Is Reality a Simulation?

From the opening pages

The simulation hypothesis begins with an idea simple enough to explain over coffee and troublesome enough to ruin the rest of the afternoon: the world we experience might not be the most fundamental level of existence. Everything familiar—tables, thunderstorms, tax offices, childhood memories, galaxies, your left knee—could belong to an environment generated by some deeper artificial system. From inside such an environment, its regularities would simply look like the laws of nature. A falling stone would still fall, breakfast would still become cold when ignored, and Tuesday would retain its remarkable ability to arrive without permission. The hypothesis does not require daily life to look strange. A competent simulation would presumably be most impressive when nothing in particular looked suspicious.

That modest description already contains the main source of confusion. There are three very different claims one can make. A simulated reality may be possible. It may be probable that ours is simulated. Or there may be evidence that ours actually is simulated. These claims are frequently blended together because they all contain the useful word “simulation” and therefore appear to be close relatives. They are not. Saying that something is possible establishes very little about how often it occurs. Saying that it is probable requires reasons for assigning it a high likelihood. Saying that there is evidence for it requires observations that genuinely favor it over competing explanations. A dragon might be logically possible under generous assumptions; this does not oblige the local fire department to establish an anti-dragon unit.

The modern philosophical discussion is closely associated with Nick Bostrom’s 2003 paper “Are You Living in a Computer Simulation?”, published in The Philosophical Quarterly. Bostrom considers future civilizations capable of producing enormous numbers of what he calls “ancestor-simulations”: detailed simulations containing beings like their historical predecessors. He then argues that at least one of three broad possibilities must be true. Humanity-like civilizations almost always fail before reaching the necessary technological stage; civilizations that do reach it almost never run large numbers of ancestor-simulations; or observers like us are overwhelmingly likely to be simulated rather than biologically original.