Computer-generated randomness can be as truly random as needed. Such randomness is useful in genetic algorithms for nudging a system out of a rut in state space and forcing it to explore other regions.
Given enough time, a system using genetic algorithms can derive unexpected results that often prove to be novel solutions to otherwise intractable problems.
While the GA approach to computer programming is tremendously promising for engineering, it is probably quite different from what is needed for self-evolving systems.
Despite being rapid ways to explore state space, genetic algorithms are not holorithmic, and holorithms are probably essential to self-evolution since they implicate the potentially self-evolving system into the broader patterns imbued within the seeming randomness of natural context.
That "rich, natural" randomness is part of what shapes the self-evolving system and allows it to transcend the dimensions its own mechanistic determinism.
Michael Webb, 2001
[ home ]