On simulations of evolution

There has been extensive discussion on Panda's Thumb and Uncommon Descent of Dave Thomas's excellent program which simulates certain aspects of evolution. See:

http://www.pandasthumb.org/archives/2006/08/take_the_design.html

and

http://www.uncommondescent.com/index.php/archives/1316

Salvador objects that Dave has designed the selection process in his program to solve the problem - and therefor it is not a genuine demonstration of the ability of blind mutation to solve a problem.

I think Salvador is quite wrong. But it is partly down to the way Dave and others describe his programme. It is misleading to describe simulations of evolutionary processes in terms of searching for one or more solutions to a problem. Evolution is not trying to solve a problem, achieve a target or produce solutions. That vocabulary implies:

* You know when you have solved the problem and then you stop.
* The fitness function is a device for getting to an end
* The problem is something separate from the fitness function

If you use that vocabulary then the likes of Salvador can always say that you designed the fitness function to solve the problem. If you recognise that there is no problem other than doing as well as possible on the fitness function then the objection goes away.

In evolution there is a fitness function - reproduction. Evolution produces some organisms that do better than others with respect to that function. There is no separate end goal and it doesn't stop.

Dave's programme can be described without using the words "solution" and "problem". It just generates patterns that compete by finding shorter paths. What is interesting is the process of variation and selection leads to patterns that are innovative and give the illusion of design.