Source: Astral Codex Ten | Author: Scott Alexander | Published: 2026-02-19

Summary

Scott Alexander examines the disconnect between public perception that crime is rampant and actual crime statistics, which show historically low rates. His thesis: people are actually concerned about disorder — litter, graffiti, homelessness, shoplifting, tent encampments — but use “crime” as a proxy because complaining about disorder directly risks social criticism. He systematically reviews data on each form of disorder, finding that most metrics are either improving long-term or only modestly above recent lows (e.g., shoplifting ~33% above 2005 lows, homelessness ~25% above generational lows).

Key Insights

The most interesting move is the three-way framing of competing explanations: (1) a recent 2020-era bump in disorder that hasn’t reversed despite crime declining, (2) the impossibility of pre-1950 disorder comparisons since the material prerequisites (plastic, spray paint) didn’t exist, and (3) the mid-20th century as an anomalous “golden age” of low disorder that created unrealistic baselines. The data quality problem is itself revealing — we have decent crime statistics but remarkably poor disorder statistics, which Alexander argues is partly because disorder is harder to measure and partly because there’s less institutional incentive to track it. The shoplifting section is particularly good at separating signal from noise, cutting through retail industry claims of a “crisis” by showing the FBI data tells a much more modest story.

Our Take

This is Alexander at his most useful — taking a widely-held intuition (“everything is getting worse”), steelmanning it, and then subjecting it to actual data. The proxy argument is compelling: it explains why crime policy debates feel so disconnected from crime data, and why “tough on crime” rhetoric resonates even when crime is falling. The main limitation is that disorder does plausibly have nonlinear effects on quality of life that aggregate statistics miss — a single tent encampment near your commute route affects your perception more than a citywide average would predict. Alexander acknowledges this but doesn’t fully engage with it. Still, the piece is a valuable corrective to catastrophist narratives and a reminder that vibes-based assessment of social trends is unreliable.