Artificial selection is not Intelligent Design

When evolutionary biologists argue with creationists or advocates of Intelligent Design about the effectiveness of natural selection, the biologists often use the example of artificial selection. And almost invariably, the ID advocate or creationist tries for a “gotcha” by saying “but artificial selection is an example of intelligent design, and is nothing like natural selection!”

We’re going to go into why this is not a “gotcha” but is a mistaken analogy on their part.

Artificial selection is a very old practice, dating back to inadvertent selection when wild animals were domesticated, with the wildest ones being eaten or chased away, or they successfully escaped. With the start of agriculture, farmers often bred from their best animals (or plants). Charles Darwin paid close attention to the ways animals had been changed by breeding, both for food production or as pets. He was particularly attentive to the results of pigeon breeding. “Natural selection” was named in analogy to artificial selection.

How much design is in Design?

Saying that artificial selection is Intelligent Design is wrong because plant and animal breeders, until very recently, had no detailed understanding of the genomes of their organisms or how to make detailed changes in the genomes to affect the phenotype. All they could do was to choose who to breed from, based on their phenotype (such as the amount of beef produced).

In the 20th century this got more sophisticated, but was still not Intelligent Design. Real statisticians got involved, and selection was based on numerical indices computed from multiple characters. I remember visiting a poultry breeding operation and hearing how they measured 10 characters, including not just the production characters but such important traits as susceptibility to viral disease. This was in the 1970s, and they had a modest mainframe computer in their offices, with databases of traits. It was the natural outcome of work in the 1930s and 1940s by quantitative-genetics-inspired animal breeders at such centers as Iowa State University.

But they did not know the detailed genetics of these traits. The selection still depended on observing the traits and not directly knowing any of the genetics, or much about the developmental biology of the traits.

On to Intelligent Design

By the 2010s animal and plant breeding was entering a new era, though with difficulty. In 2014 I was invited to give a talk (on natural selection) in Vancouver BC, at the 10th World Congress of Genetics Applied to Livestock Production. I was talking about the mathematics of natural selection. It was apparent from the muted audience response that this was now of lesser interest to people working on livestock production.

Instead the hot topic of the Congress was QTLs – “quantitive trait loci” – how could they find individual genes whose alleles affected the production traits and disease resistance traits. If these could be found, they could be used by screening potential parents of the next generation, and maybe even by genetically engineering their genotypes.

The fact that lots of effort is now going on towards entering this new era of genomics should be enough to make it clear that even thousands of years of animal and plant breeding, with lots of human intelligence applied, is not the same as Intelligent Design. So the success of that history of breeding does say a lot about it as exemplifying the power of selection, without genetic engineering or even understanding the effects of changes in genotypes.

A comparable example

It occurs to me that another similar flawed critique is when creationists and ID advocates argue that the success of Genetic Algorithms in solving engineering problems is irrelevant, because that is an example of Intelligent Design. It isn’t, if the algorithm has random mutations, random recombinations, and random mating, with fitnesses based on the final phenotypes rather than on detailed knowledge of how they were achieved.

In ID books such as Robert Marks, William Dembski, and Winston Ewert’s “Introduction to Evolutionary Informatics” there is a great effort to show that in simulations of natural selection, they only succeed because the design information is built in. For example, in Richard Dawkins’s famous teaching example of the “Weasel”, the target phrase is part of the program.

Yet in other cases, such as Dave Thomas’s genetic algorithm for a Traveling Salesman problem (see his 2006 post here at Panda’s Thumb ) no particular solution is built in. All Marks, Dembski and Ewert could do was point to places where some fine-tuning of the algorithm had been made to make it more successful, but this is not the same as building in the answer to the particular problem.

So neither the success of animal and plant breeding, nor the success of genetic algorithms, can be sensibly attributed to Intelligent Design of the answers.