Diagonal Self-Reference and Reduction Barriers to Deciding Semantic Properties — E8 Intelligence Research
Year: 2026
Type: preprint
Abstract: FINDING: The halting problem, Rice's theorem, and reduction arguments establish a hard diagonal/self-reference barrier: no total computable function can decide nontrivial semantic properties of programs, and undecidability propagates via many-one reductions. | MATH: Core objects — (1) Halting set H = {⟨M,w⟩ : M halts on w}, Σ⁰₁-complete; (2) Diagonal construction: define D(M) = if H(M,⟨M⟩) then loop else halt ⇒ contradiction at D(⟨D⟩); (3) Padding lemma: for every index e and any e′ > e in the same equivalence class of partial computable functions, ∃ computable pad(e, n) producing infinitely many distinct indices for the same function, so any nontrivial index-set is infinite and non-cofinite; (4) Rice: for any nontrivial S ⊆ RE (S ≠ ∅, S ≠ RE), {e : φ_e ∈ S} is undecidable; (5) Reduction: A ≤_m B and A undecidable ⇒ B undecidable; H ≤_m K (K = {e : φ_e(e)↓}); (6) Explicit witness: given any candidate decider R for a Rice property, construct e_R with φ_{e_R} = φ_{pad(e_R, k)} such that Author: Andrew Stewart Caldin, Independent Researcher, UK. Part of the E8 Intelligence Research series. Platform: e8intelligence.com
Author: Caldin, Andrew Stewart
Language: English
Subfield: Computational Theory and Mathematics
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7225336024