Universal Random Granular Substrate: Tests of a Granular-Spacetime Hypothesis against Precision Gravity and Cosmology

Preprint summarizing the current state of the Universal Random Granular Substrate (URGS) programme, an exploratory, independent-researcher attempt to model spacetime as a random (Poisson-sprinkled) granular substance whose stable vibrational packets are matter. Across 22 numerical and analytic assignments, the later half pre-registered with stated pass criteria and look-elsewhere estimates, the original postulates were tested against particle-physics constants, precision gravity and cosmology. Main results: No URGS mechanism derives the fine-structure constant, the electron mass or the hierarchy; a 1.3% near-coincidence from a Chain-Theory-inspired counting rule was falsified by a pre-registered muon test. All mechanical gravity variants (compaction, incompressible, micromorphic, spacing-field, random-link tensor elasticity) fail: PPN γ = 0 or −1, or a Planck-mass, non-Fierz–Pauli graviton. A curvature-only (Regge) energy on random grains reproduces Schwarzschild: γ = 1.0025 ± 0.0006 (the offset from 1 is discretization error at our resolution, not a prediction), light bending 1.737″, Mercury perihelion 42.0 ± 1.5″/century. This is a consistency test with exact boundary data and is discretized general relativity. Solitons on random grains Lorentz-contract as expected. The causal-set everpresent-Λ magnitude is recovered within a factor ~2, but Zwane–Afshordi–Sorkin Model 2 fails DESI DR2 BAO + θ* (0/20,000 realizations match ΛCDM). A proposed 'subspace leakage' contribution to cosmological redshift is bounded below ~0.8% (statistical) or ~2% including systematics, from supernova time dilation. Revised postulates R1–R5 reduce URGS to causal sets + Regge calculus + solitons. Negative results are presented as the main deliverable. Numerical work assisted by AI tools.