---
title: "Ch 15 — The Thirty-Year Experiment"
description: "Part II chapter reading holonic manufacturing as a long, well-documented, largely negative experiment on architectures shaped like the VSM that organizational…"
url: "https://wiki.cybernetic-futures.org/punch-and-judy/part-02-outside-the-model/ch15-thirty-year-experiment"
modified: "2026-11-12T15:26:25.910Z"
language: "en"
site: "Cybernetic Futures"
---

# Ch 15 — The Thirty-Year Experiment

[Chapter content](https://wiki.cybernetic-futures.org/punch-and-judy/part-02-outside-the-model/ch15-thirty-year-experiment/ch15-thirty-year-experiment-content) · Overview diagram

# Ch 15 — The Thirty-Year Experiment We Never Looked At

**Part II — Outside the model.** The chapter argues that holonic manufacturing ran a thirty-year, partly negative experiment on architectures with the VSM's shape, and that organizational cybernetics has no recorded engagement with the outcome.

## The question
What did the nearest neighboring field to ours learn when it specified, built and ran an architecture aimed at the same problems, and why did we never look?

## The finding
Production 2000+ was installed in 1999 on a cylinder-head line at Mercedes-Benz Untertürkheim, passed its performance test, ran in day-to-day operation for five years and was not repeated; the chapter counts holonic architectures that have actually run a production line as close to one. Two looser charges are refused: holonic manufacturing did not succeed where the VSM failed, since it did not diffuse either, and it is not the VSM under another name, descending from Koestler and control engineering rather than from Beer.

PROSA's two admired properties are a closed vocabulary of holon types and a failure mode specified in advance. The chapter says the difference from Beer is narrower than earlier drafts claimed: he has a closed set of types and derives each subsystem from what fails without it, and even describes the degraded "decerebrate cat". What he did not do is state the degradation as a requirement a candidate design must meet, with the surviving performance fixed beforehand.

What the record removes is one explanation the tradition has leaned on: that limited industrial reach is a writing-up problem. It does not show that specification is worthless, nor that the model's known defects are harmless. The count of one is the chapter's own, assembled from the three sources that would have reported a further deployment, and it is stated as falsifiable. Bussmann's own account blames economics, not the technology, and the chapter lists the rival explanations that remain open.

Cybersyn is set beside the holonic record as the one VSM-based architecture specified, built and run: its performance is not documented, no success criterion was stated before it ran, and Beer's own assessment was ambiguous. Four further unread literatures are listed — qualitative comparative analysis, measurement, sociological systems theory and Troncale's system process theory.

## What the chapter puts at risk
1. Run a three-way comparative post-mortem of the holonic record, the sociotechnical designs of Chapter 13 and Cybersyn — twelve researcher-days for the holonic record and about six more for the others.
2. Pre-register rival explanations for non-diffusion, including Bussmann's weak business case, Beer's piecemeal-incorporation account of Cybersyn, and his general explanations.
3. Record for each case which explanations its sources can test and which they cannot; if one explanation fits all three, retire "write it down better" as the remedy.
4. Recheck the holonic count before print — it is explicitly falsifiable and one counterexample retires it.

## Beer locators
BOTF 1981: 264, 270, 311, 312-14, 336, 339-40, 346; DC 1966: 243; DTS 1985: 66, 130, 144; HOE 1979: 142-43, 177, 179-81, 231, 235, 259, 265, 297; TBYT 2009: 150, 183-84.

## Status
Draft, *Punch and Judy* v0.5.0 — **not project canon**, and not published. The manuscript marks the Beer locators in this chapter as taken from Pure VSM and not yet checked against the PDFs. It also flags: the count of one as the authors' own count of a negative, resting on three sources with no systematic census to check it against; the deployment criterion as deliberately strict, so a looser reading would return a different number; the Bussmann lessons as one account of one plant; the 1978 Troncale source as unread; and its own corrections to earlier drafts, including that the Rockwell packing cell was a university laboratory rig.

## Related
[Punch and Judy — Overview](https://wiki.cybernetic-futures.org/punch-and-judy/overview), [Recursion](https://wiki.cybernetic-futures.org/pure-vsm/concepts/recursion), [System Two](https://wiki.cybernetic-futures.org/pure-vsm/concepts/system-two), [System Four](https://wiki.cybernetic-futures.org/pure-vsm/concepts/system-four), [System Five](https://wiki.cybernetic-futures.org/pure-vsm/concepts/system-five), [Cybersyn](https://wiki.cybernetic-futures.org/pure-vsm/entities/cybersyn), [Ch 13 — The Autonomy Design](https://wiki.cybernetic-futures.org/punch-and-judy/part-02-outside-the-model/ch13-autonomy-design), [Ch 14 — The Evidence: Ostrom](https://wiki.cybernetic-futures.org/punch-and-judy/part-02-outside-the-model/ch14-evidence-ostrom)