We continue our trek through Christoph Adami’s book, The Evolution of Biological Information. We now come to his fifth chapter, “Evolution of Complexity.” In this chapter, he talks about the “Artificial Cell Model” which is able to demonstrate the evolution of a complex metabolic system. His cells operate in a simple model of chemistry, and are able to import precursor molecules provided outside of a cell into more complex molecules by the use of enzymes encoded in their genomes.
The evolution of a complex metabolic system seems impressive, but what is going on here? Ordinarily, the metabolites within a cell serve some particular purpose. Their particular configuration is not simply random or arbitrary but has to be some particular configuration. But that is not how Adami’s model works. In Adami’s model all metabolites are considered useful and longer metabolites are considered more useful than shorter ones. There is no real model of purpose or utility to the metabolites.
The Viewpoint of Design
From an intelligent design perspective, this means they lack information. In order to exhibit information, the enzymes and metabolites would have to display specified complexity. But when just about anything is accepted, there is no specificity. Such models reward undifferentiated complexity, and it is easy to evolve that precisely because there is no specificity.
The rewarding of complexity has a long history in the context of computer models of evolution. Models tend, by default, to reward simplicity. It is actually relatively easy to evolve simplicity by removing and deleting from the genome.
This is why, for example, the Tierra simulation shows a lot of evolution by shrinking and simplifying the genome. But the author of Tierra, Thomas Ray, did not like this. He was trying to evolve complexity, so he made a change, rewarding programs for being longer. Ray still did not get the complexity he wanted, but he did manage to stop the genomes from shrinking.
The World of Avida
Avida does the same thing. In the world of Avida shorter genomes would be better as they would take less time to copy. But again, the authors of Avida did not want that. So they reward programs just for being longer to make up for this.
However, Avida takes this a step further. It rewards performing digital logic calculations. These calculations serve no purpose. There is no restriction about which calculations would be useful. Rather, Avida rewards valid calculations for the sake of calculations.
I wrote about another example a decade ago here, pykaryotes. It likewise modeled a metabolite system, but it simply randomly decided that certain metabolites would increase fitness. That system, yet again, rewarded complexity for complexity’s sake.
Why do all these models do this? Why are they rewarding undifferentiated complexity? It is because if you don’t do that, the evolutionary models do not generate complex systems. If anything they end up trending towards simplicity. If the model rewards complexity for the sake of complexity, this works. But then the complexity is not interesting. Crucially, it’s not really information.