Race And Justice: Much More Than You Wanted To Know

The exhaustive, even-handed review of racial bias in the US criminal-justice system, tier by tier: police stops (some bias, varies by jurisdiction), arrests for violent crime (match victimisation surveys - little bias), drug arrests (roughly 2x after confounders, possibly less), police shootings (no bias vs. the dangerous-encounter baseline in NYC; clear bias in 1970 Memphis), prosecution/conviction (slightly favour blacks), and sentencing (~10-20% harsher, with capital punishment clearly biased). Concludes the bias is real but limited and mostly detectable only in aggregate, rejecting both "no bias" and "systemically racist by design", while noting heavy disparate impact via class.

Quality 80 · Excellent. A model evidence-synthesis on a fraught topic - careful, numerate, fair, and willing to contradict both sides. Among his strongest MMTYWTK entries. Excellent.

Claude’s paradigm shift 32 · Slight. A synthesis/review of existing studies rather than a new concept, though the even-handed framing was valuable. Slight.

Real-world impact 2 · Minor. A model, numerate evidence-synthesis on a highly material and fraught topic (racial bias across the criminal-justice system), willing to contradict both sides. Real topical relevance, but its contribution is to understanding rather than any change in policy or practice — low RWI.