Warranted Uncertainty
Warranted uncertainty is the range of doubt an agent is entitled to given their evidence and rational skill.
Not All Spread Is Failure
The model allows a range of reasonable credence around perceived evidence. A person with weaker tools or noisier evidence may be warranted in a wider range of uncertainty. Rationality is not a demand for artificial sharpness.
The Omega Range
In the interactive chart, omega represents warranted uncertainty as a visual angle around Perceived Evidence (EP). As Deep Rationality increases, the warranted spread narrows: better tools reduce fog. As Deep Rationality decreases, the range widens: the agent should be less precise because their optics are less reliable.
The dashboard expresses the same idea as warranted slack: a probability-scale allowance before a raw Perceived Evidence to Assigned Credence gap (EP-CA) is treated as excess misalignment.
Slack Before Blame
Warranted uncertainty also creates diagnostic slack. If Assigned Credence (CA) differs modestly from Perceived Evidence (EP), but the case is noisy and the agent has limited skill, the model should not immediately treat the gap as excess misalignment.
The improved model therefore distinguishes Core Irrationality (the raw EP-CA distance) from Excess Core Irrationality, the part of that distance that remains after warranted uncertainty is allowed.
When Uncertainty Becomes Cover
Uncertainty becomes suspect when the agent uses it selectively. If uncertainty is invoked only to resist unwelcome evidence, but disappears when the evidence supports a favored conclusion, it is no longer humility. It is core-rational evasion.