Fernando, Chrisantha, George Kampis, and Eörs Szathmáry. “Evolvability of natural and artificial systems.” Procedia Computer Science 7 (2011): 73-76.
URL1 URL2
Evolvability in its simplest form is the ability of a population to respond to directional selection. More interestingly it means that some lineages show open-ended evolution by accumulating novel adaptations, and that some lineages complexity can increase indefinitely. Unlimited heredity is a precondition for such rich open-endedness, another one seems to be (analogous to) chemical combinatorics. The richness of matter seems to be a source of challenges and opportunities not yet matched in artificial algorithms. However, some “artificial” systems can be more evolvable than natural ones because for the former the whole population is not under the constraint to survive in the wild. A form of artificial selection may happen even in the brain of replicable patterns that yield complex adaptations within the lifetime of the individual.