THE REFERENCE LIBRARY
A growing collection of ideas for understanding the world — and your own thinking.
Broad vs. Narrow Framing describes the human tendency to make decisions in isolation (narrow framing) versus comprehensively (broad framing). Humans are naturally narrow framers due to cognitive biases like WYSIATI, which leads to inconsistent preferences and is mathematically inferior to the broader, rational-agent approach.
A broken leg illustrates a rare event omitted from a statistical model that could justify changing its prediction. The excerpt also considers other possible exceptions, while reporting that clinicians’ discretionary overrides often reduced overall predictive accuracy in the studies discussed.
The broken window fallacy is the illusion that destruction and subsequent spending on recovery are a net benefit to society, ignoring the unseen opportunity costs. It exposes the tendency to focus on visible gains while overlooking invisible losses.
The proposed Bronze Medal Effect links satisfaction with the alternative a competitor imagines: missing gold or missing a medal altogether. The account reports different counterfactual thought ratings among bronze and silver medalists.
The bronze medal effect is the proposed pattern that bronze medalists may feel more satisfied than silver medalists: bronze invites comparison with no medal, while silver invites comparison with gold.
A financial bubble grows when buyers pay more for an asset because they expect others to pay still more. Rising prices attract attention and reinforce optimism; if those expectations reverse, selling can feed on itself. Similar waves of enthusiasm and retreat can occur in fashion and ideas.
Warren Buffett and Charlie Munger exemplify a successful partnership built on mutual respect, contrarianism, and lifelong learning. Buffett's peer-emulation exercise helps identify and internalize desirable qualities while avoiding bad habits.
Buffon's Needle is a geometric probability problem asking for the chance that a needle, thrown randomly onto a floor with parallel lines, crosses a line. Its solution uses additivity of expected value to derive that the probability is 2/π for a needle as long as the line spacing.
A framework for using change management techniques in Root Cause Analysis (RCA) and systems analysis, emphasizing alignment of goal, problem, practical steps, roles, and rewards.
Roca staff describe building relationships with police, courts, schools, and social services to support young people. Hearing how other participants see the system can expose assumptions that keep organizations apart.
Learning communities connect people through shared questions, aims, and meaning. Leaders can help build those relationships as part of the organization’s work.
A framework for developing personal mastery to overcome ingrained cognitive defaults by cultivating self-accountability, self-knowledge, self-control, and self-confidence, and using rituals to create positive inertia that drives consistent good judgment.
Complex societies depend on trust, which is reinforced through feedback loops and iterated interactions like the Prisoner's Dilemma. Law and contracts serve as substitutes for trust or mechanisms to encourage cooperative behavior.
Building, changing, and maintaining systems commonly creates risks. The root problem is the failure to recognize and manage those risks.
The terms 'bullish' and 'bearish' oversimplify market outlooks by ignoring asymmetric outcomes. A situation can be bullish in probability (70% chance of a 1% gain) but bearish in expectation due to a low probability of a large loss (30% chance of a 10% decline), illustrating the importance of considering both probability and magnitude.
Avoid overinvesting in a single option when information is incomplete. Instead, run small, low-risk experiments on multiple options to preserve optionality and gain insight before committing major resources.
The burden of proof fallacy is an informal fallacy where one party in a persuasive dialogue fails to recognize or incorrectly assigns which side has the obligation to provide evidence for their claim.
Buridan's donkey illustrates how randomness can break decision stalemates when two equally attractive options lead to inaction. By introducing a random nudge, an impasse is resolved, allowing a choice to be made.
Treat attention as a limited daily budget. Estimate the demands of important tasks, decide which matter most, and leave room for breaks or changes rather than assuming every planned task will fit.
Displaying successes while hiding failures can make an intervention appear more effective and less risky than it is. Taleb illustrates this selection problem with delayed harms from historical radiation treatments.
This framework addresses the tension between optimistic commercial managers and cautious technical staff when developing business cases. It emphasizes the need for transparent processes, cross-functional collaboration, and respecting all viewpoints to avoid groupthink and make balanced decisions.
A structured decision framework from value investing that applies six sequential filters to evaluate a business: understandability, competitive advantage, management quality, price, disconfirmation, and downside consequences.
Business Intelligence (BI) encompasses technologies, applications, and practices for collecting, integrating, analyzing, and presenting business data to support better decision-making. It focuses on tools for data curation, summarization, visualization, and reporting, enabling organizations to derive insights from data.
A stock-and-flow model of inventory management that demonstrates how delays in perception, response, and delivery can cause oscillations in inventory levels and ordering behavior, illustrating the dynamics of supply chains.
A thinking tool that models a public corporation not as a single legal entity, but as an ecosystem of seven distinct stakeholder groups—shareholders, employees, regulators, customers, creditors, suppliers, and competitors—each with unique constituent interests that influence organizational decision-making and information flows.
Presenting bonuses separately can make an offer seem more attractive than presenting the whole package at once. The sales technique draws on the mental-accounting principle of segregating gains.
Nudging people to increase retirement contributions may seem beneficial, but it is important to assess whether it leads to new savings or merely shifts money from other accounts or increases debt. Studies from Denmark and the military suggest nudging produces mostly new savings without significant debt increase.
Instead of buying mediocre businesses at bargain prices, focus on acquiring high-quality businesses at reasonable prices. This principle, advocated by Charlie Munger, emphasizes long-term value over short-term gains.
Caching keeps likely-to-be-needed items in a small, fast, or nearby store. Least Recently Used (LRU) replaces the item untouched for the longest time, a useful rule when recent use predicts future demand. The chapter applies caching to computer memory, libraries, desks, and content delivery, and explores a related model of human recall.
The calculated risk heuristic is a decision-making approach where one accepts a risk after weighing potential benefits against possible harms, often based on experience or intuition. It is exemplified by Captain Kirk's decision to proceed despite uncertain survival odds.
A smooth curve can be approximated locally by its tangent line. The derivative describes the local rate of change; for position measured over time, it gives instantaneous velocity. Ellenberg uses zooming in on a curved path to explain the idea.
Beware of overconfidence due to the Dunning-Kruger effect; embrace doubt and learn from mistakes to improve competence.
Calibration is a learnable skill for accurately aligning confidence judgments with actual probabilities, and it improves quickly with focused practice in a specific domain, though it only partially transfers to other domains.
Calibration refers to how well subjective probability judgments match objective relative frequencies. Miscalibration can take forms including overprediction, underprediction, overextremity (too radical), and underextremity (too conservative), often driven by heuristics and task difficulty.
In the studies described here, American weather forecasters were well calibrated when predicting rain but tended to overpredict rarer events. Feedback, historical base rates, forecasting tools, and experience are possible explanations; good overall calibration can still conceal errors for particular kinds of weather.
Calibration compares reported uncertainty with observed variation. The Birge ratio compares dispersion across measurements with their reported standard errors. A ratio substantially above 1 can flag underestimated uncertainty or a failure of the model’s other assumptions.
This concept distinguishes genuine bullshit calling—which fundamentally undermines misleading claims—from pedantic, irrelevant corrections (the 'well, actually' guy) that serve only to demonstrate intellectual superiority. It emphasizes relevance, good-faith dialogue, and protecting audiences over self-aggrandizement.
Repeated audits with similar findings can signal that management is relying on auditors to monitor problems it should address. Reviewing the recurring themes and their causes may produce more useful action and free audit resources.
A decision analysis exercise where a college president chooses to abolish or maintain a speech code, evaluating outcomes in campus disruptions, state funding, and enrollment using weighted utility analysis.
Taleb argues that social-science models can cause harm when decision-makers ignore model error. He contrasts them with empirical regularities and with physical theories whose useful domains can be tested precisely.
Cap-and-trade is a market-based approach to controlling emissions by setting a cap on total emissions and allowing trading of permits, creating incentives for cleaner energy and innovation.
Capitalism rewards specialization and efficiency, and persistent execution of a superior system can allow a small player to outperform established giants, as exemplified by Sam Walton's Walmart defeating Sears.
Increasing one input (e.g., tennis courts) can make another input (e.g., instructors) more valuable, and vice versa. This concept emphasizes collaborating with people who have complementary skills rather than similar ones, leading to greater gains.
Card counting is a blackjack strategy where players track the ratio of high to low cards to gain a statistical edge over the house. It demonstrates how small objective insights can disproportionately improve decision outcomes, even when precision is low.
The Card Problem demonstrates confirmation bias: people tend to test a rule by seeking confirming evidence rather than attempting to falsify it.
This exercise asks a manager to compare a San Francisco job offer with staying in Indiana using weighted career and family outcomes. The author reports preferring to stay, but the solution and its utility assignments are absent; the stated probabilities alone do not support that conclusion.
Cargo-cult thinking copies the visible practices of a successful activity without understanding what makes it work. The model asks whether a practice has a relevant function in your setting, rather than whether it resembles what successful people do.
Taleb emphasizes probability as belief about possible alternatives, with mathematics helping examine uncertainty. This perspective encourages revisable estimates and skepticism toward confident explanations; it does not make statistical evidence irrelevant.
Carneades demonstrated the uncertainty of knowledge by arguing convincingly for and against justice on successive days, teaching that nothing can be accepted with certainty and that probabilistic reasoning provides a guide to conduct.
Carroll's Paradox demonstrates that logical inference requires an infinite regress of justifying steps unless we assume a basic rule of inference (such as modus ponens) is valid. It highlights the necessity of accepting meta-logical assumptions to reach any conclusion.
Taleb criticizes fitting ethical arguments to actions after the fact. He proposes stating rules before acting and examining incentives: does an adviser share the downside, or present a private benefit as a public good?
A trigger can explain when a failure occurred without explaining why the system was vulnerable to it. Taleb argues that focusing on the final event can distract from the conditions that allowed a disruption to spread.
A chemical catalyst speeds a reaction by providing a pathway with lower activation energy. Used as a social analogy, a catalyst reduces barriers to a process, whether its consequences are beneficial or harmful.
A catalyst speeds a chemical reaction by providing a pathway with a lower activation-energy barrier. It is regenerated in the catalytic cycle and does not change the reaction’s equilibrium. In organizational change, this offers an analogy for removing barriers, though people and programs do not behave exactly like chemical catalysts.
Aristotle's categorical logic is a system of reasoning based on four types of categorical claims (A, E, I, O) about categories and their members, using syllogisms to draw valid inferences.
A mathematical framework focusing on relationships between objects rather than their isolated properties, revealing hierarchical structures and insights through abstraction.
A base rate can influence judgments more when it suggests a causal story. In Ajzen’s exam experiment, participants used class pass rates more strongly than statistically equivalent rates created by sampling students who had passed or failed.
The conjunction fallacy can also arise from a causal link between two events, where an event A provides a plausible cause for event B, leading people to judge the conjunction A&B as more probable than B alone, even when A is not representative of the broader context.
Causal fallacies are errors in reasoning about cause and effect, often involving incorrect assumptions that correlation implies causation or ignoring confounding variables.
A structured five-step method for defining and analyzing a problem's cause-and-effect system. It involves identifying dynamic factors, mapping their direct or inverse relationships, diagramming them to uncover feedback loops (unstable or self-stabilizing), and analyzing the system's integrated behavior.