Deep Rationality (SD)
Deep Rationality is the skill dimension: the quality of the agent inferential tools.
The Skill Axis
Deep Rationality (SD) refers to the agent’s ability to process evidence well. It includes statistical literacy, causal reasoning, Bayesian updating, comparison of hypotheses, base-rate sensitivity, and awareness of bias-producing structures.
What SD Improves
Higher Deep Rationality (SD) narrows the gap between Objective Evidence (E0) and Perceived Evidence (EP). Better tools do not guarantee truth, but they improve the odds that the agent perceived evidence resembles the evidential situation.
The Limit of Skill
Skill alone is not enough. A person may be statistically sophisticated and still use that sophistication defensively. Deep Rationality improves perception; Core Rationality governs whether the agent follows that perception honestly.
Deep Rationality Is Not Intelligence Worship
Deep Rationality (SD) is not a flattering label for being clever. It refers to specific evidence-processing capacities: using base rates, comparing hypotheses, checking source reliability, noticing selection effects, and updating in proportion to diagnostic evidence.
This matters because intelligence can make rationalization more sophisticated. A person with high verbal ability but poor calibration may produce impressive arguments that do not actually track the evidential field. SD should be judged by tracking quality, not by rhetorical shine.