This is not quite what Lamport is saying. Lamport is not talking about the process of writing programs, but about reasoning about programs. You build an analog circuit by composing components (that would be programming), but you reason about what it does and how it does it using ordinary differential equations. Lamport notes that state machines are a universal model for discrete dynamical system, and they can be manipulated mathematically using notation analogous to ODEs for continuous dynamical systems. Lamport does not suggest that you should program using a state machine abstraction, nor does he offer any notation for doing so.