Opinion | Are We Living in a Computer Simulation? Let’s Not Find Out

Since the 1990s, researchers in the social and natural sciences have used computer simulations to try to answer questions about our world: What causes war? Which political systems are the most stable? How will climate change affect global migration? The quality of these simulations is variable, since they are limited by how well modern computers can mimic the vast complexity of our world — which is to say, not very well.

But what if computers one day were to become so powerful, and these simulations so sophisticated, that each simulated “person” in the computer code were as complicated an individual as you or me, to such a degree that these people believed they were actually alive? And what if this has already happened?

In 2003, the philosopher Nick Bostrom made an ingenious argument that we might be living in a computer simulation created by a more advanced civilization. He argued that if you believe that our civilization will one day run many sophisticated simulations concerning its ancestors, then you should believe that we’re probably in an ancestor simulation right now. His reasoning? If people eventually develop simulation technology — no matter how long that takes — and if they’re interested in creating simulations of their ancestors, then simulated people with experiences just like ours will vastly outnumber unsimulated people.

If most people are simulations, Professor Bostrom concluded, the odds are good that we ourselves are simulations. Our world would be just one simulation of many, perhaps part of a research project created to study the history of civilization. As the physicist (and Nobel laureate) George Smoot has explained, “If you are an anthropologist/historian and want to understand the rise and fall of civilizations, then you will need to make very many simulations involving millions to billions of people.”

The theory that we are living in a computer simulation may sound bizarre, but it has found adherents. The technology entrepreneur Elon Musk has said that the odds that we are not simulated are “one in billions.” Professor Smoot estimates that the ratio of simulated to real people might be as high as 10¹² to 1.

In recent years, scientists have become interested in testing the theory. In 2012, inspired by Professor Bostrom’s work, physicists at the University of Washington proposed an empirical experiment of the simulation hypothesis. The details are complex, but the basic idea is simple: Some of today’s computer simulations of our cosmos produce distinctive anomalies — for example, there are telltale glitches in the behavior of simulated cosmic rays. By taking a closer look at the cosmic rays in our universe, the physicists suggested, we might detect comparable anomalies, providing evidence that we live in a simulation.

Similar experiments were proposed in 2017 and 2018. Professor Smoot captured the promise of these proposals when he declared, “You are a simulation and physics can prove it.”

So far, none of these experiments has been conducted, and I hope they never will be. Indeed, I am writing to warn that conducting these experiments could be a catastrophically bad idea — one that could cause the annihilation of our universe.

Think of it this way. If a researcher wants to test the efficacy of a new drug, it is vitally important that the patients not know whether they’re receiving the drug or a placebo. If the patients manage to learn who is receiving what, the trial is pointless and has to be canceled.

In much the same way, as I argue in a forthcoming paper in the journal Erkenntnis, if our universe has been created by an advanced civilization for research purposes, then it is reasonable to assume that it is crucial to the researchers that we don’t find out that we’re in a simulation. If we were to prove that we live inside a simulation, this could cause our creators to terminate the simulation — to destroy our world.

Of course, the proposed experiments may not detect anything that suggests we live in a computer simulation. In that case, the results will prove nothing. This is my point: The results of the proposed experiments will be interesting only when they are dangerous. While there would be considerable value in learning that we live in a computer simulation, the cost involved — incurring the risk of terminating our universe — would be many times greater.

Consider the following hypothetical proposal for an experiment at the Large Hadron Collider, the world’s largest particle accelerator: “This experiment is unlikely to succeed in producing an interesting result, but if it does succeed in producing an interesting result, it may cause the annihilation of our universe.” Would conducting this experiment be justified? Of course not.

As far as I am aware, no physicist proposing simulation experiments has considered the potential hazards of this work. This is surprising, not least because Professor Bostrom himself explicitly identified “simulation shutdown” as a possible cause of the extinction of all human life.

This area of academic research is rife with speculation and uncertainty, but one thing is for sure: If scientists do go ahead with these simulation experiments, the results will be either extremely uninteresting or spectacularly dangerous. Is it really worth the risk?

Preston Greene is an assistant professor of philosophy at Nanyang Technological University in Singapore.

The Times is committed to publishing a diversity of letters to the editor. We’d like to hear what you think about this or any of our articles. Here are some tips. And here’s our email: [email protected].

Follow The New York Times Opinion section on Facebook, Twitter (@NYTopinion) and Instagram.