Warren,
You appear to be erecting a strawman. One which, btw, has been addressed in numerous threads already.
Mutations are random with regards to fitness. It is only by considering the phenotypical effects (if any) on the organism in the context of its environment that a given mutation can be,
post facto, determined to be "adaptive". In other words, you're wrong again.
Besides, that's not what is being discussed on this thread - these are software demonstrations of complexity from simplicity, not mutation and selection.