Legal Opacity and Protest Enforcement in U.S. Cities

Protest enforcement in the United States varies significantly across jurisdictions, even within cities in the same state, but the role of the legal environment in shaping that variation has been largely overlooked. This paper investigates whether measurable features of local legal text predict policing outcomes at protest events, through two linked analyses. First, we manually code 11 protest-relevant legal variables for California and Texas, and test whether three frontier LLM families (represented by Claude Opus 4.7, GPT-5.5 and GPT-5.4, and Gemini 3.5 Flash and Gemini 3.1 Pro) can replicate our coding under a standardized prompting protocol. LLMs perform well on clearly defined extractive variables but require human verification on ambiguous statutes that span multiple provisions. Second, we merge city-level protest data from the Crowd Counting Consortium with municipal ordinance scores from LOCUS across 179 cities in 38 states, and test whether four legal dimensions (opacity, enforcement discretion, paternalism, and problem salience) predict three enforcement outcomes. Opacity is the only dimension that consistently predicts enforcement, with more opaque cities showing significantly higher police presence at protests (β = +0.033, p = 0.003), a relationship that survives outlier removal and geographic robustness checks. These findings suggest that legal clarity functions as an institutional constraint on enforcement power, and that clarity may matter more than strictness for shaping how government authority is exercised at protests.