Philosopher Nick Bostrom's famous Simulation Argument suggests it's highly probable that we live inside a supercomputer. But one philosopher takes this hypothesis to task, arguing in a new paper that there are other post-human scenarios that need to be taken into account.
Before we get started, it's important to note that this discussion is limited to the philosophical arguments in support of the simulation hypothesis. But the day is coming when physicists may be able to prove or disprove it more scientifically.
According to Bostrom's Simulation Argument, only one of the following three propositions can be true given the potential for a technologically mature "posthuman" civilization to come into the possession of enormous computing power:
- The human species is very likely to go extinct before reaching a posthuman stage
- Any posthuman civilization is extremely unlikely to run a significant number of simulations of their evolutionary history
- We are almost certainly living in a computer simulation
If the first proposition is true, it's likely that we'll go extinct before reaching posthumanity (in which case there will be no so-called "ancestor simulations"). If the second is true, "then there must be a strong convergence among...advanced civilizations so that virtually none contains any relatively wealthy individuals who desire to run ancestor-simulations." This seems unlikely.
But if the third proposition is true, then we almost certainly live in a computer simulation. One way of looking as it is through the lens of probability; if there's one "real" world, and a million simulated worlds, it more probable by several orders of magnitude that we're in a simulation.
But as Bostrom himself notes: "In the dark forest of our current ignorance, it seems sensible to apportion one's credence roughly evenly between [these three propositions]."
And it's here where philosopher Paul Franceschi from the University of Corsica in France takes issue with the argument.
A 'Reference Class' Problem
Franceschi says that Bostrom didn't get the reference class right.
"It consists of human simulations," he told io9. "The original argument refers to a reference class which is that of computer simulations of human beings, of a very high quality, and by nature indiscernible from the genuine ones." But there's more to simulations than just this, he argues — Bostrom failed to account for a much broader class of posthuman simulations.
A certain ambiguity exists in the mere notion of simulations, he says, and a question subsequently arises about the applicability of the Simulation Argument to other possible types of human simulations or immersive virtual reality experiences. To that end, Franceschi describes three other kinds of simulations:
- Aware-simulations: A type of simulation that's in every respect identical to those described in Bostrom's original argument, i.e. simulations that are almost indiscernible from genuine humans, the only difference being that they're aware of their own nature in the simulation.
- Rough-simulations: Some virtual simulations at a slightly lower quality, with regard to the perfect ones hinted at in the original argument.
- Cyborg-type simulations: Simulations indiscernible from human cyborgs with, say, neural implants (possibly with full or partial uploads); think of The Matrix.
Franceschi breaks down the assumption that we likely live in a simulation into three points: (1) the notion that simulations greatly outnumber genuine humans (disproportion), (2) the fact that we are probably simulants (self-applicability), and (3) the fact that we're totally unaware that we're being simulated (unawareness).
But by virtue of his new posthuman references classes, Franceschi argues that new conclusions can be produced:
- The original argument: As noted, it entails disproportion, self-applicability and unawareness. This conclusion is worrying because it suggests we're simulants blind to our true nature as living things.
- Aware-simulations: The argument only entails disproportion (and not self-applicability or unawareness). It's a reassuring conclusion because it suggests simulants are (mostly) aware of their existential situation.
- Rough-simulations: Like the previous item, it only entails disproportion. This conclusion is also reassuring.
- Cyborg-type simulations: This also entails disproportion and self-applicability (and not unawareness). This conclusion is reassuring, too — it suggests that many simulants have a "real world" aspect to them.
By having alternate choices of different reference classes, and at a greater level of extension, different conclusions can be drawn from the premises — conclusions that produce reassuring conclusions. Put another way, it can't possibly be correct that every posthuman simulant is unaware of their true nature, or that other types of simulations don't exist.
"Now given that there does not exist in the Simulation Argument an objective criterion allowing to choose the reference class non-arbitrarily, we can choose it at different levels of restriction or of extension."
In this context, he claims that the disturbing conclusion which is associated with the original argument turns out to be an arbitrary conclusion. At the same time, there are several other reference classes which have an equal degree of relevance to the argument itself — reference classes which suggest a reassuring conclusion.
Read the entire study at the pre-print PhilSciArchive: "The Simulation Argument and the Reference Class Problem: the dialectical contextualist's standpoint"
Image: Vanilla Sky