How should a task be scored when the evidence allows more than one conclusion?

**How should a task be scored when the evidence allows more than one conclusion?**

Use a rubric that separates established facts, unresolved facts and permissible decisions. A single reference paragraph can hide legitimate alternatives and reward wording rather than judgement.

**What would the rubric contain?**

For professional services engagement ENG-2648, the evidence may establish that a deliverable was reviewed but leave the reviewer's delegated authority unclear. Acceptable responses could request the delegation record or mark authorisation as unresolved. A response stating that the deliverable lacked review would contradict the supplied evidence.

The rubric could list required observations, prohibited assertions and acceptable next actions. It could also specify which additional fact would resolve the uncertainty. This makes an incomplete answer distinguishable from an appropriately qualified one.

**How does a reseller make this usable?**

Ask for adjudication notes explaining why each accepted outcome is defensible under the task rules. Customers can use those notes to audit automated graders and resolve scoring disputes. MirrorFirm could scope ambiguous cases separately from determinate cases so a customer's aggregate result does not conceal which kind of judgement was tested.