Causal networks

The idea of using networks to represent interdependencies of events seems to have developed with the systematization of manufacturing in the early 1900s—notably in the work of Frank and Lillian Gilbreth—and has been popular since at least the 1940s. Early applications included switching circuits, logistics planning, decision analysis and general flowcharting. In the last few decades causal networks have been widely used in system specification methods such as Petri nets, as well as in schemes for medical and other diagnosis. Since at least the 1960s, causal networks have also been discussed as representations of connections between events in spacetime, particularly in quantum mechanics (see page 1027).

Causal networks like mine that are ultimately associated with some evolution or flow of activity always have certain properties. In particular, they can never contain loops, and thus correspond to directed acyclic graphs. And from this it follows for example that even the most circuitous path between two nodes must be of finite length.

Causal networks can also be viewed as Hasse diagrams of partially ordered sets, as discussed on page 1040.