History could be reenacted, raising the question of whether we live in a simulation

In "Too Hard For Science?" I interview scientists about ideas they would love to explore that they don't think could be investigated. For instance, they might involve machines beyond the realm of possibility, such as devices as big as galaxies, or they might be completely unethical, such as experimenting on children like lab rats. This feature aims to look at the impossible dreams, the seemingly intractable problems in science. However, the question mark at the end of "Too Hard For Science?" suggests that nothing might be impossible.

The scientist: Nick Bostrom, director of the Future of Humanity Institute and the Program on Impacts of Future Technology, Oxford Martin School, at the University of Oxford.

The idea: "Rerun human history," Bostrom suggests. "Evacuate all humans, except some hunter-gatherer bands, and also erase all traces of advanced civilization. Then wait and observe how history unfolds — expect this stage to take several thousand years. Note down the results, and repeat."

"With enough repetitions, we would get some really nifty statistics that would enable us to understand world history in a much deeper way that is currently possible," he continues. "We could give authoritative answers to questions such as how accidental or deterministic are the large-scale patterns of history? Will the same basic technological discoveries be made in almost all reruns of history? How often does an effective world government emerge and, if so, how stable is it? Do people always end up with roughly similar distribution of scientific, religious, and moral ideas?"

"Even more interesting would be if we could step aside and let the experiments continue a bit beyond our current point of development," he says. "We would then be able to answer questions such as how likely is it that a human civilization at our stage of development will reach a post-human condition? How likely is that some advanced technology will cause our extinction? If some runs of civilization makes it through the critical times ahead and reaches some wonderful post-human stage of development, what did they do to get there?"

The problem: "That should be pretty obvious," Bostrom says.

The solution? Bostrom does not see any potential ways to rerun history now or in the near future. "An extremely advanced civilization that has started an open-ended process of space colonization might be able to do something like this," he says. "Alternatively, a technologically mature civilization with machine super-intelligence might be able to create extremely detailed and realistic computer simulations of human history, with slightly varying initial conditions. Doing it in simulation would be much faster and cheaper than doing it directly in physical reality."

"However, to create genuine simulations of one's ancestors, ones that are realistic enough fully to capture all the general dynamics that are relevant in shaping the course of history, might require that the people in the simulation be rendered in sufficient detail that they can think, experience, and be conscious like we are," Bostrom conjectures. "This might require simulating each neuron in every brain, among other things. However, it need not involve simulating every molecule or elementary particle, something that would be utterly infeasible, since the goal is not to replicate exactly one particular run of history but rather to recreate the general patterns and behavior that systematically shape human-like histories."

"If we think that some of the mature civilizations that may exist somewhere in our apparently infinite universe are interested in creating such ancestor simulations, and if we think that each such civilization could over time create millions and millions of ancestor simulations, and if we assume that a person in a simulation cannot tell that they are in a simulation, then we have reason to wonder about our own position in this scheme of things," Bostrom concludes. "If most people with our kinds of experience exist in simulations, then we should think that we are probably now n a simulation. Here we start to approach the simulation argument — but that is perhaps a topic for another occasion."


If you have a scientist you would like to recommend I question, or you are a scientist with an idea you think might be too hard for science, email me at toohardforscience@gmail.com.