computation

network

a child's definition of a net: "a lot of holes tied together with string"

Stuart Kaufman has described the formation of networks as a phase transition that occurs as the number of connections is increased between a random graph of points. As a general feature, when the number of connections reaches half of the number of points, the majority of the points become linked in a giant cluster. Kauffman believes that we should think of the genetic program not as a serial algorithm but as a parallel distributed regulatory network .

As a general model, the network applies not only to the "intertwingled" pieces of text in a hypertext but also to the linked computers in a connected system such as Internet, to pattern-recognition systems such as the immune system, or to organisms such as the slime mold, that are made up of individual cells responding to gradients and forming larger and more differentiated entities. The ability of the brain to synchronize and coordinate activities in different parts, called reentry, is another networked process. Theorists of complexity describe the behaviour of such systems as emergent.

Problems are assumed to become intractable when they become tangled, yet models of rhizomes and networks that value links are a kind of countermodel.

Read More

replication

Charles Darwin showed how "organs of extreme perfection and complication, which justly excite our admiration" arise not from God's foresight by from the evolution of replicators over immense spans of time. Freeman Dyson reminds us that life consists of both metabolism and replication, and that the two may have started separately. (In its pure state, replication can only be parasitic.) He subscribes to Lynn Margulis' theory that RNA is the oldest and most incurable of our parasitic diseases. (see prokaryote / eukaryote) Genetics is concerned with the replication and variation of of genes in a population (and their impact on adaptation. (See genotype / phenotype )

John von Neumann proved that a machine could be designed that could replicate itself. The logical problem is how to avoid infinite regress that would require the instructions: "how to build machine (how to build machine (how to build machine (etc.)))" If the instructions merely stated "how to build the machine" it would work once and then stop. The new machine would be unable to replicate itself.

The self-replicating machine requires a certain threshold of complexity with a controller that is able to use the same instructions for its own operation as well as for replication. Thinking of the instructions as a "blueprint" the machine is both able to carry out the instructions and copy them as separate operations.

Read More