Emergent Area Laws and Entropic Attraction in a Bare Graph Dynamics Surrogate

We study whether a deliberately bare QGEFT surrogate can support two linked but conceptually distinct signals: a vacuum branch whose Blind-Observer entropy is organized more naturally by boundary area than by enclosed volume, and a finite-size matter-attraction channel generated without inserting any explicit $1/r$ potential into the action. The model is an annealed sparse graph whose edges carry diagonal $SU(3)$ phases and whose local energy is given by a triangle holonomy term. Vacuum configurations are interrogated by a topology-only entropy proxy, while the gravity-facing branch introduces two explicit heavy defects through a soft quadratic degree penalty. Across a size sweep $N=256,512,1024$ , the clearest vacuum signal appears at $N=1024$ where the entropy proxy is better fit by boundary area than by enclosed volume, with RA²≈0.848 and RV²≈0.173 Two null ensembles, based on phase shuffling and Erdős-Rényi rewiring, do not reproduce the same positive area-oriented trend. In a representative gravity run at $N=256$ with target mass degree 48 and coupling λ=0.05. the heavy pair repeatedly re-enters direct-contact or near-contact graph states, with dₘᵢₙ=1 dfinal=2, and mean separation d≈1.853, while shared-neighbor overlap rises into the mid-teens. The narrow claim supported by the current data is not that continuum gravity has been derived. It is that, within the bare QGEFT surrogate, a reproducible area-law vacuum branch can coexist with a reproducible finite-size entropic-attraction channel in which explicit heavy defects reorganize the surrounding sparse graph through shared-neighbor backreaction.