The entries in this combined essay record a change in how engineers, doctors and investigators think about failure. The older view looked for a broken part or a careless person. The newer view finds serious accidents arising from ordinary conditions lining up, in systems whose own complexity makes some failures close to inevitable. Most of this group documents that shift.
Failure as a property of the system
The most influential model is Charles Perrow's Normal Accidents, written after Three Mile Island: where a system is both tightly coupled and interactively complex, serious accidents are to be expected however careful the operators. James Reason's Swiss Cheese Model describes how it happens: each layer of defence has holes, and an accident occurs when the holes momentarily align. Diane Vaughan's Normalization of Deviance, drawn from the Challenger disaster, shows how small departures from standard that pass without harm become the new standard.
The engineering entries supply the mechanisms. A Single Point of Failure will eventually set the whole system's reliability. Common-Cause Failure defeats redundancy when backups share a design, a supplier or a location. The Law of Cascading Failure describes small failures spreading through dependent parts, and the Bathtub Curve describes the life of components, failing early from defects and late from wear. Clausewitz's Friction is the military version: countless small difficulties no plan can list.
A newer view of safety
The most recent entries push further. The Root Cause Fallacy holds that naming one cause for a complex failure hides the conditions that made it possible. Safety-II, from Erik Hollnagel, proposes studying why things usually go right rather than only why they occasionally go wrong. High Reliability Organizing, drawn from aircraft carriers and nuclear plants, describes organisations that sustain near-error-free operation through preoccupation with failure and deference to front-line expertise.
How people perceive risk
A third cluster concerns judgement. Dread Risk finds people overreacting to rare, vivid dangers and underreacting to familiar ones. Risk Homeostasis proposes that people offset safety measures with riskier behaviour. Geoffrey Rose's Prevention Paradox explains why public-health measures that save many lives seem pointless to each individual, since most people who take them would have been fine anyway. Taleb's Black Swan and Antifragility argue that rare extreme events dominate outcomes and are explained away as predictable afterwards.
Where it applies
These models reshaped aviation and are reshaping medicine. Checklists, incident reporting without blame, and the study of near misses all follow from the Swiss cheese and normalization-of-deviance accounts. Engineers design against common-cause failure by diversifying backups, not merely multiplying them. Public-health officials meet the prevention paradox every time they explain a vaccination or screening programme, and Kremer's O-Ring Production explains why one weak step in a chain of complementary tasks sets the quality of the whole.
Where it is argued over
The classic dispute is between Perrow's pessimism and the high-reliability school's optimism: whether some systems are bound to fail, or whether the right organisation can run them safely. Heinrich's Triangle, the idea that minor injuries predict serious ones in a fixed ratio, rests on data that no longer exist, and modern safety research finds serious accidents often have different causes from minor ones. Critics of risk homeostasis in its strong form point to the large measured reduction in deaths from seat belts. The Swiss cheese model is itself criticised by Safety-II advocates as too linear.
Why it is worth knowing
The shift this group records is from blaming individuals to understanding systems, and it applies well beyond industrial safety. A failed project, a hospital error or a financial collapse usually has the same shape: several ordinary weaknesses, each tolerated, lining up at once. Knowing that changes the questions an inquiry asks, from who made the mistake to what made the mistake possible, which is where prevention actually lies.
The entries behind this essay: Evidence and Model in Risk & Reliability. Every subject has its essays on the essays page.