This weekly roundup thread is intended for all culture war posts. 'Culture war' is vaguely defined, but it basically means controversial issues that fall along set tribal lines. Arguments over culture war issues generate a lot of heat and little light, and few deeply entrenched people ever change their minds. This thread is for voicing opinions and analyzing the state of the discussion while trying to optimize for light over heat.
Optimistically, we think that engaging with people you disagree with is worth your time, and so is being nice! Pessimistically, there are many dynamics that can lead discussions on Culture War topics to become unproductive. There's a human tendency to divide along tribal lines, praising your ingroup and vilifying your outgroup - and if you think you find it easy to criticize your ingroup, then it may be that your outgroup is not who you think it is. Extremists with opposing positions can feed off each other, highlighting each other's worst points to justify their own angry rhetoric, which becomes in turn a new example of bad behavior for the other side to highlight.
We would like to avoid these negative dynamics. Accordingly, we ask that you do not use this thread for waging the Culture War. Examples of waging the Culture War:
-
Shaming.
-
Attempting to 'build consensus' or enforce ideological conformity.
-
Making sweeping generalizations to vilify a group you dislike.
-
Recruiting for a cause.
-
Posting links that could be summarized as 'Boo outgroup!' Basically, if your content is 'Can you believe what Those People did this week?' then you should either refrain from posting, or do some very patient work to contextualize and/or steel-man the relevant viewpoint.
In general, you should argue to understand, not to win. This thread is not territory to be claimed by one group or another; indeed, the aim is to have many different viewpoints represented here. Thus, we also ask that you follow some guidelines:
-
Speak plainly. Avoid sarcasm and mockery. When disagreeing with someone, state your objections explicitly.
-
Be as precise and charitable as you can. Don't paraphrase unflatteringly.
-
Don't imply that someone said something they did not say, even if you think it follows from what they said.
-
Write like everyone is reading and you want them to be included in the discussion.
On an ad hoc basis, the mods will try to compile a list of the best posts/comments from the previous week, posted in Quality Contribution threads and archived at /r/TheThread. You may nominate a comment for this list by clicking on 'report' at the bottom of the post and typing 'Actually a quality contribution' as the report reason.
Jump in the discussion.
No email address required.
Notes -
Here's a little bit of incomplete thinking about the classic "13/53" number, which is a ballpark figure (varying year to year) that represents the fact that black people are overrepresented by a factor of about 5x in crime. I see a lot of people tend to interpret this number as "black people are 5x more likely to commit crimes", but that might not actually be the case.
Concretely, there's two ways this stat could come about:
a. There are 5x as many black criminals per capita and each black criminal commits crimes at 1x the rate of white criminals.
b. There are 1x as many black criminals per capita and each black criminal commits crimes at 5x the rate of white criminals.
There is of course a continuum between them, but I think it's useful to focus on the two endpoints because the endpoints have totally different policy responses and also suggest totally different causes.
For example, the policy response to (a) is that we need more police to catch a lot more black criminals. The policy response to (b) is that we need longer prison sentences for the criminals we have in order to prevent the same guy from doing 4 more crimes.
They also suggest different causes. Scenario (a) suggests something (HBD, special kinds of poverty not reflected in census stats) causes blacks to have a higher criminal propensity, whereas (b) suggests police might just be extra lenient towards black criminals thereby giving them more time on the street in which they commit more crimes.
Interestingly, while the theory of police abandonment will get you cancelled today, it was very much the theory pushed by black community leaders in the 90's. It was one of the things leading to "3 strikes" laws (long prison sentences for the 3'rd crime in order to get rid of the very worst criminals).
I have recently discovered some weak evidence in favor of theory (b) while going down an internet rabbithole on a totally different topic. Specifically, look at the first graph in this analysis:
https://github.com/propublica/compas-analysis/blob/master/Compas%20Analysis.ipynb
The "decile score" of the x-axis is a reasonably predictive index of a convicted criminal committing new crimes. The dominant features in the model generating the index are things like "# of previous crimes", "was the current crime violent", etc. As can be seen from the graph, white criminals are overrepresented on the left tail (little repeat crime risk) of the graph, whereas black criminals are spread evenly. Of course, this evidence is very weak - it's only about criminals up for parole in a certain region of Florida.
Does anyone know of more data on this?
I've posted about this before, but this doesn't actually follow. The US already has very long prison sentences, and a lot of prisoners. Our justice is already quite punitive; the problem is that we don't actually arrest anyone for most crimes. Generally having a highly certain and rapid pipeline from crime -> arrest -> trial -> punishment is a stronger deterrent than just a longer sentence. The policy response should depend on the reason why 1 person is doing a lot of crimes (and I believe that crime does roughly follow an 80/20 rule, with a few people having a rap sheet many pages long). Are they doing a lot of crimes because we find them and let them off easy, or because we never find them in the first place? If the latter, then the solution to b) is also to hire more/more competent police (and more lawyers and judges to speed up the legal system).
I'm somewhat skeptical about looking at the data this way. I think you could also interpret the data as saying, "whether black criminals recidivate is less predictable based on the observables that are included in the score." If you scroll down a bit, you'll see results for the "violent recidivism" score, which shows a consistent decrease for both groups, although it is more extreme for whites.
It would probably be much easier to just look up recidivism rates directly. This paper finds a recidivism rate of about 44.8% for whites and 50.6% for blacks, which isn't an enormous difference given the wide range of other factors that probably affect who ends up in prison in the first place or the other causes of recidvism (e.g. white ex-prisoners have an employment rate of about 66% to 60% for blacks, which is an almost identical difference to the difference in recidivism. Of course, causality could run the other way--if you are in prison, you aren't employed). It's also worth noting that recidivism is pretty high in both cases.
Anyway, I haven't read the whole thing in detail, but the abstract does claim that education level and post-release employment are the most important predictors of recidivism. And black prisoners tend to have lower education than white ones (they're also slightly more male and young, which are also predictors of criminal activity in general), although the differences in all 3 cases aren't enormous.
I do some back of the envelope arithmetic here, based solely on example numbers involving many crimes/criminal, and get totally different results: https://www.themotte.org/post/329/culture-war-roundup-for-the-week/58654?context=8#context
Can you be quantitative about what, specifically, you disagree with in that analysis?
We may have long prison sentences, but violent criminals do not spend much time in jail.
If as you say they have a "rap sheet many pages long", that means we did in fact let them off too easy the first time. How could a crime be on their rap sheet if we didn't find them earlier?
The paper you link is entirely unrelated to the question I'm asking, namely "how fat tailed the distribution of number of crimes per criminal?"
At any point, did you google to see if there is any empirical research on what deters criminals? This is an empirical question. All you have done is model the direct effect of incarceration, without accounting for whether the threat of punishment (or the memory of past punishment) might prevent a crime from taking place to begin with. In fact, you explicitly assume these effects to be 0. If you had searched around, you might have found something like https://sci-hub.ru/https://www.journals.uchicago.edu/doi/epdf/10.1086/670398 which summarizes a bunch of research, and which says in the first sentence of the abstract:
If you tweak the parameters of your napkin math, how much does the conclusion change? What if you assume that a criminals' chance of being caught is not independent for each crime (i.e. some criminals are better at getting caught)?
A rap sheet includes arrests, which doesn't mean there was enough evidence to convict. If you get arrested for 20 different crimes in 2 years, it's likely you're guilty of at least some of them, but you may not be able to be convicted of any one of them.
The distribution is probably similar for crimes where no arrest is made, which is most of them. Smarter serial criminals may even be able to avoid getting arrested entirely, which of course only makes it easier for them to commit even more crimes.
The thing you linked to is primarily about recidivism probability. What is its relevance in this context? And why is it evidence for your hypothesis?
It's also a question of only peripheral importance to the actual topic of discussion, namely that of incapacitation. The paper you linked is irrelevant because it's not even attempting to measure the crime prevented by incapacitation.
Given how many misinterpretations of my comments you seem to be making, I'm beginning to think they might be deliberate.
Just on the off chance you are discussing this in good faith, let me quote a sentence in which I very explicitly do not assume deterrence is 0: "In this scenario, for doubling clearance rates to work even as well as harsh prison sentences, it would need to cut num_criminals [deterrence] by 70%."
You could try reading the first comment I wrote, which explains clearly that a) I'm looking at one particular graph which is an approximation of P(crime/criminal) b) it's weak evidence and c) I'm asking if someone has better data.
I did dig a bit deeper and found this somewhat dated study: https://bjs.ojp.gov/content/pub/pdf/rpr94.pdf
Basically the 272k criminals released from jail in 1994 are believed to have committed 100k violent crimes and 208k property crimes within 3 years.
As per table 12, 55% of people who were released from prison in 1994 had 7 or more arrests previously and about 75% of this group would be arrested again within 3 years. 44% of the group had been in prison at least once before, and this group also has a 75% 3 year rearrest rate after getting out.
These numbers sure seem in the same ballpark as my napkin math, which you haven't stated any particular disagreement with.
The word "incapacitation" does not appear in your original comment, but it does include . Maybe you should be more explicit.
You wrote:
You can't talk about policy response with such an incomplete picture. (The real correct response to an 80/20 rule of crime is to focus on finding the serial criminals, not harshly punishing everyone).
Moreover, your napkin math has nothing whatsoever to do with how crime is distributed among criminals--it just compares different policing and sentencing strategies. The distribution of crime at no point enters into your calculation.
What is P(crime|criminal)? The probability that a criminal committed a crime is 1. If you mean probability of recidivism, that is exactly what the study I gave you measured. Someone may be commenting in bad faith here, but it's not me.
I asked some questions about them, like how sensitive they are to the parameter values or what happens if you allow for criminals to be better or worse at not getting caught. Do you want me to disagree? I can do that. 2/3 is not that high of a clearance rate; on the flip side, 1/3 is not that low. You state that the effect of deterrence would have to be 70% (although only account for prospective criminals, not those who have already been arrested in the past), but don't actually give any reason to suggest that this is unrealistic. 20-35 is probably too wide of an age range; something like 17-25 would be more realistic.
I'm sure your math returns the numbers you say they do. I just don't think they're very useful. They're also so rough that I fail to be impressed by the alleged matching to the data. I don't follow the logic from the stats you quoted to your estimate; can you make this argument in more detail?
The concept however is clearly spelled out: "...we need longer prison sentences for the criminals we have in order to prevent the same guy from doing 4 more crimes."
"Suppose the average criminal commits crimes at a rate of 3/year between age 20 and 35, meaning that in the absence of policing his career will consist of 45 crimes."
I suppose it was slightly badly phrased, I should have described it as a "representative criminal" instead of "average criminal". But yes - my napkin math shows that in the regime of high #s of crimes/criminal, locking them up forever is a very effective strategy.
The question of distribution of crimes/criminal is how much crime actually comes from that regime. You previously said you think it's a lot:
Interesting - it looks like my 33% is not too far off from the actual number of 41% for violent crime. The "high" numbers you're providing are only for murder, which is a red herring - most crime isn't murder.
You're the one making the claim deterrence is the best. Kind of strange how you haven't actually provided any estimates of elasticity here.
Not in this thread, because I don't see any reason you wouldn't ignore what I say and misrepresent me as you've already done repeatedly.
Among many other things, yes. You might have been thinking primarily about incapacitation, but your post covers a wide range of points, including speculation on the cause of the long tail and the difference between black and white crime rates (which isn't explained by the mere existence of a long tail). You need to chill out with the accusations of misrepresentation until your writing improves.
Is 3 a year that high? My impression was that the "long tail" included people with hundreds of crimes in their career but for some reason I can't find good data.
By raw numbers, most crimes are never going to get a sentence of "lock them up forever" anyway so it's not a very good hypothetical. (Although, you aren't really locking them up forever-your math for total prison time only counts the time they spend in prison before turning 35; if we could know exactly what age each person stops being a criminal, our job would already be much easier!).
Yes, but the numbers actually matter.
It's not wildly off for violent crime, but I think it's pretty high for property crime (at least in the US). Of course it's true that most crime isn't murder, but murder generally has the best data (lots of other crimes aren't reported to the police). Anyway, the point was not that 1/3 and 2/3 are individually wildly wrong--the point is that the difference could easily be much more extreme, which would have obvious implications for your napkin math. Hence why I asked, 3 times now, for any discussion at all of how your estimates change based on parameters.
Murder also gets a lot of attention for being so bad. You've mostly been discussing "crimes" as a monolithic entity, but is 1 person with 50 misdmeanor charges for public intoxication, loitering, and petty theft as important as 1 murderer?
Again, you are the one who made an argument and said that it supports your hypothesis. I think there are a lot of very large gaps in this argument. One of those gaps is that you acted like the number you got for what deterrence would have to be is unreasonable, but didn't provide any evidence.
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link