6 ideas
Many systems naturally evolve toward a 'critical' state poised on the edge of instability, where they become organized into a precarious balance without any external tuning. In this state, the addition of a single element can trigger a chain reaction of any conceivable size, because the system has built up internal stresses distributed across all scales.
When grains of sand are dropped one at a time onto a pile, it eventually reaches a critical slope where adding a single grain can cause avalanches ranging from tiny slips to massive collapses. The size of the next avalanche is fundamentally unpredictable because the pile contains 'fingers of instability' of all lengths woven through its structure.
In critical systems, event sizes follow a power law rather than a normal distribution, meaning there is no characteristic or 'typical' size for an event. Doubling the size of an event makes it a fixed number of times less likely, so catastrophes are simply scaled-up versions of small events governed by the same statistical rule.
Large catastrophes do not require large or unusual triggers; the same minor event that usually produces nothing can occasionally produce a disaster because the difference lies in the system's hidden internal configuration, not the trigger. Searching for a proportionally large cause behind a major collapse is therefore often a mistake.
Viewing wars, market crashes, mass extinctions, and revolutions as critical-state phenomena means their timing and magnitude are products of accumulated tension released by accidental triggers, not inevitable forces. The same conditions could have yielded vastly different outcomes, so history is contingent and non-repeatable rather than law-governed in its specifics.
Because critical systems are unpredictable at the level of single events, the useful question shifts from 'when will the next big one occur' to 'how often do events of a given size occur over time.' Understanding the statistical law of a system replaces the futile attempt to forecast specific catastrophes.