Detail page for the cheat sheet. Legend: ✓ read · ◐ secondary/abstract · † unread · ✗ no source.
Introduction. Ashby defines variety as the number of distinguishable elements of a set, or its logarithm to base 2 (An Introduction to Cybernetics, ch. 7). The law comes in ch. 11 in a fixed sequence:
- §11/5. The stipulation: “let us consider, then, only those tables in which no column contains a repeated outcome.”
- §11/6. The non-logarithmic proof, ending “Only variety in R’s moves can force down the variety in the outcomes.”
- §11/7. The logarithmic form, with the sentence “only variety in R can force down the variety due to D; variety can destroy variety” (p. 207).
- §11/9. The stipulation is replaced by repetition k, giving V(O) ≥ V(D) − log k − V(R). k is not bounded by anything in the theory.
- §11/10. The law is a priori: “The law states that certain events are impossible… It has nothing to do with the properties of matter.”
- §11/11. Ashby names the homology with Shannon’s Theorem 10: “his ‘noise’ corresponds to our ‘disturbance’, his ‘correction channel’ to our ‘regulator R’.”
All ✓. Pages checked on 22 Sept 2026 against the Principia Cybernetica reproduction in the calibre library, which follows the Chapman & Hall second impression (1957) page for page:
- §11/5: p. 204
- §11/6 and §11/7: p. 206, with the sentence at p. 207
- §11/9 and §11/10: p. 208
- §11/11: pp. 209–11. The sentence “R’s capacity as a regulator cannot exceed R’s capacity as a channel of communication” is at p. 209; the Shannon homology is at p. 211.
Ashby 1958 (Cybernetica 1(2): 83–99, free PDF at pespmc1.vub.ac.be/books/AshbyReqVar.pdf) restates the law in entropy form and ties it to Theorem 10 ✓. What it adds:
- “R’s capacity as a regulator cannot exceed its capacity as a channel for variety.”
- A purely error-controlled regulator cannot be 100 per cent efficient.
- Capacity can be raised by teams of regulators.
- For very complex systems, aim at control rather than understanding.
- It argues that the limit is causal, not merely definitional. This is relevant to the tautology objection below.
- The 1958 paper does not contain “only variety can destroy variety”, nor “variety can destroy variety” ✓ (checked 22 Sept 2026 in the retyped Principia Cybernetica text). Its verbal statement is: “If now Vd is given, Vo’s minimum can be lessened only by a corresponding increase in Vr. This is the law of requisite variety.” That text carries no journal pagination.
Beer’s rendering. “Only variety can absorb variety” appears in Brain of the Firm (2nd ed., 1981), in the Chile chapters. Page found on 22 Sept 2026: Brain of the Firm, 2nd ed. (Wiley, 1981), p. 308, in the section “Externalities” of the Chile part (OCR scan in the calibre library) ✓. In the same paragraph Beer paraphrases the good regulator theorem: “the cybernetic theorem that declares a regulator to be effective only insofar as its model of what is regulated is adequate”. Variety engineering (attenuate disturbance, amplify response) is Beer’s design translation. Its formal location is The Heart of Enterprise (1979); the chapter and page are not confirmed ◐.
Versions of the sentence in circulation
- Ashby 1956 §11/7: “only variety in R can force down the variety due to D; variety can destroy variety” ✓
- Ashby 1956 §11/6: “Only variety in R’s moves can force down the variety in the outcomes” ✓
- Beer 1981, p. 308: “Only variety can absorb variety” ✓
- Martin Pfiffner, The Neurology of Business (Springer, 2022), p. 143: a box headed “Definition – Ashby’s Law of requisite Variety: Only Variety can absorb Variety”, footnoted to Ashby 1956. The same wording appears again at p. 151 ✓ (read 22 Sept 2026). A recent textbook example of Beer’s sentence cited under Ashby’s name.
- Seth 2015 quotes a “only variety can force down variety” form ◐
The tautology objection, in print
- Zeleny, M. (1986). The law of requisite variety: is it applicable to human systems? Human Systems Management 6: 269–271. doi:10.3233/HSM-1986-6401. Zeleny cites Ashby’s own description of the theorem as “primarily a statement about possible arrangements in a rectangular table”. He argues the law is self-evident in the abstract and “quite irrelevant and empty for the purposes of management” ✓.
- Dewhurst, D. (1991). The status of requisite variety. Kybernetes 20(2): 61–64. doi:10.1108/eb005883. He calls the law “a limited tautology”, not an empirical law, and warns that treating it as empirical fact may constrain controller design ◐ (abstract).
- We found no published reply to either. The chapter’s position covers both sides: the law is a hard quantitative bound and it is a priori, so it cannot say anything about a particular firm without a counting convention.
Important authors. W. Ross Ashby (1903–1972), British psychiatrist. He wrote Design for a Brain (1952) and An Introduction to Cybernetics (1956) at Barnwood House, Gloucester, and later worked at the Biological Computer Laboratory, Illinois. Claude Shannon (1948) supplies Theorem 10, from which the law is derived.
Importance for cybernetics / VSM. Beer derives recursion, the channel-capacity principles and the Three–Four homeostat from this law. It is the tradition’s strongest formal claim precisely because it reduces to information theory.
Importance for the chapter. It carries the sections “Two sentences and one verb” and “What the law actually says”. It also supplies the a priori point that explains why nobody measures variety.
Sources in the reading list
- Ashby 1956, chs 7 and 11: the definitions and the exact wording ✓
- Ashby 1958: the restatement, and “two observers differ” ✓
- Beer 1981, the absorb sentence: p. 308 ✓
- Shannon 1948, Part I, Theorem 10 †