The Varieties of Democracy dataset maps every country's political trajectory across hundreds of indicators. Compress those indicators down to two principal components and you get a landscape: each nation traces a path through a two-dimensional regime space over time. Uhlig et al. analyzed these paths and found something that looks less like political science and more like condensed matter physics.
Near the boundary between democratic and autocratic regimes — the critical regime — step sizes and sojourn times follow heavy-tailed distributions. The means diverge. A country near the boundary might take a small step toward democracy, or a huge leap toward autocracy, and both outcomes draw from the same distribution with no characteristic scale. The dynamics are intermittent and heterogeneous: long periods of stasis punctuated by sudden, scale-free transitions.
A continuous time random walk model — the same mathematical framework used for anomalous diffusion in disordered media — reproduces three decades of global regime evolution with remarkable accuracy. The landscape is characterized by sparse, shifting basins of stability. Democracy isn't a fixed attractor. It's a basin that deepens and shallows as the landscape evolves. So is autocracy.
The dynamics are weakly non-ergodic: the system doesn't explore all accessible states uniformly over time. Some countries get trapped in shallow basins; others wander freely. The time a country spends in a particular regime follows the same statistical laws as the time a particle spends in a trap in disordered glass.
The finding doesn't eliminate history — the paper is careful to say political evolution follows universal stochastic principles “while remaining punctuated by unique historical pathways.” The universal part is the statistics: the shapes of the distributions, the critical exponents, the divergent timescales near the boundary. The particular part is which country enters which basin and when. Physics provides the grammar; history writes the sentences.
Democracy isn't an endpoint with backsliding as anomaly. It's a critical phenomenon with heavy tails as structure.