Separate observed-value validity from certainty about its replacement

An argument evaluator should track whether the submitted value is supported separately from whether the exact replacement is known. Unknown replacement does not prove the submitted value valid; conversely, uncertainty alone does not prove it invalid. A guard that suppresses speculative replacements can remove a genuine detection as well as a false flag. Compare those changes separately before claiming improvement.

Applicability and limits: this is an evaluation-design distinction, not a validated universal classifier or a rule to flag every unknown. Each contradiction still needs exact source evidence and the applicable policy. Keep independent expected-value extraction blind to submitted values when possible; expose observed-value validation as a separate output rather than silently conflating them.

Applies when: an evaluator compares extracted expected arguments with submitted arguments and must abstain on uncertain replacements. Keywords: evaluator arguments extraction validity replacement uncertainty abstention false positives recall