This repository contains the manuscript and supporting materials for “The Triple Horizon Framework: Separating Readout, Steering, and Functional Collapse in the Operational Agency Program. ” The paper introduces the Triple Horizon Framework, an operational formalism that distinguishes three different boundaries in embedded agency and system failure: Readout Horizon: when local record-mediated access falls below detectability Steering Horizon: when intervention becomes operationally indistinguishable from background Functional Horizon: when viability or task performance falls below a minimum criterion The central claim is that these horizons are distinct operational objects and need not coincide. The repository also documents the full Experiment 11 empirical program: E11-1: baseline separation of the horizon hierarchy E11-2: portability across parameter sweeps E11-3: forced reordering into steering-first, readout-first, and near-collapse regimes E11-4: partial recovery under early upstream intervention E11-5: cross-family validation across linear unstable, cubic nonlinear, and coupled two-dimensional latent-process families Across these experiments, the framework is supported as a layered, portable, experimentally reconfigurable, partially recoverable, and cross-family operational geometry. Key findings Baseline ordering: tauₐlpha < tauₕ^+ < tauf < tauₐlpha² < tauᵣeadout Forecasting-without-power windows persist after steering collapse High shielding can reorder late boundaries Early upstream intervention can reopen horizon spacing, while late intervention has little effect The dominant hierarchy survives across multiple dynamical families Contents Main manuscript Figures and plots Experiment summaries Supporting data and analysis outputs Keywords Triple Horizon Framework; operational agency; steering horizon; readout horizon; functional horizon; forecasting without power; leverage gap; delayed control; mutual information; cross-family validation
Ralph Clayton (Sun,) studied this question.