At his blog, Construction Physics, Brian Potter, drawing on work by economist Brian Arthur, draws a connection between modularity in technology and evolution, concluding:
The technological search process benefits greatly from modularity: being able to break a technology down into subcomponents with specific functionality, and determining whether those subcomponents are functioning properly… It’s interesting to see that biological evolution, which creates and operates in an entirely separate domain, uses a similar sort of trick: using genetic recombination (in the form of sexual reproduction and horizontal gene transfer) to make the search process more modular and gain information more rapidly.
Potter is drawing an analogy between modularity in technology and the genetic recombination in sexual reproduction.
But Does the Analogy Work?
He is considering a scenario where different mutations in different genes are tested independently before being put together by genetic recombination. This works because each gene is largely independent of the others. If a mutation improves a particular gene, it will likely still constitute an improvement when combined with beneficial mutations in other genes. That’s the benefit of modularity.
The fundamental problem with this analysis is that sexual reproduction does not make anything modular. Rather, its effectiveness depends on its already being modular. The key benefit of modularity is that, because a system is divided into multiple largely independent parts, one can make improvements in one part without disrupting the others. That makes it far easier to improve a system.
Genetic recombination and horizontal gene transfer do not cause a system to be modular. Rather, they take advantage of it. Because biological systems have a high degree of modularity, different combinations of genes will usually work. Likewise, transferring a genetic component from one organism to another via horizontal gene transfer works. That is all because the design of life is highly modular.
Life could, instead, have been designed without being modular. Every change in one part of the genome could have a complex, diffuse effect on everything else. In that case, the scenario imagined by Potter, in which mutations are tested independently from each other and then combined, would not work. Rather, there would be no reason to expect that, just because a mutation was beneficial by itself, it would still be beneficial in combination with other mutations.
Not Something that Just Happens
As any engineer will tell you, modularity is not something that just happens. Modularity is the result of design. In fact, engineers must work hard to maintain modularity. In the short term, it is often easier to make changes that disregard modularity. It is only on a longer term analysis that modularity is beneficial. Since natural selection only has a short-term view, we would not expect it to maintain modularity.
Furthermore, the challenge to evolutionary theory is not a case of making small tweaks to largely independent genes. Rather, evolution’s difficulty is explaining the evolution of complex systems that require the interaction of multiple components. This is precisely the case where trying to evaluate mutations independently will not work. The difficult problem of finding working combinations remains.
Recombination is able to exploit the modularity of biology. It does not explain that modularity but rather requires it. That modularity is not something that just happens but rather a product of design. Even with modularity, the difficult challenges to evolutionary theory remain.