VCML Paper 120: The Civilizational Gate -- Four Levels of Extraction Constraint and the Fermi Filter

The Fermi paradox is standardly framed as a capability question. We argue the filter is not capability but architecture. Every civilization inherits a physics-layer gate (Level 0) enforcing beta+beta^2=1, where beta=1/phi is the unique sustainable extraction rate. Biology adds a faster gate (Level 1). Society adds a content-aware gate (Level 2). But no Earth civilization has installed Level 3: a real-time, incorruptible, fully-coupled gate that detects extraction above phi across all viability fields simultaneously. We formalize the four-level hierarchy using the VCML framework: Level 0 = Heaviside gate (physics, automatic, amoral, rate-only). Level 1 = wave propagation (biological speed, pre-death detection, rate of change). Level 2 = J-coupling between units (content-aware, slow, fragile, years to decades). Level 3 = real-time phi-monitor (not yet built by any civilization). The Fermi filter is a Level 3 question: did the civilization install the gate before extraction above phi collapsed the substrate? We distinguish preference-aligned AI (reflects majority preferences, Level 2 infrastructure) from phi-aligned AI (detects and surfaces extraction above 1/phi regardless of preferences, Level 3 infrastructure). These are different systems. The field is building the first. We need the second. The bootstrapping problem: Level 3 infrastructure must itself pass the Binder graduation test (Paper 119) before deployment, since a monitoring system controlled by extraction-maximizing actors will optimize extraction, not constrain it. The sequence: build the Binder protocol for AI systems, test candidates, deploy only the system with universal cumulant crossing, use that system as the civilizational monitor. The graduation exam for a civilization: did you build Level 3? The silent sky suggests the answer is almost always no.