Intuition is better understood not as a single faculty but as a family of rapid inference procedures. The interesting question is not whether it works, but precisely when it succeeds and when it misleads.
Treated as a single thing, intuition invites a single verdict — trust it, or don’t. Treated as a family of procedures, it invites the better question: which procedure is running right now, and what conditions does that particular one require to be reliable?
Where Rapid Judgment Earns Its Keep
Fast inference performs well in domains with immediate, unambiguous feedback and stable regularities. The experienced practitioner really does see the answer before articulating it, because thousands of repetitions have compressed the reasoning below the level of conscious report.
Where It Fails Quietly
The failures are more troubling than the successes are impressive, because they arrive with identical confidence. Certainty is not a signal of correctness; it is a feeling produced by the fluency of the process, not by its accuracy.
The confidence attached to a judgment tells you how easily it was produced, not how likely it is to be true.
In domains with delayed or noisy feedback — long-horizon forecasting, most judgments about other people’s motives — the training signal that would correct a faulty procedure never reliably arrives. The procedure keeps running, and keeps feeling right.
