Did you see that study about how children whose criminal fathers who randomly ended up spending more time in prison ended up doing better?
If less than 1% of the population is in prison then presumably those people are close to being the top 1% in terms of being dysfunctional in an antisocial way. Likely a net negative to people around them, whether at home or at work.
"a big reduction in crime would make all that much difference in their lives. I’ve read some people who say they think it would create a utopia! That’s crazy."
In some places it would be the difference between kids playing the streets and families in parks versus being afraid to go outdoors. What do you think about places with rampant public disorder, are they over-incarcerating people? Maybe they just need more policing, but you don't seem to be making that argument.
Besides urban density, if we're being utilitarian perhaps other things to consider is how crime affects people's politics(making progressives look insane?), fertility(too unsafe for kids?) and long-term thinking in general(world seems too unpredictable with random crime?).
People's subjective feeling of safety doesn't seem to be that sensitive to changes in the crime rate. https://news.gallup.com/poll/1603/crime.aspx Most years, the crime rate goes down and people think it's gone up, which seems like it must be a media thing. If we decreased crime by 63% by incarcerating all repeat offenders (not that I think that would be the effect size), there might be just as many headlines about the remaining 37%, plus more headlines about danger from police/wrongful conviction.
On the walking alone at night question there's been a significant decline from the disorderly post 60s era. I'd be more interested in revealed preferences than whatever people say though. If crime is bad enough to cause drastic actions like urban flight then surely it's bad enough to change people's urban behavior.
I think on the civil disorder point, you really should re-read Scott's conclusion. If you want a rule to not be broken, whether it's a tiny civil infraction like littering, or a more serious crime, the critical thing is that punishment must be swift and certain -- not necessarily severe. You want potential offenders to believe they will be caught, and that the consequences will be unpleasant. But they don't have to be _that_ unpleasant! You don't have to lock people up for a year for smoking on the subway. But an agent should be able to initially issue a warning and take down their ID, or impose a slightly higher consequence like booting them out of the subway so they have to pay a new fare, if they're being uncooperative.
The question of whether we should enforce civil order is _almost entirely separate_ from the question of whether we should, at the margin, incarcerate a few more people or release a few people, and from the question of whether sentences should be longer or shorter.
And spending less money on prisons would free up resources to spend more on police, station agents who just have eyes on things, etc. As trains get automated, we could also move money from employing people to sit at the front of the train, to instead patrol up and down the train suppressing petty disorderly conduct.
My basic conclusion is that at our current margins, we should probably be _imposing consequences_ a lot more often, and that it would cost a lot of money to hire and train people who could do that. But that the _vast majority of those consequences_ should not involve prison, and even for actual crimes for which we currently imprison people, we probably could get a lot of the benefits with surveillance, which would cost less both to the public and to the people being punished. At a very back-of-the-envelope level, my suspicion looking at the kinds of numbers the studies in Scott's piece are coming up with is that the money saved by not incarcerating people, and instead having them still in their communities being at least marginally productive citizens, would fund at least _much_, if not all, of the salaries for additional police and other order-enforcement staff.
(I also generally endorse David Brin's argument in The Transparent Society that we should basically just accept that surveillance technology exists, and maximally employ it through the state, with strong oversight and FOIA style ability for citizens to check up on how it's been used. Refusing to use it within the state simply cedes control of it to private actors who will apply it to _worse_ goals than the state would.)
I've argued recently that we can assume crime rates are not that important, based on the fact that urban/dense areas have lower death rates, lower injury rates, equal rates of PTSD diagnosis, and higher crime than rural areas. Whatever the second-order negative effects of crime are, they're smaller than the positive effects of high density.
Put another way, crime and leukemia kill about the same number each year. If you argue, like some neo-reactionaries do, that focusing on murder understates the amount of crime by a factor of 5, that promotes the comparison from leukemia to lung cancer. Since we spend significantly more on incarceration than on cancer research, we can a priori expect a marginal increase in incarceration to be less effective than a new research grant.
An area that seems like it might be particularly tractable is reductions in very long/life sentances - in the USA "life without parole" really does mean "you will be in prison until you die". I am unsure exactly how you'd go about changing this but I think you might be able to make progress on shortening these sentances via pardons/commutions.
Older people are unlikely to reoffend so from a cost benefit analysis this is not a good use of money. I realize people care more about "justice" than utilitarian calculations like that, but I think the average person would feel uneasy about sentancing a man to serve 40 years for a crime he committed 30 years ago. It won't be uncontroversial but I don't think you'd get quite the same opposition as you would for more immediate crimes - perhaps there's something to be said for the UK approach of sentancing murderers to "life" and then letting them out in 15 years time conditional on good behaviour, although there are still "whole life" terms that really do mean exactly that.
Maybe there are some people who "need" to be locked up for the rest of their life for pragmatic or moral reasons, but I feel like the typical case is someone who murdered another person while young and impulsive and is now very aware that that was both immoral and stupid, and I think the average person (read: voter in a democracy) is willing to be forgiving in that situation.
Seemingly your cost-benefit analysis still omits other significant effects that make prison look better? For example: it only takes one report of a rape in an area for large numbers of women to be afraid to go out alone. Instead they may choose different activities - GDP may stay the same, or even rise if they take Uber rather than walking - but their subjective feeling of well-being is significantly impacted.
Did you see that study about how children whose criminal fathers who randomly ended up spending more time in prison ended up doing better?
If less than 1% of the population is in prison then presumably those people are close to being the top 1% in terms of being dysfunctional in an antisocial way. Likely a net negative to people around them, whether at home or at work.
"a big reduction in crime would make all that much difference in their lives. I’ve read some people who say they think it would create a utopia! That’s crazy."
In some places it would be the difference between kids playing the streets and families in parks versus being afraid to go outdoors. What do you think about places with rampant public disorder, are they over-incarcerating people? Maybe they just need more policing, but you don't seem to be making that argument.
Besides urban density, if we're being utilitarian perhaps other things to consider is how crime affects people's politics(making progressives look insane?), fertility(too unsafe for kids?) and long-term thinking in general(world seems too unpredictable with random crime?).
People's subjective feeling of safety doesn't seem to be that sensitive to changes in the crime rate. https://news.gallup.com/poll/1603/crime.aspx Most years, the crime rate goes down and people think it's gone up, which seems like it must be a media thing. If we decreased crime by 63% by incarcerating all repeat offenders (not that I think that would be the effect size), there might be just as many headlines about the remaining 37%, plus more headlines about danger from police/wrongful conviction.
On the walking alone at night question there's been a significant decline from the disorderly post 60s era. I'd be more interested in revealed preferences than whatever people say though. If crime is bad enough to cause drastic actions like urban flight then surely it's bad enough to change people's urban behavior.
I think on the civil disorder point, you really should re-read Scott's conclusion. If you want a rule to not be broken, whether it's a tiny civil infraction like littering, or a more serious crime, the critical thing is that punishment must be swift and certain -- not necessarily severe. You want potential offenders to believe they will be caught, and that the consequences will be unpleasant. But they don't have to be _that_ unpleasant! You don't have to lock people up for a year for smoking on the subway. But an agent should be able to initially issue a warning and take down their ID, or impose a slightly higher consequence like booting them out of the subway so they have to pay a new fare, if they're being uncooperative.
The question of whether we should enforce civil order is _almost entirely separate_ from the question of whether we should, at the margin, incarcerate a few more people or release a few people, and from the question of whether sentences should be longer or shorter.
And spending less money on prisons would free up resources to spend more on police, station agents who just have eyes on things, etc. As trains get automated, we could also move money from employing people to sit at the front of the train, to instead patrol up and down the train suppressing petty disorderly conduct.
My basic conclusion is that at our current margins, we should probably be _imposing consequences_ a lot more often, and that it would cost a lot of money to hire and train people who could do that. But that the _vast majority of those consequences_ should not involve prison, and even for actual crimes for which we currently imprison people, we probably could get a lot of the benefits with surveillance, which would cost less both to the public and to the people being punished. At a very back-of-the-envelope level, my suspicion looking at the kinds of numbers the studies in Scott's piece are coming up with is that the money saved by not incarcerating people, and instead having them still in their communities being at least marginally productive citizens, would fund at least _much_, if not all, of the salaries for additional police and other order-enforcement staff.
(I also generally endorse David Brin's argument in The Transparent Society that we should basically just accept that surveillance technology exists, and maximally employ it through the state, with strong oversight and FOIA style ability for citizens to check up on how it's been used. Refusing to use it within the state simply cedes control of it to private actors who will apply it to _worse_ goals than the state would.)
Thanks for the rigorous analysis!
I've argued recently that we can assume crime rates are not that important, based on the fact that urban/dense areas have lower death rates, lower injury rates, equal rates of PTSD diagnosis, and higher crime than rural areas. Whatever the second-order negative effects of crime are, they're smaller than the positive effects of high density.
Put another way, crime and leukemia kill about the same number each year. If you argue, like some neo-reactionaries do, that focusing on murder understates the amount of crime by a factor of 5, that promotes the comparison from leukemia to lung cancer. Since we spend significantly more on incarceration than on cancer research, we can a priori expect a marginal increase in incarceration to be less effective than a new research grant.
An area that seems like it might be particularly tractable is reductions in very long/life sentances - in the USA "life without parole" really does mean "you will be in prison until you die". I am unsure exactly how you'd go about changing this but I think you might be able to make progress on shortening these sentances via pardons/commutions.
Older people are unlikely to reoffend so from a cost benefit analysis this is not a good use of money. I realize people care more about "justice" than utilitarian calculations like that, but I think the average person would feel uneasy about sentancing a man to serve 40 years for a crime he committed 30 years ago. It won't be uncontroversial but I don't think you'd get quite the same opposition as you would for more immediate crimes - perhaps there's something to be said for the UK approach of sentancing murderers to "life" and then letting them out in 15 years time conditional on good behaviour, although there are still "whole life" terms that really do mean exactly that.
Maybe there are some people who "need" to be locked up for the rest of their life for pragmatic or moral reasons, but I feel like the typical case is someone who murdered another person while young and impulsive and is now very aware that that was both immoral and stupid, and I think the average person (read: voter in a democracy) is willing to be forgiving in that situation.
Seemingly your cost-benefit analysis still omits other significant effects that make prison look better? For example: it only takes one report of a rape in an area for large numbers of women to be afraid to go out alone. Instead they may choose different activities - GDP may stay the same, or even rise if they take Uber rather than walking - but their subjective feeling of well-being is significantly impacted.