In 1971, the economist Thomas Schelling — later a 2005 Nobel laureate — published Dynamic Models of Segregation, a model of red and blue tiles each mildly preferring same-colour neighbours (satisfied if at least 30% match) and relocating when unsatisfied. Running it by hand on a chessboard with pennies and dimes, Schelling found that very mild preferences produced strong aggregate segregation, with two stable regimes (integrated and segregated) and a sharp transition between them. Tipping points and regime shifts — the phenomenon where a stable-looking system suddenly flips to a qualitatively different state once a control parameter crosses a threshold — became the framework's vocabulary, popularized by Malcolm Gladwell in The Tipping Point (2000) and now standard for talking about social, ecological, and economic transitions.
Tipping points arise when positive feedback dominates negative feedback in a system's dynamics: the system has multiple locally stable states, the boundary (the separatrix in dynamical-systems language) between them is unstable, and a control parameter — temperature, population, neighbourhood composition, asset price — moves the system through state-space until it crosses the boundary and the system flips. Hysteresis is the most consequential feature: once flipped, the system does not flip back at the same threshold, and the control parameter has to move further in the reverse direction than it did to cause the flip — a eutrophic lake driven into algal dominance by nutrient loading does not return to clarity when nutrients drop back to the original level. Three other signatures recur — sharp transitions concentrated near the threshold while the system looks stable far from it, slow erosion of resilience before the flip (with rising autocorrelation, increasing variance, and critical slowing down — Scheffer et al. 2009 — as early-warning signals), and cascading effects whereby tipping in one part of the system can drive tipping in adjacent parts. The canonical examples reach across domains: lake eutrophication, coral-reef collapse to algal dominance, and savanna-to-desert transitions in ecology; Schelling segregation, neighbourhood gentrification, and Kuran-style social-norm cascades like the 1989 Eastern Bloc collapse and the 2010–2011 Arab Spring; the Greenland and West-Antarctic ice sheets, permafrost methane release, and Amazon dieback in climate science (the tipping-elements literature of Lenton et al. 2008 identified nine such Earth-system components, a count a 2022 update by Armstrong McKay et al. raised to about sixteen); financial bubbles, bank runs, and currency crises in economics. Standard objections are sharp — tipping points are easier to identify in retrospect than to predict in advance, the early-warning-signal literature has real failures alongside its successes, and the metaphor is over-applied to phenomena that are actually smooth diffusion processes (the 80/20 rule, the innovation-adoption curve) rather than genuine multiple-state regimes. The honest version of the framework is more useful for being more constrained.
Climate science has made tipping points a central concept of the present century — IPCC AR6 identifies several elements with potentially irreversible dynamics under plausible warming, and the insurance industry and central banks (the Network for Greening the Financial System) increasingly model tipping risk in financial-stability assessments. The vocabulary now spreads further: vaccination herd-immunity is a tipping-point dynamic; viral content in social media follows tipping statistics; preference falsification (Kuran 1995) explains the fast 2010s reversal of public opinion on same-sex marriage; the AI-safety concern about capability surprise — that systems will cross thresholds suddenly rather than gradually — is structurally a tipping question, with mixed empirical support (Schaeffer et al. 2023 argue some apparent emergence is a metric artifact). When you see a system that looks stable, the most useful question is often what slow variables are eroding the basin of attraction and how close the boundary is.