Introduction. Ashby defined variety as the number of distinguishable states of a set, and regulation as the blocking of disturbance from reaching a system’s essential variables. The law of requisite variety says that a regulator can reduce the variety of outcomes only to the extent that its own variety matches the variety of the disturbances: in Ashby’s words, “only variety in R can force down the variety due to D; variety can destroy variety” (An Introduction to Cybernetics, 1956, §11/7, p. 207). Conant and Ashby’s good regulator theorem adds that any regulator that is maximally successful and maximally simple must operate as a model of the system it regulates; its states must map onto the states of the regulated system. Both results are consequences of their definitions. Given a set of disturbances, a set of responses, and a table of outcomes, the law is a counting fact about the table, and the theorem is a fact about mappings. Neither can be refuted by observation. What can be right or wrong is the assignment of the terms to an organisation: what counts as a disturbance, what as a response, what as an acceptable outcome, and whether these can be fixed before the result is known. Beer’s variety engineering (attenuate disturbance, amplify response) and his rendering “only variety can absorb variety” (Brain of the Firm, 2nd ed., 1981, p. 308) are paraphrases of Ashby, not the original.
Important authors. W. Ross Ashby (1903–1972), a British psychiatrist, wrote Design for a Brain (1952) and An Introduction to Cybernetics (1956) while director of research at Barnwood House Hospital, Gloucester, and later worked at the Biological Computer Laboratory of the University of Illinois under Heinz von Foerster. Roger Conant, at the University of Illinois at Chicago Circle, extended Ashby’s programme with information-theoretic measures of regulation and system decomposition. Ashby is the cybernetician Beer cites most; the law is the load-bearing theorem of the VSM.
Importance for cybernetics and the VSM. Requisite variety is the axiom from which Beer derives everything: recursion (variety must be absorbed at every level), the channel principles (channel capacity must exceed the variety carried), and the Three–Four homeostat (an internal regulator matched to an environmental model, which is the good-regulator theorem in organisational dress). The VSM community treats the law as established and applies it qualitatively; it has not produced a measurement practice in which the varieties of disturbance and response are counted independently of outcome, so “variety” in practice has drifted toward a synonym for complexity or flexibility. Whether that drift is repairable is an open question the community has not posed.
Importance for the article. §2.3 classifies Ashby’s law and the Conant–Ashby theorem as theorems, analytic given their definitions, and draws the parallel with the free-energy principle: a formally true core surrounded by empirical claims that borrow its authority. §7 (Demonstration II) follows from that classification: v01 had proposed to test whether requisite variety “does empirical work,” and C1 §4 showed that this tests an operationalisation, not the law; v02 quotes Ashby exactly and moves the test to the model’s channels (claim IIa), with a named rival operationalisation (queueing, Galbraith’s information-processing capacity, or a domain model). A reviewer will press on two points: whether the disturbance, response and outcome sets in IIa can really be fixed before outcomes are seen (rival 3 in §7.2 says they cannot), and whether the VSM community’s qualitative use of the law leaves anything for a comparative measurement study to adjudicate. The co-author must be able to defend the theorem/operationalisation split without conceding that the law is empty.
Sources in the reading list.
- the definitions of variety and regulation and the exact wording of the law, chapters 7 and 11
- the theorem’s statement, its premises, and the conclusion that regulation theory cannot be separated from modelling
Other important sources and authors.
- Ashby, W. R. (1952). Design for a Brain. Chapman & Hall, London. — the homeostat and ultrastability, the mechanisms Beer’s homeostat language descends from
- Ashby, W. R. (1958). Requisite variety and its implications for the control of complex systems. Cybernetica, 1(2), 83–99. — Ashby’s own restatement of the law with its implications for control, shorter than the book
- Ashby, W. R. (1962). Principles of the self-organizing system. In H. von Foerster & G. W. Zopf (Eds.), Principles of Self-Organization (pp. 255–278). Pergamon, Oxford. — the same volume as Beer (1962); Ashby’s argument that “self-organisation” needs an external selector
- Conant, R. C. (1976). Laws of information which govern systems. IEEE Transactions on Systems, Man, and Cybernetics, SMC-6(4), 240–255. — the information-theoretic decomposition of regulatory work that a quantitative channel study would build on
- Shannon, C. E. (1948). A mathematical theory of communication. Bell System Technical Journal, 27, 379–423 and 623–656. — channel capacity, the quantity Beer’s channel principles presuppose
- Wiener, N. (1948). Cybernetics: Or Control and Communication in the Animal and the Machine. Wiley, New York. — the founding text; feedback and the animal–machine analogy that Beer inherited