In 2003, Nick Bostrom penned a landmark paper entitled, “Are you Living in a Computer Simulation?” It shook the world. Some saw it as a piece of science fiction; others got on board with a radical view of human existence. He based his views on the possibility of advanced technology, the kind that can produce a supercomputer. If there is a substantial chance that our civilization will get to the posthuman stage and run many ancestor-simulations, how come you are not living in such a simulation?
Bostrom knew of the predictions of this advancement, and he took it a step further. Such computing power was used by “later generations” of humans to run simulations of reality. Yes, the word is in its plural form. Within simulations dwell conscious beings, themselves also simulations. His conclusion was that our minds as we know them are not part of the original race: they belong to simulations created by its descendants. The implication is clear: we are living amid these simulated minds created by another civilization. Human biology is thrown out the window while the laws of physics remain. Those who deny this proposition, in Bostrom’s view we cannot depend on the possibility of future descendants running even more simulations.
Bostrom was engaging in futuristic speculation in the paper, like the people reading it, looking for answers to metaphysical questions. In his words, the paper “suggests naturalistic analogies to certain traditional religious conceptions, which some may find amusing or thought-provoking.”
The paper opens with an assumption or argument, followed by empirical reasons for believing in his hypothesis. He relies on probability theory and what he denotes as a weak indifference principle. As a philosopher, he is grounded in the philosophy of mind and the concept of substrate-independence. Mental states supervene on any of a broad class of physical substrates. At this point come computational structures and processes provided by a system. It is akin to conscious experience.
While not an essential property of consciousness, carbon-based biological, neural networks in the brain implement it. He then asks the hidden question: can computer processors do the same thing? His argument takes this turn: we need not assume the necessary truth of substrate-independence whether from an analytical or metaphysical perspective. We would see that a computer running a simulation program would have consciousness. This is how to create a “mind” in a computational process. It would thus behave in the normal human fashion with subjectivity.
A weaker assumption about substrate-independence that is still widely accepted is considered by Bostrom: the human brain could be replicated structurally in detail, right down to the firing of individual synapses. It pertains to the smaller neurotransmitters, chemicals and other nerve growth factors that come into play in human cognition. According to the theory, even these small-scale elements affect human subjectivity by influencing computational activities either directly or indirectly. In short, “if there can be no difference in subjective experience without there also being a difference in synaptic discharges, then the requisite detail of simulation is at the synaptic level (or higher).”
Bostrom posits the super computer creating conscious minds, but does it have limitations? Well at present it is not even possible. We don’t have the software nor the requisite power. He reviews arguments about future technological progress that will overcome the current computational shortcomings. It could come sooner than we think – in a few decades. It could be hundreds of thousands of years! The simulation hypothesis works short or long term.
Within the human imagination is a mature stage of technological innovation that has the power to use cosmic resources. We may envision a high degree of novel physical phenomena beyond the current laws of physics. But while our current level of knowledge sees limits on information processing, we can fathom the lower bounds as seen in the work of Eric Drexler on the design of a system only in the dimensions of a sugar cube. Seth Lloyd did his own calculations to determine the upper bounds of computation.
In the end, we can hypothesize or estimate the amount of computing power (non-biological processes) needed to simulate the nervous tissue in the mind of a human being. A supercomputer could even generate memories. Thus philosophy takes on biology.
Bostrom employs logic and mathematics to assess the scope and granularity of a simulation. It is not yet feasible to virtually recreate the universe, even at the quantum level. It could be in the future, using existing laws of physics of course. The simulated “humans” would thus act normally without evident glitches. The inner core of the earth is irrelevant and the cosmos can take on a smaller dimension. In Bostrom’s words, “macroscopic objects in inhabited areas may need to be continuously simulated, but microscopic phenomena could likely be filled in ad hoc.: It would all be observable through an electron microscope but there would be no way to make a confirmation as it would not detect all parts of the created universe. If we could design a system(s) using known principles to detect unobserved phenomena at the microscopic level, it would be an exception to the assumption of a lack of verification.
Also in simulation theory is the idea of a continuous representation of computers that operate at the individual level of logic. This is not yet within the realm of computing power but a “posthuman similar” would have enough power to track human belief systems. This means that the computer could fill in missing details during human observations at the microscopic level. This type of “editing” would keep the simulation plausible or the simulation could even be “rerun”. The simulation would have to be indistinguishable from physical reality for the humans dwelling within it. More needs to be learned about the computational requirements to do this.
We can posit a computer with the ability to simulate the entire mental history of man, or the idea of “ancestor-simulation”. It would take but one millionth of the available processing power no doubt. Hordes of super computers are possible given the technology of a posthuman civilization. A fraction of their power could run multiple simulations.
At the heart of the matter is the belief that we are in a simulation although we don’t have the wherewithal or information to test our own experiences. Enter the weak indifference principle. Bostrom uses several cases to explain and apply it. For one, if all human minds are qualitatively identical, they enjoy the same experiences and information. The second case may be that human minds are similar to a typical structure. However, they may be qualitatively distinct as well. For Bostrom, either way, the simulation argument holds true given there is no additional biological information. These cases may be “trivial special instances” in his words. He uses DNA and genetic sequencing to elucidate what this means. The “bland indifference principle” comes into play at this point amid the other principles, in contrast to Laplean ones and immune to Bertrand’s paradox.
The Doomsday Argument has been around and those adhering to it may worry that the bland principle of indifference could negate it. Bostrom does not agree with their concern. A stronger and more controversial premise is at hand: one should reason as if part of a random sampling from all those who will ever have lived (past, present, and future). This is “even though we know that we are living in the early twenty-first century rather than at some point in the distant past or the future”. The bland indifference principle applies only to cases where no information exists about which group of people we belong to.
Bostrom ponders making a bet on the existence of a simulation using the bland principle of indifference. He says that everyone will win if the odds support the rational belief that if everyone placed a bet on living a simulation, they would know where most people are. If they bet to the contrary that no simulation is possible, they all lose. As such, it behooves us to heed the principle.
There are issues of extinction and a possible posthuman stage. Also at issue is technological development and the collapse of civilization, proposition (1) of Bostrom’s paper. Literally it reads, the human species is very likely to go extinct before reaching a “posthuman” stage. What if a dangerous technology was developed such as molecular nanotechnology? Evil nanobots could cause the end of all planetary life.
The simulation argument has another alternative, proposition (2). Posthuman civilizations don’t care to run an ancestor-simulation at all. But the fraction of it could be negligible as well. They might not have the resources, desire, or interest. Certain laws might be enforced to prevent it from happening. There might be an “ethical prohibition” or ban against running such a simulation, or more than one.
Bostrom doesn’t see a moral issue here despite the fact that we view human existence as having ethical value. Maybe some posthuman desires are “silly” or “unfathomable” with little scientific value. We don’t know as it is likely that posthuman societies will differ from the current one. We don’t know if some “wealthy independent agents” will be free to act on their desires.
This brings Bostrom to a third alternative, proposition (3) and an intriguing one in his view. If we are in fact living within a simulation, the cosmos as presently observed, is but a bit of total physical existence. It is even possible that the laws of physics as we know them won’t apply. A virtual or simulated world would still be “real” in some sense, while it might not be fundamentally real. Simulated civilizations might contain posthumans who would use powerful computers as “virtual machines” (according to computer science). They could even simulate these devices with multiple iteration steps. Thus the first and second premises decline in value. The posthumans could in fact also be simulations along with their creators. The cost of such an undertaking is beyond comprehension.
Reality could have many levels and its own expanding hierarchy that allows for more. At one point, the simulation could be terminated. The religious implications are clear. Are these posthumans “omnipotent” like gods and can interfere with human existence, within the bounds of physical laws? Likely, they would have a superior intelligence than humans and the ability to monitor the workings of the universe. There could be a race of demigods at the lowest level ruled by more powerful gods.
A “naturalistic theogony” results for Bostrom upon contemplation of these themes. It would look into the hierarchy and any imposed constraints on lower level beings from higher ups. He calls the bottom the “basement level”. Rewards and punishments could dictate behavior, maybe a kind of novel moral criteria that applies to simulations. Bostrom even postulates the possibility of an afterlife to continue the religious connection. As with religion as we know it, ethics or moral behavior comes into play. It is a virtuous circle for the philosopher, or an “ethical imperative” from “nowhere” but of interest to everyone involved.
Simulations could be “selective” with only one person or a small group of humans. Bostrom calls the rest of mankind “shadow-people” or zombies since they are simulated at only a sufficient level to lack suspicion. They would take less enterprise and a lower cost to produce compared to “real people”. He considers that these shadow-people would be indistinguishable from real humans but lacking consciousness. Where does one find themselves in this mix? It depends upon their number. It would require one hundred billion times as many simulations of a single mind (“me-simulations”) as ancestor simulations to create innumerable editions that would include most everyone.
What if the simulators used only certain aspects of the human mentality in their simulated beings. They might have false memories, per Bostrom, that would be typical of human experience. Then the matter of evil becomes problematic with a farfetched solution. What of the existence of suffering in the world? If there is no suffering, memories of it would be illusions. In his view, you can only entertain the hypothesis while not suffering.
Bostrom ends his seminal paper with questions about the implications of living in a simulation. He does not find them to be radical at all. It just takes some empirical study of the universe we know and see and any revisions of reality that would ensure. They would likely be small or subtle.
He concludes that the truth of a simulated reality should “have no tendency to make us ‘go crazy” or prevent humans from living for tomorrow.” His results are tripartite, given in three propositions. To reiterate, the first states that the human species is very likely to go extinct before reaching a “posthuman” stage. The second asset is that any posthuman civilization is extremely unlikely to run a significant number of simulations of their evolutionary history (or variations thereof). Then the third posits that we are almost certainly living in a computer simulation.
For Bostrom, it follows that the belief is false that there is a significant chance we will one day become posthumans who run ancestor-simulations unless we are currently living in a simulation. The third proposition, he hopes, would decrease the probability of the first, positing that computation constraints make it likely that the simulators would eventually terminate a simulation prior to it reaching the posthuman level. Thus, postulation two would hold true.
Pushing on with his own brand of logic, Bostrom deems it valuable to learn more about posthuman motivations and the constraint of available resources. We might find that we are indeed evolving in the direction of a posthuman civilization. Then the idea of humans as simulations will have some empirical ground. This civilization would be “technological mature” with enormous computing power.
Three more propositions emerge at the conclusion of the paper. (1) The fraction of human-level civilizations that reach a posthuman stage is very close to zero; (2) The fraction of posthuman civilizations interested in running ancestor-simulations is very close to zero; (3) The fraction of all people with our kind of experiences living in a simulation is very close to one.
Logic rules in the end. If (1) is true, we will almost certainly become extinct before reaching the level of post-humanity. If (2) is possible, there must be a strong convergence among the courses of advanced civilizations such that virtually none contains a relatively wealthy individual with the desire to run ancestor-simulations, while being unconstrained to do so. Finally, if (3) is true, we almost certainly live in a simulation. This is where Bostrom wanted to end the entire argument.
In the “dark forest” of human ignorance, it makes sense to hypothesize the plausibility of all three propositions. Finally, according to Bostrom, “unless we are now living in a simulation, our descendants will almost certainly never run an ancestor-simulation.”