10.15.2007 Discover Magazine
Probably not. And no, he's not looking at your underwear.
There are certain questions about virtual reality (VR) that I’ve been asked a few times a day, every day, for over a quarter century. The e-mails still come in, from a kid in Korea or a grandmother in Australia: Will VR ever get so good that we will no longer be able to tell it’s VR? Is it possible we are already living in VR? Recently even The New York Times got into the act, interpreting an argument from philosopher Nick Bostrom to mean that “it is almost a mathematical certainty that we are living in someone else’s computer simulation.”
When these questions come up, I usually try to redirect the inquirer’s attention to the world of actual VR research, because that topic is richer than most people realize. But readers of this column know I am as friendly as can be to weird speculations, and it is interesting to think about the metaphysical side of virtual reality.
The concept of “VR so good you can’t tell” can mean different things. It might mean that a person who started off in natural reality can be fooled by a simulation, or it might mean that a being who was created as part of a simulation can become conscious.
Let’s consider the second possibility. Can consciousness arise inside a computer simulation? As it happens, I’m that rare creature, a cybernetic daredevil who is usually a dualist: I get thrills from figuring out how to make computer programs reproduce tricks the brain can do, and yet I often believe consciousness is something special that cannot be simulated like other phenomena. Suppose I’m wrong, though, and a simulated character can become conscious. Could it know it is living in a simulation?
Robotics researcher Hans Moravec originated the argument that we are probably already living in VR: If it is possible to build virtual realities sophisticated enough to give rise to sentient residents, it’s likely there would be many such VRs. After all, once we built the first car or the first laptop computer, millions upon millions more followed. (And even if humanity never builds superlative VR machines, some alien civilization somewhere will do it, if it is possible.) If you are a self-aware creature, then, there are two possibilities: You live in natural reality, or you live in one of these super-VRs. Since there is only one of the former and a lot of the latter, the chances are quite strong that you, and indeed all of us, are living in a simulated world.
Although this pseudostatistical style of reasoning doesn’t prove anything, it does say something about the relative likelihood of a particular metaphysical truth. That may seem like a strange way to think, but an even stranger development, to my mind, is that recent results may give us empirical evidence about whether we are living in a simulation. If you believe that thinking about metaphysics in a pseudostatistical way is sensible, then these results make it seem much less likely that we’re living in a VR than it did back when people first started asking me those questions.
When Moravec first made his case back in the 1980s, the popular way of thinking was that there is one and only one natural reality. These days, that answer is becoming less popular all the time, because of a seemingly unrelated field: quantum computation.
Experiments over the past decade show that quantum computers (which process information using the quantum states of particles rather than transistors) really can work. And as it happens, one interpretation of quantum mechanics that used to be somewhat obscure has suddenly become popular because it’s better adapted for explaining quantum computation, at least to human brains.
I’m speaking, of course, of the many worlds interpretation. In this view, each world has a copy of your quantum computer, and they all run at once; that’s why they outperform regular computers that can function in only one reality. When you get an answer out, it’s the same thing as discovering which of the many worlds you are in.
The rising fortunes of the many worlds interpretation seem to have emboldened champions of other ideas about multiple realities. Some string theorists now talk about a “landscape” of realities in which physics is different in each reality. Cosmologist Max Tegmark and the late philosopher David Lewis have offered yet other ways of thinking about many realities instead of one.
You might object that even if there are a large number of realities, there still ought to be many more VRs, because each reality could have many VRs inside it. But that’s not so. Many of the multiple-realities theories suggest an unbounded number of worlds, and if the number of realities can be infinite, then there can’t be more VRs than realities; infinity is as many as there can be. Even if there are only a finite number of realities, there is no guarantee that each reality would host VRs. Virtual realities take up time, energy, and space—and a given reality has limited supplies of those things. All of this suggests that VRs are unlikely to be more common than natural realities.
Of course, this whole discussion begs the question of what we mean when we distinguish a VR from a natural reality. If a simulation is perfect in every way, it is by definition indistinguishable from the thing it simulates. So there must be some difference between a natural reality and a virtual one, or else there’s nothing virtual about it. Maybe the VR is self-evidently low resolution. The ones we can build today certainly are! But that’s not the only possible difference.
The usual sort of difference that people are interested in is the existence of an entity that can look into the lives of players in the VR, a powerful player who is usually but not necessarily hidden. It’s similar to believing in a god. The rhetoric of VR thought experiments often plays up this angle. Some people imagine this creature as a pimply nerd in the sky who is running a cosmic copy of The Sims, who are us. Perhaps with that image in mind, one woman commented to me that she worried that this being might be able to see whether her underwear was clean on a given day.
A pimply video-game-playing kid is an especially unlikely “god” ruling over our reality.
Strangely enough, there are some recent empirical results that may influence whether we should believe in this underwear-obsessed dude. Before I describe them, though, I need to introduce the idea of a spectrum of possible gods, running from wimpy to omnipotent. All the supernatural beings of religion and science fiction fit somewhere on the spectrum.
The very wimpiest god can hypothetically see into our world but can’t do anything at all to interact with us in any way, in any world—including hypothetical afterlives. As far as we’re concerned, it’s meaningless to think of this god as one who exists. The second-wimpiest god might be able to perform just a trick or two that seem supernatural to us, but underwear spying is just one trick of many, and therefore unlikely.
Paradoxically, pseudostatistical reasoning suggests that the most omnipotent god won’t notice your underwear, either. Such a god can see and manipulate all possible versions of you and your world (including your wearing clean, filthy, or no underwear at all). If there is something special enough about your underwear to merit notice, there’s another version of you in another reality wearing even more special underwear. The Top God is infinitely less likely to focus on the particular pair of underwear you are wearing today than you are likely to focus on a particular grain of sand on a beach. He is just as moot as the weakest god.
So as far as underwear spying goes, it’s only those gods in the middle of the spectrum who should concern us. A god who spies would do so, presumably, only if he (or she for that matter) experiences surprise at the unveiling of the future and is able to see into only a narrow range of realities. The Greek gods were like this.
The empirical results that influence how we might think about God-as-video-game-player are the successful demonstrations of quantum cryptography, in which a sender and receiver can be assured that no natural observer has eavesdropped on a message. This system works because a component of the message is ruined by quantum effects as soon as it is read. For a god to eavesdrop on a quantum cryptography session and then cover his tracks, would, as it happens, require near omnipotence. When the first quantum cryptography experiments were done, I felt a little relieved and sad at the same time, because we then knew that one kind of potential exotic or supernatural form of life that might have been watching us either did not exist or wasn’t paying attention.
Continuing the pseudostatistical arguments, a god that can exist only within a narrow portion of the spectrum of possible gods is less likely than a god that can exist over a larger portion of the spectrum. The game-playing kid feels the squeeze from both ends. He has to be both weak enough to be able to focus on a particular pair of underwear and strong enough to be able to cover his tracks after eavesdropping on a quantum cryptography session—or else be willing not to peek at any messages we decide to keep secret.
That probably—no guarantees—places him within a razor-thin niche on a wide spectrum of possible gods. So I can’t swear that we’re not living in a simulation, but I can offer some assurance: A pimply virtual reality operator is an especially unlikely god.