◌  Simulism

6 March 2009

Here’s some stuff I think is underappreciated about simulism. It will probably only interest nerds; everybody else is excused to go bowling and drink wine coolers, or whatever it is you people do.

0. As a preparatory meditation on humility, let’s consider just how thin a conjecture simulism is. It’s very close to unprovable and unfalsifiable, and its predictive power so far is nil. Similar propositions, like the five-minute universe (and arguably solipsism), have had little traction. The only thing that makes it more plausible than Plato’s cave or Zhuāng Zǐ’s butterfly as a real theory instead of a metaphor is the recent proliferation of machines that look like maybe someday they could run big simulations. This pre-statistical intuition or induction is its only support. And if you like induction, look at the dozens of remarkably similar religious claims that there’s a universe outside the universe and consider how variously invalid, unsound, inconsequential, and plainly wish-fulfilling they are. So let’s not get carried away. (So far only one of Facebook’s 1.75e8 users lists simulist as their religion, so I guess we aren’t.)

1. With a reasonably broad definition of simulation, most simulations we can see are not on computers: they’re in consciousnesses. What we find it natural to call a computer sophisticated enough to run our universe would likely be, on our terms, alive, or of undefined animacy. (Functional equivalence and all, but let’s avoid the theological saw that x ∈ Y implies x < Y for a goddy kind of <. What a simulation implies about the simulator in general seems exceptionally hairy to me, which is what the next point is about.) So I think it’s not much better to call the universe a computer simulation than to call it, say, a speculation or a memory. If we say it’s a dream, we’re even using a word that’s often understood as a metaphor.

2. Arguing about upper bounds on the complexity of simulators in the parent universe is a fuzzy business indeed. Why assume, for example, that P ≠ NP there? Or that they have time? If these are unfair questions, simulism is unfair. More conservatively: they could have any number of tricks for doing hard stuff like gravity. We can imagine in our universe a computer game that involves soap film stretching around big tangles of wire (which, idealized, is hard); instead of crunching it in-game, you could pause it, dip some real wire in real soap, and scan that as good enough. (And if someone notices, you rewrite their brain, and pretty soon simulism looks like solipsism. Sigh.) With enough of this kind of stuff, it’s not a simulation so much as a set of simulations interpolating between observations.

   Moreover, reflect that in good OS emulation these days, the guest OS is actually touching the hardware much of the time, and that while the simulations called golf and tabletop RPGing (say) are on simulated land, the simulations called windsurfing and LARPing (say) use land in their parent universe – we could be partly directly in the host universe, so gravity etc. is free. And so on: as far as I’ve seen, we just don’t have the leverage to say anything about the nature of the simulator except that if it’s in a universe like ours it probably has a sizable cooling system.

3. Very little thinking has been published about simulism’s ethical implications. Consideration is in order, and may be very practical in a few decades if we want to start simulating conscious things without worrying that we might be evil gods.

4. If Dark City had been as popular as The Matrix, simulism would be much easier to explain to a lay audience. The Matrix made the simulation sexily digital but fudged the reasons behind it; it was mostly Baudrillard-y social comment and something to transcend. Dark City was much more about the simulation and gave it a relatable (if sfnal) motive.