Richard Gunther
Recursive Systems
Biography
Richard Gunther is an independent practitioner based in Newcastle, County Down, and the founder of Recursive Systems, where he develops software for organisational diagnosis using Stafford Beer's Viable System Model. The system treats model output as provisional: structure goes onto the map only when it passes an evidence-gated test, and unresolved gaps become diagnostic questions rather than invented answers. He has tested this workflow on two public-record cases. Alongside this work he advises manufacturers on Amazon marketplace strategy and builds the software supporting that practice, which is where his interest in applying the VSM to real organisational problems began. He studied Management Science at Bayes Business School in London, with periods in Hong Kong and Austin, and first encountered the VSM in an undergraduate dissertation on virtual organisations. His next step is testing it with practitioners in a live organisation, and he hopes to meet people willing to challenge the method.
Sessions
From Evidence to Inquiry: Keeping LLM Inference Provisional in Recursive VSM Diagnosis
Purpose: Clemson’s VSM ToolBox and Pérez Ríos’s VSMod showed how software can support Stafford Beer’s Viable System Model (VSM); LLMs introduce a different cybernetic problem: plausible inference can harden into organisational fact and propagate across recursion levels. This project builds a traceable path from evidence to hypotheses, questions and candidate interventions. Methods: The prototype was applied to two public-record cases: Robin Hood Energy and the Charity Commission inquiry into New Wineskins Charitable Trust and U-Turn Move on Homes. It proposes operations, relations and recursive structure, promoting each only when an evidence-gated test passes; otherwise it abstains. Structural gaps become an agenda of focused questions; answers and corrections persist through reanalysis. Testable structural hypotheses and candidate prescriptions each carry a check that could confirm or refute them. When an LLM-attributed record includes a quotation and citations, the quotation must match the cited source text verbatim. Findings: Both demonstrations produced evidence-linked maps, diagnostic observations and explicit gaps. In one, an entered answer that no role monitored the external environment or future was retained verbatim as evidence of a System 4 gap; no function was invented to fill it. VSM operates as a reasoning grammar, not only a diagram. Implications/Contributions: The contribution is a persistent evidence-to-inquiry loop that turns abstention into diagnostic work while model suggestions remain provisional. Results demonstrate software traceability, not diagnostic validity or organisational impact. Practitioner testing comes next. When should AI-proposed structure become a diagnostic premise?