Apply that to a simulation
That's been done many a times.
In fact, I even worked on such a project. It's called genetic algoritms.
Here's an example:
BoxCar2D
, what do you get, almost nothing,
Actually, what you get are highly optimized systems.
just random change added and if it survives it stays
Yes. If it survives, it stays. It means it performed
better then those that didn't survive.
Repeat that process, and what you end up with are populations of individuals who are very good at what they do.
Because evolution works.
, thing is evolution can't stack improvements over improvements as they claim it can do.
That makes no sense.
Mutations accumulated. So yes, improvements stack up.
Just like in the example I gave you of the genetic algoritm.
You start out with random polygons wich might or might not have wheels attached, which might or might not break off at the first bump on the track, that spin at different speeds.
But what you eventually end up with, are "cars" that are completely geared towards performing good on the specific track they evolved on. Almost as if they were designed to ride that track. And you know what? They actually were designed to ride that track. The designer however, is not a "who". It's a process. A process of
mutate, survive, reproduce, repeat.