The Creation Wiki is made available by the NW Creation Network
Watch monthly live webcast - Like us on Facebook - Subscribe on YouTube

Genetic algorithms require a designer to specify desired outcome (Talk.Origins)

From CreationWiki, the encyclopedia of creation science
Jump to: navigation, search
Response Article
This article (Genetic algorithms require a designer to specify desired outcome (Talk.Origins)) is a response to a rebuttal of a creationist claim published by Talk.Origins Archive under the title Index to Creationist Claims.

Claim CI113:

Genetic algorithms and computer evolution simulations do not show that intelligent design is unnecessary. On the contrary, those programs must be designed themselves, and they require a designer to specify the outcome.

Source: No source given.

CreationWiki response: (Talk.Origins quotes in blue)

1. See the response to the claim that evolutionary algorithms smuggle in design in the fitness function. Modeling evolution requires modeling fitness differences. Fitness functions often are not designed but are taken from the real world.

Even when a fitness function is based on the real world, it is still just a mathematical model and it is therefore designed. Try telling a computer programmer that has spent many hours or even days writing a program that simulates some real world phenomenon that he did not design the simulation. No mater how much a fitness function is based on the real world, it still required some looking at the real world and designing a simulation based on it.

The exact outcome is not specified in genetic algorithms, only the general requirements. And evolutionary simulations have no intended outcome for the simulated organisms at all.

The general requirements are the outcome mentioned in the claim. Talk Origins has misworded the claim substituting outcome for problem. All genetic algorithms focus on a specific problem and fitness is always evaluated based on that problem, the intended outcome a solution to the problem.

They only work because fitness is defined in relation to a specific problem, in the real world fitness is a much vaguer concept. In fact Evolutionists define fitness based on an organism's ability to reproduce. Such a broad concept of fitness is far removed from the specific problems used in genetic algorithms.

2. The fact that the programs are designed themselves is not relevant so long as they accurately model real phenomena. They will not model everything, of course; that is why they are called models. But the fact that modeling evolution is done by designers has no more implications for intelligent design than the fact that painting seascapes is done by designers, too.

Talk Origins is missing the point. The point is that, because these evolution simulations are themselves designed, the claim that they show intelligent design to be unnecessary is false. It is absurd to look at a program that it took one or more intelligent programmers many hours to write and say that it proves that intelligent design of life is unnecessary.