One of the oldest problems in philosophy is that of the external world. It present an epistemic challenge forged by the skeptics: how do I know that what I seem to be experiencing as the external world is really real for real? Early skeptics often claimed that what seems real might be just a dream. Descartes upgraded the problem through his evil genius/demon which used either psionic or supernatural powers to befuddle its victim. As technology progressed, philosophers presented the brain-in-a-vat scenarios and then moved on to more impressive virtual reality scenarios. One recent variation on this problem has been made famous by Elon Musk: the idea that we are characters within a video game and merely think we are in a real world. This is, of course, a variation on the idea that this apparent reality is just a simulation. There is, interestingly enough, a logically strong inductive argument for the claim that this is a virtual world.
One stock argument for the simulation world is built in the form of the inductive argument generally known as a statistical syllogism. It is statistical because it deals with statistics. It is a syllogism by definition: it has two premises and one conclusion. Generically, a statistical syllogism looks like this:
Premise 1: X% of As are Bs.
Premise 2: This is an A.
Conclusion: This is a B.
The quality (or strength, to use the proper term) of this argument depends on the percentage of As that are B. The higher the percentage, the stronger the argument. This makes good sense: the more As that are Bs, the more reasonable it is that a specific A is a B. Now, to the simulation argument.
Premise 1: Most worlds are simulated worlds.
Premise 2: This is a world.
Premise 3: This is a simulated world.
While “most” is a vague term, the argument is stronger than weaker in that if its premises are true, then the conclusion is logically more likely to be true than not. Before embracing your virtuality, it is worth considering a rather similar argument:
Premise 1: Most organisms are bacteria.
Premise 2: You are an organism.
Conclusion: You are a bacterium.
Like the previous argument, the truth of the premises make the conclusion more likely to be true than false. However, you are almost certainly not a bacteria. This does not show that the argument itself is flawed. After all, the reasoning is quite good and any organism selected truly at random would most likely be a bacterium. Rather, it indicates that when considering the truth of a conclusion, one must consider the total evidence. That is, information about the specific A must be considered when deciding whether or not it is actually a B. In the bacteria example, there are obviously facts about you that would count against the claim that you are a bacterium—such as the fact that you are a multicellular organism.
Turning back to the simulation argument, the same consideration is in play. If it is true that most worlds are simulations, then any random world is more likely to be a simulation than not. However, the claim that this specific world is a simulation would require due consideration of the total evidence: what evidence is there that this specific world is a simulation rather than real? This reverses the usual challenge of proving that the world is real to trying to prove it is not real. At this point, there seems to be little in the way of evidence that this is a simulation. Using the usual fiction examples, we do not seem to find glitches that would be best explained as programming bugs, we do not seem to encounter outsiders from reality, and we do not run into some sort of exit system (like the Star Trek holodeck). Naturally, this is all consistent with this being a simulation—it might be well programmed, the outsider might never be spotted (or never go into the system) and there might be no way out. At this point, the most reasonable position is that the simulation claim is at best on par with the claim that the world is real—all the evidence is consistent with both accounts. There is, however, still the matter of the truth of the premises in the simulation argument.
The second premise seems true—whatever this is, it seems to be a world. It seems fine to simply grant this premises. As such, the first premise is the key—while the logic of the argument is good, if the premise is not plausible then it is not a good argument overall.
The first premise is usually supported by its own stock argument. The reasoning includes the points that the real universe contains large numbers of civilizations, that many of these civilizations are advanced and that enough of these advanced civilizations create incredibly complex simulations of worlds. Alternatively, it could be claimed that there are only a few (or just one) advanced civilizations but that they create vast numbers of complex simulated worlds.
The easy and obvious problem with this sort of reasoning is that it requires making claims about an external real world in order to try to prove that this world is not real. If this world is taken to not be real, there is no reason to think that what seems true of this world (that we are developing simulations) would be true of the real world (that they developed super simulations, one of which is our world). Drawing inferences from what we think is a simulation to a greater reality would be like the intelligent inhabitants of a Pac Man world trying to draw inferences from their game to our world. This would be rather problematic.
There is also the fact that it seems simpler to accept that this world is real rather than making claims about a real world beyond this one. After all, the simulation hypothesis requires accepting a real world on top of our simulated world—why not just have this be the real world?