Overcoming Your Biases

The empirical foundation here is the Good Judgment Project led by Philip Tetlock at the University of Pennsylvania, which spent years systematically measuring forecasting accuracy across thousands of geopolitical questions. Tetlock's research identified the practices that distinguish 'superforecasters' from average forecasters, and almost none of them are about intelligence or domain expertise. Superforecasters break problems down into smaller estimable pieces (Fermi estimation), seek the outside view by finding base rates from comparable past events (reference class forecasting, developed by Daniel Kahneman and Bent Flyvbjerg for megaproject planning), update beliefs in small increments as new evidence arrives (Bayesian updating), and actively look for disconfirming evidence (the consider-the-opposite technique demonstrated by Lord, Lepper, and Preston in 1984). Each of these techniques exists because System 1 cognition produces predictably biased outputs that System 2 cannot reliably intercept without procedural scaffolding.

The structural-intervention paradigm has been validated across domains. Gary Klein's pre-mortem, in which a team imagines a future failure and reasons backward to its causes, produces about a 30% increase in identified risks compared to conventional brainstorming because it exploits prospective hindsight — the brain finds it easier to generate explanations for events presented as having already happened. Atul Gawande's 'Checklist Manifesto' documents how aviation-style checklists reduced surgical mortality by 47% in a WHO-led trial, not by adding new knowledge but by ensuring known steps were not skipped under cognitive load. Decision journals, championed by Annie Duke and Daniel Kahneman, fight hindsight bias by recording predictions and reasoning before outcomes are known, providing an honest record to compare against later. And the bias blind spot, identified by Emily Pronin, predicts a sobering ceiling on introspection-based debiasing: people who can flawlessly identify biases in others remain blind to the same biases in their own real-time thinking.

The practical synthesis is that bias mitigation must be architectural. Build structured decision processes that you must follow whether or not you currently feel biased; recruit external checks (red teams, peer review, devil's advocates) that operate independently of your introspection; precommit to evaluation criteria before you encounter the evidence; and use base rates as the starting point for every forecast, requiring strong specific evidence to deviate from them. As you work the scenarios, notice that the wrong-answer options often describe willpower-based or knowledge-based debiasing strategies that research has consistently shown do not work. The right answers describe procedural interventions that work because they do not depend on you noticing the bias in the moment. For broader treatment, see Metacognition on the discipline of thinking about your own thinking, and Cognitive Biases: Memory & Self on biases that distort self-knowledge specifically.