If a cellular automaton is allowed to run forever from a start state that can generate unlimited novelty, e.g. double breeder or decaying agar, will intelligent structures eventually evolve? More generally, will evolution occur?

The general idea is that our corner of the universe, say, our solar system at quantum resolution, is a finite computable system, and certain cellular automata are universal computers, so given enough time and unlimited novelty, a pattern simulating our corner of the universe, accurate enough to simulate our intelligent life, will form by chance. Yes, this will probably take longer than the lifetime of the universe and a computer larger than the universe. Perhaps less if evolution, survival and propagation of the fittest, occurs.

The key is probably whether stable structures can form that are mostly defended or isolated from interference from other nearby (or distant) structures. Intuitively, consider trying this experiment on 4-dimensional or higher grids. There's more space up there for things to miss each other. In two dimensions, even crossing glider streams is difficult.

If intelligent life were to evolve in this simulation, would they ever be able to deduce that the underlying fabric of their existence is a simulated cellular automaton? Probably not.

## No comments :

Post a Comment