Knowledge · Decisions
Evidence, Uncertainty, and Belief
What should you believe—and how strongly?
What Would Count as Evidence?
Separate claims, observations, assumptions, and interpretations, then learn why evidence gains force by discriminating among alternatives.
Belief Comes in Degrees
Use base rates, diagnostic evidence, and Bayesian updating without pretending that every uncertainty deserves a precise number.
Measurement Shapes the Fact We See
Examine definitions, instruments, samples, missingness, and changing methods before treating a statistic as a fact about the world.
Causes Need Counterfactuals
Move from sequence and association to counterfactual comparison, causal mechanisms, credible designs, and limits on transportability.
One Study Is a Fragile Object
Reconstruct a study’s design, result, and interpretation, then identify the limitation most capable of changing its conclusion.
A Body of Evidence Can Still Mislead
Judge search, independence, replication, heterogeneity, and publication processes before treating a literature as cumulative evidence.
Models Forecast by Leaving Things Out
Inspect assumptions, validation, calibration, and scenario design before treating a model output as a fact or forecast.
Knowledge Depends on Other People
Examine testimony, credentials, consensus, disagreement, incentives, and correction in a world of unavoidable epistemic dependence.
Inquiry Under Identity and Information Pressure
Examine motivated selection, repetition, misinformation, and social incentives, then design a revision process that does not depend on heroic neutrality.
Decide, Then Stay Revisable
Separate confidence from decision thresholds, expose value judgments and asymmetric losses, and finish with an action that can learn and change.