The cost of abstaining

This overview explains how an advice policy can choose the exact fallback and still spend time deciding not to rewrite. Zero accepted rewrites does not mean zero advice cost when features and decision logic still execute. Start with a direct exact control costing ten illustrative units. It produces the required answer under the declared contract. This is the baseline request the gated route must match. The gate spends two units deciding to abstain, then runs the same ten-unit control. The answer remains exact and no rewrite was applied, but the complete request costs twelve. Abstention avoids an unsuitable action; it does not erase the gate's own cost. If all three requests take this fallback path, the direct total is thirty and the gated total thirty-six. Every answer can be identical while the gated route is slower on every request. These are teaching units, not reconstructed historical measurements. An abstention policy can preserve the exact fallback while still spending time obtaining features and deciding to abstain. Safe advice and profitable advice are different tests. Suppose a toy request takes 10 cost units without advice. A gate spends 2 units to decide that no rewrite is worth applying, then runs the unchanged 10-unit control. The final output is correct and the decision avoids a risky transformation, but the request costs 12. An out-of-range input should also abstain, with its reason recorded separately from insufficient predicted gain. The retained frozen D9 policy abstained on all evaluation workloads and preserved exactness. Its charged gate remained slower, so no rewrite policy was promoted. Do not present zero accepted rewrites as a zero-cost path. Suppose a toy gate is validated only for a declared range of inputs. An outside-range input should not be treated as if the same evidence applies. This is a schematic eligibility rule, separate from any specific retained D9 guard. The gate abstains from rewriting and uses an exact eligible fallback. This can be the correct safe behavior for the request, even if it is slower than calling that fallback directly. Report how often the gate acts or abstains, whether every output stays exact, and how much the complete route costs. Safe abstention and profitable routing require different evidence. Keep the retained all-abstained result and its slowdown visible. Charge the decision to abstain. Compare the full advised path with calling the fixed exact arm directly.