Maybe we’re not living in a simulation

Nick Bostrom has this interesting simulation argument thing (see also his discussion in our Ideas of the Century series). It holds that at least one of the following three propositions must be true:

  • Almost all civilisations at our level of development become extinct before becoming technologically mature.
  • The fraction of technologically mature civilisations that are interested in creating ancestor simulations is almost zero.
  • You are almost certainly living in a computer simulation.

Okay so I was just eating a packet of crisps, and I came up with a couple of counter-arguments. The first was that the argument doesn’t get off the ground unless we’re not living in a simulation (because if we are living in a simulation then propositions 1 and 2 might not even make sense about the “real” universe). The second was that it’s possible to construct an identical argument with “baby universes” (thereby demonstrating that we’re not living in a simulation).

Annoyingly other people seem to have got there first with these arguments, which isn’t in the least bit surprising (of course). And Nick – who incidentally is super smart and a nice guy – has provided his response (see the FAQ here – arguments 4 and 16).

I’m not entirely convinced by his responses, but regardless, suppose we combine arguments 4 and 16 together? Like this:

If we’re living in a simulation or a baby-universe we have absolutely no way of knowing from our experience of that simulation or baby-universe, whether in the originating world (the “real” world), it is easier to construct simulations or baby universes, or whether – for whatever reason – an advanced people would be more inclined towards one rather than the other. It follows then we’re in no position to make a judgement about the relative probability of a simulation versus a baby-universe. Therefore, there is no reason to suppose that at least one of the propositions above must be true (and therefore the simulation argument fails).

Right, I’ve spent about 10 minutes thinking about this and I didn’t really know what the simulation argument was until about an hour ago, so I suppose my argument must be wrong.

So where’s it wrong?

Leave a comment ?

53 Comments.

  1. > If we’re living in a simulation or a baby-universe we have absolutely no way of knowing from our experience of that simulation or baby-universe, whether in the originating world (the “real” world), it is easier to construct simulations or baby universes, or whether – for whatever reason – an advanced people would be more inclined towards one rather than the other. It follows then we’re in no position to make a judgement about the relative probability of a simulation versus a baby-universe. Therefore, there is no reason to suppose that at least one of the propositions above must be true (and therefore the simulation argument fails).

    First approach: In our current universe, it is much much easier to construct simulations than baby-universes, yes? Scientists routinely run simulations rather than experiments, and even ordinary civilians spend thousands of hours on simulations (we call them ‘video games’).

    So, if our parent-universe or simulator is different, then that must be deliberate: they must deliberately have chosen to make our baby-universe or simulation biased in favor of simulation.

    Why would they do that? Whatever the reason, this is an additional assumption which is unjustified by any evidence. If you want to avoid the SA, you might as well adopt the assumption ‘advanced civilizations would just not do such things on ethical grounds*’.

    Second approach: simulation is more general than baby-universes. If you or I could write a world simulation, then it would be easy to add on an ability to create pocket universes; very straightforward from a programming point of view – it’s sort of like network programming, where each universe is a computer and they interact/communicate over some abstract channel.

    However, the reverse is not true: being able to make baby-universes does not lead to easy simulations. (You might be able to encode some sort of computation in the initial state of the baby universe, let it run to completion, and see what the results are, but this doesn’t seem easy at all.) You just get baby-universes, whatever one does with those. And inside baby-universes, there may be empirical evidence indicating a baby-universe – such as the ability to create more baby-universes, which by argument 1 indicates that we are in a baby-universe, since why would a simulater make its simulation inaccurate and unfaithful? If the presence of a baby-universe ability is evidence *for* being in a baby-universe, then the absence of a baby-universe must be evidence *against* being in a baby-universe (and thus evidence for being in a simulation).

    Hence, we should be biased in favor of simulations: there could be arbitrarily many of them nested within one another (my processor is a virtual machine for Linux which is a virtual machine for Firefox, which is a virtual machine for… an emulated MS Windows running VirtualBox hosting OpenBSD…), while there’s just the one parent-universe.

    * if video game characters were real and felt pain as much as you or I, I suspect first-person shooters would quickly be outlawed

  2. Gwern – Thanks for this.

    I don’t think either of your rebuttals work for the following reasons.

    Your first rebuttal

    1. The general point here is that the onus is on the person advancing the SA to demonstrate the falsity of an objection. And you can’t do this just by claiming that some possibility – in this case the proposition that people in the parent-universe would favour baby-universes over simulations – seems unlikely. You have to do better than that.

    2. Notwithstanding that point it’s also the case that the original SA allows for claims such as “advanced civilisations would not do such things on ethical grounds”. If that’s what you thought then you’d assert the truth of proposition 2, which means that the SA is true (because the SA claims that one of the three propositions is true);

    3. What my argument does is to say that one can never know how such things as ethical considerations, a preference between simulations and baby universes, etc., will play out between simulations versus baby universes. The significance of this is that one can assert that proposition 2 is false (i.e., the fraction of technologically mature civilisations that are interested in creating ancestor simulations is much greater than zero), while also asserting that proposition 3 is false, because the fraction of technologically mature civilisations interested in creating baby universes is greater than the fraction interested in creating ancestor simulations.

    Your second rebuttal

    You assert various things you can’t possibly know. For example:

    (a) being able to make baby-universes does not lead to easy simulations; (A very advanced civilisation might know exactly what conditions have to be met for easy simulations);

    (b) since why would a simulater make its simulation inaccurate and unfaithful; (A simulator might have all sorts of reasons we’re in principle unaware of);

    (c) then the absence of a baby-universe must be evidence *against* being in a baby-universe (and thus evidence for being in a simulation). (Entirely possible that a simulator might create a baby-universe that is set up in such a way that it cannot spawn baby universes. Assuming the laws of physics allow it, of course, which we can’t know, since the laws operating in our universe ain’t necessarily the same as the laws operating in a parent universe).

    So my argument stands. But thanks for the response. Appreciated! 🙂

  3. Jeremy,
    I got involved with this when it was posted on the Ideas of the Century. I don’t remember “baby universes” being mentioned. What are they?

  4. Baby universes are universes that we might create from within our own universe. Apparently they’re not ruled out by the laws of physics of our universe:

    http://www.casavaria.com/sentido/science/2006/06-0802-new-universe.htm

    So you can set up an equivalent argument as the SA, but for baby universes. Call it an SB argument.

    Then if you add it to the argument about the consequences of not knowing which we’re in – i.e., we have literally no way of assessing whether it’s more likely we’re in one rather than the other – then you get a refutation of the SA.

  5. Jeremy,
    Me again. Forget the previous question. I’m still in kindergarden wrt this problem. I can’t get beyond my own question of, so what?
    I’m going to be plebian about this and start with c being false. So, assuming there are other civilizations and those might have some of the same propensities we do then I’ll even say one or both of the first two statements are true. What makes that interesting?
    What do we get out of this?

  6. I think there are various things here. Partly you can’t just say “we’re not living in a simulation”, therefore, one or both of propositions 1 and 2 must be true, because if we were living in a simulation you’re not going to know it.

    So you’d have to offer reasons for thinking that either of propositions 1 and 2 are true – or as it’s set up by Nick Bostrom, do it in terms of probabilities.

    But, of course, if you’re doing it in terms of probabilities then you’re going to get the conclusion that there is some chance we’re living in a simulation. (Only since I’m still not convinced I haven’t refuted the argument, I don’t think that’s clear, but assuming the argument isn’t flawed, then that seems to be the consequence).

    That’s an interesting consequence partly because it shows that a very simple argument can have very large metaphysical and epistemological consequences.

  7. Stop. A philosopher can’t use the term “simulation” as a given in an argument or analysis. It needs hauling out of the waters of computerese and gutting on the deck, in the light of day.

    We require a minimal description of the way we are to use the term simulation. If we are concerned with levels of autonomy then we could describe a number of situations that place different emphases on autonomy for its participants – from theatre (audience and actors) and training programs, through to forced labour.

    It would seem that forced labour is the model of simulation being used here. The fact that the force is applied without anyone noticing is incidental. The sort of simulation described here is that of the labouring participants bringing pleasure to the viewers (or “audience” at a stretch), who enjoy or require the mimicry or enactment of some vision of theirs. For this type of “simulation” we might ask whether it is the viewers who have lost their autonomy, and not the participants.

  8. John – Sorry I don’t see the force of your point.

    We are the simulation; or there’s no simulation. The SA works perfectly well on that basis. Of course there’s more to be said in the detail, but I don’t see how any of it is relevant for this blog post.

  9. Jeremy,
    “A simple argument?” You’ve got to be kidding.
    By a stretch of imagination I’ll give you the possibility of a conscious computer being. From that to creating a being looking out on and being able to take in Lake Tahoe, with its resplendant beauty and the millions of bits of detail that make up its magnificance, is quite different. What’s being suggested is that computer software can emulate 4 billion years of evolution. No!

  10. But Ralph, we’re talking here about a massively advanced civilisation, maybe many many thousands of years more advanced than we are.

    Think how far we’ve come in the 60 years or so computers have been around. Then multiply that by 100 or 1000 or whatever figure you like (and obviously there will be a trade off vis-a-vis proposition 1), then it doesn’t seem in the least bit implausible.

    As Nick Bostrom puts it:

    “But even if it takes 10 million years, it makes no difference to the argument.”

    Also, the argument *is* simple even if you think one of its premises is implausible (that there will ever a civilisation advanced enough to create simulations).

  11. Jeremy,
    I felt as though I did factor in the millions of years of advancement. I hope we’re not confusing that many years with an infinite number. It’s a little too glib for me to look at the last sixty years and extrapolate at the same rate to a million or even 500 years. You’re suggesting anything that can be imagined will happen, on the basis of a century of incredible technological advances. Although I don’t believe it, we may come to a screeching halt.
    As you said in an earlier response to me, if you’re talking about probabilities then there’s some probablility that we’re living in a simulation. A very, very minute possibility, which is different than the third statement’s “You are almost certainly living in a computer simulation.”

  12. Also, Jeremy,
    Let’s hope that along with the force of technology, morality will
    make comparable strides. Where is morality in the three statements?

  13. Ralph

    Even if you’re right about how hard it is to construct a simulation in our universe – and I don’t think you are – you’re making the mistake of assuming that it would be a hard thing to do in a parent universe. But, of course, we have no way of knowing that.

    Regardless, you’re not actually falsifying the SA. You’re just endorsing Proposition 1 (unless you think that there are in principle reasons why simulations aren’t possible and there is zero probability that advanced civilisations will go extinct).

    The morality thing is a non sequitur. Either the SA is right or it isn’t.

  14. I have battled unsuccessfully with this paper before, and had no desire to drag through it all again. Now I find I have again spent some hours and made many rough notes many of which have already been posted by others. Currently my feeling is that we have before us a magnificent philosophical problem, robustly logical but an argument based on conjecture, which is difficult to refute. That however is all that it does. It does not really assist in our current attempts understand the hard problem of consciousness which must be solved before we can even scratch the surface of what the argument suggests. It is currently impossible I feel sure, to look that far into the future with any accuracy and with the cognitive acumen that humans presently possess. Scientifically the argument is a blind alley. Does anybody writing really think they are living in a simulation whatever that may mean, sheer common sense replies of course not, that is not even an argument to the best explanation.

    If we are living in a Computer simulation, So What? Constant unvarying deception is no deception. If the illusion is undetectable then it is not an illusion it is the only reality that can be known. So Descartes’ Daemon, provided he is consistent, presents us so far as we are concerned, with the only reality we can know. He does not deceive us. I suggest we accept that, until we have good solid scientific evidence to the contrary. If anything of that nature exists I have yet to see it.

  15. Don,
    I like what you said. Do you mind if I ride on your coat tails?
    Jeremy,
    Thanks for bearing with me. L’m not entirely satisfied. But superficially at least I do find the argument a really bright one.

  16. Ralph – Just a small thing. You really don’t have to thank me for chatting with you (you’ve done so previously as well). It’s a pleasure, not a chore! 🙂

  17. “If we’re living in a simulation or a baby-universe we have absolutely no way of knowing from our experience of that simulation or baby-universe, whether in the originating world (the “real” world), it is easier to construct simulations or baby universes, or whether – for whatever reason – an advanced people would be more inclined towards one rather than the other. It follows then we’re in no position to make a judgement about the relative probability of a simulation versus a baby-universe. Therefore, there is no reason to suppose that at least one of the propositions above must be true (and therefore the simulation argument fails).”

    The problem you seem to be drawing out is that baby universes could mess up the probabilities: the reason for concluding that we are almost certainly in a simulation is that if such simulations were possible, we might reasonably expect them to vastly exceed actual civilizations in number. However, if there are baby universes as well, there may also be many of THOSE compared to actual civilizations. So we don’t know which of the two we are in.

    For reference, Bostrom originally calculates the fraction of people living in simulations as
    f= pNH/(pNH + H), where p is the probability of a civilization reaching a post-human level, N is the average number of simulations they run, and H is the population up to that point. The implicit assumption is that pNH + H at least approximates the total number of people. But if baby universes are possible, then what we really need on that denominator is pNH + pBH + H, where B is the average number of baby universes a posthuman civilization generates. Which screws up the estimation of the overall fraction. Nice.

    However, this still supports the conclusion that “we are not living in the “root” universe”, where the root universe is the bottom of the “tree” of baby universes/simulations (if such a thing makes sense). Which may or may not be worrying, given that it seems much more reasonable that we be living in a baby universe than a simulation!

    I think you should email Nick with this, as I think this is a genuine loophole in the maths. The other example he gives, that of a terraforming version, isn’t a problem like the baby universe one is because it can’t be combined with simulations in the way that this one can.

  18. Thanks Michael.

    “However, if there are baby universes as well, there may also be many of THOSE compared to actual civilizations. So we don’t know which of the two we are in.”

    Yes, that’s the point.

    “However, this still supports the conclusion that “we are not living in the “root” universe”, where the root universe is the bottom of the “tree” of baby universes/simulations (if such a thing makes sense). Which may or may not be worrying, given that it seems much more reasonable that we be living in a baby universe than a simulation!”

    Exactly. We’re not living in the root universe, but if it’s not a simulation, so what…?

    I can’t email Nick, though, I don’t know anything about this – I’ve not even read the original article! – and I’d feel way too sheepish. 🙂

    Glad you think my idea isn’t obviously daft, though.

  19. Re:- Sabella Aug 11th
    No not at all, be my guest. I believe you are a mathematician. That being the case I refer you to part IV of Bostrom’s Paper published in Philosophical Quarterly (2003), Vol. 53, No. 211, pp. 243-255. see http://www.simulation-argument.com/simulation.html I cannot make much of this maybe I have not tried hard enough. Can you enlighten me? My Mathematical knowledge does not seem to embrace it very comfortably.

  20. Don,
    I’ll give it shot later today. If it’s probability, and I suspect it is, don’t expect too much from me. I never had nor taught a course in the subject.

  21. Don,
    It’s arithmetic, not probability so I can’t play the “I don’t know anything about probability” card.
    I’m not too far into it, but I thought I’d try to reassure you by telling you It’s not clear to me either what’s going down.
    The first equation he has, the f(sim) looks suspiciouly like nonsense, but I’m probably missing the point. One thing that has me wondering about the equation is the H-bar drops out, i.e. the numerator and den. have factors of H-bar and can be cancelled. So, if the fraction makes sense one should be able to arrive at it in an easier way.
    I’m on holiday, a (combined)family of 20+ reunion with me the patriarch so my presence is required away from the computer. I’ll get back to you.

  22. Don,
    I’ve gotten as far as I think I can go, which isn’t all the way.
    First, his notation is poor. He uses the f’s in two different ways, the first f(p) is the number (not a fraction) of post-human civilizations out of all human type civils. Later, he defines f(I) similarly. Then f(sim) is a true fraction, in fact less than 1.
    Now, f(sim) according to how it’s defined should be the quotient of a-simulations over all those with human type experiences. How it starts off, in the numerator:
    the product of f(p) with N-bar gives the total number of a-simulations there will be, What multiplying that by H-bar (the avg. number of people living in such a civ. before it produces simulators) is beyond me. There probably are a few steps left out, or worse. I notice he does get rid of the H-bar later.
    If someone out there is following this and knows how to derive f(sim), I’d love to see it.
    Tell me what you think of the above, Don.

  23. With all due respect, Sabella, his maths makes perfect sense.

    His notation is entirely clear. He defines each term before he uses it. Also f(p) *is* a fraction, as he claims: it is the fraction of civilizations that survive to become post-human. I can see why you might become confused, however, as he does skip a step you might expect in calculating f(sim) (although it isn’t necessary): namely, he has effectively already cancelled out a factor of M from top and bottom, where M is the actual number of civilizations there are. Then f(p) times M is the number of post-human civilizations, which is what I think you thought f(p) itself was; f(p)MNH-bar is the (approximate) number of people in simulations, and the MH-bar on the bottom is just all the people not in simulations.

    As for the use of H-bar, that is because the simulations are *ancestor* simulations, so by hypothesis, the number of people in them is the number of people up to that point in the parent simulation. Technically, its use on the denominator is an underestimate, but I don’t think that makes much difference to the argument.

  24. MPJ,
    I wasn’t going to argue the point of notation figuring we were coming from different places, but your third or fourth sentence changed my mind. How can you say the notation is “entirely clear” when f(p)can be interpreted in 2 different ways? M never was mentioned as far as I can see.
    Michael, you understand this better than I do, if you will, please explain “f(p)MNH-bar is the (approximate) number of people in simulations.” I have f(p)MN-bar as the total number of simulations for all appropriate civilizations. How do you get what you have in quotes? And while you’re at it please explain the denominator. Exlain it the way you would to a bright 12 year old.
    While we’re together, f(sim), by the way I interpreted f(p) turns out to be:
    total no. of simulations/(the total number of simulations + 1), which is clearly unintuitive.
    Using your (and I assume the correct) def. of f(p) we get:
    total no. of simulations/(the total number of simulations + M), which admittedly is much better. I think I can buy that. So, my question is, and I don’t know if you can answer this, why bring in the H-bar at all? He divides it out along with the M. It doesn’t appear to me to serve any purpose and if he hadn’t used it you wouldn’t have to explain f(p)MNH-bar to me.

  25. MPJ,
    Sorry, but I have to retract something. I wrote:
    “Using your (and I assume the correct) def. of f(p) we get [f(sim) equals]:
    total no. of simulations/(the total number of simulations + M), which admittedly is much better.”
    But, now I fear, not good enough. The den. is adding simulations and civilizations which can’t be right.
    Can you find where I went wrong?
    I’m afraid I’ve given you an awful lot to explain.

  26. “If we are living in a Computer simulation, So What? Constant unvarying deception is no deception. If the illusion is undetectable then it is not an illusion it is the only reality that can be known.” Don Bird.

    Don, you’re not even wrong. That argument is like stating that if a magician performs an illusion, like sawing a lady in half. If the audience is unaware of how the illusion is created – or if they are unaware that the magician is an illusionist and does not have supernatural powers. Then the lady really has been sawn in half.

    An illusion is still an illusion, whether revealed or not. When it is not revealed it seems like reality. The pleasure in a magic trick is seeing the expected rules of reality breached. Given the simulation argument, it is possible that what we take for solid concrete reality is an unrevealed illusion. The magician/creator will not let us see his slight of hand. A magic trick is not reality because we don’t know how it is done – or can’t see the hidden reality in the trick.

  27. “Right, I’ve spent about 10 minutes thinking about this and I didn’t really know what the simulation argument was until about an hour ago, so I suppose my argument must be wrong.

    So where’s it wrong?”Jeremy Stangroom

    I don’t know how you’ve never heard this argument up to now. It’s a common enough sci-fi plot. Bostrum gets the credit for nailing it within a philosophy department.

    Bostrum’s firt two propositions are not central to the argument.

    The third proposition is crucial. If it’s re-phrased it reveals more of its’ implications: The reality that you are living in is almost certainly constituted by a mechanism.

    The argument reveals the nature of reality.

    In a virtual world we have created ourselves, like the computer game The Sims. The virtual world is constituted by rules. To the beings within the game, the rules are invisible computer code. The beings do exist. We can see them on our computer screen. They live with as much autonomy and realisation as the rules of the game allow. They can even know we exist, if the rules allow it.

    Importantly; reality and rules have a shared etymological root.

    Forget about concrete worlds and the observed universe. The Sims are beings – they exist in an ontic space. For them to exist in their ontic space, the space must have rules to allow their existence. The space has/needs rules/a reality. To a computer generated Sim, even on an ancient version of the game running on an old computer, the Sim’s experience of their reality is as real as our experience of ours. In fact you can stretch the argument to cover ontic spaces like films and books. The characters in books and films, exist by the rules within the space. They love, fight, hate, have feelings, think, do things. They can even breach the third wall and become concious of our reality. If the author alters the rules to allow them to see our reality. In our reality, the rules do not allow beings in films to step out into our reality. But in film, like Woody Allen’s The Purple Rose of Cairo, it is possible within that specific ontic space. The author of the space is allowed from outside its’ reality to change the rules – to dictate its’ reality.

    In the Sims, logicians and mathematicians and all kinds of scientists could be magicked into existence. For all their cleverness, their logic and probability theory can not in itself determine the true nature of their reality or the realities that exist beyond theirs.

    We can only be sure that logic and maths exist within our reality. Our understanding of being and existence can only exist within our reality. Our ideas of advancement,technology, intelligence, even causality can only exist within our reality.

    We are, with certainty, beings existing in an ontic space, which has a reality/rules that has been constituted by a being/beings, who exist in an ontic space beyond our own. Possibly even, in the ontic space our creators exist in, their is no being and there is no existence.

    We may be an alien teenager’s toy. likely, we’re something else. We created the Sims, because we needed them – for entertainment and enjoyment. Our creators may have created us for similar or even more crucial needs for their existence.

  28. What is your model of a “simulation”?

    It looks like it is the one we have today, where we live, here and now. Where some people can be persuaded by politicians to think that policies are doing them good. We don’t need “baby universes” or other extravagent video-game mythologies to illustrate the point about simulation.

    There are other types of simulation, of course where personal autonomy is retained. Such as actors and audience, or an enactment (such as a training program) or rehearsal where the actors are also the audience.

  29. @Sabella

    You can’t interpret f(p) in two different ways! He defines it right there. It’s his paper, you’ve got to use his definition 😛
    “f(p): Fraction of all human-level technological civilizations that survive to reach a posthuman stage”

    That’s how he defines it. If you don’t see how his use of it leads to his derivation of f(sim), then that doesn’t mean you should reinterpret f(p)! That just means that either his derivation is wrong, or you misunderstood.

    M was introduced by me. Just as a helper for explaining the derivation. When Bostrom does it, he doesn’t bother doing it like that. His fraction is correct, I was just illustrating a more intuitive way of arriving at it. Here we go:

    f(p)M = no. of posthuman civilizations, by definitions of f(p) and M.
    f(p)MN = no. of ancestor simulations run by posthuman civilizations, by definition of N, and above.
    f(p)MNH-bar = no of people in said ancestor simulations, since each such contains H-bar people, in virtue of it being an ancestor simulation (H-bar was basically the number of the civilizations ancestors).
    MH-bar is the number of people not in simulations, since M is the number of simulations, and H-bar is an underestimate for how many people there are in each.

    Okay, so that’s the individual terms. Now I propose that
    f(sim) = f(p)MNH-bar / (f(p)MNH-bar + MH-bar),
    that is, the number of people in simulations (derived above), over the number of people in simulations, plus the number of people NOT in simulations, again, as derived. So there is the fraction. BUT we can just cancel M from top and bottom to give
    f(sim) = f(p)NH-bar / (f(p)NH-bar + H-bar),
    which is Bostrom’s formula.

    So, to answer your specific questions, the way I got the bit I had in quotes was by multiplying by M. Bostrom doesn’t do this because he doesn’t need to. It cancels later. The answer is the same.

    As for H-bar, it does serve a purpose; namely in calculating f(sim) correctly! Once we’ve done that, we can see that in fact H-bar cancels out as well. This is why at (*) H-bar has been cancelled out already. H-bar is crucial to working it out, because without H-bar you aren’t talking about quantities of people at all!

    And in your second post, you misread. That M should be an H-bar. And the first term is in people. So there is no adding of simulations and civilizations, only people.

    Hope this helps,
    M

  30. Re: KRD August 13th

    I do not think my definition of the perfect illusion is the same as a so called illusion conducted by a stage performer for the following reasons.
    1/ We know with the stage performer, that what we are to see, will not be a true state of affairs.
    2/ We know that sawing a lady in half is just, not done.
    3/ We are not really fooled into thinking that what we see is as purporting to be a state of affairs is in fact the case.
    4/ Most importantly we are not ourselves actually a part of the illusion.
    5/ We know our reality, from non-reality

    The perfect illusion, which I am suggesting is one in which we are completely embedded. All our perceptions are consistent there are no contradictions it is all we know and understand. This is our reality. Surely it is what would be the case, were we a computer simulation or even some sort of duplication of reality. The only entity which knows the truth of things, is the the one responsible for creating the illusion.
    If we suspect, we could be something like a brain in a vat, or some sort of conscious self aware construction there seems little we could do about it, and would we want to? In any case we could never be sure, provided consistency prevails. That is why I say “so what”? Let’s get on with life as we know it, be it real or simulated.

  31. MichaelPJ,
    You explained the situation extremely well. If you’re not a professor, you should be one, but I do have just a few things left to say:
    So the term f(p)MNH is the number of p.h. civs times the avg no of a-sims run by each such civ times the no. of ancestors of the civ.
    This sounds like every sim must include every ancestor. If that’s right, it wasn’t the way I perceived what was going on. I can’t conceive of the purpose for doing this. Anyway, this isn’t for you to explain, unless I have the wrong interpretation and you want to correct me.
    Also, looking at the f(sim) equation before cancelling the M’s, since M in fact does divide the den. it must div. the second term which must be therefore MH. But we don’t really want all of M, only that fraction of it that are p.h. civs, viz. f(p)M. Therefore each term in num and den has factors f(p), M and H. Now dividing all of these out, we get:
    f(sim) = N/(N+1)the sense of which I fail to appreciate.
    Finally, about the definition of f(p). There are 2 ways (at least) of interpreting it. After all, I interpreted it in two ways. If you look in a dictionary you’ll find the def of fraction is not only the quotient of two non-zero integers but also as a part of a whole. Notice the way I used it above. The fraction of all M which are p.h. civs. is f(p)M. It turns out if you leave out the M’s and use the second definition of fraction in f(sim) you get the same thing. In fact, that Bostrom uses the phrase “The actual fraction . . .” before the f(sim) equation and without the M or any such indication of a total number of civs. leads me to believe his definition of f(p) is the one you deny exists. Why else use the word “actual?”
    By the way, my name is Ralph.

  32. @Michael, @Ralph

    I’d just like it to be known that I have not the faintest clue what either of you are going on about!

    But I suspect it’s good stuff. 🙂

  33. Hey Jeremy,
    That’s a nice switch.

  34. Ralph I seem to have got you involved in profound discussion with Michael concerning the maths. Due to a bout of influenza I have been unable to follow the exchanges. However I have now printed them and will read shortly. If I come up with anything worth saying (probably unlikely I think) I will do so.

  35. Hey Don,
    I was wondering about you. Here you got me involved with a real heavy hitter, and I thought, then just deserted me.
    Take care of yourself.

  36. @Ralph

    Oh. Sorry about the name, I just kind of went with what I had!

    The whole ancestor simulation thing is just how Bostrom does it. You might not agree that it’s likely that posthuman civilizations would be at all interested in running ancestor simulations, but that’s precisely one of the possible conclusions he draws. Even a passing interest would result in a lot of simulations, since their computing power is so great.
    Also, considering ancestor simulations might explain why WE’RE in it!

    Yes, the second term in the denominator is MH. It’s supposed to be everyone who’s not in a simulation, and that includes those who’re in other civilizations. So it should indeed be MH.
    Once you cancel out the terms, you do indeed get
    f(sim) = f(p)N/ (f(p)N +1)
    I’m not sure what you don’t understand about this; it’s just the same fraction! Algebra doesn’t change anything.

    f(p)M would be the actual *number* of posthuman civilizations, and I would expect Bostrom to say it like that. Trust me, that’s not what he means. Apart from anything else, he isn’t an idiot, and the maths doesn’t make sense unless he means that!
    What’s the “actual fraction” of males in the population? Would you expect an answer like 0.5, or 50 million?

  37. Michael PJ,
    I didn’t mean to make it sound you should know to call me Ralph. Just prefer a first name basis after twenty notes to one another.
    It sounds like you have more than a casual knowledge of Bostrom.
    The whole question of purpose and morality of a-sims. is of great interest to me. Perhaps you would clarify something for me. Is my reasoning wrong? I envision the use of these sims. being “run” on a timewise one-to-one basis,i.e. 10 mins. of sim. would take up 10 mins. of a person who is “viewing” it. If this is taking place, say, 50000 years from now, then a complete run would be close to a 100000 years long. So, obviously, there must be a way to pick and choose people or time periods to view. Which leads me to thinking that if a person watches say 10 years of her/his life viewing these things, using all ancestors in every run is introducing a huge error in the analysis of f(sim).

    Yes, I came to the conclusion after sending my last comment that it should be MH. Too bad. I liked cancelling everything out.
    I’m not sure why I think the equation f(sim) = f(p)N/(f(p)N + 1)
    should in a sense explain itself to me, but I think its simplicity for something I would expect to be more complex leads me to believe that. Thinking of it in those terms, the fact that the function is only dependent on the fraction of phc’s of all htc’s and the avg. number of a-sims run is noteworthy. The number of ancestors doesn’t come into it. We used H to get the equation, but in the end how many ancestors there are is of no importance. Why this should be is that each run takes in all ancestors. So, if an a-sim was defined with all ancestors to be included in it, then H would be superflous, and the whole thing a lot easier to understand. But that’s me.

    Yes, I would say .5 for the actual fraction. I’m not sure where that gets you. I might have used the same example to further my point. Michael, let’s put a halt to the “fraction definition” argument. Just let’s leave it at I’m willing to take your word for it that Bostrom meant both f’s to be defined in the same way.

  38. Sabella,
    It looked like you were ignoring some pertinent points by welcoming those that were not.

  39. J J,
    Probably but Huh? A little less cryptic.

  40. John J
    I get it. Yes, I did ask if there was anyone who can explain things to come forward. Michael knows this stuff really well, and through him I’ve come to understand it, I think. That was the idea of asking.

  41. “1/ We know with the stage performer, that what we are to see, will not be a true state of affairs.
    2/ We know that sawing a lady in half is just, not done.
    3/ We are not really fooled into thinking that what we see is as purporting to be a state of affairs is in fact the case.
    4/ Most importantly we are not ourselves actually a part of the illusion.
    5/ We know our reality, from non-reality” Don Bird.

    Don the only way we have of knowing an illusion is an illusion is seeing through it, if the misdirection fails or knowing the illusionist is preforming a trick.

    If the audience had never seen or heard of a magician before the performance, they would assume they’re seeing the rules of reality being broken. They would be amazed and terrified. Sensing reality as constant and concrete is a sense. If it’s made to appear less concrete we experience the terror of a psychotic.

    Our experience of reality is a constructed simulation by our senses of the actual world. It’s not the actual world. Something like light; we only see within the visual spectrum, we can’t see the infra-red and ultra-violet that the world bathes in. We know it’s there by other means of observation than our direct senses. We will never see the reality of that light as it can not be relayed and reconstructed in our reality.

    The schizophrenic (literal trans broken minded) in a psychotic episodes has a broken simulator. Representations of the actual world are misrepresented. They see things and hear things as real, that are not happening in the real world. Within the psychotic’s mind the rules of reality melt. They can’t distinguish internal brain chatter from external reality. In the film Vanilla Sky, the central character goes through something like a psychotic episode. In the end, it’s not a psychotic episode but a failure in the simulators software – Tom Cruise calls for technical support. A psychotic will take anti-psychotic medication – reality returns. Tom Cruise’s character in Vanilla Sky is unaware he is in a simulation. The break down in the software only makes itself apparent through nightmares and ever more surreal episodes of his experienced reality.

    The Cartesian brain in a jar; I prefer to call it Cartesian cinema, so I can make the joke: “this movie’s crap, let’s slash the seats”. The Cinema is the simulator. It’s wrong to think of it as a physical jar somewhere or a physical cinema somewhere, even in an abstract sense. The cinema or jar cannot be visualised as a physical object. When we think of them as physical spaces and apply the logic of our reality to them you come up with we silly arguments claiming they can’t exist because of some logical flaw, like the demon in the cinema would need another demon in its’ head on into infinity. Neo in the Matrix or Tom Cruise in Vanilla Sky cannot see through the illusions of the simulators they’re in, until Neo takes the blue pill, or Tom Cruise locates the simulators technical support. We have no way of telling if we’re like Neo or Tom Cruise, unless some agent from outside of the simulation informs us. Even then the Deus Ex Machina will only appear in our reality as one of us – or something we can comprehend; like a talking burning bush. They could be lying. And how is Neo to know once he’s gone from the Matrix simulation that he is not in another simulation. A sim can only exist within a simulation.

    We do exist – but ultimately we may only exist in a simulation. That we are some kind of simulacra.

  42. Well, I’m still stuck on defining “technologically mature”.

    And, apparently, these discussions have not given those running our particular simulation reason to intervene. Not that this proves anything, though I do still wonder if there will be a giant message appearing in the sky that says “GAME OVER, HIGH SCORE…”

  43. Opps, the explanation of “technologically mature” right toward the top of the paper… never mind.

  44. “the argument doesn’t get off the ground unless we’re not living in a simulation (because if we are living in a simulation then propositions 1 and 2 might not even make sense about the “real” universe).”

    -The argument states that at least one of the possibilities is true, since the 3rd possibility accounts for the idea that we are in a simulation, the argument stands. An absence of logical sense in the first two possibilities wouldn’t matter, as argument 3 would therefore become automatically 100% correct.

    “it’s possible to construct an identical argument with “baby universes” (thereby demonstrating that we’re not living in a simulation).”

    -Again, Bostrom’s argument stands, as number two would turn out to be true… This conclusion would account for all other possible methods of inducing consciousness with future technology.

    In my personal opinion, I give a wide margin of merit to the notion that we’re not living in the “root civilization”, simply because of the overwhelming odds against us since the possibility of forming a conscious mind is very likely, and given the multiverse theory, which is widely accepted due to how well it works to explain 11 dimensions. The amount of human civilizations that would have existed must have amounted to an overwhelming number, making the likelihood of one of them deciding to mass produce artificial consciousness almost definite. This/these civilization(s) would obviously have no problem with making more then one of these “conscious pools”, and I believe they would make many more than the total needed to give enough merit to the chance that one of these civilizations would create a “conscious pool”.

    If you are confused, here is what I’m saying:

    the total number of conscious minds(found by finding the number of human civilizations needed, and than formulating a total number of conscious minds that would be alive in these civilizations) that would be required in order for one of these civilizations to be 100% to create consciousness.(compared to the whole amount of human civilizations)
    over
    the total number of conscious minds that the average one of these special cases would create

    this equation would calculate the likelihood that we exist in an “original” world.

    -The only argument I can find to refute my argument is that I’m making false assumptions about the future, that creating artificial consciousness is not possible/limited to a certain set of finely tuned circumstances. I could also be wrong about my assumption that the multiverse theory is true.

    The implications of actual proof that we aren’t living in a “real” world would be that we as a civilization are less likely to encounter mass human extinction, as logically, the more likely it is that we’re not “real”, the more likely it seems that we are the history of our creators

  45. you can’t disprove the simulation because you’re using the simulation’s logic system…

  46. 1. Who cares if it is a simulation?
    2. Why does life have to be so sophisticated? Why can’t life be life and be done with it. If you are a Christian like myself we believe that God created everything.
    3. And if this is a simulation it would take millions of years to get to the technological level to make it so amazing and perfect.
    4. If this was a simulation then why would they allow us to wonder about this, and why would they allow us to use their technology to make simulations?
    5. If this was a simulation then we would have to have a real form somewhere. Also are UFOs also apart of the simulation?
    I am no philosopher, but I think life is life and that we who have open minds become a bit delusional because we come up with theories and believe them with no hard evidence.

  47. It’s a fad now four years later, hitting the mainstream. I didn’t hear about it until maybe 2013.

    One must admit that it’s possible, but the number of reasons why it’s unlikely are staggering. I am pretty shocked by those claiming that the probabilities are enormous. Really?

    There are some good reasons, though, why a post-human civilization might create simulations. I can name a few:

    * They are looking for something, and are simulating the world (and perhaps scrapping the data) trying to find it. Perhaps an idea, a way of thinking, a discovery.

    * A simulation is not entertaining. But it would not surprise me if a post-human race takes an interest in poetry, novels, art, etc., and runs simulations to produce great works of art.

    * They are also living in a simulation, and discovered that fact, and then proceeded to create simulations themselves (simulations within simulations).

    But there are many reasons why it’s unlikely. I doubt that aliens would simulate this, because I also doubt in space travel. I doubt that a society would build such a computer. But most of all, a post-human race would probably be busy or focusing on something else. We should not assume that they would find it useful or entertaining.

  48. I have stumbled here via a documentary about quantum mechanics which left me investigating quantum entanglement. The question in my head is whether or not the current problems with consistency in our physics models arise from our inability to understand the universe as a result of the species-subjective nature of human investigations into reality or do they arise from some other source?

    The most likely explanation is that we have not arrived yet, we are not ‘post-human’ enough in Bostrom’s terms.

    Another possibility might look like this: living in a simulation could put as at the mercy of algorithms that are under no obligation to provide logical mathematical consistency in our terms. Entanglement could be part of the programming of a simulated universe along with dark energy, faster than light universal inflation and dark matter in a manner we are always doomed to find inexplicable. For the gamers out there – think about the ‘new improved physics model’ that accompanies so many new sims.

    Attempts to tackle the simulation proposition as a mathematical assessment of computing power are entirely dependent on the assumption that we have access to the same mathematical tools as those possessed by our creators.

    This leads me to the question: can failures of our universe to conform to our logic, our physics, be considered as evidence in support of a simulation theory of reality?

    But using the simplest approach to the simulation idea it seems to me that the real question is, please forgive me for this, are we the creation of one or more gods? This is well covered ground and Bostrom’s proposition surely belongs there with more than 2 millenia worth of argumentation.

    It seems to me that the simulation approach to reality is the old question of faith wearing the ipad as a cassock.

    And please no disrespect intended if I am missing the point of this discussion. I find Bostrom’s approach fascinating and the discussion important.

  49. All three propositions are assumptions and therefore none of them can categorically be perceived as true.

    The real question is what would be the difference between being part of a dreaming consciousness as opposed to being part of a computer simulation.

    A dreaming consciousness would be one consciousness. In reality there would not be a subject/object relationship since there is only one subject, the consciousness of the dreamer.

    A computer simulation by an intelligent being, or beings, would in essence be separate from the mind and consciousness of the intelligence that created it and would result in a subject/object relationship between the simulation and the originating intelligence.

    A simulation versus a baby universe could be a matter of possibility versus probability.

  50. It’s completely obvious we’re not living in a simulation. The idea is stupid, for many reasons.
    Scientists consistently overestimate the ability of machines, and likewise underestimate the ability of humans (for instance, a machine that can walk a tightrope, juggle, or even just make breakfast, has yet to be built).
    Some scientists also underestimate the phenomenon of consciousness, and believe it can be recreated in machine form. But living things are fundamentally different to machines. Living things feel and die – objects do not. Being alive is a prerequisite to being conscious. Being conscious is a prerequisite to being able to feel. The fact that we can feel pain/pleasure is proof by itself that we are not machines – if proof were needed (which it really shouldn’t be).
    Consider the immense problems scientists would have building a machine that could run and catch a ball simultaneously. All the talk of ‘immense progress’ over the last century, and of a supposed ‘singularity’, glosses over the problems we have creating a machine that can perform almost any task we take for granted. So ‘in the future’ we’ll be able to create ‘conscious simulations’ that can do everything we can? YEAH, RIGHT! I’ll just type it up on my quantum computer!

  51. I think Musk probably believes that some very advanced civilization built the simulation and that it overcame all these technical problems. While there are certainly many philosophical arguments against being able to have a material consciousness (as opposed to having an immaterial mind), there are also excellent arguments for just such a possibility.

    But, as you note, there is the reasonable question about whether or not such a thing would be technically possible even for an incredibly advanced civilization. This is analogous to debates about other high tech speculation, such as about warp drive, transporters and anti-gravity.

  52. It is analogous to other as-yet impossible technology, as you say, but those things might be at least possible with sufficient advances in physics/engineering. I’m not sure the same can be said about consciousness – it may be that it can’t exist without a living body (ie, machines might be unable to be conscious at all).

  53. True, consciousness could be a purely organic thing.

Leave a Comment


NOTE - You can use these HTML tags and attributes:
<a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>