Observational Consequences of Entropy-Geometric Branch Weighting: Boltzmann Brain Suppression, Inflationary Predictions, and the Curved-Space C=1 Extension

Paper 1 of this series derived an entropy-weighted branch measure from the C=1 path-integral formulation and established the corresponding observer measure, effective free-energy functional, and arrow-of-time selection rule. In the present work we investigate phenomenological consequences of that framework without introducing any additional postulates beyond those retained in Paper 1: the Born rule, the Observer Measure Principle, and the Past Hypothesis. We first analyse the Boltzmann Brain problem. Using Carroll's estimate of the Helmholtz free energy required to assemble a functioning brain-like structure from thermal equilibrium, ΔF_BB ≈ 10⁴⁵ J — consistent with the free-energy framework used in multiverse-measure studies of this problem (De Simone, Guth, Linde, Noorbala, Salem & Vilenkin 2010) — we obtain a characteristic entropy cost ΔS_BB/k_B ≈ 2.66×10⁶⁷ at the present CMB temperature. The entropy-weighted branch measure alone, independent of any additional assumption, yields p_BB/p_OO ≲ exp(−2×10⁶⁷). If this mechanism acts independently of the standard Born-amplitude suppression already present in the underlying quantum probabilities — an assumption we state explicitly and do not derive — the combined bound strengthens to exp(−3×10⁶⁷), with additional suppression expected from fluctuation-determinant effects. When the Observer Viability Criterion is strictly imposed, Boltzmann Brains are excluded exactly regardless of either estimate. The framework is extended to curved spacetime through a minimal generally covariant formulation in which the entropy functional enters the path-integral measure but does not contribute directly to the classical stress-energy tensor. We derive inflationary phenomenology associated with entropy-geometric branch weighting and obtain the leading-order predictions ΔP/P = 4Ξ_inf, f_NL = 0, g_NL = 18Ξ_inf for the preferred coupling β = −3/2. The vanishing bispectrum and nonzero trispectrum constitute a distinctive signature of the framework. For representative GUT-scale parameters, Ξ_inf ∼ 1.6×10⁻⁷, all predicted deviations remain well below current observational sensitivity. Finally, we discuss a local entropy-reduction mechanism arising from Euclidean instanton selection. This mechanism provides a route to localised low-entropy structure formation but does not derive, replace, or explain the global Past Hypothesis, which remains a foundational postulate of the framework. Paper 2 of 3 in the EBM+C=1 series.