INTELLIGENT DEEPENING
Go deeper where it matters.
The answer to missing evidence is not to inspect everything blindly — and not to guess. It is to choose the next bounded observation that can separate the remaining possibilities.
QUESTION
What is known, what is missing, and what observation matters next?
ATLAS retains known/unknown boundaries and can deepen structural, relational, project-shard, information-gain or experimental axes where admitted.
The canonical selection law maximizes information gain per bounded observation cost. This is a reusable strategy, not a universal claim that every domain already has a proven optimal policy.
FAILED CLOSE
If the evidence still does not close, the result stays bounded.
UNKNOWN and FAILED_CLOSE are meaningful results.
Deepening exists to improve qualification, not to force an answer.