I'd bet that Maine's high clearance rate is skewed by the totally insane number of murders in Cabot Cove and the efforts of Jessica Fletcher to solve them.
I usually hesitate to dismiss people's arguments out of hand, but are we taking people like Bouie too seriously here? At this point, it seems like there's just a segment of the progressive wing whose political agenda is basically "tear down lots of systems." Not that their intentions are bad-- I think they genuinely want to improve lives. But they just won't be happy if you try to help people in a way that doesn't obliterate existing systems.
My experience in reading stuff online is that this specific breed of progressives rarely engage with any arguments based on statistical data/published studies that go against their political objectives. They just have a different epistemology in which that data is trumped by some sort of collective narrative. On the other hand, there are probably tons of left-leaning people (like me) who find this epistemology ridiculous, but a lot of these people probably also don't buy into the "tear systems down" view of the world.
So my question is who's the marginal person that we're trying to persuade here? A lot of people who would buy into data-based evidence maybe don't need too much convincing anyway, whereas the people who need to be convinced won't believe the data. I'd love to see MY's take on the competing types of epistemology that we now have on the left wing. (The right wing, of course, is completely gone on this front.)
This post is a mix of some really great ideas, and one really bad solution. When I read it in my email and clicked on the comment box, the first thing I thought was I wonder what is Graham going to have to say. If anyone doesn’t subscribe to a sub stack, they need to go do it right now.
You can tell what sort of cop shows that Matt watches. He is watching NCIS and not “The Rookie” or even CSI when it comes to how Detectives actually solve crimes.
I know a few police officers and detectives. One of the benefits of being a veteran and relatively conservative.
Identifying the criminal is more based on core police knowledge. Knowing the players. It’s rarely a mystery who did it. Gang banger one was shot in the driveway. Well, it was probably gang banger 2 since that was the one Gamgbanger 1 shot at last weekend. Or at least his friend. Because Group A has been moving in on Group Bs corner.
It doesn’t involve forensics or social psychology. Besides, the forensics are performed by technicians not police officers anyway.
All that forensics stuff is what you need to convict somebody, not necessarily solve the crime.
I’m going to guess, that a large number of unsolved murders, or actually solved, or at least the police have a suspicion. There’s just no evidence. No witnesses. The gun has been hidden. No ballistics. Even the victims injured friend won’t rat shooter out.
But this basically just duplicates what Graham said below as far as how bad idea it is to have a dual track system system.
Patrolling and detecting are really just complementary skills, with one leading to the other. Both do something.
Officers and enlisted are not like this at all. First of all, officers don’t actually do anything, with the exception of pilots..They’re basically management, with just enough training to be familiar with the people they’re leading. A more apt comparison would be Jr enlisted verse senior enlisted (privates vs sergeants)
Having said all that, Matt is pretty much spot on on almost all the other subjects. I wasn’t aware about the $950 shoplifting misdemeanor limit in California, but that makes sense. As someone who generally looks down on San Francisco and their self-made problems, all I can do is shrug. Elections have consequences.
Anyway, Matt Yglesias, if you are reading this… You really should read Graham’s substack. I know you’re a busy guy, but you’ve actually got some really smart people in your comments. I am not one of them.
And one antidote .I grew up in the mid cities of Los Angeles county. Downey specifically. Back then, and even then today, Downey was always a more nicer / upscale city compared to its bordering cities. We were surrounded by Paramount, Compton, Southgate, Bell Gardens, pico Rivera, Bellflower, and Santa Fe Springs. This was back in the 80s, and pretty much every one of those cities I listed had gangs. Downey had none. To get to the point, Besides for Johnny broilers and the old Lockheed plant, Downey was famous in the area for having really dick cops. The city was small enough that if you were driving through there or hanging out and you didn’t belong, the Downey cops would definitely pull you over if you looked the least bit suspicious.
30 years later, Downey is actually still pretty nice. It’s nickname now, is the Mexican Beverly Hills. That’s where a lot of successful Latinos have settled. And, I have heard that but from a friend who still lives there, that the cops still have a reputation for being dicks.
Anyway, I’m getting ready to go to Argentina for work. Yes, I know it’s officially closed, but it’s open for critical work, so I got a special visa. Hope everyone has a great day.
Once again, forgive any grammatical errors, I am dictating this on my phone while still laying in bed, and my eyes suck.
The differential clearance rates between various cities are interesting, but I want to see them cross-checked against false conviction rates.
We know that police departments, like other bureaucracies, can always respond to numerical goals with malicious compliance. You, the reformer, think you have told them to convict more murderers. They, the bureaucracy, hear that you have incentivized the production of more murder-convictions, whether they get the right guy or not.
There's got to be follow-up, to make sure that the departments and prosecutors are not juking the stats. And the exoneration of innocent convicts has to lead to consequences for the people who put them away, if they cut any corners.
Otherwise, you are simply replacing the old false negatives (uncleared cases) with a boatload of false positives (frame-ups).
Granted, any diagnostic system will make some errors in both directions, and there's got to be an acceptable rate of each. The best departments are probably going to miss some convictions, but the worst will be full of passionate intensity to meet the new quotas.
So, I like the proposal for reform. And reformers need to anticipate unintended consequences.
Ever since it was invented in Victorian England, the modern police force was always designed as a blue collar career. The level of training is sometimes minimal. The FBI which has "investigation" right in its name only hires from people with both police experience and college degrees. Solving crimes is way more work than mere patrolling and breaking up bar fights. As with other public sector jobs, you get what you pay for. Sometimes less.
The data that I want to see is the extent to which murders are being committed by people who have already had significant, recent contact with the criminal justice system - like, they're awaiting trial on a gun possession or assault charge, or they were released quickly after a previous carjacking. In DC at least, this seems like a thing -- and not one that's a function of police quality, but rather of decisions around bail, minors, and prosecution.
It’s easier to solve a crime if you have detailed intelligence files on the area, including on people who have not committed any crimes themselves. It’s easier to solve a crime if there’s a lot of video footage of the area where the crime was committed. It’s easier to solve a crime if people can be compelled to cooperate with investigators, and if it is considered a crime to lie to investigators.
There are downsides to these things as well. simply pouring money into manpower or OT without improving the tools available to investigators has its own problems—mainly encouraging a parade of interrogations fishing for someone who happens to be in a fragile enough state that they will give a weak confession and generate a clearance.
All a long winded way of saying that the main reason we don’t have effective investigators is that we are not super comfortable with effective investigators. We actually have a lot of roadblocks to investigation built into the foundation of our legal system. They are supposed to safeguard our ‘freedom’ or some such.
Matt Y seems to have spent a lot of time reading research and data, but near zero time talking to actual police officers. He comes across as very confident in his opinions on a subject that he only understand superficially.
Graham's posted the following: "To be honest, I don't put a lot of stock in city-to-city comparisons of clearance rates. "Clearance" does not include just arrests. It also includes "clearance by exception." If this is the a valid point of view on clearance, then it undermines the main building block of MY's argument.
There's 3 or 4 comments below where people "want to see the data". My professional job title is "data scientist"; I want to see the data, too. But it drives me crazy when people skip straight to seeing the data without doing the work on understanding what the data actually represents. Talking to subject matter experts, especially those who actually created the data set, is a crucial first step, and it feels like MY isn't doing that when it comes to policing.
One of the things I noticed back when everyone was worrying about the valorization of police in tv shows, is that almost all tv and movie cops are detectives. Other than like the show Blue Bloods. But, detectives are like, what 5% of cops?
I remember a few years ago with the popularity of the show Bones, there were a bunch of students in college hoping to be forensic anthropologists, who were disappointed to find out the were, like, only a few people with that job in the entire state. But if we could create a college major in criminology that focused on investigations, that actually had good career outcomes, I'm sure there'd be many students interested in it.
I don't know who to attribute this quote to, but Yuval Noah Harari says it often in interviews ... "Things are better than ever before. Things are still quite bad. Things can get much worse."
So much about this policing debate seems to be anchored to just one of these three states.
So does proactive policing work? It seems yes. Are there downstream consequences for the "hot spot" communities? Seems plausible although likely more a set of near-term vs. long-term tradeoffs. Could radical change to the structure and goals of existing police departments lead to near-term disruption and worsening conditions (i.e., what Matt proposes)? Also seems likely. Does this make me more of an incrementalist? Yes.
Rape clearance rates are much lower than murder clearance rates. In 2019, only 136 men began Georgia prison sentences for rape. Conversely, about 70,600 girls are born every year in Georgia. This leaves two possibilities: 1) only 0.2% of women are raped during their lifetimes or 2) the clearance rate sucks.
Many rapes go unreported, but the authorities do an abysmal job with the rapes that are reported. Only 1/3 of reported rapes are prosecuted, and only 1/6 of prosecuted rapes are prosecuted end in conviction. Rape is a difficult crime to prosecute because it typically occurs between acquaintances, so there is rarely forced entry or other circumstantial evidence corroborating the victim. It's usually just a swearing contest.
I'm not sure how much good additional detectives would do here. A swearing contest is just a swearing contest and, in many cases, investigation just can't shed much light on what happened in private. The solution is more prosecutions and the problem is the career incentives of prosecutors. Prosecutors don't want to bring big cases that are hard to win. Furthermore, when a crime carries a long sentence, jurors will hold prosecutors to a very high standard and indulge fairly small doubts about guilt. I strongly believe that the prosecution of acquaintance rape should be more like the prosecution of drunk driving: more prosecutions, lower penalties. There should be no shame in losing a rape trial. If more rapes are reported repeat assailants will have long rap sheets. Former victims can testify as "similar transactions" witnesses in Georgia even if no conviction occurred. If a critical mass of women come forward and have their voices heard, the repeat offenders will be ferreted out and convictions will increase.
Right now, the median sentence for rape is zero years in prison. Pursuing one or two-year sentences plus registration as a sex offender would be a vast improvement.
Interesting post but I did want to raise one question about the recommendation to do more benchmarking of police departments, or as Matt puts it, "be more like this other department over here."
Having dealt with lots of crappy data, I do wonder about the wide range in clearing murder cases across cities, states, etc. that Matt cites. Perhaps these data reported to the UCR are perfectly reliable, but hmm. . . After all, as far as I know, there's no uniform rigorous way of defining and reporting case clearance across police jurisdictions, and it wouldn't surprise me if the varying ways departments do it lead to incommensurate results across jurisdictions. (This is touched on a bit in the article on "Clearance Rates and Criminal Investigations" in the Encyclopedia of Political Science (https://books.google.com/books?id=HIE_zF1Rv7MC&printsec=frontcover&source=gbs_ge_summary_r&cad=0#v=onepage&q&f=false vol 1, starting at page 182).
Matt cites the Murder Accountability Project and presumably those people know what they're doing. But when I look at their post "Do Most Murders Go Unsolved in Your Town?" (http://www.murderdata.org/) and I see a chart of the trendline for percentage of major police departments that fail to clear most homicides. For the late 1960s, the percentage of such departments is effectively zero and then it increases pretty much in a straight line to around 75% currently. Really? Police departments have gotten that much worse at solving murders? Or are the data across time not that reliable?
No knock on these people, but data reliability and comparability are tough nuts to crack, so the reader always has to approach some things with at least a smidgen of skepticism.
I'd bet that Maine's high clearance rate is skewed by the totally insane number of murders in Cabot Cove and the efforts of Jessica Fletcher to solve them.
I usually hesitate to dismiss people's arguments out of hand, but are we taking people like Bouie too seriously here? At this point, it seems like there's just a segment of the progressive wing whose political agenda is basically "tear down lots of systems." Not that their intentions are bad-- I think they genuinely want to improve lives. But they just won't be happy if you try to help people in a way that doesn't obliterate existing systems.
My experience in reading stuff online is that this specific breed of progressives rarely engage with any arguments based on statistical data/published studies that go against their political objectives. They just have a different epistemology in which that data is trumped by some sort of collective narrative. On the other hand, there are probably tons of left-leaning people (like me) who find this epistemology ridiculous, but a lot of these people probably also don't buy into the "tear systems down" view of the world.
So my question is who's the marginal person that we're trying to persuade here? A lot of people who would buy into data-based evidence maybe don't need too much convincing anyway, whereas the people who need to be convinced won't believe the data. I'd love to see MY's take on the competing types of epistemology that we now have on the left wing. (The right wing, of course, is completely gone on this front.)
This post is a mix of some really great ideas, and one really bad solution. When I read it in my email and clicked on the comment box, the first thing I thought was I wonder what is Graham going to have to say. If anyone doesn’t subscribe to a sub stack, they need to go do it right now.
You can tell what sort of cop shows that Matt watches. He is watching NCIS and not “The Rookie” or even CSI when it comes to how Detectives actually solve crimes.
I know a few police officers and detectives. One of the benefits of being a veteran and relatively conservative.
Identifying the criminal is more based on core police knowledge. Knowing the players. It’s rarely a mystery who did it. Gang banger one was shot in the driveway. Well, it was probably gang banger 2 since that was the one Gamgbanger 1 shot at last weekend. Or at least his friend. Because Group A has been moving in on Group Bs corner.
It doesn’t involve forensics or social psychology. Besides, the forensics are performed by technicians not police officers anyway.
All that forensics stuff is what you need to convict somebody, not necessarily solve the crime.
I’m going to guess, that a large number of unsolved murders, or actually solved, or at least the police have a suspicion. There’s just no evidence. No witnesses. The gun has been hidden. No ballistics. Even the victims injured friend won’t rat shooter out.
But this basically just duplicates what Graham said below as far as how bad idea it is to have a dual track system system.
Patrolling and detecting are really just complementary skills, with one leading to the other. Both do something.
Officers and enlisted are not like this at all. First of all, officers don’t actually do anything, with the exception of pilots..They’re basically management, with just enough training to be familiar with the people they’re leading. A more apt comparison would be Jr enlisted verse senior enlisted (privates vs sergeants)
Having said all that, Matt is pretty much spot on on almost all the other subjects. I wasn’t aware about the $950 shoplifting misdemeanor limit in California, but that makes sense. As someone who generally looks down on San Francisco and their self-made problems, all I can do is shrug. Elections have consequences.
Anyway, Matt Yglesias, if you are reading this… You really should read Graham’s substack. I know you’re a busy guy, but you’ve actually got some really smart people in your comments. I am not one of them.
And one antidote .I grew up in the mid cities of Los Angeles county. Downey specifically. Back then, and even then today, Downey was always a more nicer / upscale city compared to its bordering cities. We were surrounded by Paramount, Compton, Southgate, Bell Gardens, pico Rivera, Bellflower, and Santa Fe Springs. This was back in the 80s, and pretty much every one of those cities I listed had gangs. Downey had none. To get to the point, Besides for Johnny broilers and the old Lockheed plant, Downey was famous in the area for having really dick cops. The city was small enough that if you were driving through there or hanging out and you didn’t belong, the Downey cops would definitely pull you over if you looked the least bit suspicious.
30 years later, Downey is actually still pretty nice. It’s nickname now, is the Mexican Beverly Hills. That’s where a lot of successful Latinos have settled. And, I have heard that but from a friend who still lives there, that the cops still have a reputation for being dicks.
Anyway, I’m getting ready to go to Argentina for work. Yes, I know it’s officially closed, but it’s open for critical work, so I got a special visa. Hope everyone has a great day.
Once again, forgive any grammatical errors, I am dictating this on my phone while still laying in bed, and my eyes suck.
The differential clearance rates between various cities are interesting, but I want to see them cross-checked against false conviction rates.
We know that police departments, like other bureaucracies, can always respond to numerical goals with malicious compliance. You, the reformer, think you have told them to convict more murderers. They, the bureaucracy, hear that you have incentivized the production of more murder-convictions, whether they get the right guy or not.
There's got to be follow-up, to make sure that the departments and prosecutors are not juking the stats. And the exoneration of innocent convicts has to lead to consequences for the people who put them away, if they cut any corners.
Otherwise, you are simply replacing the old false negatives (uncleared cases) with a boatload of false positives (frame-ups).
Granted, any diagnostic system will make some errors in both directions, and there's got to be an acceptable rate of each. The best departments are probably going to miss some convictions, but the worst will be full of passionate intensity to meet the new quotas.
So, I like the proposal for reform. And reformers need to anticipate unintended consequences.
In general, I think we need more fine-grain studies of individual departments and fewer generalizations about "policing in America."
Ever since it was invented in Victorian England, the modern police force was always designed as a blue collar career. The level of training is sometimes minimal. The FBI which has "investigation" right in its name only hires from people with both police experience and college degrees. Solving crimes is way more work than mere patrolling and breaking up bar fights. As with other public sector jobs, you get what you pay for. Sometimes less.
The data that I want to see is the extent to which murders are being committed by people who have already had significant, recent contact with the criminal justice system - like, they're awaiting trial on a gun possession or assault charge, or they were released quickly after a previous carjacking. In DC at least, this seems like a thing -- and not one that's a function of police quality, but rather of decisions around bail, minors, and prosecution.
It’s easier to solve a crime if you have detailed intelligence files on the area, including on people who have not committed any crimes themselves. It’s easier to solve a crime if there’s a lot of video footage of the area where the crime was committed. It’s easier to solve a crime if people can be compelled to cooperate with investigators, and if it is considered a crime to lie to investigators.
There are downsides to these things as well. simply pouring money into manpower or OT without improving the tools available to investigators has its own problems—mainly encouraging a parade of interrogations fishing for someone who happens to be in a fragile enough state that they will give a weak confession and generate a clearance.
All a long winded way of saying that the main reason we don’t have effective investigators is that we are not super comfortable with effective investigators. We actually have a lot of roadblocks to investigation built into the foundation of our legal system. They are supposed to safeguard our ‘freedom’ or some such.
Matt Y seems to have spent a lot of time reading research and data, but near zero time talking to actual police officers. He comes across as very confident in his opinions on a subject that he only understand superficially.
Graham's posted the following: "To be honest, I don't put a lot of stock in city-to-city comparisons of clearance rates. "Clearance" does not include just arrests. It also includes "clearance by exception." If this is the a valid point of view on clearance, then it undermines the main building block of MY's argument.
There's 3 or 4 comments below where people "want to see the data". My professional job title is "data scientist"; I want to see the data, too. But it drives me crazy when people skip straight to seeing the data without doing the work on understanding what the data actually represents. Talking to subject matter experts, especially those who actually created the data set, is a crucial first step, and it feels like MY isn't doing that when it comes to policing.
One of the things I noticed back when everyone was worrying about the valorization of police in tv shows, is that almost all tv and movie cops are detectives. Other than like the show Blue Bloods. But, detectives are like, what 5% of cops?
I remember a few years ago with the popularity of the show Bones, there were a bunch of students in college hoping to be forensic anthropologists, who were disappointed to find out the were, like, only a few people with that job in the entire state. But if we could create a college major in criminology that focused on investigations, that actually had good career outcomes, I'm sure there'd be many students interested in it.
"California voters who worried about mass incarceration voted to turn shoplifting offenses under $950 into misdemeanors"
Correct me if I'm wrong but in Texas thefts below $2500 are classified as misdemeanors, right? I think there may be more to this story.
I don't know who to attribute this quote to, but Yuval Noah Harari says it often in interviews ... "Things are better than ever before. Things are still quite bad. Things can get much worse."
So much about this policing debate seems to be anchored to just one of these three states.
So does proactive policing work? It seems yes. Are there downstream consequences for the "hot spot" communities? Seems plausible although likely more a set of near-term vs. long-term tradeoffs. Could radical change to the structure and goals of existing police departments lead to near-term disruption and worsening conditions (i.e., what Matt proposes)? Also seems likely. Does this make me more of an incrementalist? Yes.
Solving murders is a "boring" way to reduce crime? Based on the popularity of podcasts and TV shows I'd say there is huge public interest.
Get a couple of rival TV producers to fund different cities, start real reality shows, and see if anyone can de-throne Dick Wolff.
Rape clearance rates are much lower than murder clearance rates. In 2019, only 136 men began Georgia prison sentences for rape. Conversely, about 70,600 girls are born every year in Georgia. This leaves two possibilities: 1) only 0.2% of women are raped during their lifetimes or 2) the clearance rate sucks.
Many rapes go unreported, but the authorities do an abysmal job with the rapes that are reported. Only 1/3 of reported rapes are prosecuted, and only 1/6 of prosecuted rapes are prosecuted end in conviction. Rape is a difficult crime to prosecute because it typically occurs between acquaintances, so there is rarely forced entry or other circumstantial evidence corroborating the victim. It's usually just a swearing contest.
I'm not sure how much good additional detectives would do here. A swearing contest is just a swearing contest and, in many cases, investigation just can't shed much light on what happened in private. The solution is more prosecutions and the problem is the career incentives of prosecutors. Prosecutors don't want to bring big cases that are hard to win. Furthermore, when a crime carries a long sentence, jurors will hold prosecutors to a very high standard and indulge fairly small doubts about guilt. I strongly believe that the prosecution of acquaintance rape should be more like the prosecution of drunk driving: more prosecutions, lower penalties. There should be no shame in losing a rape trial. If more rapes are reported repeat assailants will have long rap sheets. Former victims can testify as "similar transactions" witnesses in Georgia even if no conviction occurred. If a critical mass of women come forward and have their voices heard, the repeat offenders will be ferreted out and convictions will increase.
Right now, the median sentence for rape is zero years in prison. Pursuing one or two-year sentences plus registration as a sex offender would be a vast improvement.
Interesting post but I did want to raise one question about the recommendation to do more benchmarking of police departments, or as Matt puts it, "be more like this other department over here."
Having dealt with lots of crappy data, I do wonder about the wide range in clearing murder cases across cities, states, etc. that Matt cites. Perhaps these data reported to the UCR are perfectly reliable, but hmm. . . After all, as far as I know, there's no uniform rigorous way of defining and reporting case clearance across police jurisdictions, and it wouldn't surprise me if the varying ways departments do it lead to incommensurate results across jurisdictions. (This is touched on a bit in the article on "Clearance Rates and Criminal Investigations" in the Encyclopedia of Political Science (https://books.google.com/books?id=HIE_zF1Rv7MC&printsec=frontcover&source=gbs_ge_summary_r&cad=0#v=onepage&q&f=false vol 1, starting at page 182).
Matt cites the Murder Accountability Project and presumably those people know what they're doing. But when I look at their post "Do Most Murders Go Unsolved in Your Town?" (http://www.murderdata.org/) and I see a chart of the trendline for percentage of major police departments that fail to clear most homicides. For the late 1960s, the percentage of such departments is effectively zero and then it increases pretty much in a straight line to around 75% currently. Really? Police departments have gotten that much worse at solving murders? Or are the data across time not that reliable?
No knock on these people, but data reliability and comparability are tough nuts to crack, so the reader always has to approach some things with at least a smidgen of skepticism.