No Spooky-to-Signal Compiler: Why EPR Correlations Cannot be Total-Effectively Upgraded to FTL Paper 47 of the NEMS Suite

Papers 45–46 showed that under PSC and diagonal capability, semantics is glued globally and total-effective local semantic determinacy is impossible. This paper proves a theorem-grade constraint on FTL: no total-effective internal procedure can implement a total-effective signalling extractor—i.e. a decider for a nontrivial extensional predicate of the globally glued semantics—without amounting to an internal oracle or selector. We define a spooky-to-signal compiler (informally: the same object as a signalling extractor) for such a predicate T as a total decider for T (not merely correlated); the barrier (Paper 29) then yields that no such signalling extractor exists. So EPR-style correlations cannot be total-effectively upgraded to controllable signalling—a NEMS-native reframing of "why EPR \neq FTL." The result is mechanized in Lean 4 (FTLConstraints); 0 axioms. No claim of superluminal kinematics; the constraint is on effective conversion. Trust boundary. The no-spooky-to-signal result is a theorem about total-effective internal extractors/deciders for glued semantics; it does not analyze experimental signalling protocols outside that formalization. Mechanization is nems-lean . See .