer. And instead of saying, "You know, you have a bit of a point there," you're just making stuff up. And I'm getting a bit annoyed. Granted I annoy easily sometimes but this is one of those times.

    I presume that 'the sims' are the humans in the simulation.
    The hypothesis is that the sims are us, so tautologically they're as self-aware as you are.
    noAxioms

    The sims are programs. What are their outputs (he asked again).

    If 'the sims' is a reference to the simulation software, program, or process, well that's a different answer since people are not hypothesized to be any of those things.noAxioms

    That is exactly what Bostrom is hypothesizing!

    Could we agree to disagree? Could you accept that you can't answer any of these questions except by making stuff up? This has been an interesting convo but this last post did not have any content IMO. Just handwaving about questions you can't answer. I don't want to leave in a huff, but I might have to leave in a minute and a huff, as Groucho said.
  • noAxioms
    1.4k
    I asked WHAT is the output.fishfry
    We can only speculate as to the purpose of running this kind of simulation, and thenature of the output depends on that purpose. Maybe it is a sort of detailed history book. Maybe it is pictures. Maybe it's just a stored database. Maybe the purpose is simply to see how long humanity lasts until it goes extinct, in which case a simple number might be the output.
    I did mention the nature of the output later in the post above, such as the example of the output of google maps for instance, a very useful interface for display of simulation results.

    So the sims have an inner life (one of Bostrom's hidden assumptions)
    You define 'the sims' below to be the programs in the GS world. I see no assertion that either a program (a static chunk of software on perhaps a disk somewhere) or a computer process (the execution of said program on some capable device) with no inputs would have what you might consider to be an 'inner life'. Bostrom doesn't say this, and neither do I.

    but the simuilators have no knowledge of it?
    They have knowledge of it in the same way that I have knowledge of my wife having an inner life. If that's going out on a limb, then one is presuming solipsism. But my presumption of my wife having inner life does not let me know what it's like to be her.
    The simulation can report what each person thinks and feels. The simulation has to have access to this because physics is dependent on what people are thinking. So it can report that Bob at time X is paying attention to his laser experiment and is feeling frustrated that he cannot get the setup just right, and his bladder is getting full. It can show his point of view if that helps. Make up your story. What interface tech exists for them is speculation on our part. Humans are notoriously bad at predicting 'future'/higher tech.

    I put 'future' in scare quote because maybe the simulation is being run in the year we call 1224 or something. Maybe in the GS world, advancements came much sooner, and in our simulated world, things happened much slower, and we're far behind them despite 8 more centuries to learn. If that is the case, the Gregorian calendar is only meaningful in our world, and they number their years differently.

    So YOU know how consciousness works.
    Geez, another strawman. I make no such claim. Bostrom presumes that consciousness is physical/computational. That assumption is no more an explanation of how consciousness works than is the non-explanation by anybody else.

    So step one, they figure out how to implement consciousness using computers; and step two, they entirely ignore that and focus on behavior.
    I didn't say they figured out how consciousness works, nor did I say they focus only on behavior. The simulation needs to know what each persons mental focus is, what his intent is, because physics as he describes it depends on it. One doesn't need to know how consciousness works to do this.

    And again, how is that behavior communicated to them?
    There's no 'them' to communicate to. OK, observers in the GS world can watch, (very similar to the google map interface), but they don't affect anything since that would constitute external input. The running of any sim doesn't require observation of any kind, but why run it if nobody's going to pay attention to the outcome? Yet again, the output is dependent on the purpose of running the thing, and we can only speculate on the purpose.

    An MRI does not provide access to internal mental states. You know that.
    A full classical scan of a person provides access to internal physical states, and that's all that's needed to simulate the person, per naturalism. But such a simple simulation would not have physics supervening on mental states like the sim Bostrom proposes, so the one he speculates is far more complicated and requires access to mental states, not just physical states.

    You're just speculating
    Yes, with that quote, I was. I don't know the purpose of the sim, and I don't know what tech is available to the entities running the sim, so I can only speculate as to how they would choose to 'observe' it.

    The sims are programs.
    Ah, not us, but the program in the GS world. Apologies for getting that wrong. Sims then typically not conscious, especially since it typically lacks input.

    Could you accept that you can't answer any of these questions except by making stuff up?
    Me saying what the output would be is definitely making stuff up. Me knowing what a simulation is and how it typically works is not making stuff up, since I did it regularly.

    Our opinions definitely differ, but I'm trying not to assert opinions. I'm trying to interpret what Bostrom's opinion is, and how he attempts to back it.
  • Barkon
    147
    Simulation doesn't have to be contrary to the norm of reality, simulation can coincide with the norm. For example, each star may have a system, and each system is separate from the other - no system contains another system. In this way the universe is simulated systematically. Minds only 'load in' the presence of their solar system, and other systems aren't 'loaded in', but will be if mind becomes local.
  • fishfry
    3k
    We can only speculate as to the purpose of running this kind of simulation, and thenature of the output depends on that purpose. Maybe it is a sort of detailed history book. Maybe it is pictures. Maybe it's just a stored database. Maybe the purpose is simply to see how long humanity lasts until it goes extinct, in which case a simple number might be the output.
    I did mention the nature of the output later in the post above, such as the example of the output of google maps for instance, a very useful interface for display of simulation results.
    noAxioms

    Ok, I'll concede that the sim program has some kind of graphic output that lets the simulators watch the ancestor simulation in action.

    BTW Google maps is not a simulation, it's a Geographic Information System. And it takes inputs, such as the zoom and recenter operations from the user.

    You define 'the sims' below to be the programs in the GS world.noAxioms

    Yes, what else could we be talking about? Bostrom: "Are YOU living in a computer simulation?" My emphasis. Me. You. Each of us. We are a program being run by the simulators. What else do you think he's talking about? The simulators run programs that that are us, in some magic way.

    I see no assertion that either a program (a static chunk of software on perhaps a disk somewhere) or a computer process (the execution of said program on some capable device) with no inputs would have what you might consider to be an 'inner life'. Bostrom doesn't say this, and neither do I.noAxioms

    Process. Executing program. If I said program, I should have written process, or executing program. I actually kind of doubt I said this, but if I said program and not executing program, I meant executing program.

    What on earth else do you think Bostrom means? Are you living in a computer simulation? What else can he mean?

    They have knowledge of it in the same way that I have knowledge of my wife having an inner life. If that's going out on a limb, then one is presuming solipsism. But my presumption of my wife having inner life does not let me know what it's like to be her.
    The simulation can report what each person thinks and feels. The simulation has to have access to this because physics is dependent on what people are thinking. So it can report that Bob at time X is paying attention to his laser experiment and is feeling frustrated that he cannot get the setup just right, and his bladder is getting full. It can show his point of view if that helps. Make up your story. What interface tech exists for them is speculation on our part. Humans are notoriously bad at predicting 'future'/higher tech.
    noAxioms

    Agreed, to a point. It's odd that Bostrom thinks the computers instantiate self-awareness in the sims, yet show little interest in it. Well it's a small point, virtually nothing in Bostrom's thesis holds up anyway.

    I put 'future' in scare quote because maybe the simulation is being run in the year we call 1224 or something. Maybe in the GS world, advancements came much sooner, and in our simulated world, things happened much slower, and we're far behind them despite 8 more centuries to learn. If that is the case, the Gregorian calendar is only meaningful in our world, and they number their years differently.noAxioms

    Bostrom clearly thinks the simulators live in (our) future and we are simulations of their ancestors. Though of course you're right, there's no reason that would be true. Maybe we're the Jetsons and not the Flintstones.

    Geez, another strawman. I make no such claim. Bostrom presumes that consciousness is physical/computational. That assumption is no more an explanation of how consciousness works than is the non-explanation by anybody else.noAxioms

    Ok.

    I didn't say they figured out how consciousness works,noAxioms

    Bostrom says that. That's the one great revelation I had from this thread. Bostrom explicitly states that the sims are self-aware, and blithely justified is as "it's widely believed."

    nor did I say they focus only on behavior. The simulation needs to know what each persons mental focus is, what his intent is, because physics as he describes it depends on it. One doesn't need to know how consciousness works to do this.noAxioms

    Ok.

    There's no 'them' to communicate to. OK, observers in the GS world can watch, (very similar to the google map interface), but they don't affect anything since that would constitute external input. The running of any sim doesn't require observation of any kind, but why run it if nobody's going to pay attention to the outcome? Yet again, the output is dependent on the purpose of running the thing, and we can only speculate on the purpose.noAxioms

    Of course they are watching, they are running an ancestor simulation. Of course I have no idea why Bostrom chose that particular reason, since with all our impressive computing power, WE don't run ancestor simulations. Maybe we're their pr0n hub. They like to watch us mate. That's more likely than that the history majors are running ancestor simulations.

    A full classical scan of a person provides access to internal physical states, and that's all that's needed to simulate the person, per naturalism. But such a simple simulation would not have physics supervening on mental states like the sim Bostrom proposes, so the one he speculates is far more complicated and requires access to mental states, not just physical states.noAxioms

    Right. You can map all the neurons and you would not know what someone's thinking. Although impressive work in that direction is being done by the cogsci crowd, so I could be proven wrong soon enough.

    Yes, with that quote, I was. I don't know the purpose of the sim, and I don't know what tech is available to the entities running the sim, so I can only speculate as to how they would choose to 'observe' it.noAxioms

    Ok. I'll concede that they have a graphic or numeric output that can be observed. And they can talk about their mental states, as we often do. "I think I'm hungry."

    Ah, not us, but the program in the GS world. Apologies for getting that wrong. Sims then typically not conscious, especially since it typically lacks input.noAxioms

    ARGHHHHHH! The sims are conscious. That's on page one of Bostrom's paper. We are the sims. After all this, are we not at least agreed on this?

    Me saying what the output would be is definitely making stuff up. Me knowing what a simulation is and how it typically works is not making stuff up, since I did it regularly.noAxioms

    Ok.

    Our opinions definitely differ, but I'm trying not to assert opinions. I'm trying to interpret what Bostrom's opinion is, and how he attempts to back it.noAxioms

    That's the funny thing. You have said you don't agree w/Bostrom. And for some reason, that makes you want to put great effort into explaining his wrong position to me.

    One more thing: Last night, you said: "If 'the sims' is a reference to the simulation software, program, or process, well that's a different answer since people are not hypothesized to be any of those things."

    Have you retracted that yet? It's Bostrom's thesis that people ARE hypothesized to be those things. "Are you living in a computer simulation." Bostrom speculates that WE are sims.

    Surely we agree on that, at least, yes? No?
  • noAxioms
    1.4k
    BTW Google maps is not a simulation,fishfry
    No, but it has an interface which is the beginnings of what one might look like for viewing simulation states. Yes, the controls to the tool constitute input to the tool, but since viewing simulation results has zero effect on the simulation itself, it doesn't count as input to the simulation, only input to one of many read-only tools to view the data produced by the simulation.

    Google maps can only show you specific places. You can go into a few select buildings, but your view is mostly confined to streets. With the simulation, there is no restriction of views only where the van was, taking a picture every 10 meters or so. You can go inside walls and watch the rats eat the wiring if you want, even if it's totally dark in there.

    You define 'the sims' below to be the programs in the GS world.
    — noAxioms
    Yes, what else could we be talking about?
    I thought they were the people, not the programs.
    But you defined it earlier to mean 'the simulation processes", of which there may be many running at once, each simulating a different world.

    Note: You yet again redefine 'sims' to be the people below. Using the word in both ways is the source of so much of our disconnect.

    Bostrom: "Are YOU living in a computer simulation?" My emphasis. Me. You. Each of us. We are a program being run by the simulators.
    'Living in a computer simulation" is different from being that computer simulation. The two exist in different worlds. They're not the same thing. The simulation runs in the GS world. We exist in this (simulated) world. That's the distinction I've been trying to stress. I'd try to use your meaning, but all sorts of strawman conclusions can be drawn when one equates the two very distinct things, such as "the simulation program is conscious'" which it isn't even though you and I are. Simulation programs tend to be very simple, endlessly running the same relatively small list of instructions again and again over a relatively large data set.

    I meant executing program.
    I know. It is still a mistake to say you are an executing program, for the reasons stated just above and in prior posts.

    It's odd that Bostrom thinks the computers instantiate self-awareness in the sims, yet show little interest in it.
    Presuming 'sims' is the people with this comment, else it makes no sense.

    It's a very weak point in his argument in my opinion, so he avoids it. To run a good ancestor simulation like this, it would require far less resources to have a good AI imitate (rather than simulate) each of the people. We're talking about something far better than passing a Turing test since each person needs to not just type like a human, but to act and defecate and bleed like a human. Now your ancestor sim can go on at perhaps a thousandth of the resources needed to do it at the level of simulation of consciousness of each person. But his hypothesis requires this, so he's forced to posit this implausible way of achieving the goal he's made up. The ratio is likely waaaay more than 1000-1.

    He tries to address this by waving away my '1/1000th' guess with 'we don't know the real number'. He calls the imitation people (as opposed to fully simulated ones) 'shadow people', and discounts this strategy, and yet gives every simulated person a shadow body and populates the world with shadow animals and plants and such, none of which is actually simulated like the brains are. Go figure.

    Bostrom clearly thinks the simulators live in (our) future and we are simulations of their ancestors.
    The initial state of the sim had perhaps some real ancestors (depends what date they selected), but we (the descendants of those initial people) are not in any way their ancestors, and thus the simulators are not in our future, only the future of some past year they selected for their initial state.

    Yes, I agree with you that Bostrom seems to imply that history would play out more or less the same, in which case he's just fooling himself, or, if there's a script, it's not a simulation at all, but just a CG effect for a movie script, which doesn't involve people that need to make their own choices.

    Bostrom says that. That's the one great revelation I had from this thread. Bostrom explicitly states that the sims are self-aware, and blithely justified is as "it's widely believed."
    And I buy that. Yes, the simulated people (and not the simulation processes) are self aware. But he doesn't explicitly say that anybody knows how 'consciousness works'. You don't have to. You put matter together like this, and the thing is conscious. That's what the sim does. It just moves matter. It doesn't need to know how the emergent effects work.

    That's more likely than that the history majors are running ancestor simulations.
    Agree. Or the biologists, which is a history major of sorts. What will they get from a sim that starts at a state resembling some past state, but evolves in a completely different direction? Not much. What if you run a thousand of them, all with different outcomes. Now you have statistics, and that's useful. Output would look like a history book. 'Watching' specific events from a selected point of view probably won't be too useful for that, but such a view would be useful to find the initial cause of some avoidable calamity (like a war) which helps our future people know what to look for to prevent their own calamities.

    Point is, that's a good starting point to resolve the 'why would such a sim be run'? I also still say that imitation, not full simulation, would be a far less costly way to achieve any of the goals mentioned. Only Bostrom requires it, but he can't force the 'future' people to do it an inefficient way.

    You can map all the neurons and you would not know what someone's thinking.
    But they kind of already do. They can put a thing on your head, measuring only external EM effects on your scalp (like an EEG) and they can see you make a decision before you're aware of it yourself. Point is, one doesn't need to know 'how consciousness works' in order to gean what the sim needs, which is mostly focus and intent. What is our guy paying attention to? Why? The sim needs to know because the physics of that thing is dependent on it., It changes from when nobody is paying attention to it. This is done for optimization purposes, and for faking non-classical effects in a classical simulation.

    The sims are programs.fishfry
    ARGHHHHHH! The sims are conscious. That's on page one of Bostrom's paper. We are the sims.fishfry
    Aaand the definition changes again. You said the sims are the programs. The programs are processes running in the GS world. We are humans living in this simulated world. Maybe we should stop using 'sims' as shorthand for this ever moving target.
    Be explicit. Use either 'simulated people' (us) or simulation process (the program running in a different world).

    Bostrom does not use the word 'sims', so it isn't on any page of his paper.
    He says on page 1 (the only reference to 'conscious' on that page): "Suppose that these simulated people are conscious". He is proposing that the people in the simulated world, and not the program running in the simulating 'future' world, is what is conscious. This is consistent with what I've been saying.

    He goes on later to presume substrate independence, which is that consciousness is not necessarily confined to carbon based biological forms. But the simualted people in his proposal are based on simulated carbon-based simulated biological forms. But he must say this to emphasize the standard objection that by definition, no computer can instantiate something conscious.
    Nowhere does he state that something as simple as a simulation process is itself conscious.

    That's the funny thing. You have said you don't agree w/Bostrom. And for some reason, that makes you want to put great effort into explaining his wrong position to me.
    Yea, that's right. There's indeed not much point in this since your personal beliefs conflict, so you won't consider it on its own grounds.

    Bostrom speculates that WE are sims.
    Surely we agree on that, at least, yes? No?
    You keep changing what 'the sims' means, and Bostrom doesn't use the word, so I cannot say yes or no.
    Bostrom does indeed speculate that it is more likely than not that we are simulated people: that we are composed of simulated matter being manipulated by a simulation process running in some other world. He nowhere speculates that we are that simulation process itself.
bold
italic
underline
strike
code
quote
ulist
image
url
mention
reveal
youtube
tweet
Add a Comment