The Good Regulator Theorem comes from a 1970 paper by Roger Conant and W. Ross Ashby titled “Every Good Regulator of a System Must Be a Model of That System”.
The claim
A regulator is “good” if it achieves its goal (say, keeping an outcome stable) in the simplest possible way. Conant and Ashby showed that the simplest optimal regulator must map the states of the system it regulates onto its own actions, in effect containing a model of the system. Regulation and modelling are two sides of the same thing.
Implications
The theorem gave a formal basis to an intuition common in cybernetics: to steer something, you must in some sense represent it. It was used to argue that brains must model the world, that managements must model their organisations, and that a government must model the society it governs. Stafford Beer drew on it in the Viable System Model.
The theorem has been criticised for being narrower than its title suggests, and has been refined since. The internal model principle in control theory makes a related claim, and Karl Friston’s free energy principle can be read as a modern descendant: a system that survives behaves as if it holds a model of its environment.
Further reading
- Roger C. Conant and W. Ross Ashby, “Every Good Regulator of a System Must Be a Model of That System”, International Journal of Systems Science (1970)
Mentioned in
- Article Cybernetics as a Discipline
- Person W. Ross Ashby
- Concept Free energy principle