Snippable genes — the optimum size of sequences for evolution

John Holland has shown how an information system which guides a real world system may evolve and “learn” by gradually building effective models of functioning, in the form of “genes”.
In Holland’s analysis, he describes the genes which we know in organisms as a special case of a much more general phenomenon which allows learning to take place through the exchange and evolution of individual processes that contain encoded solutions to small, limited problems, and are then replicated and spread.

Perhaps the most important element of Holland’s analysis is his discovery of mathematical reasons why the learning, and spreading, and successful evolution of the genes, will occur most successfully to the extent that the genes are small and independent.

Composable units!

This particular argument is merely one example of a more general argument which shows that small independent “lumps” of coherent problem-solving information, the smaller they are, and the more independent, the more likely they are to survive and spread into the gene pool. Sequences, too, are most likely to spread when they are small.

All this is mirrored in biological genetic evolution. Biological evolution also works because the evolution of processes is taking place in very small increments. The evolving system is not asked to take unfeasibly large jumps, but rather, very small, and advantageous steps, which leave everything working, while making one improvement at a time.

A gene is small, interchangeable, and can be transplanted effectively from one system to another — in many cases with success.
This is the secret of biological evolution. I believe it will also turn out to be the secret of the evolution of the genes controlling the living structure of the Earth and of the built worlds on Earth.

#book/The Nature of Order/2 The process of creating life/20 The spread of living processes throughout society#

Notes mentioning this note


Here are all the notes in this garden, along with their links, visualized as a graph.