Michelle Lee
SSP205, Spring 2005
Commentary on Bostrom's "Are You Living in a Computer Simulation?" (November 2001)
http://www.simulation-argument.com/classic.html

Bostrom's "Are You Living in a Computer Simulation?" is an exhaustive, thorough prose proof of several propositions involving the human species and posthuman stimulation. However, it is my belief that Bostrom's argument lacks a crucial link early on human- equivalent computational power with successful (i.e. plausible) simulation of the human—or other—worlds.

Certainly his preliminaries are uncontroversial. Not only is substrate independence a common assumption in the philosophy of mind, as Bostrom indicates, but it is also a common and provable assumption in the field of computer science. The Java Virtual Machine, for example, was specifically created to be a substrate(platform)- independent metaplatform to enable a new age of web-based applications. Another solid concept is the proposition that simulation-capable hardware and software will eventually be available given "unabated" technological progress.

Superhuman computational power, even such capable of simulating a human mind, doesn't necessarily equal the ability to run a sufficiently complex simulated world within that mind. At present, human simulations seem to take place in dreams (single agent) or social simulations (multiagent). Neither of these are strong enough to be indistinguishable from reality.

Single-agent human simulation

Lucid dreaming is a dream state in which we are able to freely recall the circumstances of waking life, to think clearly, and to act deliberately upon reflection—all while experiencing a dream world. While rare (only 20% of the population, by self-report, experiences lucid dreaming more than once monthly), is a neurologically verified phenomenon and has been demonstrated to be a learnable skill[1]. However, there remains a hard stop in the experiencing of a dream world: apparently it is impossible to dream one's own death. Therefore we do not fully experience waking life in a human dream, and can show that our dream is not our reality.

Multiagent Human Simulation

Social simulations are the creation of multiple individuals. At least one party creates the simulation through scripting of an environment and its people, and another party enters such a simulation. Such simulations of a total society have been described to plausible levels of detail in literature and in film—from the Starfleet's Enterprise to Margaret Atwood's The Handmaid's Tale. However, there exists no record outside of fiction in which all participants and all observers. Then we have not yet disproved the possibility that the human species is participant and a posthuman machine is creator and observer in a simulation. What we have shown, however, is that there has to date been no historical record of a human-created total simulation.

To successfully assert that "posthuman civilization" can run a plausible simulation, and then that we have gullibly lived in such a simulation until now, we must establish that the space between:

  1. achieving enough computational power to create a simulation, and
  2. creating a deeply plausible simulation
is continuous and therefore trivial.

In my opinion, Bostrom doesn't conclusively establish this. Can we?

References

[1] LaBerge, Stephen. "Lucid Dreaming: Psychophysiological Studies of Consciousness during REM Sleep". In Bootzen, R. R., Kihlstrom, J.F. & Schacter, D.L., (Eds.) Sleep and Cognition. Washington, D.C.: American Psychological Association, 1990 (pp. 109-126). Also available online at http://www.lucidity.com/SleepAndCognition.html.