This paper proposes JX-Graph 1.0, a preliminary conceptual framework for auditing hidden interventions in digital platforms using graph-theoretic and information-theoretic tools.We formalize a time-varying (near-regular) graph model and introduce four diagnostic metrics:Network tension (Φt) – imbalance across interactions.Symmetry divergence (Δsym) – structural asymmetry of user orbits.Jump-history entropy (JHET) – unpredictability of behavioral shifts.Description-length change (ΔMDL) – complexity differences in encoding system states.Through a simple case study on a cycle graph (C₆), we illustrate how localized interventions trigger characteristic shifts across all four metrics.The framework is explicitly proof-of-concept: empirical validation is limited, scalability challenges remain, and comparisons with existing anomaly detection methods are absent. Nevertheless, the contribution lies in posing a new research question—how can platform interventions be audited in a transparent, mathematically principled way?We invite critical feedback and further development, aiming to refine the theoretical foundation and extend the methodology to large-scale, real-world data.
AKIOHOTTA (Tue,) studied this question.