While it may not be the principal account for incarceration rates, the paper does note that 17% of the current prison population is in prison on drug-related charges. Additionally, it notes that 20% of the growth in prison populations can be attributed to drug-related charges, and that about 20% of the flow of prisoners can be attributed to drug-related charges.

That 20% number is still a massive number of people incarcerated on drug-related charges. It's still a massive problem and solving it will put a huge dent in the insane incarceration rate that we see.

Of course a problem as large as the over-incarceration of the US population is going to be complicated and is going to require many solutions rather than a single silver bullet. I'm all for prosecutorial reform as well as drug reform; I'm willing to bet I'll need to discover several other things that we need to reform in order to bring our incarceration rates down.

Put another way, if I'm optimizing some code and I see 20% of CPU-time is being spent on a trivial operation that is definitely a section of the code that I'm going to evaluate and attempt to readdress even if it doesn't fully resolve the performance issues.