Why φ? The Golden Ratio as Universal Efficiency Limit in Computational Finitism

The golden ratio φ appears throughout Computational Finitism: in the Fibonacci weights of the Golden-Fano Automaton, in the φ-fractal scaling of information dynamics, and in the antifragile entropy ≈ 0.988. Its status, however, remains ambiguous. Is φ a geometric property of the measurement interface, a convenient coordinate imposed by observers, or an intrinsic efficiency limit of the universe's information processing? We argue that both readings are correct, and that for universal efficiency limits the distinction dissolves. The ontic reading is grounded in Hurwitz's theorem (φ is the "most irrational" number, least approximable by rationals), the KAM theorem (golden tori are the last to break under perturbation), and the Pell equation (optimal discrete rotation on a pixelated lattice). The epistemic reading is grounded in the base-independence of the substrate. We propose a formal duality: for constraints that any long-lived discrete system must satisfy, the ontic and epistemic descriptions are equivalent. The duality makes a falsifiable prediction: φ should emerge spontaneously in discrete systems that were not designed with φ in mind. We connect this to the Fano geometry and the percolation threshold = 0.325, and identify the precise division of labor: Fano geometry governs structure, φ governs dynamics. "The golden ratio is not a magic number. It is the unique solution to the anti-resonance problem. The universe uses φ because it has no choice."