Although born a world apart, Phillips' and my careers had striking parallels, warm interactions, and similar motivations. We both did research on electrical engineering, automatic control, computer simulation, economic stability, econometrics, and inflation-unemployment.

Control theory

We started corresponding in the summer of 1956 about bringing the tools of operations research and electrical engineering to bear on improving economic stability. Bill had already published several articles applying tools from electrical engineering to the stability of single loop economic models.

It was becoming clear, however, that economic stabilisation policy called for the analysis of economic systems incorporating multiple loops and multiple control variables. Fortunately operations research on industrial operations done at Carnegie Mellon University by a team composed of (in alphabetic order): Holt, Franco Modigliani, John F. Muth, and Herbert A. Simon, had shown that the optimal solution to that problem could be obtained under dynamic uncertainty (Holt et al. 1960). Where the criterion in a decision analysis could be adequately approximated with a quadratic function, and the dynamic system could be approximated by linear difference equations, the inversion of an infinite matrix would yield optimal linear decision rules. That model was discovered independently three times: by our operations research group, by economist Henri Theil, and by a group of process control engineers. Later it was adopted as the ‘classic’ model for engineering control.