On the Simulation Argument

In the movie The Matrix, the main character, Neo, awakens from an artificial reality. He eventually discovers that most of humanity has been captured by a race of elite machines, who cruelly imprison human minds within an artificial reality known as the Matrix. The release of the movie back in 1999 hopefully provoked both philosophical and non-philosophical minds alike to question the true nature of reality, which I think is a positive thing.

How likely is it, then, that life as I know it is merely a Matrix-like computer simulation? I’m not exactly prepared to assign a specific probability to this scenario, but, as an ardent questioner of reality, I realize it’s possible. I’ve come to learn that not only is it possible, but it might be more probable than I initially realized, at least according to philosopher Nick Bostrom of Oxford University.

In fact, if I decide to accept a few seemingly reasonable assumptions it is highly probable that I’m living in someone else’s computer simulation right now. Pardon my language, but how’s that for a Cartesian Mindfuck?

This claim may sound outrageous, but allow me to clarify, it’s actually logically sound. First, in order to understand Mr. Bostrom’s Simulation argument we must lay out a few key assumptions.

#1 The first assumption is that of ‘substrate independence’, which means that conscious minds could, in actuality, exist not only through biological neurons, but also on something like a silicon-based computer processor. (If we assume substrate independence, in essence, we admit that it’s theoretically possible to upload a human mind onto a sufficiently fast computer.)

#2 The second assumption is that we can estimate how much computing power would be needed to implement a human mind within a virtual reality.

Ok, now that those two assumptions are out of the way: enter the simulation argument.

The simulation argument suggests that one of the three following propositions is true (please note, however, it does not claim to demonstrate that we live in a simulation, but simply that one of the following three propositions must be true.)

#1 The chances that a species at our current level of development can avoid going extinct before becoming technologically mature is negligibly small.

#2 Almost no technologically mature civilizations are interested in running computer simulations of minds like ours.

#3 You are almost certainly in a simulation.

At face value, it’s pretty hard to argue that neither of these three propositions is true given the assumptions that were laid out.

Consider proposition #1. The question is: will humanity survive long enough to develop this technology that we assumed is feasible to develop or will we kill each other off first? Even if human technological development comes to a screeching halt, biological evolution does not. If humans manage to survive for a long enough time scale, then, eventually we’ll continue to evolve into some species of “posthuman”, who may develop the necessary technology to run the simulation if we are unable to. Anyway, we must also question the likelihood that our evolutionary cousins will not kill each other off before they have the technology to run supercomputer simulations.

Consider proposition #2. Perhaps our potential simulators see no reason to run a computer simulation of the past. Our betters may have an evolved ethical sensibility and would find running a simulation cruel. I hope this is true.

Consider proposition #3. If this proposition is true, is it likely that the creator of our simulation would be as compassionate as the creators of The Truman Show by eventually letting us know? I’m not sure. What I do know is that if we are indeed living in a simulation and are unaware of it, there is nothing we can really do except to continue living our simulated life as programmed, much like Truman did. Perhaps this explains some of the strangeness we perceive in this world.

So why haven’t I hit the panic button yet? The reason is because I don’t accept the first assumption. I don’t believe that human consciousness operates in a similar fashion to a computer. To put it simply, we don’t know diddly-squat about many of the complex intricacies of the human mind. We can’t even sufficiently define things like ‘consciousness’ and ‘sentience’, let alone recreate them in machines. Belief in transhumanism is merely a way to assuage the human fear of death. And, in that sense, it has become the New Religion.

Further Reading

1) Here’s a link to Nick Bostrom’s technical article about the Simulation argument “ARE YOU LIVING IN A COMPUTER SIMULATION?”

2) Here’s a link to a less technical piece about the Simulation argument, written by Mr. Bostrom, titled “The Simulation Argument: Why the Probability that You Are Living in a Matrix is Quite High


5 Comments on “On the Simulation Argument”

  1. Mark says:

    The validity of the disjunction depends on an additional unnoted assumption in the argument, that the average number of human-type individuals per simulation is approximately equivalent to the average number of human-type individuals per real civilization, without this the disjunction doesn’t necessarily hold true and from a sample of one civilization how can we speak to the truth of this assumption, especially as we don’t know whether our knowledge of this world represents knowledge of a simulated civilization or that of a real civilization.

    There are also other implicit assumptions in the argument; that simulators of human-type civilizations are not interventionist in that technological progress within a simulation is natural (to enable the above assumption), and that simulations of human-type civilizations are not deceptive in that extrapolations from the present to future technologies are realistic.

    • Greg Linster says:

      @63bdbc8359b8e37ec063bd61d1c49e03:disqus Thanks for the thoughtful comment!  I’m not sure I understand why the assumption you assert is necessary.  Why does the number per real civilization matter?  Couldn’t the real civilization simulator creators simulate as many people as they want in the simulation?  

      • Mark says:

        The disjunction posed by Bostrom rests on his formula for the ‘actual’ fraction of simulated human-type* individuals, f = FNH/(FNH + H), with f the fraction of all human-type observers that inhabit simulations, F the fraction of real human-type civilizations that go on to run simulations of human-type civilizations, N the average number of human-type simulations run by those post-human civilizations that choose to run such simulations, and H the average number of historical individuals per post-human civilization that runs such simulations (a factor C has been cancelled and omitted from the formula, where C is the number of real civilizations i.e. f = CFNH/(CFNH + CH)). The denominator of this formula represents the historical sum of simulated (CFNH) and non-simulated (CH) human-type individuals, but implicit in this is the assumption that the average number of individuals per human-type simulation and the average number of individuals per human-type real civilization are equal, as H is used in each factor, but there is no reason for this to be the case.

        It is easy to see why this is a mistake; in a universe with 2 real civilizations, a post-human one with a historical population of 1A that runs N human-type simulations and another human-type civilization with a historical population of 10A that does not run simulations, Bostrom’s formula gives an incorrect fraction f = N/(N + 2) when the fraction is in reality f = NA/(NA + 10A + 1A) = N/(N + 11). A correct formula would be f = CFNH/(CFNH + CR), where R is the average of individuals per human-type, real civilization and H is the average of individuals per human-type simulation. In a slightly more palatable form the formula is f = N/(N + R/FH) and here it is easy to see that for the disjunction to be true R/H ~< 1  must be true, else it is possible, if R/H is large, for all three propositions in the disjunction to be false.

        For the purposes of salvaging his argument Bostrom must prove that large values of R/H are impossible for any hypothetical simulation hierarchy, an impossibility. It is not enough to simply argue that the probability of R/H being large is small because the core of the argument is the fact that the formula represents the 'actual' fraction of simulated individuals and so the parameters within the formula must accord to their actual values, else f is not the actual fraction of simulated individuals and the bland indifference principle Cred(SIM | f = x) = x is invalid, i.e. f is not the probability that we are simulated. The bland indifference principle is contingent on the validity of R/H ~< 1 but we have no way to evaluate the truth of this statement.

        The original formula doesn't quite give the same problem because the only relevant parameters are N and F once H is cancelled (and C), i.e. f = N/(N + 1/F)**, and the disjunction represents the full range of possibilities inherent in the formula, but in the corrected formula the full range of possibilities aren't reflected in the disjunction.

        * Human-type refers to civilizations who's histories include a period that roughly accords to our level of technological progress i.e. for the purposes of this analysis we can discount those real and simulated civilizations that have not reached our level of technology at this present stage or will never reach it.
        ** Possibly of interest to you and not too obvious is that the original formula represents the fraction of simulated civilizations rather than the fraction of simulated individuals.

  2. Nateorf says:

    The hole that I find in the simulation argument is that it fails to address the supposed simulation designers. If we discovered that we were simulations, that’s our origin explained. Now explain their origin. They may be simulations too, and if they are, then where does “reality” originate? 

    Also, the theory does not address “biological” history. It has no room in it for dinosaurs and fish and unicellular life. What post human designer (or any designer for that matter) wants to start out with single celled simulations?

    I could be wrong. In which case we still have to ask why our simulation designers left us with all of this evidence of past, now extinct life. That sounds similar to Fundamentalist Christians claiming that dinosaurs inhabited the world in the time of Moses.

    I do not claim to know every detail of Mr. Bostrom’s argument, but I believe that he makes the mistake of not addressing the evidence of a non-simulated world and assuming that any technologically advanced civilization with the ability to run simulations would be anything like us.

    • Greg Linster says:

      @9b979aef55f82a9470295c7184e4180d:disqus Thanks for the thoughtful comment.  I’ll attempt to provide answers to your two questions.

      1) I see it as a Kantian problem of sorts, i.e., where our simulators originated is out of bounds for us given the constraints of our simulated reasoning abilities.

      2) Again, from a Kantian perspective, we’d have no way to know why our originators would start “biological” history the way they did.

      In regards to your last point, it’s perfectly reasonable to assume that our simulators could create evidence of a non-simulated world in our simulation, right?

Leave a Reply