1. only state prisons are considered
2. only people in prison for a year or more are considered
there's nothing in the comment showing that shorter term inmates are a smaller population, nor that they aren't mostly non violent offenders, nor that federal prisons aren't mostly non violent offenders.
In total, the comment just shows that there are violent and nonviolent felons in prison
I'm the author of the linked comment. I originally focused on state prisons because the article was about events in a state prison, so P(person did something bad|person is currently in state prison) was the relevant statistic. But let's repeat the analysis counting federal prisons and local jails as well.
In general, people with sentences longer than a year go to prison, and sentences shorter than a year go to jail. Roughly 57% of incarcerated people in the US are in state prisons, 9% in federal prisons, and 33% in jails [0]. In my earlier comment I noted that 52% of state prisoners are in for violent crimes, 18% for property crimes, and 18% for drug crimes. For jails, based on this source [1] I estimate 18% of the jail population is in for violent or weapons offenses; 3% for weapons; 30% for property crimes; and 31% for drug crimes. This source [2] suggests around 18% of federal prisoners are in for violent crimes; 17% for weapons/explosives/arson offenses; 11% for property crimes; 46% for drug crimes; and 8% for immigration reasons.
In my original comment, for state prisons, I counted violent crimes, property crimes, and 1/2 of other non-drug crimes to estimate that 76% of the prison population did something genuinely bad. For jails the number will be lower, both because there are fewer violent offenders and because the non-violent offenses tend to be less severe; if we count violent crimes, 1/2 of property crimes, and 1/2 of other non-drug crimes, we get 44%. For federal prisons, if we count violent crimes, property crimes, and weapons/explosives/arson offenses we get 46%. We can weight those three numbers by the sizes of the different groups to arrive at an overall figure of 61% for all incarcerated people in the US. (Of course, these numbers depend heavily on what you consider a "bad" crime.)
So of the entire incarcerated population of the US, about 1/2 to 2/3 are incarcerated for something genuinely bad. While you're correct to point out that this is lower than the 3/4 figure for state prisons, Broken_Hippo is still wrong to claim that "most" prisoners are in for things like non-violent drug offenses.
