Is this world a computer simulation?

Eliezer S. Yudkowsky (
Mon, 06 Sep 1999 12:40:56 -0500

david gobel wrote:
> > Eliezer S. Yudkowsky wrote:.. there's.. at least a 50% chance of this being...
> > a 100% probability. My intuitions, though, tell me that this Universe
> > is real down to the quark level, which I don't think would be necessary
> > if this were a simulation. Then again, in this sort of situation it's
> > kind of hard to trust one's intuitions - but I still do.
> Reality sure is funny stuff...

Not really. Quantum physics and relativity, or whatever turns out to be the ultimate logic of reality, are normal. Humans are "funny stuff".

> I answered this question to my own satisfaction in 1994 when our first
> Worlds program actually worked...especially when we included the softbot
> with extensive knowledge of bartending. The answer is: a virtual world is an
> actual world under all conditions where the simulated cognitives are unaware
> that their world is a simulation.

I know that. Everyone with half a brain knows that. If anyone wants to start another damn argument about whether a perfect simulation is actually real, I say we shoot them. I consider the issue settled. It's distinct from the question of whether a perfect simulation is possible (under whatever random circumstances), but if it's perfect, it's real.

> Once interior consciousness is available,
> it is real. I have been aware that we are a simulation since 1994.

Yes, but the term "simulation" implies a simulator - one who has purposes and might intervene in the simulation. *Especially* once we get close to the Singularity and start trying to break out, or using extreme amounts of computational power.

The question isn't so much "What is our ontology made of?" but "Is anyone going to interfere with us?" This is qualitatively distinct from questions of external interference by aliens, because it implies that our past and our future may be created and specified by unknown goals.

There's also the possibility that this Universe uses the lowest possible resolution consistent with internal believability; the only time quarks are actually simulated is inside particle accelerators. This could have weird effects as we build computers closer and closer to the molecular level. Maybe all those so-called "programming bugs" are because the simulation doesn't want to give you the hundred-million-ops-per-second necessary to actually *run* your program the way you wrote it.

The one thing I do know is that this simulation allows the natural existence of Specialists, meaning, with any luck, that neurohacking isn't outlawed. And I know from observation that you can actually get better analysis of these sort of issues with that kind of neurohacking. So the basic paradigm of "Confusist Singularitarianism" - "Bring about the existence of a greater-than-human intelligence and ask it what the heck is going on" - should still be workable no matter what happens. We will *not* remain in an existential fog forever, no matter *what* barriers are imposed by the simulation.

What do we do if we *are* in a simulation? Well, "How do you outwit a Power?" is of course a standard question in Applied Theology. (The fact that I'm one of maybe a dozen people far enough in the future to know the standard questions doesn't change that.) The answer, of course, is "The same way you outwit the laws of physics"; you treat the behaviors of the Power as physical laws and then try to find loopholes the same way you would with any other physical law.

           Eliezer S. Yudkowsky
Running on BeOS           Typing in Dvorak          Programming with Patterns
Voting for Libertarians   Heading for Singularity   There Is A Better Way