Last week a friend pointed me to an interesting paper by Nick Bostrom, titled “Are You Living In A Computer Simulation?” Attempting to even discuss this paper has been equally interesting and not a little amusing. As one might expect, reactions to the subject, even without getting any farther than just the title, tend to be rather knee-jerk, often dismissing the idea out of hand, not necessarily because it’s wrong, but simply because it’s uncomfortable.
As usual, I recommend reading the very accessible paper before this cloudy commentary. The argument is pretty straightforward to describe. First, Bostrom considers what he calls “post-human” civilizations, that start out with humans, or with civilizations like us, but eventually develop sufficiently powerful computing technology to be able to run “ancestor simulations” of humans that exhibit consciousness. (This artificiality of intelligence obviously requires “substrate-independence,” mentioned only briefly in the paper, and which I imagine loses many skeptics, religious groups in particular, right out of the gate. I think John Searle’s imagery is the most amusing here, scoffing at the idea of a mind made out of “beer cans or streams of toilet paper.” But let’s keep moving.)
Then, via some simple mathematical calculations, he argues that at least one of the following propositions is true:
- Any human-like species (including us) will almost certainly go extinct before reaching a “post-human” stage; or
- Any post-human civilization is very unlikely to run ancestor simulations; or
- We are almost certainly living in a computer simulation.
Bostrom provides some interesting interpretations of each of these propositions. Most amusing is his description of post-human civilizations conditioned on (2) being true, where apparently the governments of the far future are not exactly libertarian: “virtually all posthuman civilizations [must] lack individuals who have sufficient resources and interest to run ancestor-simulations; or else they have reliably enforced laws that prevent such individuals from acting on their desires.”
The implications of proposition (3) are the most interesting. The thrust of the argument for (3), given the negations of (1) and (2), is that if most post-human civilizations do run ancestor simulations, then most conscious humans, including ourselves, must be simulated. This leads immediately to the potential for “nesting” behavior of such simulations, where it is possible that not only is our civilization being simulated, but the civilization simulating us is also being simulated, etc.
This idea reminded me of two interesting related subjects, both of which motivated this post. First, the nesting of worlds is a concept that struck me with some force when I was a kid, when I read Wilbur Daniel Steele’s short story, “The Man Who Saw Through Heaven.” You should be able to find it online somewhere, and it is definitely worth a read. I won’t say any more so as not to spoil a good story.
Second, the argument that Bostrom gives in his paper reminded me of the Sleeping Beauty problem, which I think has a similar flavor:
Sleeping Beauty volunteers for an experiment. On Sunday, she is put to sleep. A fair coin is then tossed. If the coin comes up heads, Beauty is awakened and interviewed on Monday, and then the experiment ends. If the coin comes up tails, she is awakened and interviewed on Monday, put to sleep, and awakened and interviewed again on Tuesday. The drug used to put Beauty to sleep erases her memory, so upon awakening she does not know how many times she has been put to sleep, nor what day it is. During each interview, she is told these details of the experiment, and asked the following question: “What is your credence now for the proposition that the coin came up heads?”
This problem may smell a lot like the Monty Hall problem, with all of its associated debate. Indeed, opinions about this problem fall into one of two categories: “thirders” (like me) think that the answer is 1/3, and “halfers” think that the answer is 1/2. Unlike Monty Hall, however, this problem seems to still be unresolved. Debate continues, including an interesting 2007 paper that argues a “hybrid” position… also written by Nick Bostrom! Apparently my sense that Sleeping Beauty and “living in computer simulations” are related was justified.
The paper is titled, “Sleeping Beauty and Self-Location: A Hybrid Model.” Here Bostrom does a good job of outlining the problem and the arguments from both sides, and presents an interesting argument that suggests that, in a sense, both sides are correct. (I disagree, but that’s for another post.)