Ambiguous terms
Words that reasonable reviewers can interpret differently, especially when no definition is supplied.
OracleMangle is not a forecasting product. It measures whether an event question is clear enough for automated research workflows, API routing, and repeatable review.
Words that reasonable reviewers can interpret differently, especially when no definition is supplied.
Labels like material, significant, substantial, or clear when the workflow needs a measurable standard.
Whether the question has a single authoritative source or relies on broad, delayed, or conflicting evidence.
Questions that combine multiple requirements without explaining partial completion or deadline edge cases.
Patterns that have repeatedly produced contested outcomes across the labelled event-question corpus.
A practical signal for whether to continue automation, log the question, rewrite it, or queue manual review.