This record presents a demonstration-grade empirical evaluation of a simple divergence metric derived from the Appearance/Ground (A/G) framework. The metric is applied across three distinct substrates—financial bankruptcy, biological malignancy, and metabolic disease progression—to examine whether a common structural signature of failure can be observed across domains. The study is intentionally positioned as an auditable baseline rather than a deployment-ready predictive system. Feature assignment and stratification thresholds are computed in-sample, and all limitations related to evaluation posture are explicitly acknowledged in the manuscript. A reference implementation and raw source data are included to support full reproducibility and inspection. Empirical results show consistent risk separation across domains, with increasing divergence nearer to labeled failure horizons in the financial datasets. The accompanying software package contains runnable scripts, documentation, and canonical raw data partitions to enable independent verification or extension. This work is intended as a methodological contribution illustrating a substrate-invariant divergence geometry, rather than a finalized applied model.
Sanjit Mehat (Thu,) studied this question.