Thursday, September 17, 2020

[qusfmonq] The most difficult game of go

Two superintelligent beings take turns altering the initial state of a cellular automaton.  Perhaps it is Conway's Game of Life, though better would be a cellular automaton with three states per cell, so it looks superficially like go 囲碁.  (Previously, fictional chess for the superintelligent.)

The objective is to create an initial cellular automaton state such that, when it is run, it simulates an entire universe, perhaps a universe not unlike our own, in which intelligent life, perhaps life not unlike us, comes to exist.  (Perhaps those simulated intelligent beings experience an illusion of free will, though their lives are actually fully controlled by the rules of the cellular automaton.)  Two of those simulated intelligent beings play a game of go within their simulated universe.  The outcome of that go game determines the outcome of the outer game being played by superintelligent beings.  We assume that the cellular automaton is Turing-complete, so playing the outer game requires superintelligence of a level at least able to solve the Halting Problem.  After the go game has concluded, perhaps the superintelligent players pack up their game, ending the universe it was simulating.  (Little did the residents of the simulated universe know that their universe would end with this go game.)  Perhaps the superintelligent beings also ponder whether their own universe is itself a game simulated by beings higher up, and if so, how high the simulation hierarchy goes, and what the objective of the ultimate game is.

Easiest is for the initial initial state to be (pseudo) random and very large, perhaps infinite, so that a universe simulator is guaranteed to exist inside it, and some go game is eventually guaranteed to happen inside the simulated universe when it runs.  The superintelligent players try to modify the initial state, a vast field of infinitude, each to their favor.

No comments :