Self-Organized Criticality

Definition

Per Bak's sandpile model shows that a system fed energy or material slowly and steadily — grains of sand dropped one at a time onto a growing pile — drives itself, without any external tuning, to a critical state balanced on the edge of collapse. In that state, adding one more grain can trigger nothing, a small local slip, or an avalanche that reorganizes the whole pile — and the sizes of these avalanches follow a power law: many small events, exponentially fewer large ones, with no characteristic "typical" size. The system never resets to zero; it maintains itself perpetually near the critical threshold.

In the Book

Chapter 4 introduces Bak's sandpile model and the simplified "block-stacking" version anyone can run on a grid by hand: pile blocks on squares, and once a stack reaches four it topples, redistributing blocks to neighboring squares in a chain reaction. Chapter 5 grounds this empirically in real earthquakes: Beno Gutenberg and Charles Richter found that plotting the logarithm of earthquake frequency against magnitude gives a straight line — the Gutenberg-Richter law, a textbook power law where a magnitude-8 quake is roughly twenty billion times more energetic than a magnitude-1, yet the same simple law spans that entire range. Bak and Chao Tang built a computer model of the San Andreas Fault showing why: unlike the traditional picture of strain building up and releasing all at once, a fault in a self-organized critical state undergoes slips "on all scales," never fully resetting. The chapter extends the same pattern to the fossil record's mass extinctions (the Big Five, including the K-T event) and to experiments with real long-grain rice piles at the University of Oslo, which reproduce the power law and — examined edge-on — turn out to be fractal.

Why It Matters

This concept explains why catastrophic events in many systems (earthquakes, extinctions, financial crashes, cascading failures) are not rare anomalies needing a special large cause — they are the same mechanism as the small, routine events, just further out on the same power-law tail. It tells you that a system fed a steady trickle of stress will find its own critical point without anyone tuning it there, and that "it's been quiet for a while" is not evidence the system is safe — quiet periods and the next avalanche are generated by the identical underlying process.