Is the world around us real — or are we living in a simulation, like characters trapped inside some space alien’s video game?
That sounds like a question you might hear at a midnight screening of "The Matrix," but lately it’s become the subject of serious academic debate. High-profile proponents of what’s known as the “simulation hypothesis” include SpaceX chief Elon Musk, who recently expounded on the idea during an interview for a popular podcast.
“If you assume any rate of improvement at all, games will eventually be indistinguishable from reality,” Musk said before concluding, “We’re most likely in a simulation.”
Astrophysicist Neil deGrasse Tyson agrees, giving “better than 50-50 odds” that the simulation hypothesis is correct. “I wish I could summon a strong argument against it, but I can find none,” he told NBC News MACH in an email.
Reality comes under attack
The current assault on reality began with a 2003 paper by Nick Bostrom. In it, the University of Oxford philosopher laid down some blunt logic: If there are long-lived technological civilizations in the universe, and if they run computer simulations, there must be a huge number of simulated realities complete with artificial-intelligence inhabitants who may have no idea they’re living inside a game — inhabitants like us, perhaps.
These beings might imagine themselves real but would have no physical form, existing only within the simulation.
If computer-loving aliens truly exist, Bostrum argued, “we are almost certainly living in a computer simulation.” And then people like Tyson and Musk found their minds blown.
Now scientists are searching for ways to put the simulation hypothesis to the test. Bostrom is eager to see more concrete developments of his idea. Experiments that could distinguish physical reality from a simulation “are what would be needed for it to be a bona fide scientific assertion,” he told MACH.
Scott Aaronson, a computer scientist at the University of Texas at Austin, is more expressive about what such experiments could mean. “If there were bugs in the program running our universe, like in the Matrix movies, those could clearly have observable effects,” he says. “Just like God appearing in a thundercloud could be pretty good empirical evidence in favor of religion.”
Looking for gaps in the sim
Any such bugs in our Matrix world would have to be extremely subtle, or else we would have noticed them by now. Silas Beane, a nuclear physicist at the University of Washington in Seattle, proposes that we may be able to ferret out previously overlooked flaws by uncovering the mathematical structure used to build our simulated reality.
He points out that scientists in his field use a lattice-like set of coordinates to simulate the behavior of subatomic particles. Maybe the aliens (or whoever built our simulation, if it exists) used that approach, too. If our reality is built on top of a lattice, there’d be a fundamental coarseness to it, since there could be no details in our mock-universe smaller than the resolution of the simulation.
Even if the resolution limit is too small for us to observe directly, Beane says, we may be able to detect it experimentally. In a paper he wrote with two colleagues, Beane proposes that a simulation lattice could affect the behavior of ultra-energetic particles known as cosmic rays, affecting their orientation and maximum intensity.
Instruments like the Telescope Array, a network of 500 detectors scattered across 300 square miles of Utah desert, watch for cosmic rays as they crash into Earth’s atmosphere from deep space. The detectors have already discovered particles as much as 100 quintillion times as energetic as visible light. That seems like a great place to start hunting for bugs in any simulation.
It would be a delicate task: High-energy cosmic rays are rare, and the deviations from ordinary physical effects might not be obvious. But Beane and company are encouraged that making such a measurement is feasible, at least in principle. “There always remains the possibility for the simulated to discover the simulators,” the authors write.
Is our world badly rendered?
Another way to sleuth for glitches in the simulation is by looking inward rather than outward. In a recently proposed test, former NASA engineer Thomas Campbell and his colleagues point out that human video game designers typically maximize the efficiency of their programming by generating only the parts of the virtual world that players can see. If our Matrix overlords are similarly focused on efficiency, they may be meticulous about simulating details while we’re watching an event, but allow a looser style of simulation when they think nobody is looking.
Following that line of thought, Campbell is focusing on subtle quantum physics experiments, where gaps in the simulation might be most obvious. He has conceived several tabletop optics arrangements that would shoot a laser beam through an elaborate sequence of slits, mirrors and detectors. Photons of laser light would follow different paths contingent on whether they are behaving like waves or like particles, which in turn depends on the structure of the setup.
Or rather, it should depend only on the setup. If reality is rendered at the moment we are watching, Campbell theorizes, his experiment could yield results normally considered impossible, such as being able to predict whether an individual photon passes through or bounces back when it hits a half-reflective mirror. That outcome would "represent an unambiguous indicator that our reality must be simulated," he writes.
As a huge bonus, Campbell claims the experiment could also explain the weird way that events in quantum physics seem to be influenced by the observer: It may be a quirk of the simulation we live in, not a fundamental aspect of reality.
Marcus Nowack, a computational physicist at Lawrence Berkeley National Lab with a keen interest in the simulation hypothesis, sees problems with these attempts to outsmart the Matrix. For instance, Campbell assumes that a simulation would be for our benefit only, “but what if the simulator does not simulate us just for us, but rather to observe how everything plays out?” And Nowack notes that Beane’s approach would come up empty if the lattice of reality is too fine for us to detect — or if the wily simulators have built in systems to defeat any test we might run.
The bottom line, Nowack says, is that it’s impossible to test the simulation hypothesis as a whole. The best we can do is explore a “limited neighborhood” of notions about how the simulation might work, and hope that the designers are too lazy or too indifferent to prevent us from discovering their handiwork.
Simulations all the way down?
That assessment seems to combine the worst of both worlds: We don’t know if we’re living in a simulation, but merely knowing that we might be in a simulation seems pretty depressing. Tyson calls it “a creepy concept.” Bostrom adds that it “seems to foster a sense of absolute dependency.”
But there are also constructive ways to look at the simulation hypothesis. Aaronson sees it as a fresh way to contemplate “the ancient mysteries of where our universe comes from, who or what created it, and why.”
Nowack also finds it a fruitful thought problem about where human research might be headed. “I simulate many phenomena that represent only a tiny subset of all the physical things going on around us,” he says. “I’m interested in the computational effort a world simulation would need, and the computer involved.”
The rapid advance of AI research and computer modeling raises the possibility that one day we humans might create our own hyper-realistic simulations containing self-aware digital beings. That possibility is both inspiring and disconcerting. It also introduces a new set of brain-hurting questions. Would these simulations-within-a-simulation be the end? Or could our simulated beings keep going and create yet another layer of simulation, and so on?
“There could be an infinite stack of simulations if there were infinite serial computing power available at the bottom level and in each higher level,” Bostrom says. Fortunately, in a finite universe things can never get quite that crazy, he says: “As far as we can tell, the serial computing power available to a simulator in our universe is finite, in which case we could only create finitely many levels of simulations.”
Phew. So maybe that’s one small consolation: We might be in a simulation, or a simulation in a simulation, but at least we can be pretty sure that it’s not simulations all the way down.