[FRIAM] random v stochastic v indeterminate

[┣glen┫]

Exactly.  And even though we're conflating the model of evolution with the real thing, I find it difficult to believe the "space" operated on by evolution is entirely convex or even connected.  So, (point) mutation alone may *never* reach some regions, regardless of infinite individuals, infinite generations, or infinite space and time.

On 08/12/2017 09:07 AM, Marcus Daniels wrote:
> "Can we truly say that the crossover had nothing to do with the "innovation" ... that it only preserved the innovation and the mutation caused it?  A neutral mutation can't be considered an "innovation", right?"
> 
> A function related by rotation might be a candidate for crossover.
> 
> f(x,y,z,...) -> good
> f(y,z,x,...) -> good
> f(z,x,y,...) -> good
> f(x,z,y,...) -> bad
> 
> Going through the combinations just by using mutation takes forever.  But splicing at different points would help.   One could imagine for motor functions these symmetry or shift detectors could be important.   (Here it is just 1 dimensional.)

-- 
␦glen?