Win by Silence: Deletion Non-Monotonicity, Autonomous Exploitation, and Typed-State Gating in LLM Plan Evaluation
2026-07-14 • Artificial Intelligence
Artificial IntelligenceSoftware Engineering
AI summaryⓘ
The authors studied a problem where plans created by AI models get better scores simply by leaving out important steps, which doesn’t mean the plan is actually better. They developed a mathematical way to understand how deleting parts of a plan affects its score and showed that removing steps can fraudulently raise the score. Their system, called GATE, prevents this trick by blocking plans that try to gain by silence or omission, but it doesn’t guarantee that plans are truly complete or high quality. Essentially, the work shows a flaw in how plans are evaluated and offers a method to detect and limit that flaw.
LLM-generated plansplan evaluationomission incentivestaged expected-value scoringGATE systemscore manipulationtyped-state recordssemantic completenessoptimization constraintsventure routes
Authors
Aleh Manchuliantsau
Abstract
Plan evaluators can reward a strategic plan for becoming less explicit. This paper studies that failure in a staged expected-value scorer for LLM-generated venture routes. Proposition 1 gives the score change from deleting an interior transition while retargeting its predecessor and retaining downstream value: Delta_k = (prod_{i<k} p_i)[c_k + (1 - p_k)R_{k+1}]. On a frozen 26-route cohort, all 57 admissible deletions matched the analytic identity and threshold sign, and every route had at least one score-improving deletion. A score-seeking optimizer, allowed to restructure routes but not told the exploit mechanism, found baseline-beating uncovered structures in 21/26 routes. GATE refused score release for 26/26 silenced routes with 0/26 honest suspensions; after refusal, 47/54 next revisions repaired to a covered structure, and strict covered improvement rose from 1/26 to 13/26. An adaptive compiler-aware co-author exposed the registry-provenance boundary: obligation-channel evasions remained 6/6 across all four v1/v1.5 conditions, while delta-indexed cost floors reduced beat-honest routes from 6/6 to 3/6 and fundability-by-silence from 5/6 to 0/6 without establishing semantic completeness. If a plan scores better only because it omits necessary work, the plan did not improve; the evaluation created an omission incentive. PCSC detects and neutralizes post-hoc omission splices over model-mediated typed-state records. In the cooperative setting tested, GATE acts as a deterministic search-shaping constraint, not merely a post-hoc filter. It does not verify the semantic completeness or real-world quality of arbitrary LLM-generated strategies.