18 hours ago
- Brian Arthur's simulation of technological evolution shows that modularity—combining simple components into increasingly complex modules—greatly simplifies the search for useful technologies.
- Biological evolution uses random mutation and natural selection, and like Arthur's simulation, it benefits from modularity at the genetic level to increase the rate of information acquisition.
- Sexual reproduction allows genetic variation without reducing average fitness, enabling faster spread of beneficial mutations compared to asexual reproduction, which suffers from clonal interference.
- In simulations, sexual reproduction reaches maximum fitness much faster than asexual reproduction (e.g., 33 generations vs. 200 generations for a 200-gene genome).
- The rate of fitness increase in sexual populations is proportional to the square root of genome length, while in asexual populations it slows to a fraction of a gene per generation.
- Sexual reproduction effectively tests each gene independently, turning the search for a fit genome into parallel searches for each gene, analogous to modularity in technology.
- Information acquisition in evolution can be measured in bits; sexual reproduction acquires information more rapidly than asexual reproduction.
- The analysis assumes genes contribute independently to fitness; real gene interactions (epistasis) complicate the search process, but the simple model reveals key dynamics.
- Asexual organisms like bacteria use horizontal gene transfer to share beneficial mutations, mimicking some advantages of sexual reproduction.
- Modularity in both technological and biological evolution narrows the search space and accelerates the accumulation of useful information.