Simulation or Mere Semblance (of an Argument)?
E. John Winner
**In the following discussion, I am indebted to a paper written by Brian Eggleston, when he was an undergraduate systems analysis student at Stanford University. 
Much of the recent interest in the idea that we all might be living in a computer simulation, run by an advanced civilization, arises from an argument put forth by Nick Bostrom, philosopher at Oxford University.  As long as we remain within the realm of higher order probabilistic logic, the argument seems persuasive. If we apply common sense, however, we can easily identify serious problems with it.
First, let us separate the general argument from its initial construction, and refer to it as “a bostrom argument for a simulated reality.” Confronting Nick Bostrom’s original version of the argument requires a facility with probabilistic logic, and he cleverly hedges his, suggesting that while the case he makes could be true, the very fact that it hinges on probabilities means that there is no way to know whether it must be true. Bostrom, then, is really engaging in a kind of thought experiment. Nonetheless, a number of well-known figures – notably, science popularizer Neil de Grasse Tyson and billionaire technologist Elon Musk – have accepted and even promoted this idea that reality is a simulation. (In some quarters, the bostrom argument is deployed as a kind of pep-talk, whose point is to urge continued technological research.) Consequently “a bostrom argument” should be understood simply as a probabilistic argument that our reality is, in fact, a simulation programmed by a higher intelligence.
Bostrom’s own simulation argument rests on two assumptions: (A) Substrate-independence – the notion that consciousness may arise independently of the material in which it appears. “It is not an essential property of consciousness that it is implemented on carbon-based biological neural networks inside a cranium: silicon-based processors inside a computer could in principle do the trick as well.” (It should be noted that this is, in fact, actually highly controversial. A silicon based consciousness might be possible, but there is a danger in carrying this principle too far. Surely, it is no accident that a panpsychist like David Chalmers is a bostrom argument advocate. After all, if consciousness is independent of its material habitation, then it may be independent of any materiality at all.)
(B) The Principle of Indifference – Quoting Eggleston, “When there is no independent reason to believe one proposition over another, the probability that the proposition is true is equal to the number of possible ways that the proposition could turn out to be true divided by the total number of possible outcomes.” That is, they are both equally likely to be true, barring further information; even if they are contradictory. In traditional logic, this resolves into an exclusive “either/or” proposition, on the basis of non-contradiction. In probabilistic logic, it helps form a basis for decision. When I flip a coin, heads or tails are both equally likely. If I’m betting on the outcome, however, I will have to apply further reasoning to decide the outcome on which I should bet. The principle of indifference suggests that the outcome will determine the validity of the reasoning.
Both of these principles carry over into the more popular forms of bostrom arguments, in sometimes quirky ways. The quirkiest is the manner in which the principle of indifference is assumed to somehow guarantee the original bostrom argument’s presumption of the future developments of both the human species and its technology. This reveals that there are two hidden assumptions in bostrom arguments that need to be revealed: (i) human history is a single lane highway; (ii) this highway forming a one-way incline, pointed in the direction of progress. As we become smarter and smarter, our technology will become better and better, until we at last surpass ourselves and evolve into “post-humans” capable of simulating anything with our computers. When those engaged in a bostrom argument admit that neither of these assumptions concerning the future may prove true, they do so in apocalyptic tones: we must either evolve into a more advanced condition, with an advanced technology, or we are doomed.
These assumptions acknowledged, we move onto the argument itself. Bostrom himself introduces it as resolving the following trilemma:
- The fraction of human-level civilizations that reach a post-human stage (that is, one capable of running high-fidelity ancestor simulations) is very close to zero, or
- The fraction of post-human civilizations that are interested in running ancestor-simulations is very close to zero, or
- The fraction of all people with our kind of experiences that are living in a simulation is very close to one.
According to Bostrom himself, at least one of these claims must be true.
Using probabilistic logic, the bostrom argument proceeds by interpreting the first of these propositions as so highly unlikely, it converts to its contrary – the fraction of human civilizations achieving post-human evolution with capability for high-level simulations is quite high. Thus, by the principle of indifference, and given the principle of substrate independence, the third proposition has the highest probability of being true. Setting aside the second possibility in Bostrom’s proposed trilemma (that future evolved post-humans choose not to engage in ancestral simulations, a possibility proponents and critics agree has a low probability and would implicate nothing in the context) , a bostrom argument, in more traditional form, reads something like this:
1.’ The probability of the human species surviving to evolve into a post-human civilization with world-simulating capabilities is quite low; or the probability that we all now live as simulations is quite high.
2.’ However, assuming that rapid advances in computer technology continue unabated in the future, the probability of “the probability of humans surviving to evolve into a post-human civilization with world-simulating capabilities is quite low” is itself low. The probability of humans evolving into a post-human civilization with world-simulating capabilities is thus high.
3.’ Whether we are or are not living as simulations is probabilistically indeterminate.
4.’ But if the probability of humans evolving into a post-human civilization with world-simulating capabilities is high, then it likely has already occurred.
5.’ Thus, the probability that we all now live as simulations is quite high.
There are several fatal problems with this argument.
First, we need recognize that many of these probabilities collapse into universal claims. If we allow that there is any fraction of the current human population not living in a simulation, we must confront an absurdity – at least one contemporary human is currently living in the fundamental reality of the post-human simulators, which will not come into being for a couple centuries. Thus, the probability claims on fractions of existents (e.g., “the probability is very close to one”) must collapse into the universal claim that we are all now living in this simulation. (The probability of which isn’t “close to one,” it would equal one; that is, it must be a certainty.) Most attacks on a bostrom argument point out this weakness. For one thing, such a universal claim does not permit the deployment of probability arguments in support of it – we either are or are not living in a simulation. Further, a universal claim like this is well open to truth-value analysis, and thus triggers demands for evidence, verifiability, falsifiability, etc. Further still, if we all live in a simulation, the only way we could know this is if we have been programmed to do so, and there is no external reasoning or reasoner to which we can appeal in order to determine this. Finally, if we all live in a simulation, there is nothing to be done about it. Everything we do has been programmed beforehand, so all that we can do is act in the same way, with the same motives, with the same sense of agency, that we always have. The notion that a bostrom argument can be used to urge further technological advance is both ironic and self-subverting. If we’re programmed to do this, we’ll do it anyway. If we’re programmed not to do this, we won’t, regardless of what we tell ourselves. (But of course, we’ve been programmed to make and respond to these arguments, so…)
The strangest thing about the bostrom arguments, however, is their weird sense of history. I noted that a base assumption of such arguments is that history is a one-lane highway, but we must also see that they also entail that the human species has already traveled that highway. What we experience as the present is “in fact” the past of the future post-humans, who have programmed this simulation. This is where bostrom arguments exhibit a rather strange faith in the certainty provided by probabilistic logic. If a future event enjoys a high probability (“very nearly one”), then it is as if it has already happened. That’s fine for playing the lottery (the probability of losing is very nearly one, so don’t bother). But when we start speculating on possible future humans developing the capacity to interfere with the past, problems and paradoxes abound. The writers of and audiences for time-travel fiction are quite familiar with many, if not most of these, but those advancing bostrom arguments seem blissfully unaware of them.
Undoubtedly, the most important problem here is that treating probabilities as certainties, predicated on the assumption that history is a one-lane highway – that there is no possible divergence or detour or fork in the road ahead – is so blinkered, so myopic, that it verges on delusional. One minor war at the wrong time in the wrong place (say, a brief nuclear exchange between North Korea and the US); one economic crash redefining the developmental strategies of major economies; or one major scientific discovery that improves our environment or our health care, and the map to the future may get wholly re-drawn. We’ve already seen this, several times, but even recently: the global rightward turn among voters has radically re-determined the kind of progress we can anticipate, at least in the foreseeable future.
So the grand historical narrative that the first wave of bostrom arguments depended on is dubious at best, delusional at worse. Not surprisingly, then, the narrative has been re-written. The more popular versions today, inspired by such science fictions as the Matrix films, hold that it is not a future post-human civilization that has constructed this simulation, but a super-intelligent alien species, on another world or even in another universe. 
The very fact that the narrative has been so substantially re-written should give those tempted by bostrom arguments reason for pause. If the fundamental historical assumptions of such argument have to be changed to keep them standing, then that raises the question of whether the structure of the arguments is itself sound. It was thought that clay would support the brick and mortar, but now we find it necessary to mix cement? And using ingredients imported from science fiction?
Alas, this new version, while seemingly unassailable, is in fact weakened in part by that very unassailability. There may be no glaring paradoxes to threaten it, but neither are there any empirical grounds for believing it.
That there may be other intelligent “alien” life forms in the universe is not a hypothesis, strictly speaking, but an allowable probability, given the size of the universe and the possible number of life-sustaining planets. Occasional research efforts, like sending “Welcome!” greetings in exploratory space ships are little more than gestures signaling our openness to such a possibility.
The presumption of super-alien programmers of a simulated universe is an entirely different matter, since it hinges on no empirically grounded possibility beyond our own ability to build computers. Hence, it cannot be verified. Nor can it be falsified, since any falsification procedure presumably would be pre-programmed by the super-aliens, for their purposes, which remain forever inscrutable. The super-alien simulation “hypothesis” can only unravel of shown to be incoherent or otherwise logically defective. (One can easily imagine, for instance, a regress-style problem. If the super-aliens are programming us, then who is programming them? And what constitutes the fundamental or base reality, by which simulations are identified as such?)
Notably, the whole of the above paragraph applies to divine appeals as well. “You can’t prove there’s no god! It’s possible that he exists!” Well, yes, but I can show that claims concerning such a supernatural intelligence are internally inconsistent and often self-contradictory. And the fact that super-alien programmer claims and the arguments for them are so similar to those that we find in religious apologetics should also concern those tempted by bostrom arguments. 
But for the sake of amusement, let us re-write our reading of the original bostrom argument, substituting super-intelligent aliens from another world (SIAFAW) for post-humans:
1.’’ The probability of SIAFAW with world-simulating capabilities is quite low; or the probability that we all now live as simulations is quite high.
2.’’ However, given the possibility of there being many universes with uncountable numbers of planets like our own, the probability of ‘the probability of – SIAFAW with world-simulating capabilities is quite low’ is itself low. The probability of SIAFAW with world-simulating capabilities is thus high.
3.’’ Whether we are or are not living as simulations is probabilistically indeterminate.
4.’’ But if the probability of SIAFAW with world-simulating capabilities is high, then it has already likely occurred.
5.’’ Thus, the probability that we all now live as simulations is quite high.
But with the temporal element stripped from the argument, on what basis can we assume the probability of SIAFAW simulators is very high? The problem is that we now have a probability resting on a mere logical possibility. The original argument assumed that history is a one-way highway to future progress, but the SIAFAW argument doesn’t assume anything other than that there are super-intelligent aliens, either from our own universe or another one. Not only do we have no reason for thinking this is true, we have no grounds upon which to calculate its probability. What was once an argument, then, becomes little better than a series of assertions that are little more than appeals to non-divine divinities.
From the abstract:
This paper argues that at least one of the following propositions is true: (1) the human species is very likely to go extinct before reaching a “post-human” stage; (2) any post-human civilization is extremely unlikely to run a significant number of simulations of their evolutionary history (or variations thereof); (3) we are almost certainly living in a computer simulation. It follows that the belief that there is a significant chance that we will one day become post-humans who run ancestor-simulations is false, unless we are currently living in a simulation.
It would seem from this summary that Bostrom wants to argue that our descendants will not be evolved post-humans, and thus we are not living in a simulation, but the remainder of his paper argues the opposite.
Proposition 1 is problematic, given that extinction is an unknown, predicated on a myriad of incidentals, most of which we cannot know. Given the possible developments that Bostrom admits might make Proposition 1 moot, nothing hinges – either assertively or probabilistically – on Proposition 1. Thus, Bostrom’s argument falls apart. There is no trilemma, but rather, a mere thought experiment.
3. However if we set this second proposition aside, then we don’t have a trilemma, we have an exclusive disjunction: Either the human species goes extinct before evolving into post-humans, or we are living in a simulation.
It is notable that in his response to Bostrom, Eggelston raises the question that Bostrom claims he has no interest in: how do we live our lives if we accept that we live in a simulation? I’ve indicated that I find this question unresolvable. But Bostrom’s threat of extinction surely carries its own ethical implication in one respect: accepting we live in a simulation, we should do everything to achieve post-human evolution within it; we thus either justify the simulation, which our programmers would surely admire, or we would prove ourselves as not simulations but become post-humans in the process, capable of such simulations. (The SIAFAW variant of this has us advancing our technology to somehow communicate with the SIAFAW in the future.)
Essentially, this is Pascal’s Wager for the computer age – believe in God and even if God doesn’t exist, you will live a better life (and you might get to heaven). Believe otherwise, and you will live miserably, and if God does exist you will burn in Hell.
So, this is the real bostrom argument underlying all this simulation hypothesis chatter: Promote post-human evolution, and realize it in either reality or simulation, or the human race is doomed.
4. One of Eggleston’s major points is that without this assumption, Bostrom’s original argument begs the question of indifference. After all, the principle of indifference determines a probability prior to any outcome. Hence, presumption of any outcome cannot be leveraged as added information to the indifference between the choices. Hence, the probability we live in a simulation can only be salvaged by assuming the probability that there are other worlds with the level of civilization needed to engage in simulation, whose inhabitants then do so.
5. What does it get us in science to pre-suppose gods or super-aliens? It’s one thing to be a believer and to look at the wonders science reveals and exclaim, “how beautiful is god’s creation!” It’s another thing entirely to initiate scientific research on the basis that the natural world simply must (or must be made to) reveal some god’s existence, and the same is true of super-aliens. I’m no longer a proselytizing atheist – those who wish to believe in God or gods may do so if they choose. But I see no need and have no interest in trying to mold science or philosophy according to a “possibility” that is little more than a hope. And of course, the same is true of super-alien simulators.