The Status of Requisite Variety, Revisited

Purpose — Requisite variety is regarded either as the fundamental law of Cybernetics or with suspicion. Dewhurst (1991) judged it "better regarded as one possible strategy rather than as a law." This paper attempts to settle its status by asking not whether it is a law but which kind of law-like statement it is.

Approach — The approach is conceptual, drawing on the philosophy of scientific laws, in which no agreed test awards the title "law"; on Einstein's distinction between constructive and principle theories; on Ben-Menahem's account of constraints; and on Beer's Cybernetic account of a law as the invariant a homomorphism selects.

Findings — Requisite variety is not an empirical law, and not an optional strategy. It is a principle-theory constraint in Einstein's sense: a prohibition on what any regulator can do, of the same kind as the second law of thermodynamics. Provable from its own definitions, the inequality states nothing measurement can confirm or refute; its content is the partition it presupposes: which variables count as disturbance, regulator, and outcome. This dissolves Dewhurst's objections: "tautology" names the provability such a constraint must have, and the subjectivity of the elements describes the partition, not a defect. It relocates criticism from the inequality, which cannot fail, to the partition.

Originality — The contribution is the reclassification of Ashby's law of requisite variety. Native to Cybernetics, it rests on Beer's account of lawhood and Ashby's treatment of variety as observer-relative.