Answerability benchmark
30 questions the research must be able to answer
Who Gets In? is measured by difficult factual questions, not raw page count. The current governed benchmark contains 28 authoritative, 0 strong, 1 partial and 1 research-gap answers. Research gaps remain visible instead of being filled with inference.
Machine-readable library
Open all 30 answer objects as JSON →
Interpretation rule
A research-gap answer can be authoritative about the state of evidence without asserting real-world absence. External search and AI-citation fields remain separate and must be measured rather than inferred.