This discussion on wealth is an offshoot of Serious Conversations parts 53 and 54. We are considering the book The Origin of Wealth by Eric D. Beinhocker. (I do not profit from clicks). (Ed.: we will be taking the general format of outlining the major points of the chapter and then discussing what we believe to be important or intriguing points.)

This chapter is on networks and how networks can explain behavior in the economy (and a multitude of other human endeavors). A network is made of edges, which connect different nodes. For example in a company each division is a network, and each employee is a node. The edges are meetings, emails, phone calls between the employees.

The connectedness of networks can explain why a certain product will explode in use suddenly. Its connectedness reaches a tipping point where the ratio of edges to nodes reaches 1 (there are more meetings than people in the company analogy above). After this phase transition, a sparsely connected network becomes densely connected. The example provided was how the internet suddenly because incredibly useful in the 1990s even though it had been around for decades (i.e. the internet became useful when there were more ways to connect people (webpages or content) than there were people on the internet.

By visualizing connections between US cities, we see that random networks have far fewer degrees of separation. For example, in a regular, lattice grid, if a city was connected to its 4 closest neighbors, it might take as many as 30 connections to go from the east coast to the west, but if instead each city were randomly connected to four others, some connections would be long but some would be short. The number of small connections would be the same as the number of medium or long connections. Thus, you could connect the east and west coasts using one long connection and several short ones.

Boolean networks are discussed next. Boolean networks have nodes that can either be in one of two states, 0 or 1. Boolean networks are guided by three variables: number of nodes, connectedness, bias of the rules.

The number of states a network can be in scales as 2^N (N=number of nodes). This means that the potential for novelty increases exponentially. For example, say 10 people work at a coffee shop, that is only 4 orders of magnitude smaller than the number of people who work at Boeing, but the complexity of making a coffee to making a jet is many more than four orders of magnitude difference.

This scaling is offset by the scaling of connectedness. If on average a network has more than one connection per node, then as the number of nodes grows, connections will scale exponentially. The number of interdependencies grows faster than the network itself. As the interdependencies grow, changes on one side of the network can ripple to the other, and the more interdependencies there are, the better the chance that a positive change in one side of the network will lead to a negative change on another part grows too. This can lead to a “complexity catastrophe.” This is why large companies can’t innovate all that well. “Densely connected networks becomes less adaptable as the grow.” Hierarchical networks can ameliorate the complexity catastrophe by reducing the complexity of the connections and thus the number of connections. It compartmentalizes networks within networks.

At some point networks will go from spontaneous order to chaos as the average number of connections increases. Bias is the parameter that describes the point at which a network will transition from order to chaos. The higher the bias, the more predictable the output from a node will be. For example if a boolean node had output of 90% 1 given a random input of 0 and 1, the node is biased towards 1. In general, the higher the bias, the more connected a network can be before it transitions to chaos.

The complexity catastrophe can explain bureaucracy. No one deliberately designs it. Instead, it happens because the individual divisions simply want to optimize their section of the network. I’m fascinated that bureaucracy is another emergent property of networks. It seems that a lot of negatives in society are simply the result of how out society is setup, and not only that it seems this might be the only way an efficient society could be set up.