The Scottish measures definitely sound sensible, but this sent me off investigating the usual issue with these stories: crime is a wicked problem, and its hard

The Scottish measures definitely sound sensible, but this sent me off investigating the usual issue with these stories: crime is a wicked problem, and its hard to discern "this intervention reduced crime" from "we kept trying different interventions until crime rates happened to fall". From that article's evidence: In 2004/05 there were 137 homicides in Scotland... By 2016/17 the number had more than halved to 62. Last year this had reduced by a further three to 59... This homicide figure was the joint lowest number of recorded homicide cases for a single 12-month period since 1976. So homicides fell from ~2.5/100k to ~1/100k, which is pretty impressive. But 2004/05 is cited because it was a spike 20% above the 2000-2008 average. London homicides also fell by about 50% over that same period before the recent spiek, without equivalent programs, which raises some questions about whether this was simply demographic. And the population pyramid for Glasgow (where 1/3 of the murders were happening) sort of supports that theory; there's a large population boom-and-hollow, for which the boom was 16-20 during the murder spike and is now into the lower-crime late 20s. London has similar demography, slightly older, with a slightly earlier crime drop. So there's some reason to believe the recent murder spike with no population change is qualitatively different from the mid-2000s peak that Scotland and London both saw. On the other hand, that "further three" claim in the article is a statistically meaningless bounce in a number that's been basically flat for since 2013/14. Almost all of the real-looking change is a murder rate that plummeted from 2008 to 2012. That's actually sort of promising: it means the falloff was faster than the decline in the 16-26 population. And the gang-alternative program mentioned launched in 2009, based on a Boston system which was also followed by a sharp decline in murder rate. Both systems work on redirecting a small population with an extremely high murder rate, so a fast response and then a plateau is actually a very plausible outcome. That technique of proactively engaging gang members has been repeated in many cities at different times, and the statistical evidence for it looks surprisingly strong. I initially suspected that the problem was engaging large networks, since Glasgow and Boston had much lower raw murder rates and at-risk populations than London, but Chicago also showed promising results. Tentatively, it looks like this really was an intervention that caused a major improvement. But I can't find any sign of a comparable program being tried in London - does anyone happen to know one? edit: Chicago had 200% London's raw homicide count, but only 25% of its population. Operation Ceasefire theoretically targets a percentage of the population, independent of lethality, so it may still have been a much lower number than London. As far as I can see, none of this ever caught on in NYC.