CONCEPT
Self-Organized Criticality
The principle — discovered by Per Bak in 1987 — that complex systems naturally drive themselves toward critical states where small perturbations can trigger cascading events of any size, following power-law distributions without external tuning or design.
Self-organized criticality (SOC) is the tendency of large, complex systems to evolve toward a state where they are maximally sensitive to perturbation — the critical point where a single small cause can trigger consequences ranging from negligible to catastrophic. Unlike traditional critical phenomena in physics, which require careful external tuning of parameters, self-organized critical systems reach criticality through their own internal dynamics. The canonical example is the sandpile: grains dropped one at a time accumulate until the slope reaches the critical angle, at which point the next grain might cause anything from a single-grain shift to a system-wide avalanche. Bak demonstrated that this mechanism explains phenomena as diverse as earthquakes, forest fires, species extinctions, and — as 2020s research confirmed — the training dynamics of artificial neural networks.
In The You On AI Field Guide
The mathematical signature of self-organized criticality is the power-law distribution. In systems at criticality, the frequency of avalanches decreases with their size