A recent fad among intellectuals is pondering whether the world we live in is really just a simulation running on the cloud computers of our distant post-human descendants. Like angels dancing on pinheads the idea boggles the imaginations of stoners and serious thinkers alike. As odd as this sounds there are a surprising number of otherwise smart people who have convinced themselves that it's almost certainly true. Should we take it seriously? Are we living in a simulation?
Short answer: No.
Slightly longer answer: There's every reason to believe in the material universe as science describes it, and zero tangible reason to believe that it's all computations.
For a full answer we have to consider the hypothesis and the many ways it fails to stand up. Formally the argument is three assertions, one of which must be the truth:
- Human civilization will cease technological progress.
- Our technologically advanced descendants will have no interest in simulating their ancestors.
- The world we live is is most likely a simulation.
Informally the story goes like this. Humans continue to advance technologically (statement 1 is false). At some point computers become so powerful that it becomes possible to simulate the entire Earth and billions of people on it very cheaply and easily. If the people of the future are interested in doing that at all (statement 2 is false) then there will be huge number of such "ancestor simulations". Since the simulations far outnumber the one real Earth history, it's most likely that our own existence is as one of those simulations.
This first problem with this argument is that it isn't science. Science is the only valid way we know to determine what's true about the world, and whether the world is a simulation or not is certainly a science question. The way science operates is that we start with an observation of something in the world, make a model that explains that observation, then confirm the model by testing its predictions beyond the original observation. The simulation hypothesis fails as science at both ends -- there's no observation that requires the hypothesis in the first place, and the hypothesis makes no testable predictions. A positive feature of the simulation is that it's indistinguishable from reality, or even if it is different we have no access to anything outside the simulation so there's nothing to compare it to. A hypothesis that cannot be tested is not scientific.
As a philosophical argument it also fails to Occam's Razor. This states that if we have two competing hypotheses that both fit the facts, we should prefer the one with fewer entities or assumptions. The simulation hypothesis doesn't simplify anything. In fact it requires the normal history of the universe, the natural evolution of humans, and the development of human civilization. Without a time and place where billions of humans were living on the real Earth the simulation theory can't work. So it's just taking everything we know about reality and tacking on "but this is a simulation of that". Because we can explain reality perfectly well without needing it to also be a program running on future computers, Occam's Razor begs us to accept the simpler explanation.
Also a dilemma (or in this case trilemma since it has three options) like this often has falsely limited choices. The most famous such argument is C.S. Lewis's trilemma arguing for the divinity of Jesus. He gives us three choices, that Jesus must have been either "Liar, Lunatic, or Lord", and then with Biblical sophistry eliminates the first two so we are supposed to accept "Lord". The problem is those aren't really the only three possibilities. An obvious forth choice that Lewis conveniently ignored is "Fictional". While the prongs of the simulation trilemma seem comprehensive, we have no reason to believe that they are. This is speculating about the far, far future. They might have capabilities or limitations that we simply can't imagine.
As an aside, it's strange that some of the same people who believe in the simulation hypothesis also believe in the singularity. One of the defining aspects of the singularity -- indeed why it has that name -- is that it's impossible for us on this side of history to even imagine what we'd find on the other side. And yet these same people start from "Moore's Law + video games" and conclude that sandbox Earths with procedurally generated populations are nigh inevitable.
Finally, if we ignore all the other problems with the argument and just take it at face value, it still fails. If we accept that those three statements are the only outcomes possible, which is likely to be true? Option 1 might be true; humans could go extinct or there could be good reasons that the kind of simulation imagined is impossible even at a high level of technology. But let's be optimistic. That means the deciding factor between 2 & 3 is whether those future people want to set up simulations. But let's consider what that means. The subjects in the simulation are as intelligent, conscious, and self-aware as we are. They aren't just NPCs in a video game; they're people. Building the simulation means trapping people inside a world where they will suffer and die for historical accuracy, all without their consent. A civilization that would allow this would be monstrously evil, and being optimistic I don't want to imagine our descendants that way. Hopefully the number of ancestor simulations they build will be zero.