The go-to example is Portugal, where drugs were decriminalized, and death rates went down. Not without challenges though, as they are now seeing issues with public drug taking and associated crime (which could be more about pandemic homelessness since we see that in US cities too). I think the harm reduction results are clear here but you also need to do the part where you invest more broadly in social programs (and with sufficient funding for homelessness services, taking a zero tolerance approach to public intoxication seems perfectly acceptable to me). See https://www.cato.org/white-paper/drug-decriminalization-port... on harm.

There is lots of evidence supporting needle exchanges as good for harm reduction. See Scotland, Netherlands, and many other EU countries.

Another one to watch is Oregon, where they did the decriminalization bit, but I think without the shift in funding to social services. Haven’t dug much here. (I think “reduce policing, increase social services” is the most sympathetic interpretation of what the moderate Democrats were thinking when they initially supported the “defund the police” slogan, which is now associated with more radical policies that are toxic to the center.)

There is a risk of “no true Scotsman” in the following point, but this is really an extremely radical policy proposal, at least it looked so until a couple years ago; there aren’t actually good examples end-to-end. So I think we need to employ epistemic humility and do local iterative experiments to prove it out (the advantage of the federal model in the US is a federal decriminalization would result in 50 natural experiments as states passed their own laws.) I don’t want to concede too much uncertainty though, the harms of criminalization are objectively very high, so you’d need to measure very high costs from the proposed new policy which I’m very skeptical that you’d see.