Governance of socio-technical systems is typically treated as a design problem: write rules, enforce them, revise when necessary. This document proposes that governance systems exhibit a structural phase transition analogous to gelation in polymer chemistry and autocatalytic closure in origin-of-life theory. Two conjectures are stated in formally attackable terms. Conjecture A posits that increasing cross-linking density among governance artefacts, combined with a functioning enactment layer of agents and authority, induces a threshold-like transition from predominantly exogenous production of governance artefacts to partially endogenous reproduction and repair — a transition detectable via Reflexively Autocatalytic and Food-generated (RAF) set analysis. Conjecture B posits that in nonstationary, partially observable environments, static governance has a finite expected survival time, while governance systems equipped with sufficiently effective correction loops can maintain viability for substantially longer. The minimal transferable unit of self-sustaining governance is characterised as a tuple (R, , L, E_) comprising seed rules, a cross-linking mechanism, an emergent correction loop, and a minimum viable enactment layer. Neither conjecture is proven. Both generate testable predictions against a live governance system (Stirps). Falsification criteria, an evidence interpretation matrix, and a deferred empirical protocol are specified. The formal proof, if it exists, is a research programme requiring percolation theory, RAF detection algorithms, stochastic stability theory, and active inference formalism applied to formally specified governance network models.
Colin Bernhard Apel (Mon,) studied this question.