An answer benchmark should show which input types were excluded. Scores are easier to reuse when the blind spot is visible.