Calibration
Calibration is the fit between confidence levels and actual reliability.
What Calibration Means
A person is well calibrated when their confidence levels match reality over time. Claims held at 70 percent confidence should be true roughly 70 percent of the time. Calibration does not mean always being right; it means knowing how right one tends to be.
Overconfidence and Underconfidence
Overconfidence assigns more certainty than the evidence warrants. Underconfidence assigns less. Both are failures of proportion. Overconfidence tends to produce dogmatism; underconfidence can produce paralysis, false humility, or chronic deference.
Calibration in the Model
Calibration supports both Deep and Core Rationality. Deep Rationality improves the accuracy of perceived evidence. Core Rationality keeps assigned credence from inflating or deflating for non-evidential reasons.
Calibration in Practice
Calibration is easiest to understand through repeated predictions. If you mark ten claims as 80 percent likely, about eight should come true over time. The point is not that any single claim must be right; the point is that your confidence language should have a reliable track record.
This matters because overconfidence and underconfidence both harm inquiry. Overconfidence turns weak evidence into premature closure. Underconfidence refuses to acknowledge strong support. Good calibration lets confidence become both courageous and corrigible.
Adjacent References
These are starting points rather than exhaustive citations. They connect this page to nearby work in probability, calibration, cognitive bias, and social epistemology.
- |Daniel Kahneman and Amos Tversky, "Judgment under Uncertainty: Heuristics and Biases" (1974).
- |Sarah Lichtenstein, Baruch Fischhoff, and Lawrence D. Phillips, "Calibration of Probabilities" (1982).