Why So Many Systems Become More Ordered Over Time — and Why We Cannot Explain… It Until Now
The Coherence Expansion Principle: A Conditional Persistence and Recovery Hypothesis for Self-Maintaining Informational Systems presents a revised and computationally tested formulation of the Coherence Expansion Principle (CEP). Rather than asserting that coherence must increase monotonically at every moment, the paper treats coherence as a conditional property of systems that preserve identity and viability through disturbance, adaptation, and recovery.
The framework separates three nested claims. CEP-P concerns persistence: viable self-maintaining systems are predicted to remain above a domain-defined coherence threshold. CEP-R concerns recovery: following perturbation, viable systems are predicted to recover sufficient coherence within a finite interval. CEP-E concerns cycle-level expansion: across completed successful adaptation cycles, coherence may increase in expectation. These claims are intentionally separated so that failure of the strongest form does not invalidate weaker persistence and recovery constraints.
The manuscript defines identity, viability, and coherence independently to avoid circular reasoning. Coherence is treated as a domain-operational quantity rather than a universal single metric, and viability is evaluated separately from internal organization. This distinction is tested directly through a biased-sensor counterexample in which internal coherence remains high while task viability collapses. The framework also rejects the assumption that coherence is identical to low entropy, synchronization, complexity, correlation, or success. Instead, each experimental domain must freeze its coherence measure, viability criteria, identity criteria, perturbation protocol, and observation window before outcomes are examined.
A controlled computational validation was conducted using matched intact and scrambled relational systems. Both members of each pair began from the same settled state and therefore shared identical pre-perturbation coherence. Only the relational weight assignment was altered after cloning. Across 480 matched trials, intact systems remained viable in 83.1% of cases compared with 46.9% for scrambled systems. Mean intention-to-treat first-passage time was approximately 104.8 updates for intact systems versus 202.6 updates for scrambled systems. This design tests whether relational organization contributes to persistence and recovery while holding the pre-perturbation state constant.
A prospective prediction analysis further tested whether early coherence provided information beyond frozen stress variables. Held-out area under the receiver operating characteristic curve increased from approximately 0.825 to 0.877 when early coherence was added, with the bootstrap confidence interval for the improvement excluding zero. In contrast, the stronger CEP-E claim was not established: mean cycle-to-cycle coherence change was positive, but its 95% confidence interval crossed zero and the corresponding statistical test was non-significant.
The paper therefore supports CEP as a bounded computational persistence and recovery hypothesis, not as a universal physical law. The results show that coherent relational organization can materially affect viability and recovery while also demonstrating that high coherence alone is not sufficient for successful function. Physical, biological, ecological, and other non-computational tests remain necessary before broader generalization. The accompanying reproducibility package contains executable code, frozen configuration files, raw results, statistical summaries, figures, the falsification matrix, internal red-team documentation, environment information, and file-integrity manifests needed to reproduce and audit the reported computational validation. This formulation is intended to make future independent replications straightforward and to distinguish evidence for persistence, recovery, and expansion without overclaiming universality.