On most days, we do not wake up anticipating that we may be suddenly thrust into the sky while popcorn shrimp rains down like confetti, as some guy roars from above: “Hey, there, I’m Jack. And you are in a computer simulation.” Instead, we wake up thinking that an atom is an atom, that our physics is inherent to this universe and not prone to arbitrary change by coders, and that our reality is, well, real.
Yet there may be another possibility. Game developers have opened up massive, explorable universes and populated them with computer-generated characters based on advanced A.I. The experiences still lack some key components of reality, but a precise simulacrum of our world and its denizens may just be a matter of time. Someday soon, you could spend your waking hours wandering through alternative worlds where most of the people you meet aren’t really people. Indeed, how will you know whether you’re a person? When virtual humans come to outnumber real ones, then as the Oxford philosopher Nick Bostrom argued in an influential 2003 paper, any given person should rationally conclude he or she is made of ones and zeros rather than flesh and blood.
In effect, we will have created parallel universes nested within our own universe rather than existing in distinct spacetime bubbles or branches of the quantum wavefunction. “Some of the conceptional problems that are raised by them are similar,” says Caltech cosmologist Sean Carroll. “You would like to know: What is the probability that I end up in one universe versus another? Can I make predictions for the kinds of thing that I would see if the scenario was true?” The basic problem is simple and deeply worrying: The observations you make in a multiverse, whether in space or in silico, need not bear any relation to the way things really are.
It is not science fiction to wonder whether we living in a simulation right now. If so, we might be able to tell. Computers operate on universal principles first articulated by Alan Turing in the 1930s. If we are simulated, the computer that runs us will be subject to the same rules as our own computers are, even if the laws of physics in its universe are vastly different. In particular, if the universe that hosts the simulation is limited, its computational resources will be limited as well. It might, for example, model spacetime as discrete rather than continuous, and we could look for that by observing very high energy cosmic rays, as MIT theoretical physicist Zoreh Davoudi argued at the 2016 Isaac Asimov Memorial Debate. We might also look for signs that the coders had to cut corners, leading to deviations from physical laws.
Even our memories are more likely to be implanted than true.
It is not self-evident, though, that we would recognize such errors. The universe may have been crafted to keep us in the dark. “It may be impossible for us to get evidence that we are not in a simulation because… the simulators will just simulate everything, including that evidence,” suggests NYU philosopher David Chalmers. Indeed, we may have been crafted to remain oblivious. “You would not expect simulated beings to reach the judgement that they are in a simulation, unless they are in one of the simulations that could support beings that could make judgments of that sort,” says philosopher Eric Schwitzgebel at the University of California, Riverside.
But if that’s the case, we run into a paradox. The argument for living in a simulation is based on the laws of physics and logic. But if we are living a simulation, we can’t trust those laws, so we have no basis to conclude we are living in a simulation. A scenario “where the belief that you’re in a sim reasonably causes you to doubt the trustworthiness of the very evidence that convinced you that you were likely to be in a sim,” Schwitzgebel says, “could be self-negating.”
A similar paradox comes up in cosmology: the Boltzmann brain. Physical theory suggests we are statistically more likely to arise from an ephemeral agglomeration of randomly moving atoms than from millions of years of evolution. Even our memories are more likely to be implanted than true. So, our theory has led us to conclude that our theory is illusory, like the man who proclaims, “I am a liar.”
This paradox suggests our theories are flawed. There could be a fundamental problem with creating lifelike simulations, for example. Bostrom says, “At the moment, our understanding of what consciousness is and in particular what the necessary and sufficient conditions are for some computational process to implement consciousness is not really clear.”
One theory of consciousness, Integrated Information Theory, holds that a computer brain simulation is not conscious, even if it reproduces everything the brain thinks and says. Neuroscientist Christof Koch of the Allen Institute for Brain Science in Seattle compares this to simulations in other branches of science. “You can simulate a black hole, but you don’t have the causal power of gravity to actually bend space time,” Koch says. “You have to have mass. Unless you have that causal power, you won’t be able to affect spacetime around you.” By similar logic, if we feel ourselves to be conscious, we cannot be in a simulation. This is not a completely accepted theory by any means, though.
Almost any resolution of the paradox would entail that there can be no true computational multiverse, merely a single real world and multiple flawed copies. Many physicists argue that the cosmological multiverse, likewise, is too paradoxical to be real. People may long for alternative realities, but romanticizing lifelike simulations and an infinite number of “yous” may not be in our best interest.