The simulation hypothesis. Is life an illusion?

Seth Shostak takes a look at the simulation hypothesis, the idea that we are in a computer simulated reality: Is Life an Illusion? | Seth Shostak.

…a future historian (or curious teenager) wielding programming skills and access to a honking big computer could construct SimEarth on steroids. They could, for example, run a simulation of 15th century European society to see what it was like during the era of the Black Plague: a so-called ancestor simulation. Unlike Keanu Reeves in The Matrix, the people in the simulation wouldn’t know that their lives were merely code running in a machine.

Now if it gets this far, you can bet that the coders won’t run merely one simulation. Someone who has Grand Theft Auto on their machine doesn’t play it just once. They play it a gazillion times. In other words, if ancestor simulations are possible, then they will greatly outnumber real societies. Consequently, it’s very probable that all humanity is in a simulation — that we didn’t get one of those rare lottery tickets that would make us a real society of real beings. Everything you do today (and have ever done) could be just an illusion coded up by some clever Klingon.

Shostak list two possible alternatives to the probability of us being in a simulation.  The first is that societies rarely progress past our current state of technology to the point of being able to build these simulations (implying that our civilization is probably on borrowed time), and the second is that most societies don’t bother with simulations, which he finds implausible.

I’m not a big fan of this type of probability reasoning.  There’s just too much about reality that we don’t understand, and when I hear that calculations show that reality is probably an illusion, or that we’re all probably Boltzmann Brains, or anything along those lines, I’m much more inclined to think the calculation is simply flawed due to our ignorance.

That being said, I think there is at least one other possibility: that all encompassing simulations are much harder to do seamlessly, or will be much more expensive in resources, than we currently imagine them to be.  The idea of a proliferation of universe simulations assumes that computing power will continue to increase indefinitely.  I’ve noted before that this isn’t guaranteed.  Silicon performance may be within a decade of its performance limits, and quantum computing isn’t guaranteed to rescue us from those limitations.

Of course, there is the possibility that our universe actually is a simulation.  If it is, what does that mean for us?  Well, I think it would mean that our reality has a purpose, something many have hoped for, although that purpose might not be a comforting one.  It also means that the intelligence behind this simulation would effectively be our God or gods, again with no guarantee that they would be benevolent ones.

The problem is, if we are in a simulation, is there any way to detect it?  Shostak discusses attempts to see calculation grids in the cosmic microwave background, but I think the only real hope we might have of ever detecting a simulation is if it has any defects.  Of course, if we do find a defect, we might never know whether it is an actual defect or just an attribute of our universe.  For example, is our inability to reconcile quantum mechanics with general relativity due to a defect in the simulation, or to us simply not understanding the universe well enough yet?

If we are in a simulation, there is probably little we can know about the outside universe, at least without the intelligence behind the simulation communicating with us.  That outside universe, and the beings in it, may be unimaginably alien.  They may be incomprehensible to us.  They may also not even be aware of our existence, or if they are, they may regard us as an unwelcome complication in their experiment, vermin gumming up the works.  Worse yet, if the whole universe is being simulated, we are such a minor player in it that we might be irrelevant noise.

If we are in a simulation, then this simulation is our reality.  And the simulation appears to have stark consequences for failing to take it seriously.  As entities within the simulation, we have little choice but to continue playing the game, and hope the controlling intelligences don’t end it.

Another interesting question is if we do start bringing up simulations, at what point do the characters in those simulations start to be deserving of rights?  It seems like we should be careful about our attitudes toward those characters, since our own simulation owners might be watching and judging how we treat them.

23 thoughts on “The simulation hypothesis. Is life an illusion?

  1. It also means that the intelligence behind this simulation would effectively be our God or gods, again with no guarantee that they would be benevolent ones

    Not necessarily. If we are not in a simulation and if we don’t freewill, then its probable if we are in a simulation then the supposed beings that are simulating us also don’t have freewill. In which case they wouldn’t be our gods. Their behavior (and thus ours) would still be dictated by the laws of their physics.


    1. You seem to be saying that an entity has to have free will in order to be a god. I guess it depends on how you define “god” and “free will”. My usage of the “god” label was only intended to mean that they would have engineered our reality and that we’d be at their mercy.


      1. Well by freewill I mean that the entity can take decisions without getting influenced by external factors. My life depends on the laws of this universe but if this universe is a simulation then its at the mercy of the laws of the enclosing reality and so on. Well we all have different definitions of god but then why would we call something a god, if it doesn’t have freewill? I mean, if the god is at the mercy of another being or the laws of physics, would we still call it god? So technically, if we are simulated, then we wouldn’t be at the mercy of the simulated beings but at the mercy of the enclosing reality and so on.


  2. Hmm a very interesting hypothesis, but one that is hard to believe because it makes an inference from a hypothetical situation — essentially the argument is “given this imaginary scenario, this follows”. It’s just too far removed from reality to take seriously.

    The concluding question of your last paragraph is also very interesting. What sorts of rights would we grant them, and when would we enforce them? I’d imagine the idea is that they have rights against us doing certain cruel things to them. How they treat each other, in the simulation, is a different matter and it seems we shouldn’t intervene in one man taking the life of another in the simulation, as that is how the simulation plays out, and letting things unfold as they will seems to be the entire point of running a simulation. But if we are ourselves in a simulation that is being monitored by higher forms, then really the simulation we are observing is just a simulation in our simulation, and so the higher forms wouldn’t have any reason to interfere in how we treated other beings in our simulation, again, that is just how things play out in this simulation. So I’m not sure that it follows from the fact that we are in a simulation ourselves that we should be careful about how we treat the simulations we run.

    But I might just be completely lost in all these layers of reality lol.


    1. Good point. If we are in a simulation, the intelligence behind it does seem committed to letting it play out, at least as far as we can determine from within the simulation. This assumes that we can trust our memories, that the simulation didn’t start a few minutes ago after being adjusted and restarted numerous times. All of this gets to be a bit solipsist if you take it too far.

      On rights, if we have a simulation with self aware entities that intensely desire to survive, do we have the right to let the simulation play out if it means they will suffer? Do we have an obligation to step in? In fact, do we have a right to create self aware entities with survival instincts if we intend to eventually erase them?

      Suppose we have uploaded minds from humans in these environments? Do they deserve a different status than the ones who started out as simulations?

      Liked by 1 person

      1. Very interesting questions. The central question, of which I do not know the answer, seems to me to be: if as conscious, sentient beings, simulations have a right to be free from harm, does an obligation on our part to prevent their suffering depend on whether or not they know that we exist? Essentially, one person’s right creates an obligation for another, but does a right to be free from harm create an obligation on us if they don’t know we exist?


        1. If I understand what you’re saying correctly, the question is do we have an obligation to reach into their reality and interrupt the flow of things to prevent their suffering, particularly if they have no idea this outer reality exists? No doubt them not knowing of our existence would make non-interference easier, and many would reason that it’s not actually our reality, so they’re not real to us.

          I could see arguments both ways, but I would have a visceral discomfort with the idea of us creating self aware entities, putting them in a harmful environment, and then not interfering. Perhaps an easier situation for non-interference arises when such entities evolved in a long running simulation. Then they are completely creatures of their own environment and it seems easier, perhaps, to leave them to their fate.


        1. I enjoyed it immensely. Iain Banks’s culture novels are a lot of fun. At the center of the stories is a post-singularity space faring culture, but this particular book features virtual hells that some civilizations construct to resemble their mythological versions, and the resulting, um, controversy.


  3. Hi SAP,

    A few comments.

    The third possibility you mention is actually folded into the first one mentioned by Shostak, although it’s a little clearer in Bostrom’s formulation. Ancestor simulations being very difficult is basically just another way of saying we never develop the technology or wherewithal to implement ancestor simulations. It doesn’t necessarily mean we’re doomed in the short term.

    It could also be folded into the second possibility. If the human race survives indefinitely, expanding out throughout the galaxy, then the collective resources of the human race would likely be sufficient to conduct an ancestor simulation of some kind at some point in the future (millions or billions of years from now). But it may be so horrifically impractical and infeasible that it may never happen simply because of the cost and time scales that would be required. As such, it is not so difficult to imagine that what prevents it is our lack of interest, the same way that while we may have the resources to combat climate change, and even though doing so is vitally important to us, it is just too difficult to marshal that kind of commitment to a difficult and expensive project to make it happen.

    I have recently been debating via email whether it is scientifically possible that the universe could be a universe simulation with physicist Marko Vojinović, who wrote an article against determinism on Scientia Salon not so long ago. Since computers are deterministic machines, he believes his argument rules that out. Indeed, he seems to think that this is common knowledge in the physics community. I really don’t buy that, but unfortunately his proof relies on advanced mathematics I am not competent to judge, so the argument fizzled out in a rather unsatisfying way.

    Finally, again, I just want to mention the MUH (sorry for being a bit monotonous on this), because I think that it is very relevant to the simulation argument. If the MUH is correct, then it becomes meaningless to ask whether we are in a simulation or not. If all mathematical objects exist, and some are universes, then a simulation of a universe is just an exploration of that object which exists independently of the simulation. The universe is then not created by the programmers but discovered by them. If they pulled the plug we would continue to exist.

    At least, that is the case if we assume the programmers do not interact with the universe. If they react with this universe from outside, then our universe is no longer one that can be described by our laws of physics alone but needs to take into account the reality outside of the simulation. So, on the MUH, only if they interact with us does the question make sense. We would still continue to exist if they pulled the plug, however, because the state of the simulation at the point where they pull the plug comprises a mathematical object and the rules of the simulation are still defined even without input from outside, meaning that so is the future evolution of the universe if we take the point of final interaction as the initial state.


    1. Hi DM,
      Appreciate the clarification on Bostrom’s alternatives. I was going by Shostak’s summary of them, where the difficult / impractical option seemed to be missing.

      Your debate with Vojinović sounds interesting. If the universe is ultimately indeterministic, can a digital computing system model it? Possibly no, but then no simulation can perfectly match external reality. To me, the question would be, would it be close enough to be instructive? Also, there isn’t any rule that says a future computing system couldn’t have analog aspects to it, which might introduce indeterminancy. We don’t usually do that, because for most computing applications it’s undesirable.

      No worries on the MUH. I would be surprised at this point if you didn’t mention it in a discussion this close to it 🙂 Within the MUH, I can see where you’re coming from in saying that we would continue to exist if the simulation was shut off. Our pattern would still be “out there”.

      Of course, if the MUH isn’t true, then we disappear from reality. If they did shut down the simulation, I hope they wouldn’t be cruel enough to announce it.


      1. Hi SAP,

        In the debate with Vojinović, I’m conceding that a computer simulation cannot be fundamentally indeterministic, so I’m really attacking the notion that we can be sure the world is indeterministic. With reference to his article, I am choosing to resolve Bell’s theorem by rejecting locality and so embracing the idea that distant events can have instantaneous consequences of the kind we seem to see in making separated observations of entangled particles. I then run into the Cauchy problem, where Vojinović says that the equations of physics would entail multiple possible futures. Here, I assume that one could use a pseudorandom number generator to help construct an arbitrary future which is consistent with the equations, however he seems to think this move is not possible. I don’t understand his argument, but it seems to be that if one were to do that, it would be to effectively convert the system of equations to local ones, which would mean you run into the other side of his argument (Heisenberg and chaos theory). I don’t buy it but since he is appealing to mathematical results I am not educated enough to understand we are at an impasse.


        1. Wow, you guys are operating above my head. Just looked up the Cauchy problem, and was starkly reminded of how lacking my mathematical knowledge is. (Intro calculus is as far as I ever got and even that is hopelessly atrophied.)

          I’m somewhat surprised to hear you reject locality, since it seems like you usually fall more into the non-realism camp, although I guess those don’t have to be exclusive.

          I’m personally undecided on determinism. It seems evident that at a pragmatic level, there are situations where it’s false, and that might make whether it’s true at a metaphysical level irrelevant. But then, today’s metaphysics might be tomorrow’s science, so who knows.


      2. I’m not really operating above your head, Vojinović is. I don’t understand the Cauchy problem. The first time I heard of it was in his article. I just don’t see how it can be that we have a set of partial differential equations which are consistent with many possible futures, but we are prevented from picking one particular future using pseudorandom number generation. At an intuitive level, it seems obvious to me that this cannot be correct. I’m not going to doubt an accepted mathematical proof, but I suspect something may be amiss with Vojinović’s interpretation.

        With regard to locality vs realism, I don’t see how one can really prefer locality, particularly if one is trying to make a case for the universe being a mathematical object or computer simulation. Any computer simulation has a real definite state at any given time, which means we require realism. It doesn’t have to mean that there is an explicit state for any particular measurement before we make it, but it does mean that there must be sufficient state to derive that state at the time of measurement.

        So, I do think that the universe is a mathematical object, but this doesn’t mean that I don’t think it’s real (because if reality is what exists and I’m a Platonist then the universe exists and is real), but more importantly this kind of realism is not, I believe, at all related to the kind of realism at issue with the Bell theorem, which pertains to the realism of state between measurements.

        I am also undecided on determinism, or rather I would be if I were not so persuaded by the MWI. My issue is only with his argument that determinism can be ruled out. Incidentally, his argument is compatible with the MWI, which though deterministic by most definitions is not deterministic by Vojinović’s definition because it entails multiple futures extending from here. As such I am specifically interested in discussing his claim to have refuted determinism in the context of a computer simulation.


        1. Ah, I was confusing realism with your views on existence. Makes sense.

          I do agree that Vojinovic’s confidence in indeterminism seems hasty. Didn’t realize that there were versions of the MWI that are not deterministic, although it does seem one of its strengths is that it bring determinism back into QM, albeit in a way that we can never access.


      3. The MWI *is* deterministic. But it is not “Vojinović-deterministic” because Vojinović has his own idiosyncratic definition of determinism. For Vojinović, determinism means that this observable universe has one possible future. The MWI has one possible future too, but that future is for the wavefunction of the universe, which from our perspective entails many possible futures within that. For Vojinović, this is not determinism.

        Liked by 1 person

        1. I actually have some sympathy with Vojinovic’s version of determinism. Of course, just because the observable universe isn’t deterministic doesn’t mean that the universe overall isn’t. But from a methodological point of view, the universe we could investigate and work in would effectively be indeterministic.


    1. Good point. Of course, it’s always possible that they’re using reasoning similar to the ones that ausomeawestin and I engaged in above. It’s also possible that suffering might be intentionally part of the simulation. Perhaps they want to see how we cope with it.


Your thoughts?

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.