“That’s the way to do it” is what Mr Punch says after each blow, in the swazzle voice,1 to an audience that has no vote. It is a verdict announced by the only party keeping score. Any tradition that scores its own results is exposed to the same temptation, and most of all when the tradition is good, because then the self-congratulation is mostly earned. We leave the phrase there and will not press it further.
This book appears in the centenary year of Stafford Beer’s birth. Centenaries have a settled shape: collected volumes, retrospectives, panels on legacy, a figure placed carefully in a lineage. That work is worth doing, and much of it is being done well elsewhere. We have chosen a different use for the occasion. Some five decades after the Viable System Model reached the form we now teach,2 we would rather examine it than commemorate it. Beer built the model by taking things apart. It seems right to mark the year by taking his apart.
We have applied the model to hospitals, ministries, cooperatives, software teams, whole national economies. When an intervention succeeded, we credited the diagnosis. When it failed, we found that the model had been applied badly, or the client lacked commitment, or the recursion had been cut at the wrong level. There is no result we have not been able to accommodate.
That is our charge. It is narrow and, we think, serious: the Viable System Model has never been put at risk. Quantitative tests of it have been published, and Chapter 18 examines the two most substantial; neither was built so that a negative result could cost the model its central claim. It has not been shown to be false. Neither has it been shown to be right, because the community around it, ourselves included, has never stated what observation would make the model give something up.3 Beer did not ask to be spared. “If mistakes have been made,” he wrote, “the models can be tested — to destruction.”4 He did not say which observation would count, and neither have we. A framework that can absorb any outcome is a language, not a theory. It is useful for describing what we have already decided to see and silent on what we might be wrong about. This thread runs from Chapter 1 to Chapter 30, and it is the only claim in the book we would defend without qualification.
What follows is not an obituary. We are not announcing that the model is dead, and we do not think it is. Nor is it a defense; the community has produced enough of those, and we have written some of them. And it is not a demolition. Part Four, “What we lose,” is no courtesy paid at the end. It sets out a narrow claim we believe the model can actually defend: that it is the only available instrument that will take a named system at a named level and say what that system needs in order to remain viable, although we have not operationalized that claim or designed the test that could falsify it. It then names six things we know of nothing else that does: a closed set of functions asserted as necessary, recursion with a criterion, a diagnosis rather than a description, a pathology catalog generated from the structure, one currency across scales and substrates, and the algedonic channel. Chapters 25 and 26 add the transducer as an object of design and autonomy stated as a quantity rather than a value. We state that claim carefully because criticism is only worth reading when it is specific about what it leaves standing. A critique that dismisses everything has told you nothing about which part failed.
Every charge in Parts One and Two ends with a receipt. It names the smallest piece of work that would settle the matter, what that work would cost in people and weeks, and what a negative result would establish, which is the part we have most often skipped. If we claim the neuroanatomical mapping we teach cannot be found in the sources we cite, the receipt is a bounded reading of those sources by someone with no stake in the answer, and a statement of what we will teach differently if the mapping turns out to be there. If we claim the participation loop in our founding case, Project Cybersyn in Chile, stayed an intention and was never built as a channel with standing, the receipt names the archive, the hours, and what it would mean for our account if evidence of such a channel is found. Some of these tests cost a few weeks. The two cheapest decisive ones are in Chapter 29; the most expensive cost rather more, and Chapter 30 prices the whole program. A criticism that names no such test is a complaint, and complaints have never cost this model anything.
We are writing for the people who teach the model, apply it in organizations, defend it in reviews and argue about it at conferences. We are two of them. Between us we have taught the mapping described in Chapter 1 to students who trusted us, quoted the law in Chapter 3 in the form we now believe to be altered, and drawn the channels of Chapter 4 on whiteboards without noticing what we had left unspecified. The first person plural in this book is meant literally. Nothing here is a charge against a community we stand outside of, and where the record is embarrassing, it is our record.
What we want from the reader is a better version of the disagreement, rather than agreement. If a chapter’s charge is wrong, the receipt attached to it should make the error easy to demonstrate. We would rather be corrected in a form we can learn from than granted the benefit of the doubt for another fifty years. The model has survived on the second of those. We think it can survive the first, and we would like to find out.
Notes
- On the swazzle, the reed a Punch performer places at the back of the mouth to make the voice, and on the history of the show whose catchphrase we borrow: “‘That’s the way to do it!’ A history of Punch & Judy,” Victoria and Albert Museum, last updated 17 April 2024, https://www.vam.ac.uk/articles/thats-the-way-to-do-it-a-history-of-punch-and-judy, accessed 29 September 2026. See also George Speaight, Punch & Judy: A History (Studio Vista, 1970). ↩
- We date that form from Stafford Beer, Brain of the Firm (Allen Lane, 1972), and The Heart of Enterprise (John Wiley & Sons, 1979). The second edition of Brain of the Firm (John Wiley & Sons, 1981) was prepared in parallel with the latter. ↩
- Chapter 9 sets out the criterion we apply here, and Chapter 18 applies it to the published tests of the model. ↩
- Stafford Beer, Brain of the Firm, 2nd ed. (John Wiley & Sons, 1981), 377. The passage continues: “They seem to me to be useful; but they are cheerfully submitted to Popperian criteria of disproof.” Chapter 9 takes up what that commitment asks of us. ↩