The critical angle of repose is the specific slope at which a sandpile reaches self-organized criticality. Below this angle, the pile is subcritical — grains added to the pile settle into stable positions, and avalanches are small and localized. Above this angle, the pile is overcritical — it cannot maintain its structure and immediately collapses. At the critical angle, the pile exhibits a paradoxical property: it is simultaneously stable (the configuration can persist indefinitely) and unstable (the next perturbation can trigger a reorganization of any size). This is not a designed state but an attractor — the pile's own dynamics drive it toward this angle and maintain it there. The critical angle is where the correlation length diverges, where power-law distributions emerge, and where the system achieves maximum sensitivity to perturbation.
The critical angle of repose for actual sand depends on grain shape, surface friction, and moisture content, typically ranging between 30 and 40 degrees from horizontal. But Bak's interest was never in the specific number. What mattered was that such an angle exists — a threshold below which the pile accumulates grains and above which it cannot sustain itself. The system self-organizes toward this threshold not through any external controller measuring and adjusting the angle, but through the grains' own settling dynamics. When the pile is subcritical, grains accumulate faster than they avalanche. The slope steepens. When the pile reaches the critical angle, grains accumulate at exactly the rate they avalanche. The pile has found its attractor.
The AI transition, mapped onto this framework, reveals the technology industry as a sandpile that spent decades approaching its critical angle. Each innovation that reduced the imagination-to-artifact ratio — each compiler, framework, cloud service, abstraction layer — was a grain steepening the slope. The slope increased imperceptibly because each grain was small. The approach to criticality was invisible because it happened gradually. But the accumulation was relentless, driven by the same competitive and creative dynamics that produce all technological change. By December 2025, the pile was at its critical angle. The natural-language interface was not uniquely destabilizing; it was the grain that happened to land on a pile already poised for reorganization.
The critical angle explains the simultaneity of response across the global developer community. If the pile had been subcritical, Claude Code's arrival would have produced local effects — some teams adopting early, others waiting, adoption curves following Rogers's standard diffusion pattern over years. Because the pile was critical, the response was immediate and correlated across the system. Developers on different continents, in different industries, with different backgrounds simultaneously recognized the phase transition — not because they were communicating but because they were grains in the same critical pile, and the pile's global state had made them sensitive to the same perturbation. The correlation length had diverged. The system was one thing.
The angle of repose has been studied in engineering contexts since the 19th century — important for silo design, grain storage, and mining operations. Bak's innovation was recognizing that the critical angle was not merely a material property but a dynamical attractor. The pile drives itself toward this angle through its own avalanche dynamics: if the slope is too steep, avalanches occur and flatten it; if too shallow, grain accumulation steepens it. The critical angle is the fixed point where accumulation and avalanching balance. This self-regulation toward criticality, operating without external control, was the key insight that generalized from sandpiles to all self-organized critical systems.
Attractor, not accident. The critical angle is where the system's dynamics converge and stabilize — the pile drives itself there and remains there through purely local interactions.
Simultaneous stability and instability. At the critical angle, the pile can persist indefinitely (stable) while remaining maximally sensitive to perturbation (unstable) — a paradox resolved by recognizing that stability and sensitivity measure different properties.
Divergence of correlation length. At criticality, perturbations can propagate across the entire pile — grains that were independent below the threshold become correlated at it.
No external tuning required. Unlike classical critical phenomena requiring careful control of temperature or pressure, the sandpile finds its critical angle autonomously through avalanche dynamics.
Threshold for phase transition. The critical angle marks the boundary between the subcritical regime (local perturbations, predictable behavior) and the critical regime (non-local effects, power-law statistics, fundamental unpredictability).