In US prisons I suppose. Most prisons in the world don't have shankings, or gangs, they're designed to rehabilitate, not punish.
[1] https://worldpopulationreview.com/country-rankings/recidivis...
The source publication concludes "that international comparisons are currently not valid" due to inconsistencies between countries' data and lack of transparency.
https://www.researchgate.net/publication/278714801_A_Systema...
> An estimated 68% of released prisoners were arrested within 3 years, 79% within 6 years, and 83% within 9 years. (https://nij.ojp.gov/topics/articles/measuring-recidivism)
Norway is often claimed to be one of the lowest in the world at ~20% within 5 years.
They're also low population, and quite wealthy from the giant oil fields national income.
If you broke the 330 million in the US into 5 million sized chunks, you likely find well performing outliers like Norway too. That's the problem when assuming a small thing should scale.
This article puts it at 25% (https://www.bbc.com/news/stories-48885846) which admittedly isn't 20, but isn't far off. Wikipedia has it at 20% after 5 years using 2016 data, but I haven't checked their sources on that (https://en.wikipedia.org/wiki/Incarceration_in_Norway)
The wiki claim has three sources listed, none of which state the claim. The first claims 20%, no timeframe (so may well be the same 2 year value stated lots of places), the second sources the 20% in 2 years and a report (which looks to be a decent study, but does not have any claim of 25% at 5 years), and the third again has the 20% at 2 years with a linked report (the report also has that data).
Not one is a very good source, and if the person speaking for an interview were right, you'd think that data would appear all the times the 20% at 2 appears.
Given that, and that the data I cited was from a careful study [1] including lots of research from many countries (including the 20% at 2 years results from Norway), I tend to think the 25% is more or less made up wishful thinking. If it were defensible, it would be cited in other studies that aggregate all this data.
When you find something online, try to determine it's original source. Someone simply saying something is not generally true unless it's from a reputable place that actually measured it.
[1] https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6743246/pdf/wel...