Standard models of authoritarianism assume a monotonic “calculus of dissent”: repression raises the expected cost of opposition, inducing more false conformity. This paper challenges that prediction. Using two decades of survey microdata, I construct behavioral indicators of economic preference falsification (EPF) by comparing stated pro-governmental preferences with revealed preferences. Across specifications, support for authoritarianism increases EPF, yet perceived repression sharply reduces it, consistent with psychological reactance rather than fear-driven conformity. Objective repression explains essentially none of the variation once perceptions of repression are controlled for, and high-capacity states face the strongest backlash because of a democratic-accountabiltiy deficit. Quasi-experimental evidence from natural experiments where repressive shocks occurred during survey fieldwork in Burkina Faso, Ethiopia, and Sudan, confirms causality. Heterogeneity analyses, residualization, placebo outcomes, non-response tests, lagged regression, and sensitivity analysis rule out alternative explanations. The results carry implications for aid and reform in fragile states where citizens’ preferences are systematically distorted.

Why do rational authoritarian regimes, equipped with precise digital surveillance tools, choose to empower volatile religious mobs? This paper challenges a critical assumption of digital authoritarianism (DA) literature, namely that regimes invest in technology to raise the material cost of dissent. I argue that regimes are adopting a new strategy—sacralized digital authoritarianism (SDA)—which fuses algorithmic capacity with sacred legitimacy to fundamentally alter the utility calculus of repression. I formalize this logic using a global game model with strategic complementarities, introducing two parameters: sacralization (λ) and digital crowdsourcing (δ). The equilibrium analysis demonstrates that by framing repression as a spiritual duty and providing digital platforms for coordination, regimes can lower the “minimum necessary grievance” (θ) required for incentivizing citizens to repress, bypassing the fiscal and political costs of traditional coercion. I test these predictions using comparative evidence from Turkey, Pakistan, China, India, and Myanmar. I find that SDA regimes successfully create a self-sustaining equilibrium of violence that is more cost-efficient than DA and solves the “problem of authoritarian control” without empowering the military. The findings suggest that the future of authoritarian resilience lies not in the precision of the police state, but in the algorithmic coordination of the faithful.

Standard models of preference falsification (PF) treat conformity as a static optimization problem: individuals instantaneously calculate the costs of nonconformity based on current social pressure. This paper challenges that assumption by formalizing PF as a path-dependent process. Integrating evolutionary dynamics with bounded rationality, I develop an endogenous model where myopic agents build a "stock of falsification capital" under cognitive uncertainty. Driven by adjacent complementarity, each act of misrepresentation lowers the psychological marginal cost of future PF, trapping agents in a self-reinforcing loop of "persona preservation." Rather than assuming forward-looking optimization akin to Beckerian rational addiction models, this framework uses adaptive learning dynamics, showcasing how agents become locked into Pareto-inefficient equilibria of dishonesty even when social pressure subsides. Formal stability analysis using the system’s Jacobian characterizes two distinct regimes: (i) social equilibria sustained purely by psychological hysteresis, and (ii) "cold-turkey" cascades, where exogenous shocks push the falsification stock below a critical threshold, triggering a discontinuous collapse (modeled as a saddle-node bifurcation). Historical cases, including the 1989 Eastern European revolutions and the persistence of authoritarian values in post-Spring Egypt, are shown to fit the model’s comparative statics. The results provide a behavioral micro-foundation for the puzzle of persistent suboptimal equilibria, such as authoritarian resilience and political polarization.