Source: Astral Codex Ten | Author: Scott Alexander | Published: 2026-02-18

Summary

Alexander argues that the US has achieved historically low crime rates — potentially the lowest murder rate in 250 years — and that this decline is genuine, not an artifact of underreporting or better emergency medicine. He marshals three lines of evidence against the reporting-bias explanation: the National Crime Victimization Survey (240,000 people asked directly about victimization) tracks police data closely; murder, which is almost always reported, has declined at comparable rates to other crimes; and car theft, which is reliably reported due to insurance requirements, shows the same downward trend. Against the medical-care explanation, he cites recent research showing gunshot injuries are becoming more severe (more wounds per victim, higher injury severity scores), roughly offsetting improvements in trauma care and keeping lethality rates stable.

Key Insights

The strongest move is the triangulation strategy — showing that three independently motivated data sources (victim surveys, homicide counts, insurance-driven theft reports) all converge on the same conclusion. Any single source can be attacked, but explaining away all three simultaneously requires increasingly contrived theories. The medical care rebuttal is also well-targeted: the earlier “declining lethality” literature had become a popular counternarrative, and the newer evidence on increasing injury severity effectively neutralizes it. The 250-year historical framing, drawing on Randolph Roth’s work, usefully puts the current moment in perspective — this isn’t just a return to some recent baseline.

Our Take

This is classic Alexander: taking a contrarian-seeming claim (“crime is actually at historic lows”) that is in fact the mainstream empirical finding, and defending it against the real contrarians who deny the data. The piece is thorough on debunking but lighter on explanation — why crime has declined so dramatically remains one of the great unsolved puzzles in social science, and Alexander wisely doesn’t overcommit to any single causal story. The one gap is international comparison: are other developed countries seeing similar declines, and if so, does that constrain which US-specific explanations (mass incarceration, lead abatement, policing changes) are plausible? Still, as a “establishing the facts before theorizing” exercise, this is exactly the right approach.