I find it kind of hilarious that both financial hardship from the pandemic and having better finances than usual due to stimulus checks (which supposedly was spent on guns) were being offered as (entirely ad hoc) explanations for the BLM effect.
I also found it hilarious how many commenters accused Scott of coming to the conclusion first and then desperately looking for a way to justify it - followed shortly thereafter by desperate attempts at dismissing the huge amounts of data Scott provided and asserting some politically convenient explanations with little to no data to support them.
I think the distinction between depolicing and a collapse of trust in police is important (though of course you have every right to be agnostic on the question).
If the murder spike was a result of police pulling back, the lesson would be that getting police to be more active would save lives. That could mean making sure police feel that they'll be supported by politicians if they make tough calls in tough situations and not giving police extra paperwork.
If the murder spike was a result of the public losing trust in police, and so not calling the cops when there's a problem, the solution is the opposite--restoring public trust in police by holding police accountable when they do something wrong and requiring careful record-keeping.
The homicide rate right now for the black population in Chicago is at the highest recorded level since the start of modern record keeping. All those people saying that crime was much worse in the 1990's? Not for black residents of Chicago. The previous high in the 1990's was about 75 homicides per 1000 people. In 2021 that hit 85.2 incidents per 1000 people. By contrast the homicide rates for whites and Latinos are nowhere near their all-time highs.
The first question I have is whether or not this holds true for other large metropolitan areas as well. If so then aggregating homicide rates across an entire city is a mistake. The hypothesis that needs to be tested is whether or not the Chauvin/Ferguson/Freddie Grey/etc. incidents resulted in radicalization of black communities specifically rather than metropolitan populations as a whole.
Edit: more specifically poor urban black communities. If there is a historically bad epidemic of black homicides then why isn't that more apparent? After all the average US homicide rate has risen to 1996 levels but still falls short of historic highs around 1991-1992. I can think of a couple of factors. While blacks make up about 12% of the US population and have for decades the percentage of blacks living in poverty has declined sharply since the 1990's (peaking at around 31% in 1992 and declining to around 17% today). Potentially related to that with regards to Chicago specifically the black population of that city has also been in decline for decades, since at least the 1980's, even as the Latino population in that city has soared. If the twofer is urban poverty than a decrease in black poverty rates plus a change in the racial composition of cities could result in an average homicide rate that is still short of all time highs even as poor inner city communities face historic levels of devastation.
Thank you for so properly refuting the garbage analysis from Artifex0. His is a stupid idea that just won't go away despite the evidence to the contrary.
> If you have some clever reason why Central America is also a bad comparison, then please find me any major country
Noted, we should do more homework as a rule. We DO appreciate you doing the (occasionally tedious) research - it's still much better that somebody does it, and it helps a lot in fostering a culture of doing it. Even if it's a slow process.
Speaking of, I noticed recently my google-fu is lagging behind the times. Can you maybe do a post on "how to research"? Including stuff you find obvious or simple, like what sites you go first, or what tool you use to quickly generate a graph. For example: I realized annoyingly late that on sci-hub you should search by DOI, and not by paper title.
> I don’t really think of this as an alternative explanation. I am agnostic to the exact causal pathway between the events of May 25 2020 and the homicide spike; all I’m trying to show is that the spike did begin around that time and seems connected.
And this?
> My claim is that this is false. It is not difficult to assign priority. The protests were the primary cause, with the other two being minor contributors at most. When I say the media is getting this issue wrong, I mean that they’re saying things like this.
In the first you're saying that there was a homicide spike as a result of the events that occurred right around Floyd's killing, and in the second you're saying specifically that it's the protests (keep in mind the comment to which you responded with the "agnostic" point is saying that it is *not* the protests). My reaction to the initial post was that you can't disentangle the protests themselves with other things that happened at the same time. You say you're agnostic between those things, but then also that it was the protests.
This leads to another issue I had, which is a question of which groups have agency and which are merely reacting to other stuff. In the "BLM did it" version of events, BLM protests lead to police doing their jobs less which leads to more crime - this is interpreted as "it's because of BLM". Assigning BLM agency to cause things to happen, and the police actions as being a natural reaction to outside events. Why not treat the *protests* as a natural reaction to outside events, same as you do with police doing their jobs less, and the *cause of the protests* as being the ultimate culprit?
When a negative event is a result of the actions of multiple people (or groups), often each person/group will characterize their own actions as being a mere *reaction* to someone else, and the *other* group as the real cause of the bad thing. The question becomes, who do you blame / put the onus on to change? Usually the answer is "whoever I sympathize with the least" but it's often not examined all that much.
Something that is deeply frustrating to me are complaints that collecting data on police stops (traffic, terry, or otherwise) is a problem. First, it's distasteful to me as an American to claim that creating a record of a stop is overly burdensome to the state security apparatus; I was taught to believe that undue search and seizure are things we should be guarding against. It's hard to guard against things that are invisible because there is no (legible to inspection) record. Second, there's increasingly smooth software to reduce the pain of paperwork (e.g. Mark43).** Third, records of police actions are useful to from an efficiency standpoint, as a lack of any stops may also be concerning.
So, data is good in my opinion. If collecting the data is painful, then we should seek to ease that pain, because deterring police work is inefficient. But arguing that silly reformers passed the Racial Identity and Profiling Act -- and that's why police can't police -- is itself rather silly. This is because it ignores that policing is part of a state's obligation to the general welfare of its populace -- policing isn't measured only from how efficiently police can deploy coercive power.
These questions should be simple for an officer to fill out or articulate.
Enough about my feelings though, what are some brass tacks about why we should care about recording data on stops? Well, let's look at traffic stops. Traffic stops are nicer for criminologists because they are usually discretionary and also usually recorded, because license plate checks themselves are recorded as part of almost traffic stops. These records are often enriched by the demos of the person(s) stopped, time/date info, and geolocation. This has allowed extensive empirical analyses to study both 1) efficiency and 2) bias in traffic stops. Findings indicate that traffic stops and searches are racially biased (when daylight allows), and this racial bias results inefficiency as searches of Black drivers are less likely to produce (drugs, guns, etc.). However, jurisdictions, units, and officers vary significantly. It's a well-known finding called the 'veil of darkness' (racial disparities disappear at night and often on highways, where it is difficult to discern phenotypical racial characteristics). The latest standout study in this vein is Feigenberg and Miller 2022 in the Quarterly Journal of Economic. Here's an imgur link of the most important point: https://imgur.com/a/renWnIu
Basically, searches are racially biased and inefficient in Texas, and equalizing them conditional on stops would yield more contraband and less racial bias. This isn't a shocking finding - it comports with the greatest hits of that literature:
Chanin, Joshua, Megan Welsh, and Dana Nurge. 2018. “Traffic Enforcement Through the Lens of Race: A Sequential Analysis of Post-Stop Outcomes in San Diego, California.” Criminal Justice Policy Review 29(6–7):561–83. doi: 10.1177/0887403417740188.
Feigenberg, Benjamin, and Conrad Miller. 2022. “Would Eliminating Racial Disparities in Motor Vehicle Searches Have Efficiency Costs?*.” The Quarterly Journal of Economics 137(1):49–113. doi: 10.1093/qje/qjab018.
Fryer, Roland G. 2019. “An Empirical Analysis of Racial Differences in Police Use of Force.” Journal of Political Economy 127(3):1210–61. doi: 10.1086/701423.
Pierson, Emma, Camelia Simoiu, Jan Overgoor, Sam Corbett-Davies, Daniel Jenson, Amy Shoemaker, Vignesh Ramachandran, Phoebe Barghouty, Cheryl Phillips, Ravi Shroff, and Sharad Goel. 2020. “A Large-Scale Analysis of Racial Disparities in Police Stops across the United States.” Nature Human Behaviour 4(7):736–45. doi: 10.1038/s41562-020-0858-1.
Roh, Sunghoon, and Matthew Robinson. 2009. “A Geographic Approach to Racial Profiling: The Microanalysis and Macroanalysis of Racial Disparity in Traffic Stops.” Police Quarterly 12(2):137–69. doi: 10.1177/1098611109332422.
Taniguchi, Travis A., Joshua A. Hendrix, Alison Levin-Rector, Brian P. Aagaard, Kevin J. Strom, and Stephanie A. Zimmer. 2017. “Extending the Veil of Darkness Approach: An Examination of Racial Disproportionality in Traffic Stops in Durham, NC.” Police Quarterly 20(4):420–48. doi: 10.1177/1098611117721665.
Withrow, Brian L., Jeffery D. Dailey, and Henry Jackson. 2008. “The Utility of an Internal Benchmarking Strategy in Racial Profiling Surveillance.” Justice Research and Policy 10(2):19–47. doi: 10.3818/JRP.10.2.2008.19.
Zingraff, Matthew, William Smith, and Donald Tomaskovic-Devey. 2003. “North Carolina Highway Traffic Study, Final Report.” Department of Justice.
The blue line is the difference over the same smoothed 7 day period the year prior. Note that this is a trailing average, so the data point "today" is the arithmetic mean of the prior 7 days (including the current date). The vertical red line is Floyd's death. There is indeed support for some non-trivial excess shootings in the earlier part of may, but it's clearly dwarfed by what comes afterwards.
I also did the analysis using the average since 2006 of that same 7 day period as the baseline, but the chart looks very similar so I won't bother to upload that too.
Don't have any nitpick, just passing by to show my gratitude for you drilling deep down on a contentious topic that I feel the MSM should have covered with nuance but instead simply used as an opportunity to push sloganeering. Quality content like this is why so many of us read you.
"what would be the explanation for why this trend would start on May 20 or something?"
Wikipedia tells me that the stay-at-home order expired May 17. I mostly believe your thesis, but you must admit that the alternative "The murder spike was caused by COVID damaging mental health and the economy (or any other cause) but started only with the end of lockdowns" also fit the timing. It probably fails on the international comparison, but you should not overstate your case...
Don't know about the homicide spike but the trends in the homicide graph from central America are beautiful to see. What has been happening in El Salvador ? That's an impressive decrease !
Above my pay grade to comment on the particulars, so...just gonna leave some appreciation for the hard work. It's nice to have a feeling of relatively firmer empirical ground "in real time" as opposed to years later when the Official History Books(tm) are written. Scott going all Scott on thorny Current Issues is both (probably) the most controversial content, but also the primary reason SSC/ACX are my favourite blogs. You just write better than most when clearly annoyed/angry/jerk-ish. "A spoonful of heat helps the light go down"
Is it not possible, that protest actually did cause a decrease in policing AND, in parallel caused an increase in homicide, BUT homicide increase was not caused by de-policing? (or at least another factor had ~the same effect size as de-policing)
My theory is that protests and news about protests can temporarily change the behaviour of people. It may result in more people being on the streets, and them being more concentrated into smaller areas. In addition, based on "Politics is The Mind-Killer/Toxoplasma Of Rage/etc" articles, they may be more agitated and more prone to violence in such cases.
So: more people outside + concentrating them in specific areas + being agitated/outraged --> increased probability of violence happening --> increase in homicides
The comparisons with other countries is a tricky one. In the original post you looked at the murder rate in European countries and saw no peak. But those countries, as you point out, have very low levels of gun ownership and also start from a lower level of homicide. That means that the number of people who can easily kill someone if they suddenly wanted to is already low. In the US, the high stocks of guns should allow more potential murderers - which may reflect in a more visible spike when factors combine to "encourage" homicide.
Perhaps looking at violent crime figures in Europe (without resulting in death) could be more reflective than just homicide rates. A quick look at the numbers from London suggests they also fell significantly in 2020 and 2021 (https://www.statista.com/statistics/864736/knife-crime-in-london/), while the murder rate doesn't show much of a change (https://www.statista.com/statistics/862984/murders-in-london/). So it doesn't look like the pandemic conspired to increase crime there (indeed, the opposite is probably true). Interestingly, the lack of relationship between murder and knife crime rates should support the idea that most knife crimes don't end up with some dead, but I suspect most gun crimes do.
I'd also caution the comparison with central America. The degree of pandemic restrictions and the level to which those restrictions were enforced varied greatly across the world. Having spent some time during the pandemic in less developed countries, I can tell you that in many places you'd barely know there was a pandemic at all (other than seeing a few badly worn masks).
In your original article on this topic, you specifically claimed that the increase in murder rates was due to a police pullback. When you were challenged about this, you claimed you are actually agnostic on the issue.
I'm disturbed by the gun suicides vs homicides graph. This post glosses right over the (granted, unrelated to the topic) GIGANTIC STEADY INCREASE IN SUICIDE OVER A 12-YEAR PERIOD FROM 2006 TO 2018. But that seems more interesting than the actual topic of the post.
Could we at least get a comment like "Don't be too alarmed, the gun suicides are displacing other suicide methods" or "Wow! What happened to suicide rates over the last 16 years?"
"So this theory requires us to believe that number of guns increasing 3.5% every year from 2015 - 2020 had no effect on the murder rate, but that guns going up 5.5% in 2020 had a very strong effect on the murder rate. Specifically, an extra two percent increase in guns must lead to a 30% increase in murder rates. Why would we believe that?"
Isn't this comparing a percentage point increase in gun sales to a percent increase in murders? That seems just as big a sin as comparing stocks and flows.
Putting both in percent increases, a 63% increase in sales was associated with a 30% increase in murders.
The rest of that section was pretty convincing, but I think that particular argument is bad and your point would be stronger if you omitted it.
There might be a correlation between BLM protests and rising gun sales:
If citizens imagined (rightly or not) police during the BLM protests, for fear of being labelled racist, were less likely to protect the public and their property against violent protesters , then citizens would feel the need to arm themselves. If police were defunded, this would provide even more reason to do so.
If the above is true, even if gunsales affected homicide rates, the latter would indirectly be affected by the BLM protests.
On the question of 'police pullback' vs 'distrust in police', you write:
> I am agnostic to the exact causal pathway between the events of May 25 2020 and the homicide spike; all I’m trying to show is that the spike did begin around that time and seems connected.
If you yourself are agnostic, you should be aware that your original essay is very firmly not so, and is prominently on the side of police pullback. You give it in your opening sentence:
> In my review of San Fransicko, I mentioned that it was hard to separate the effect of San Francisco’s local policies from the general 2020 spike in homicides, which I attributed to the Black Lives Matter protests and **subsequent police pullback**.
You have a *topic heading* "Police pullback" that opens with "My specific claim is that **the protests caused police to do less policing** in predominantly black areas" then gives multiple reasons that police would pull back, concluding: "I don’t want to speculate on which of these factors was most decisive, only to say that **at least one of them must be true, and that police did in fact pull back**."
Elsewhere:
> My interpretation is that people complied with the strict lockdown early in the pandemic, that effect was played out by May, and then separately **the protests caused a longer-term decrease in policing**.
> But there are lots of reasons to expect that the Black Lives Matter protests would **cause police to pull back** from black communities in particular.
> The New York Times had an article Deconstructing The Ferguson Effect, subtitled “The idea that **the police have retreated under siege** will not go away. But even if it's true, is it necessarily bad?”, which as far as I can tell is **as close as the New York Times has ever come to acknowledging that a politically inconvenient fact is true**.
> No country except the United States had a large homicide spike in 2020, which suggests that the spike was unrelated to the pandemic and more associated with US-specific factors, for which the BLM protests and **subsequent pullback of policing in black communities seem to me to be the most obvious suspect.**
You had to be there Scott. Everyone around Mpls knows the police are feeling unloved right now. There was stupid overreaction locally after Floyd’s death and we all know that too. There were a lot of early retirements and resignation as a result. Those that remain are understandably reluctant to leave their cars. Less enforcement leads to more crime.
There were several pieces in the NYT that made similar points to yours. They were more nuanced than your position. This sort of analysis should be nuanced. Right wing media beats the ‘BLM protests caused increased crime’ drum hard and often because their business model is to generate outrage.
Treating the BLM protests as a first cause in the chain of events is simply wrong.
I don’t understand why you are taking this tack of being a lone defender of The Truth. I honestly don’t think you are. You’ve analyzed a complex series of events by looking at one link in isolation and the result was that people felt free to use words like ‘libtard’ and reinforce their arguments with the rhetorical device of CapsLock shouting.
Yes, NYT has a moderate liberal bias. I try to keep that in mind when I read it. Is it possible that you have a moderate anti NYT bias?
> I am agnostic to the exact causal pathway between the events of May 25 2020 and the homicide spike; all I’m trying to show is that the spike did begin around that time and seems connected.
This isn't consistent with the language you use elsewhere. I don't think there's so much disagreement about the correlation.
> Also, what would be the explanation for why this trend would start on May 20 or something? There isn’t more pandemic that day. There aren’t more guns that day. It’s not even especially warm that day. I think it’s got to be an artifact.
What? This reads as really close to assuming what you're trying to show. Could you spell this reasoning out?
It's also funny you should say May 20, since it was the first day with a high above 75 F in Minneapolis that year, as well as the first with a low above 55, going by https://www.timeanddate.com/weather/usa/minneapolis/historic?month=5&year=2020. (Also, as another commenter points out, the stay-at-home order expired after May 17.)
By the way, it looks like you've put the black line for "Floyd's death" about a day and a half early. It should be on the shoulder of the first peak.
"Graham" wrote: "When an officer who is fired or prosecuted for something that is clearly unreasonable (like Derek Chauvin, whom all cops agreed was guilty)"
I am going to press [X] to doubt. I haven't been able to find any surveys that specifically ask current/former police officers what they think of Chauvin's actions. I've found articles that report the findings of interviews with cops. Those who are willing to go on the record, particularly high-ranking cops who don't do any actual policing and are sensitive to political considerations (not "real po-leece," as Det. McNulty would say) , are unanimous. But that begins to break down as soon as anonymity is provided, and as the interviews work their way down the chain to beat cops. To the extent that I've found a consensus (on a first-pass review of mainstream news articles), that consensus seems to be that Chauvin crossed a line but wasn't guilty of murder.
>> There's an alternative explanation that fits the evidence here: the killing of Floyd itself caused the crime increase by damaging trust in the police, which led to an increase in retaliatory violence.
The article linked below that comment is focused on gangs. In my general model of gangs (on which I am an Noted Expert because I watched Breaking Bad and The Wire) and their feelings about the police, trust is not the word that comes to mind.
Of course, the probability of a murder suspect getting quickly arrested greatly affects the retaliatory murders. In a city where, by magic, the correct suspect is arrested within 24h of any homicide, only very foolish gang members would try there hands at revenge killings, while in a city where murders are never prosecuted, revenge killings may be rational for a gang.
I do not see why gang members should update their estimate of that probability as a direct result of a murder committed by the police. A much more likely chain of reasoning seems to be
murder by cops => decreased trust in police in general population => depolicing => higher probability estimate (among gang members) of getting away with murder => more gang revenge killings.
Summing up all the steps in between as "damaging trust in police" seems like an oversimplification.
(The quoted statement does not directly refer to gangs, so it could also be interpreted as meaning that previously noncriminal people start committing murders because they lost their faith in the justice system. I find that unconvincing as well. I don't think the update from "Police are good people who serve and protect, so I should report crimes to them" to "Police are racist assholes, so I should not talk to them and avenge any crimes on my own" will happen after any one murder by a cop. I don't claim that the effect of eroding trust is not there, I just claim that it is unlikely to cause that spike within a few days of a single incident.)
Is homicide the type of crime that should be expected to increase as a result of reduced policing?
I would not expect my chances to get away with murder to increase significantly as a result of fewer patrol cars rolling through the neighborhood. If my fingerprints are on the shell casing, if the victim has my skin cells under her fingernails, if my own blood is at the scene, I'm going to get caught. The cops might show up to the crime scene later, but unless they're so lackadaisical that I'm able to dispose of the evidence and concoct an alibi, I'm in trouble.
Shoplifting, minor assaults, smash-and-grab thefts, and public urination should explode, though. Now there may be a broken-windows effect by which murder is a downstream consequence of reduced policing, but I wouldn't expect that to take the form of a spike.
I don't see why you dismiss the temperature thing out of hand, when a plot of murders vs. time of year is right above. There is a clear dramatic increase from late May to September every year, yes there is a "critical temperature of 78 degrees" or perhaps a "critical solar angle" ... why is that silly?
The MN data do show a large increase in baseline (red vs. blue) so "nobody was comitting extra pandemic aggravated assaults/murders" looks to be flatly false - the "un-rolled" data in particular show far, far more red events compared to blue in the winter. In any case the ratio in assaults between winter and summer in MN, about 2x, is the same for both the 2020 and baseline data.
We don't have month-by-month murder data for the country to show whether this "surge" is bigger than the usual summer surge. It looks to be proportionately the same in MN and bigger in NY.
In the bigger picture distinguishing between "bigger summer effect due to pandemic " and "it's BLM" is difficult because they happened only once and at the same time, and you really didn't try. You just said that you don't take it seriously, then made some sarcastic "wHaT is ThErE a CrItICaL TeMp LOL" statements as if it proved anything ...
> the homicide spike started at the the same time in a lot of different cities with widely varying temperatures
Maybe show this? It's clear from the MN data and NY data that they neither started nor peaked at the same time ... not that you have any sort of statistical definition of "start" to begin with or anything, nor did you show data from other cities, nor did you show that the correlation was tighter than the usual correlation in seasonal crime etc. etc.
Which is fine, it's not your day job (or is it, being an influential truthy-ness teller), but don't claim you've rigorously disproved something by saying "it's hard for me to take it seriously".
It is a sad comment in the state of modern academia and politicized issues that when someone tells me they are a PhD candidate at a top program in a politicized field, instead of my trust in their comment going up immensely, it only ticks up slightly over random stranger.
From what happened in the media and what criminologists were saying in the aftermath of the riots it is clear the discipline is very badly ideologically compromised. There are literally hundreds (thousands?) of papers who sole purpose seems to be misinterpreting data to avoid politically unflattering results.
It seems likely to me that we are nowhere near the Pareto optimal curve between policing and crime rate, just as in most governments' response to the pandemic we were nowhere near the optimal curve between economic impact and lives lost.
This situation would imply that while it may well be that police funding loss and increased restraint led to increased crime in 2021 specifically, it is still entirely possible that the best possible versions of "defunding the police", in which some large percentage of a city's police budget gets redirected to evidence based social services, could still result in a net decrease in crime.
People's day to day lives continue to get worse so crime continues to rise. I don't find it to be that complicated and I find this whole analysis and conversation to a bit moot.... I see a lot of people taking correlations and inferring causality while oversimplifying this complex social issue. Scott has done a great job at doing this, but because of that nature the conversation leaves much to be desired for me. Take this excerpt from Scott as an example of the subtle framing expressed in his phrasing through the different posts in this thread.
> I accept I should have put more work in the original post into ruling out gun sales as the cause.
Why is it "the cause" and not "a cause"? Framing is so important.
----
I recognize policing as a way to abate crime, data shows it and, for example, the NIJ recognizes its utility among many other approaches (https://www.ojp.gov/pdffiles/171676.pdf), but I know of no evidence that a lack of policing is a driving force for crime. Key word: driving.
Am I surprised to see what Scott presented and do I doubt the data? No.
Do I agree that a decreasing amount of policing can be, and this case was, a contributing cause to rising homicides? Yes, and unfortunately I've seen it anecdotally in my community.
Do I think this conversation is well framed, actionable, and effectively presenting a variety of perspectives? Not really.
How do you determine if people are arguing in good faith in an argument like this?
Thing X happens via a very simple mechanism that many people predicted, people propose a whole series of complicated different explanations.
I suppose that education and health issues we discuss often have complicated explanations has damaged our epistemic health when it comes to other areas.
Crime is a simpler area because it involves lower IQ people, simpler systems and clear incentives.
It is hard to tell if the complex explanations come from people with an open mind or just ideologues upset that their theories fall apart.
What fraction of gun homicides use legal guns? I'd have thought that a lot of guns used in crime would not have been purchased through legal channels anyway, so the gun sales graphs would be irrelevant.
I was among those who wanted to read more about the homicide spike, so a late thanks to Scott for the post.
After reading the post, I asked an active member of the German Green Party about it; I was curious about the reaction I would get. (I would say sth. like: apparantly there is data showing that homicides among blacks grew after BLM protests.) What's your guess?
He confirmed this immediatly, his reaction was something like: yes, the police in the states is only trained to act while being violent. Then there are protests that they shouldn't be so violent, and they stop/reduce policing. They simply have no idea on how to carry out their duties without resorting to violence.
I admit I was surprised he knew the data and jumped right to: police is doing less (we're both living in Germany). It's just an anecdote, and I don't think it's representative for anybody. (I would still be surprised if the next nine Party Members or Green Party Members here knew.) I also think it shows nicely that even taking a BLM - homicide spike correlation as given, the question on *who has to change what* is totally dependant on the overall understanding of the situation.
I saw some documentations comparing police training in US versus selected European countries. The differences both in total length of training, and in length and importance of training in deescalatory measures seemed huge. I didn't did deeper, so I don't know how much it holds under scutiny.
I would be interested to know more; relevant questions in this context would include:
- overall lenght of training
- weight giving to training on how to deescalate dangerous situations
- admission requirements
- overall weight given to making sure that nobody is harmed or killed (including attitude & and how people talk about that internally)
- measures to ensure that extremists, both from left and right, can't serve as police officers.
I'm aware that contexts (among others, gun ownership in general population) differ widely.
I have a big issue with using "police pullback" as a generic term because it obfuscates _how_ police pull back.
The Minneapolis police, for example, didn't say "let's reassess how we're interacting with the public and stop doing broken window policing" after George Floyd, they went full on "I'm taking my ball and going home." This is an excerpt from a store owner who used to have one of the best used book stores in Minneapolis which also happened to be close to where Floyd was killed:
"...There were 5 cops on the roof of the 5th Precinct building, watching and presumably reporting on events to a headquarters somewhere else, but there were no cops on the ground.
A little after 10 pm, some people pried a sheet of plywood part of the way off the door to a convenience store/gas station across the street from the 5th Precinct, crawled inside and grabbed some loot. There was no police response, so a few more people also crawled in and grabbed some loot...." http://www.unclehugo.com/prod/riotreport.php
The police bunkered up in their buildings and _watched_ looting and later arson go on. I find it a very disingenuous argument to blame reformers for the results of police literally watching felonies being committed without comment.
When they did actively police, many did so in the most incendiary way possible, to the point where a man was actually acquitted of shooting at police due to it being an act of self defense: Police where shooting rubber bullets from an unmarked van at random people without identifying themselves: https://www.ammoland.com/2021/09/man-not-guilty-in-self-defense-case-against-police-during-minneapolis-riots (Note that the author does his best to blame the incident on anything but the police, his account of the facts still has to admit that the police 1.) did not follow policy with regard to shooting people, 2.) did not follow policy when investigating the incident, giving the officers a chance to view the video and talk with each other before taking statements, 3.) made statements that conflicted with the videos they were shown before making those statements, and 4.) assaulted a man who had already surrendered himself.)
I think thinking in terms of police "pulling back" is at best not useful and more likely hiding the issue when it can cover activities as broad as not making as many traffic stops due to time required for paperwork and flagrantly standing and watching people looting and committing arson. Furthermore, saying police have "pulled back" when they are driving around in unmarked vans shooting people is highly misleading.
The discussion should be around changes in how antagonistic police policies and actions are, since the Minneapolis police have demonstrated that it certainly possible to both reduce the number of police interactions and increase the antagonism between police and the public at the same time.
Wanted to plug this Atlantic article from yesterday: "Six reasons why the police murder clearance rate has declined". Its a short (4 min read) interview with a crime analysts about why the murder clearance rate, or % of "solved" murders, has declined from 90% nationally in the 1960's to around 50% today. TLDR: Statistics until the 1990's are bunk and such a high % of murders weren't actually solved, firearm murders have increased from ~50% to ~80% and are much harder to solve than other murders, and distrust between cops and black Americans in areas with a high murder rate is worse than ever.
I think the reason this is getting SO much pushback is that you're looking at extremely noisy time-series data and telling a just-so story about it (handwaving things like, maybe the early-May spike was random and the late-May spike is real, or maybe it's the rolling average--without actually trying to verify how it was averaged! IMO this was a significant oversight in the original piece.)
Not that this means your just-so story is WRONG, but it makes people want to tell other just-so stories. For example: the pandemic caused an increased murder rate in the US for <pick-your-politically-charged-reason-here>, but the lockdowns caused a countervailing decrease, so the real spike took a few months to show up. You've already pooh-poohed this theory (“it took a few months for people to get cabin fever from the pandemic”), but seriously, look at the movement data for the US in "Retail and recreation" or "Public transport stations" here: https://ourworldindata.org/covid-google-mobility-trends It craters in March, with a significant recovery through May and June, exactly when murders were spiking.
To be clear, the data from Baltimore are very compelling, and the demographic breakdown of the murder spike is suggestive. I'm convinced that publicized police murders + protests caused part of the spike. But I think the time series data were the weakest part of the argument.
The 'George Floyd's killing directly lead to the violence outbreak' claim is kind of shocking to me. If a bunch of black people went around killing more cops, white people, politicians etc. than normal, that would at least make sense (while being obviously morally indefensible to non-BLM-supporters). But what are they saying here? A black man got killed, and their response was to go out and kill a bunch of other black people? Putting aside the lack of empirical evidence for the claim, the implication of the claim strikes me as profoundly unflattering for black people, and if true should inform how we view black social pathology generally.
It's frustrating to see people devote so much time and energy theorizing about homicides without any understanding of the specific factors that distinguish homicide from other types of crime, even other violent crimes, leading to "explanations" for general increases in crimes and not specifically spikes in homicides.
The single most important thing to understand about homicides is that the majority of urban homicides are committed by relatively small groups of gang members, starting as drug/turf disputes and continuing as retaliations against previously unsettled homicides. Thus they are highly correlated in a way that has been compared to disease outbreaks - one homicide leads to another, which leads to another - which is not at all the case for other crimes likes robberies. (The excellent book "Don't Shoot!" by David Kennedy describes this dynamic as well as fascinating efforts to understand and break such cycles) Any explanation must take this into account. Trying to explain a percentage increase by looking for general increases in uncorrelated behaviors across large groups of the population is a waste of time.
It seems to me that *something* tends to happen after these high-profile incidents that leads to new retaliatory cycles, or more intense cycles, or some other changed dynamic from the previous status quo (which is already pretty bad in many urban areas but is apparently not the maximum possible rate). I wish everybody would spend their creative energies hypothesizing about what that something might be (changes in the success rate of detective work that can end cycles, changes in opportunities to start a new cycle or heat up a previously cooling one, etc) and not on all the other somethings out there that don't really have anything to do with the nature of urban homicide in the first place.
When I predicted this homicide spike in early June 2020 and provided the reasoning, I specifically mentioned that homicides would increase (not all crime) and that black people would be most affected. And now we still have criminologists try to come up with alternate explanations that don't even take these two things into account.
The question is how the Floyd incident triggered the spike in homicides, not if it triggered the spike (which Scott in my opinion convincingly enough substantiates). What are the causal paths between these two phenomena - the "intermediate variables"? Scott himself suggested several possibilities in his original post, and several additional hypotheses have surfaced during the debate. The causal path/s is important, because they are linked to different "what to do" suggestions.
Could it be as simple as: Americans just being awakened to their overall unpreparedness for emergencies like a pandemic (ie lines to get paper towels and toilet paper), decided that when they saw police stations and Target stores and CNN buildings getting set on fire....that they decided to be prepared for those activities and proactively bought a gun for protection? I have anecdotal evidence from friends that would have never owned a gun prior to May 2020, but have one in their home now. The increase in gun transfer is the effect, not the cause?
Totally hypothesising here and I don't have time atm to research this, but it seems weird to me that my first guess at the cause hasn't been mentioned anywhere I can see. Take a society with lots of guns and a strong suspicion of government and introduce a new scary phenomenon that cuases mass unemployment and house arrest. Take that increase in murderous intent and repress it with lockdowns and reduced mobility for the first few months (allowing for an increase in murderous intent but a reduced opportunity for murder) and then reduce movement restrictions and see what happens.
Hopefully will gget time to return to this and post some evidence.
If you're tired of us foreigners discussing a situation in US and want to engage in looking at problems in other countries for a change, this is for you:
In Germany, former UN expert Nils Melzer just accused the government of 'systemic failure' in the oversight of police. To my understanding the main accusation was, that goverment and judiciary system don't react adequately to cases of misuse of force by police. (https://www.zdf.de/nachrichten/politik/corona-demo-polizeigewalt-100.html).
In Germany, the rule of *Verhältnismäßigkeit* (proportionate reaction) is key. It rougly means, that police of course is allowed to use force, but only to the extend to which it is necessary in a given situation. Melzer found, that there were too many cases where force was used excessively (eg. taking sb. to the ground, even if the situation was already under control). The main problem here, according to him, is not only those cases, but that they are rarely prosecuted, and that the goverment doesn't really care much about them.
In case you care, the topic of excessive police force came up in the context of anti-Covid-measures - demonstrations. Not that I think this has any relevance.
Overall, I guess police is rather well trusted in Germany (except by the usual suspects like very leftists groups). Being part of the overall 'trust' vis-a-vis police, I think that most do a great job. A general tendency to protect police officers even if they overdid it, internally in the police, as well as in the judiciary system, + minimizing the problem by the goverment, is however plausible.
Here is a case that made it to the news: a girl that happened to be close to Anti-G20 protests wanted to go home, biked right towards a police barrier she thought was still open for passage, and ended up with a broken arm and severe PTSD. The article describes at lenght how the internal investigation was diverted and delayed for five years, until it was finally closed. A parallel lawsuit at an administrative court resulted in a statement that the force used by the police was illegitimate - this however has no concrete consequences. I don't think this one piece says all that much about the overall context; I would mostly trust that the main key points of the story itself have been well researched. (https://www.zeit.de/gesellschaft/2022-07/g20-gipfel-hamburg-polizeigewalt-gericht/seite-5)
Then of course the question, how many police officers were killed. I found the number to be > 400. Since 1945. (https://www.welt.de/politik/deutschland/article236666105/Mehr-als-400-Polizisten-im-Dienst-getoetet.html). If you scroll down, the article contains an overview of each single case in the last 20 years. No killings in 2018 and 2019, two police officers killed in 2020 and two killed in 2021. I remember the last case, it was in national news with lots of declarations of outrage and sadness.
The last two bits make me wonder, whether we live in a specifically peaceful country, or whether there is lots of crime and problems, but just below the level of killing each other (where police is involved). Probably a bit of both.
I want to make the argument that the media coverage of the murder itself is what led to the increased murder rates.
It is already well known that the wide-reporting of suicides leads to an increase in suicides (and, sadly, an increase in car and commercial aircraft crashes). It is similarly known that the reporting of murders leads to an increase in murders in the areas of reporting. This is most notably observed following, again sadly, school shootings, when numerous shootings occur in succession. So, I think the argument can be made that the massive media coverage of the murder of George Floyd led, in turn, to the murder of more black men.
This would track with the logic that in reporting suicides and murders you lead to copycat suicides and murders. Where ages are reported in articles discussing suicides, the increase in suicides occurs in those within that age group/range. It seems to follow, therefore, that massive reporting of the murder of a black man would lead to copycat events. This would seem to explain why deaths of other ethnicities do not see a staggering increase. Likewise, one imagines that coverage of the murder of George Floyd was more extreme in the United States hence why the increase is only observable there (albeit I would need to actually see if the coverage was greater in the US). I would add that this explanation would also cover why media outlets have discussed other reasons; to state this is the reason would make them, in some way, complicit.
Timings wise it would also make sense. The coverage and the protests would occur contemporaneously, each feeding off the other.
I appreciate that this thought is unsourced, and low-effort in that it may have been covered by other comments, or by ACT and I have just missed it. Regardless, any discussion based on this thought would be interesting.
Has there been any attempt to ask the murderers about this?
Would it be possible to get a hold of a sample of people who attempted or completed homicides in the summer of 2020, and ask them why they did it, what pushed them over the edge etc? Or perhaps just do a deep analysis of court proceedings from those cases.
I agree with most of Scott’s analysis but I think we could do a lot more to understand the causal pathways at play.
The "stock of guns" argument seems flaky to me. No data, but I expect that a "new" gun would be vastly more likely to be involved in an actual shooting. There's various reasons including: People buy/obtain guns for a reason, a lot of the stock is tied up in large collections which won't all be used, new gun owners may not have thought ownership through and modified their guiding narratives and reactions.
The CDC WONDER database this week now has pretty solid data thru December 2021 on homicide victimizations.
I come up with black deaths by homicide going up 37.8% in 2020 over 2019, Hispanics up 28.2%, non-Hispanic whites up 20.7%, American Indians up 20.4%, and Asians flat at 0.0%.
For 2021 vs. 2019, blacks are up 43.8% and Hispanics 42.6. Whites are up 19.2%, Asians up 7.2%, and Native Americans 5.5%.
These are not murders perpetrated but victims of homicide. The FBI collects data on deaths by murder, which are a smaller but closely related number. The FBI has demographics on perpetrators of cleared murders, but a large fraction aren't cleared. Also, the FBI numbers are all snarled up involving Hispanics and whites, whereas the CDC numbers follow the usual modern government method of giving priority to Hispanics or Non-Hispanics, then looking by race at Non-Hispanics, so the CDC numbers are pretty useful.
I have qualms about the “gun sales” argument. Unlike hot weather and a pandemic, increase in gun sales do not just happen for no reason. It can happen as a result of a change in legislation, it can happen because TV had a cool show about a gun nut, it can happen because people are preparing to overthrow the government. But it happens for a reason. It is not an engine of the change, only possibly a transmission belt. So you cannot explain something *by* an increase of gun sales, only *through* it to the root cause.
(I have the same qualms when Aella quotes fatherlessness as a cause for social disparity: fatherlessness is not an intrinsic trait of a certain population, it is a consequence of other past social factors.)
To me, the protests were a breaking point, perhaps a spark that set off a fire that was already burning: a resentment of the privileged, of the status quo, of rules and civil order and bourgeois niceties that we take for granted which has been growing more and more in the last years, but accelerated during the pandemic. I see it in the recklessness of drivers, in the dismissal of property crime as victimless and deserved, in the attitude that most jobs, and expectations within those jobs (dress codes, uniforms, punctuality, etc.) are de facto exploitation.... feelings that were already festering, but the protests freed it all up - hence more lawlessness, more traffic incidents, more violence, and more murders too.
Also, this is a tangential point but any treatment of the protests as spontaneous seems off-base to me. They appeared almost simultaneously in so many cities at once, large, medium, small ones, including communities where large protests and social unrest weren't ever a thing in my memory.... suggesting a well-organized operation that had been preparing or planning this sort of action. And around it of course lots of people came out of their own accord. I don't know how to test my hypothesis about this, or whether it changes the conclusion about WHY the murder spike.... But I think it is worth thinking about and taking into account.
I’m a newbie here, but I’ll get this right out: Ever since Ferguson in 2014 I’m not sure very many amateur writers have dived into this topic of crime, race, and policing in quite the same way I have. Mostly I’ve put out Medium pieces and social media posts, but below I’ll just rattle off a few resources that might be handy for folks.
I 100% endorse Scott Alexander’s conclusions. It’s obvious it's the anti-establishment, anti-policing protests that have caused the near 29% spike in crime in 2020, just as nationwide protests led to consecutive double-digit percentage increases in 2015 and 2016. I’m forever impressed how Scott addresses critics, and I’ll repeat his request: “Please find me any major country besides the US that had a homicide spike in 2020.” It galls me how many obfuscate and point to the COVID pandemic for explanations. Talk about motivated reasoning.
• This 50% increase over six years means that from 2015 to 2021 we’ve likely seen an excess of 26,500 murders if 2014 is the baseline (50% black victims). Compare this to the 40-50 unarmed killings a year (25-35% black). This 26,500 number I calculated very simply in this 2nd tab “Murder increase 2014-2021.”
• In 1st tab “The Real BLM Effect” I calculate and hypothesize that some significant part of 180,000 lives were saved by incarcerating and arresting violent offenders, as the USA dropped from an average of 24,000 annual murders (1990-1994) to a low of just above 14,000 murders in 2014. We’re now past 21,000 murders again in 2020 and 2021. I quote Stephen Leavitt of Freakanomics fame who claimed these four factors led to violent crime decreasing: "Increased incarceration, more police, the decline of crack, and legalized abortion." (Source: http://pricetheory.uchicago.edu/levitt/Papers/LevittUnderstandingWhyCrime2004.pdf)
• In the 4th tab I ran the numbers to show that a black person killed by police is 12 times more likely to get a news story than someone non-black. That’s what you get when blacks are 25% of the people killed by cops but 80% of the news coverage. Obviously, black lives DO matter. Yes, the media is brainwashing our collective brains on this topic, and it’s why so many people believe cops are routinely going around killing black people willy nilly. It’s all we see, from CNN to the New York Times to our local affiliates. (Sources: https://www.commentary.org/articles/wilfred-reilly/no-there-is-no-coming-race-war/ & https://www.skeptic.com/research-center/reports/Research-Report-CUPES-007.pdf)
THAT’S AMATEUR NUMBERS, GIVE ME ACADEMIC RESEARCH!
7 papers and analyses indicating a Ferguson Effect certainly occurred the past few years, defined as an uptick in violence and homicides as a result of increased police scrutiny and protests. This combination delegitimizes police in the eyes of citizens, causes police to pull back, and often leads to policies detrimental to public safety (like consent decrees).
• Dataset countering BLM and #SayHerName myths I put together on Fatal Force using 3 years of Washington Post data (and many other tabs with nerdy number crunching): https://tinyurl.com/http-blm-sayhername-myths
3 MEDIUM ARTICLES I’VE WRITTEN
I humbly serve these up for critique and/or for people to use as they see fit.
#1 Titled: “Kamala Harris: Biden’s VP Pick & Your Next Candidate to Libel Police & Criminal Justice”
The larger picture shows that African Americans are not being hunted down by police. But Americans are hunting down each other. In fact, from Chicago to New York City, black Americans are more than three times less likely to be killed by police than previous generations. Yet vast black-white disparities in murder haven’t changed at all.
For at least four decades African Americans have murdered others at a rate 8 times higher than their non-Hispanic white counterparts, and are killed at a rate 6 times greater. In 2016, CDC indicated there were more than 19,500 homicides and The Washington Post reported 234 black killings by police and 465 white killings by police. This means for every 40 blacks killed by fellow citizens, there’s only 1 by a cop. For whites, that ratio is 12:1. But even in August 2019 we get yet another questionable study omitting crime statistics, and incredibly arguing that police are a major threat to the lives of blacks, with this Harvard Kennedy School headline: “Black men 2.5 times more likely than white men to be killed by police, new research estimates.”
People simply need to do the basic math. Or does the math not add up to the social injustices activists, academics, and the media have been portraying?
[end excerpt]
-
#2 Titled: “MY OPEN LETTER TO CHICAGO: Stripping Context from Media & Government Reports on Police Abuse a Likely Cause of More American and Chicago Bloodshed (Subhead: Statistical-Based Evidence Undermines Consent Decree Logic that Throws Chicago Police Under the Bus)”
THE GIST: Chicago-centric hard analysis w/ stats and graphics I created to try to help prevent a consent decree from taking shape. Roland Fryer’s 2020 paper highlighted how these consent decrees have led to potentially thousands of lives being lost. Had more than 20,000 views back in 2018.
-
#3 - Written directly after the George Floyd protests as the writing was on the wall, but few knew: Only around 10 unarmed black people were killed by cops the year before. Skeptic Magazine’s polling shows how large numbers of liberals believe 1,000 or even 10,000 unarmed blacks are killed by police each year. Note: Those folks driving our “national conversation” are in media and academia, yet their beliefs undergirding their “reform” demands are wildly off.
This goes without saying: These unarmed deaths are real lives and real tragedies. Just as the 62 police killed by felony gunfire in 2021 are, a jump from 45 in 2020. 346 officers were shot in 2021.
These are the final “unarmed and killed by police gunfire” breakdowns in the context of “Black Lives Matter” protests for 2019, the year before George Floyd’s death:
• 26 white
• 12 black
• 11 Hispanic
• 5 in the category of “other”
But look at that Skeptic chart above again. It’s also crystal clear that more than 50% of “very liberal” Americans live in a fictional reality — a morose and racist universe where the number of unarmed Black Americans killed by police might be as high as “about 1,000,” “about 10,000,” or “more than 10,000.” And it’s those “very liberal” people making the biggest stink on Twitter, trust me.
[end excerpt]
Fin. And hope to have many fun conversations here.
"One reason might be if the people buying guns in 2020 were very different from the people buying guns in previous years. For example, if previous gun buyers were collectors who had 100 guns each, but 2020 gun owners were new buyers getting their first gun, then the share of people with at least one gun would go up by more than 2% over an average year."
Maybe the suicide rate didn't increase because the extra Americans who bought guns in 2020 were not people in suicide-prone demographics. Suicides are most common among older, white men living in rural areas, and I remember hearing about how many of the people buying guns in 2020 were women, black, or living in urban or suburban areas.
Late to the conversation, but this feels worth noting given Matty's point RE: the media's characterization of the homicide increase
Right on queue, on 7/8/2022, the Wapo published an article titled:
"The staggering scope of U.S. gun deaths goes far beyond mass shootings"
The print version, where this article was the header article on the front page, had the following subheader
"45,000 fatalaties in each of the past two years (line break) Increase conicides with record firearm purchases"
The article (and notably its actual interviewee quotes) goes on to hedge a bit but clearly wants to suggest that the gun purchase increase combined with covid stress is the likely culprit.
It cites lots of data on new gun purchases without ever comparing this to the existing gun stock or the existing gun ownership rate
It even correctly cites the race/skew of homicide and suicide victims (black male skew in homicide victims and white male skew in suicide victims) but it of course fails to cite the skew in homicide perpetrators or the race/sex characteristics of the perpetrator increase.
I think an under-ratedly weird aspect of the post-protest homicide spike is just how FAST it happened. Like, I wouldn't have been that skeptical of the protest/crime-spike link ex ante, but I would have guessed it would have taken, I dunno, months to set in gradually. When people try to deny the link, I think they're sometimes pattern-matching to other social phenomena in which there are meaningful lags between cause and effect, leaving all conclusions very fuzzy and uncertain. But here the whole thing played out in the blink of an eye. It seems almost too good to be true.
I have the same dizzy sensation regarding the CHAZ/CHOP debacle (https://en.wikipedia.org/wiki/Capitol_Hill_Occupied_Protest). I would have predicted ex ante that a miniature anarchist commune would run into trouble pretty quickly, but it was literally only a matter of days before these people reconstituted a police force that shot and killed an unarmed black teen! "Reductio ad absurdum" doesn't begin to cover it.
I guess the update is that...there ISN'T much ruin in a nation?
I'd like to signal boost the contributing factor of court closures to the perception of potential criminals that they would be unlikely to face (swift enough) consequences even if they did commit crimes.
In my perception it seems like the argument about changed police behavior (in part due to BLM protests) is implicitly assuming a model where the police behavior causes perception-of-risk amongst criminals, which in turn is the actual proximate cause of (the increase in) crimes committed. I wanted to draw more explicit attention this part of the causal chain.
I'd be curious to know if there was a good way to estimate the effect size of changed court behavior (probably mostly caused by responses to the pandemic) versus changed policy behavior (probably mostly caused by responses to BLM protests). I do think that police, prosecutor, and court behavior is the main cause (feasibly manipulable contributing factor) of the magnitude of violent crime and are much more significant than small changes in gun prevalence or other pandemic-related factors, in partial agreement with Scott's thesis.
Regarding the Alexander/Yglesias disagreement about media coverage: I think both positions make a lot of sense if we model media coverage of a generic "what causes Y?" questions as a binary choice between the "A is THE ONLY cause of Y, if we control A we will completely control Y (we literally cannot conceive the concept of multiple contributing factors or causes)" template and the "wow Y is complicated, look we can sit here and make up contributing factors all day, it sure seems impossible to compare factors and draw conclusion" template. So from Yglesias's perspective the media chose the correct template for the situation, and from Alexander's perspective the templates suck.
I think it is better for our blood pressure to recognize that this is a structural problem with the way news media is written that affects every topic they cover, and that the media is not specifically trying to hide the truth on this issue.
(Note: the thesis here is reliant on new media using a small fixed number of framings / templates for everything where the choice of template is the entire analysis of the topic, not on there being exactly 2 templates.)
I find it kind of hilarious that both financial hardship from the pandemic and having better finances than usual due to stimulus checks (which supposedly was spent on guns) were being offered as (entirely ad hoc) explanations for the BLM effect.
I also found it hilarious how many commenters accused Scott of coming to the conclusion first and then desperately looking for a way to justify it - followed shortly thereafter by desperate attempts at dismissing the huge amounts of data Scott provided and asserting some politically convenient explanations with little to no data to support them.
I think the distinction between depolicing and a collapse of trust in police is important (though of course you have every right to be agnostic on the question).
If the murder spike was a result of police pulling back, the lesson would be that getting police to be more active would save lives. That could mean making sure police feel that they'll be supported by politicians if they make tough calls in tough situations and not giving police extra paperwork.
If the murder spike was a result of the public losing trust in police, and so not calling the cops when there's a problem, the solution is the opposite--restoring public trust in police by holding police accountable when they do something wrong and requiring careful record-keeping.
Regarding the racial disparity in homicides Andrew Sullivan retweeted this recently:
https://twitter.com/sullydish/status/1544375597312385024
The homicide rate right now for the black population in Chicago is at the highest recorded level since the start of modern record keeping. All those people saying that crime was much worse in the 1990's? Not for black residents of Chicago. The previous high in the 1990's was about 75 homicides per 1000 people. In 2021 that hit 85.2 incidents per 1000 people. By contrast the homicide rates for whites and Latinos are nowhere near their all-time highs.
The first question I have is whether or not this holds true for other large metropolitan areas as well. If so then aggregating homicide rates across an entire city is a mistake. The hypothesis that needs to be tested is whether or not the Chauvin/Ferguson/Freddie Grey/etc. incidents resulted in radicalization of black communities specifically rather than metropolitan populations as a whole.
Edit: more specifically poor urban black communities. If there is a historically bad epidemic of black homicides then why isn't that more apparent? After all the average US homicide rate has risen to 1996 levels but still falls short of historic highs around 1991-1992. I can think of a couple of factors. While blacks make up about 12% of the US population and have for decades the percentage of blacks living in poverty has declined sharply since the 1990's (peaking at around 31% in 1992 and declining to around 17% today). Potentially related to that with regards to Chicago specifically the black population of that city has also been in decline for decades, since at least the 1980's, even as the Latino population in that city has soared. If the twofer is urban poverty than a decrease in black poverty rates plus a change in the racial composition of cities could result in an average homicide rate that is still short of all time highs even as poor inner city communities face historic levels of devastation.
Thank you for so properly refuting the garbage analysis from Artifex0. His is a stupid idea that just won't go away despite the evidence to the contrary.
> If you have some clever reason why Central America is also a bad comparison, then please find me any major country
Noted, we should do more homework as a rule. We DO appreciate you doing the (occasionally tedious) research - it's still much better that somebody does it, and it helps a lot in fostering a culture of doing it. Even if it's a slow process.
Speaking of, I noticed recently my google-fu is lagging behind the times. Can you maybe do a post on "how to research"? Including stuff you find obvious or simple, like what sites you go first, or what tool you use to quickly generate a graph. For example: I realized annoyingly late that on sci-hub you should search by DOI, and not by paper title.
How do you reconcile this
> I don’t really think of this as an alternative explanation. I am agnostic to the exact causal pathway between the events of May 25 2020 and the homicide spike; all I’m trying to show is that the spike did begin around that time and seems connected.
And this?
> My claim is that this is false. It is not difficult to assign priority. The protests were the primary cause, with the other two being minor contributors at most. When I say the media is getting this issue wrong, I mean that they’re saying things like this.
In the first you're saying that there was a homicide spike as a result of the events that occurred right around Floyd's killing, and in the second you're saying specifically that it's the protests (keep in mind the comment to which you responded with the "agnostic" point is saying that it is *not* the protests). My reaction to the initial post was that you can't disentangle the protests themselves with other things that happened at the same time. You say you're agnostic between those things, but then also that it was the protests.
This leads to another issue I had, which is a question of which groups have agency and which are merely reacting to other stuff. In the "BLM did it" version of events, BLM protests lead to police doing their jobs less which leads to more crime - this is interpreted as "it's because of BLM". Assigning BLM agency to cause things to happen, and the police actions as being a natural reaction to outside events. Why not treat the *protests* as a natural reaction to outside events, same as you do with police doing their jobs less, and the *cause of the protests* as being the ultimate culprit?
When a negative event is a result of the actions of multiple people (or groups), often each person/group will characterize their own actions as being a mere *reaction* to someone else, and the *other* group as the real cause of the bad thing. The question becomes, who do you blame / put the onus on to change? Usually the answer is "whoever I sympathize with the least" but it's often not examined all that much.
Wonder if Canada had a big rise in murder. I can't find any up to date stats. We did have a lot of simultaneous protests.
My prediction... a smaller increase than the US of A, but an increase nonetheless. Also an increase in traffic fatalities.
It could be instructive to have some analysis of motives in these homicides.
Something that is deeply frustrating to me are complaints that collecting data on police stops (traffic, terry, or otherwise) is a problem. First, it's distasteful to me as an American to claim that creating a record of a stop is overly burdensome to the state security apparatus; I was taught to believe that undue search and seizure are things we should be guarding against. It's hard to guard against things that are invisible because there is no (legible to inspection) record. Second, there's increasingly smooth software to reduce the pain of paperwork (e.g. Mark43).** Third, records of police actions are useful to from an efficiency standpoint, as a lack of any stops may also be concerning.
So, data is good in my opinion. If collecting the data is painful, then we should seek to ease that pain, because deterring police work is inefficient. But arguing that silly reformers passed the Racial Identity and Profiling Act -- and that's why police can't police -- is itself rather silly. This is because it ignores that policing is part of a state's obligation to the general welfare of its populace -- policing isn't measured only from how efficiently police can deploy coercive power.
** See the exact requirements here: https://mark43.zendesk.com/hc/en-us/articles/4410748957965-Understanding-and-Completing-a-RIPA-Report
These questions should be simple for an officer to fill out or articulate.
Enough about my feelings though, what are some brass tacks about why we should care about recording data on stops? Well, let's look at traffic stops. Traffic stops are nicer for criminologists because they are usually discretionary and also usually recorded, because license plate checks themselves are recorded as part of almost traffic stops. These records are often enriched by the demos of the person(s) stopped, time/date info, and geolocation. This has allowed extensive empirical analyses to study both 1) efficiency and 2) bias in traffic stops. Findings indicate that traffic stops and searches are racially biased (when daylight allows), and this racial bias results inefficiency as searches of Black drivers are less likely to produce (drugs, guns, etc.). However, jurisdictions, units, and officers vary significantly. It's a well-known finding called the 'veil of darkness' (racial disparities disappear at night and often on highways, where it is difficult to discern phenotypical racial characteristics). The latest standout study in this vein is Feigenberg and Miller 2022 in the Quarterly Journal of Economic. Here's an imgur link of the most important point: https://imgur.com/a/renWnIu
Basically, searches are racially biased and inefficient in Texas, and equalizing them conditional on stops would yield more contraband and less racial bias. This isn't a shocking finding - it comports with the greatest hits of that literature:
Chanin, Joshua, Megan Welsh, and Dana Nurge. 2018. “Traffic Enforcement Through the Lens of Race: A Sequential Analysis of Post-Stop Outcomes in San Diego, California.” Criminal Justice Policy Review 29(6–7):561–83. doi: 10.1177/0887403417740188.
Feigenberg, Benjamin, and Conrad Miller. 2022. “Would Eliminating Racial Disparities in Motor Vehicle Searches Have Efficiency Costs?*.” The Quarterly Journal of Economics 137(1):49–113. doi: 10.1093/qje/qjab018.
Fryer, Roland G. 2019. “An Empirical Analysis of Racial Differences in Police Use of Force.” Journal of Political Economy 127(3):1210–61. doi: 10.1086/701423.
Pierson, Emma, Camelia Simoiu, Jan Overgoor, Sam Corbett-Davies, Daniel Jenson, Amy Shoemaker, Vignesh Ramachandran, Phoebe Barghouty, Cheryl Phillips, Ravi Shroff, and Sharad Goel. 2020. “A Large-Scale Analysis of Racial Disparities in Police Stops across the United States.” Nature Human Behaviour 4(7):736–45. doi: 10.1038/s41562-020-0858-1.
Roh, Sunghoon, and Matthew Robinson. 2009. “A Geographic Approach to Racial Profiling: The Microanalysis and Macroanalysis of Racial Disparity in Traffic Stops.” Police Quarterly 12(2):137–69. doi: 10.1177/1098611109332422.
Taniguchi, Travis A., Joshua A. Hendrix, Alison Levin-Rector, Brian P. Aagaard, Kevin J. Strom, and Stephanie A. Zimmer. 2017. “Extending the Veil of Darkness Approach: An Examination of Racial Disproportionality in Traffic Stops in Durham, NC.” Police Quarterly 20(4):420–48. doi: 10.1177/1098611117721665.
Withrow, Brian L., Jeffery D. Dailey, and Henry Jackson. 2008. “The Utility of an Internal Benchmarking Strategy in Racial Profiling Surveillance.” Justice Research and Policy 10(2):19–47. doi: 10.3818/JRP.10.2.2008.19.
Zingraff, Matthew, William Smith, and Donald Tomaskovic-Devey. 2003. “North Carolina Highway Traffic Study, Final Report.” Department of Justice.
The NYC data is actually available in incident-level granularity, if you look at shootings. You can download it yourself here:
https://data.cityofnewyork.us/Public-Safety/NYPD-Shooting-Incident-Data-Historic-/833y-fsy8/data
I did download it, and here's what a very simple analysis of it looks like:
https://i.imgur.com/kXxSKiR.png
The blue line is the difference over the same smoothed 7 day period the year prior. Note that this is a trailing average, so the data point "today" is the arithmetic mean of the prior 7 days (including the current date). The vertical red line is Floyd's death. There is indeed support for some non-trivial excess shootings in the earlier part of may, but it's clearly dwarfed by what comes afterwards.
I also did the analysis using the average since 2006 of that same 7 day period as the baseline, but the chart looks very similar so I won't bother to upload that too.
Don't have any nitpick, just passing by to show my gratitude for you drilling deep down on a contentious topic that I feel the MSM should have covered with nuance but instead simply used as an opportunity to push sloganeering. Quality content like this is why so many of us read you.
"what would be the explanation for why this trend would start on May 20 or something?"
Wikipedia tells me that the stay-at-home order expired May 17. I mostly believe your thesis, but you must admit that the alternative "The murder spike was caused by COVID damaging mental health and the economy (or any other cause) but started only with the end of lockdowns" also fit the timing. It probably fails on the international comparison, but you should not overstate your case...
Don't know about the homicide spike but the trends in the homicide graph from central America are beautiful to see. What has been happening in El Salvador ? That's an impressive decrease !
Above my pay grade to comment on the particulars, so...just gonna leave some appreciation for the hard work. It's nice to have a feeling of relatively firmer empirical ground "in real time" as opposed to years later when the Official History Books(tm) are written. Scott going all Scott on thorny Current Issues is both (probably) the most controversial content, but also the primary reason SSC/ACX are my favourite blogs. You just write better than most when clearly annoyed/angry/jerk-ish. "A spoonful of heat helps the light go down"
Is it not possible, that protest actually did cause a decrease in policing AND, in parallel caused an increase in homicide, BUT homicide increase was not caused by de-policing? (or at least another factor had ~the same effect size as de-policing)
My theory is that protests and news about protests can temporarily change the behaviour of people. It may result in more people being on the streets, and them being more concentrated into smaller areas. In addition, based on "Politics is The Mind-Killer/Toxoplasma Of Rage/etc" articles, they may be more agitated and more prone to violence in such cases.
So: more people outside + concentrating them in specific areas + being agitated/outraged --> increased probability of violence happening --> increase in homicides
The comparisons with other countries is a tricky one. In the original post you looked at the murder rate in European countries and saw no peak. But those countries, as you point out, have very low levels of gun ownership and also start from a lower level of homicide. That means that the number of people who can easily kill someone if they suddenly wanted to is already low. In the US, the high stocks of guns should allow more potential murderers - which may reflect in a more visible spike when factors combine to "encourage" homicide.
Perhaps looking at violent crime figures in Europe (without resulting in death) could be more reflective than just homicide rates. A quick look at the numbers from London suggests they also fell significantly in 2020 and 2021 (https://www.statista.com/statistics/864736/knife-crime-in-london/), while the murder rate doesn't show much of a change (https://www.statista.com/statistics/862984/murders-in-london/). So it doesn't look like the pandemic conspired to increase crime there (indeed, the opposite is probably true). Interestingly, the lack of relationship between murder and knife crime rates should support the idea that most knife crimes don't end up with some dead, but I suspect most gun crimes do.
I'd also caution the comparison with central America. The degree of pandemic restrictions and the level to which those restrictions were enforced varied greatly across the world. Having spent some time during the pandemic in less developed countries, I can tell you that in many places you'd barely know there was a pandemic at all (other than seeing a few badly worn masks).
In your original article on this topic, you specifically claimed that the increase in murder rates was due to a police pullback. When you were challenged about this, you claimed you are actually agnostic on the issue.
I'm disturbed by the gun suicides vs homicides graph. This post glosses right over the (granted, unrelated to the topic) GIGANTIC STEADY INCREASE IN SUICIDE OVER A 12-YEAR PERIOD FROM 2006 TO 2018. But that seems more interesting than the actual topic of the post.
Could we at least get a comment like "Don't be too alarmed, the gun suicides are displacing other suicide methods" or "Wow! What happened to suicide rates over the last 16 years?"
"So this theory requires us to believe that number of guns increasing 3.5% every year from 2015 - 2020 had no effect on the murder rate, but that guns going up 5.5% in 2020 had a very strong effect on the murder rate. Specifically, an extra two percent increase in guns must lead to a 30% increase in murder rates. Why would we believe that?"
Isn't this comparing a percentage point increase in gun sales to a percent increase in murders? That seems just as big a sin as comparing stocks and flows.
Putting both in percent increases, a 63% increase in sales was associated with a 30% increase in murders.
The rest of that section was pretty convincing, but I think that particular argument is bad and your point would be stronger if you omitted it.
Regarding rising gunsales:
There might be a correlation between BLM protests and rising gun sales:
If citizens imagined (rightly or not) police during the BLM protests, for fear of being labelled racist, were less likely to protect the public and their property against violent protesters , then citizens would feel the need to arm themselves. If police were defunded, this would provide even more reason to do so.
If the above is true, even if gunsales affected homicide rates, the latter would indirectly be affected by the BLM protests.
On the question of 'police pullback' vs 'distrust in police', you write:
> I am agnostic to the exact causal pathway between the events of May 25 2020 and the homicide spike; all I’m trying to show is that the spike did begin around that time and seems connected.
If you yourself are agnostic, you should be aware that your original essay is very firmly not so, and is prominently on the side of police pullback. You give it in your opening sentence:
> In my review of San Fransicko, I mentioned that it was hard to separate the effect of San Francisco’s local policies from the general 2020 spike in homicides, which I attributed to the Black Lives Matter protests and **subsequent police pullback**.
You have a *topic heading* "Police pullback" that opens with "My specific claim is that **the protests caused police to do less policing** in predominantly black areas" then gives multiple reasons that police would pull back, concluding: "I don’t want to speculate on which of these factors was most decisive, only to say that **at least one of them must be true, and that police did in fact pull back**."
Elsewhere:
> My interpretation is that people complied with the strict lockdown early in the pandemic, that effect was played out by May, and then separately **the protests caused a longer-term decrease in policing**.
> But there are lots of reasons to expect that the Black Lives Matter protests would **cause police to pull back** from black communities in particular.
> The New York Times had an article Deconstructing The Ferguson Effect, subtitled “The idea that **the police have retreated under siege** will not go away. But even if it's true, is it necessarily bad?”, which as far as I can tell is **as close as the New York Times has ever come to acknowledging that a politically inconvenient fact is true**.
> No country except the United States had a large homicide spike in 2020, which suggests that the spike was unrelated to the pandemic and more associated with US-specific factors, for which the BLM protests and **subsequent pullback of policing in black communities seem to me to be the most obvious suspect.**
These highlights posts are some of your best ones. I really appreciate the effort you put into making them.
You had to be there Scott. Everyone around Mpls knows the police are feeling unloved right now. There was stupid overreaction locally after Floyd’s death and we all know that too. There were a lot of early retirements and resignation as a result. Those that remain are understandably reluctant to leave their cars. Less enforcement leads to more crime.
There were several pieces in the NYT that made similar points to yours. They were more nuanced than your position. This sort of analysis should be nuanced. Right wing media beats the ‘BLM protests caused increased crime’ drum hard and often because their business model is to generate outrage.
Treating the BLM protests as a first cause in the chain of events is simply wrong.
I don’t understand why you are taking this tack of being a lone defender of The Truth. I honestly don’t think you are. You’ve analyzed a complex series of events by looking at one link in isolation and the result was that people felt free to use words like ‘libtard’ and reinforce their arguments with the rhetorical device of CapsLock shouting.
Yes, NYT has a moderate liberal bias. I try to keep that in mind when I read it. Is it possible that you have a moderate anti NYT bias?
What are you hoping to accomplish?
> I am agnostic to the exact causal pathway between the events of May 25 2020 and the homicide spike; all I’m trying to show is that the spike did begin around that time and seems connected.
This isn't consistent with the language you use elsewhere. I don't think there's so much disagreement about the correlation.
> Also, what would be the explanation for why this trend would start on May 20 or something? There isn’t more pandemic that day. There aren’t more guns that day. It’s not even especially warm that day. I think it’s got to be an artifact.
What? This reads as really close to assuming what you're trying to show. Could you spell this reasoning out?
It's also funny you should say May 20, since it was the first day with a high above 75 F in Minneapolis that year, as well as the first with a low above 55, going by https://www.timeanddate.com/weather/usa/minneapolis/historic?month=5&year=2020. (Also, as another commenter points out, the stay-at-home order expired after May 17.)
By the way, it looks like you've put the black line for "Floyd's death" about a day and a half early. It should be on the shoulder of the first peak.
"Graham" wrote: "When an officer who is fired or prosecuted for something that is clearly unreasonable (like Derek Chauvin, whom all cops agreed was guilty)"
I am going to press [X] to doubt. I haven't been able to find any surveys that specifically ask current/former police officers what they think of Chauvin's actions. I've found articles that report the findings of interviews with cops. Those who are willing to go on the record, particularly high-ranking cops who don't do any actual policing and are sensitive to political considerations (not "real po-leece," as Det. McNulty would say) , are unanimous. But that begins to break down as soon as anonymity is provided, and as the interviews work their way down the chain to beat cops. To the extent that I've found a consensus (on a first-pass review of mainstream news articles), that consensus seems to be that Chauvin crossed a line but wasn't guilty of murder.
https://is.gd/cHsZ2g
>> There's an alternative explanation that fits the evidence here: the killing of Floyd itself caused the crime increase by damaging trust in the police, which led to an increase in retaliatory violence.
The article linked below that comment is focused on gangs. In my general model of gangs (on which I am an Noted Expert because I watched Breaking Bad and The Wire) and their feelings about the police, trust is not the word that comes to mind.
Of course, the probability of a murder suspect getting quickly arrested greatly affects the retaliatory murders. In a city where, by magic, the correct suspect is arrested within 24h of any homicide, only very foolish gang members would try there hands at revenge killings, while in a city where murders are never prosecuted, revenge killings may be rational for a gang.
I do not see why gang members should update their estimate of that probability as a direct result of a murder committed by the police. A much more likely chain of reasoning seems to be
murder by cops => decreased trust in police in general population => depolicing => higher probability estimate (among gang members) of getting away with murder => more gang revenge killings.
Summing up all the steps in between as "damaging trust in police" seems like an oversimplification.
(The quoted statement does not directly refer to gangs, so it could also be interpreted as meaning that previously noncriminal people start committing murders because they lost their faith in the justice system. I find that unconvincing as well. I don't think the update from "Police are good people who serve and protect, so I should report crimes to them" to "Police are racist assholes, so I should not talk to them and avenge any crimes on my own" will happen after any one murder by a cop. I don't claim that the effect of eroding trust is not there, I just claim that it is unlikely to cause that spike within a few days of a single incident.)
Is homicide the type of crime that should be expected to increase as a result of reduced policing?
I would not expect my chances to get away with murder to increase significantly as a result of fewer patrol cars rolling through the neighborhood. If my fingerprints are on the shell casing, if the victim has my skin cells under her fingernails, if my own blood is at the scene, I'm going to get caught. The cops might show up to the crime scene later, but unless they're so lackadaisical that I'm able to dispose of the evidence and concoct an alibi, I'm in trouble.
Shoplifting, minor assaults, smash-and-grab thefts, and public urination should explode, though. Now there may be a broken-windows effect by which murder is a downstream consequence of reduced policing, but I wouldn't expect that to take the form of a spike.
I don't see why you dismiss the temperature thing out of hand, when a plot of murders vs. time of year is right above. There is a clear dramatic increase from late May to September every year, yes there is a "critical temperature of 78 degrees" or perhaps a "critical solar angle" ... why is that silly?
The MN data do show a large increase in baseline (red vs. blue) so "nobody was comitting extra pandemic aggravated assaults/murders" looks to be flatly false - the "un-rolled" data in particular show far, far more red events compared to blue in the winter. In any case the ratio in assaults between winter and summer in MN, about 2x, is the same for both the 2020 and baseline data.
We don't have month-by-month murder data for the country to show whether this "surge" is bigger than the usual summer surge. It looks to be proportionately the same in MN and bigger in NY.
In the bigger picture distinguishing between "bigger summer effect due to pandemic " and "it's BLM" is difficult because they happened only once and at the same time, and you really didn't try. You just said that you don't take it seriously, then made some sarcastic "wHaT is ThErE a CrItICaL TeMp LOL" statements as if it proved anything ...
> the homicide spike started at the the same time in a lot of different cities with widely varying temperatures
Maybe show this? It's clear from the MN data and NY data that they neither started nor peaked at the same time ... not that you have any sort of statistical definition of "start" to begin with or anything, nor did you show data from other cities, nor did you show that the correlation was tighter than the usual correlation in seasonal crime etc. etc.
Which is fine, it's not your day job (or is it, being an influential truthy-ness teller), but don't claim you've rigorously disproved something by saying "it's hard for me to take it seriously".
It is a sad comment in the state of modern academia and politicized issues that when someone tells me they are a PhD candidate at a top program in a politicized field, instead of my trust in their comment going up immensely, it only ticks up slightly over random stranger.
From what happened in the media and what criminologists were saying in the aftermath of the riots it is clear the discipline is very badly ideologically compromised. There are literally hundreds (thousands?) of papers who sole purpose seems to be misinterpreting data to avoid politically unflattering results.
It seems likely to me that we are nowhere near the Pareto optimal curve between policing and crime rate, just as in most governments' response to the pandemic we were nowhere near the optimal curve between economic impact and lives lost.
This situation would imply that while it may well be that police funding loss and increased restraint led to increased crime in 2021 specifically, it is still entirely possible that the best possible versions of "defunding the police", in which some large percentage of a city's police budget gets redirected to evidence based social services, could still result in a net decrease in crime.
People's day to day lives continue to get worse so crime continues to rise. I don't find it to be that complicated and I find this whole analysis and conversation to a bit moot.... I see a lot of people taking correlations and inferring causality while oversimplifying this complex social issue. Scott has done a great job at doing this, but because of that nature the conversation leaves much to be desired for me. Take this excerpt from Scott as an example of the subtle framing expressed in his phrasing through the different posts in this thread.
> I accept I should have put more work in the original post into ruling out gun sales as the cause.
Why is it "the cause" and not "a cause"? Framing is so important.
----
I recognize policing as a way to abate crime, data shows it and, for example, the NIJ recognizes its utility among many other approaches (https://www.ojp.gov/pdffiles/171676.pdf), but I know of no evidence that a lack of policing is a driving force for crime. Key word: driving.
Am I surprised to see what Scott presented and do I doubt the data? No.
Do I agree that a decreasing amount of policing can be, and this case was, a contributing cause to rising homicides? Yes, and unfortunately I've seen it anecdotally in my community.
Do I think this conversation is well framed, actionable, and effectively presenting a variety of perspectives? Not really.
I'm ready to move onto the next topic...
Canada did have more murders in 2020 than any year since 1991, though it should be noted that a significant percentage of our murders that year (3%) was the work of one mass shooter in Nova Scotia. https://www.reuters.com/world/americas/canadas-worst-mass-shooting-helped-push-2020-murder-rate-15-year-high-2021-11-25/
How do you determine if people are arguing in good faith in an argument like this?
Thing X happens via a very simple mechanism that many people predicted, people propose a whole series of complicated different explanations.
I suppose that education and health issues we discuss often have complicated explanations has damaged our epistemic health when it comes to other areas.
Crime is a simpler area because it involves lower IQ people, simpler systems and clear incentives.
It is hard to tell if the complex explanations come from people with an open mind or just ideologues upset that their theories fall apart.
Did assaults and other crimes go up or not?
I don't recall if this was addressed in the original essay.
If they did, then it's people doing it.
If they didn't, it's less policing.
What fraction of gun homicides use legal guns? I'd have thought that a lot of guns used in crime would not have been purchased through legal channels anyway, so the gun sales graphs would be irrelevant.
I was among those who wanted to read more about the homicide spike, so a late thanks to Scott for the post.
After reading the post, I asked an active member of the German Green Party about it; I was curious about the reaction I would get. (I would say sth. like: apparantly there is data showing that homicides among blacks grew after BLM protests.) What's your guess?
He confirmed this immediatly, his reaction was something like: yes, the police in the states is only trained to act while being violent. Then there are protests that they shouldn't be so violent, and they stop/reduce policing. They simply have no idea on how to carry out their duties without resorting to violence.
I admit I was surprised he knew the data and jumped right to: police is doing less (we're both living in Germany). It's just an anecdote, and I don't think it's representative for anybody. (I would still be surprised if the next nine Party Members or Green Party Members here knew.) I also think it shows nicely that even taking a BLM - homicide spike correlation as given, the question on *who has to change what* is totally dependant on the overall understanding of the situation.
I saw some documentations comparing police training in US versus selected European countries. The differences both in total length of training, and in length and importance of training in deescalatory measures seemed huge. I didn't did deeper, so I don't know how much it holds under scutiny.
I would be interested to know more; relevant questions in this context would include:
- overall lenght of training
- weight giving to training on how to deescalate dangerous situations
- admission requirements
- overall weight given to making sure that nobody is harmed or killed (including attitude & and how people talk about that internally)
- measures to ensure that extremists, both from left and right, can't serve as police officers.
I'm aware that contexts (among others, gun ownership in general population) differ widely.
>I can’t find temperatures for 2020 in particular, but here’s average temperature in New York City over the course of the year.
I found temperatures for 2019-2020, and graphed them to see how 2020 compared to the previous year.
https://i.imgur.com/R2TBdqF.png
This graph shows that the temperature in NYC was very similar to the previous year.
I have a big issue with using "police pullback" as a generic term because it obfuscates _how_ police pull back.
The Minneapolis police, for example, didn't say "let's reassess how we're interacting with the public and stop doing broken window policing" after George Floyd, they went full on "I'm taking my ball and going home." This is an excerpt from a store owner who used to have one of the best used book stores in Minneapolis which also happened to be close to where Floyd was killed:
"...There were 5 cops on the roof of the 5th Precinct building, watching and presumably reporting on events to a headquarters somewhere else, but there were no cops on the ground.
A little after 10 pm, some people pried a sheet of plywood part of the way off the door to a convenience store/gas station across the street from the 5th Precinct, crawled inside and grabbed some loot. There was no police response, so a few more people also crawled in and grabbed some loot...." http://www.unclehugo.com/prod/riotreport.php
The police bunkered up in their buildings and _watched_ looting and later arson go on. I find it a very disingenuous argument to blame reformers for the results of police literally watching felonies being committed without comment.
When they did actively police, many did so in the most incendiary way possible, to the point where a man was actually acquitted of shooting at police due to it being an act of self defense: Police where shooting rubber bullets from an unmarked van at random people without identifying themselves: https://www.ammoland.com/2021/09/man-not-guilty-in-self-defense-case-against-police-during-minneapolis-riots (Note that the author does his best to blame the incident on anything but the police, his account of the facts still has to admit that the police 1.) did not follow policy with regard to shooting people, 2.) did not follow policy when investigating the incident, giving the officers a chance to view the video and talk with each other before taking statements, 3.) made statements that conflicted with the videos they were shown before making those statements, and 4.) assaulted a man who had already surrendered himself.)
I think thinking in terms of police "pulling back" is at best not useful and more likely hiding the issue when it can cover activities as broad as not making as many traffic stops due to time required for paperwork and flagrantly standing and watching people looting and committing arson. Furthermore, saying police have "pulled back" when they are driving around in unmarked vans shooting people is highly misleading.
The discussion should be around changes in how antagonistic police policies and actions are, since the Minneapolis police have demonstrated that it certainly possible to both reduce the number of police interactions and increase the antagonism between police and the public at the same time.
Wanted to plug this Atlantic article from yesterday: "Six reasons why the police murder clearance rate has declined". Its a short (4 min read) interview with a crime analysts about why the murder clearance rate, or % of "solved" murders, has declined from 90% nationally in the 1960's to around 50% today. TLDR: Statistics until the 1990's are bunk and such a high % of murders weren't actually solved, firearm murders have increased from ~50% to ~80% and are much harder to solve than other murders, and distrust between cops and black Americans in areas with a high murder rate is worse than ever.
https://www.theatlantic.com/newsletters/archive/2022/07/police-murder-clearance-rate/661500/?utm_source=newsletter&utm_medium=email&utm_campaign=atlantic-daily-newsletter&utm_content=20220707&utm_term=The%20Atlantic%20Daily
I think the reason this is getting SO much pushback is that you're looking at extremely noisy time-series data and telling a just-so story about it (handwaving things like, maybe the early-May spike was random and the late-May spike is real, or maybe it's the rolling average--without actually trying to verify how it was averaged! IMO this was a significant oversight in the original piece.)
Not that this means your just-so story is WRONG, but it makes people want to tell other just-so stories. For example: the pandemic caused an increased murder rate in the US for <pick-your-politically-charged-reason-here>, but the lockdowns caused a countervailing decrease, so the real spike took a few months to show up. You've already pooh-poohed this theory (“it took a few months for people to get cabin fever from the pandemic”), but seriously, look at the movement data for the US in "Retail and recreation" or "Public transport stations" here: https://ourworldindata.org/covid-google-mobility-trends It craters in March, with a significant recovery through May and June, exactly when murders were spiking.
To be clear, the data from Baltimore are very compelling, and the demographic breakdown of the murder spike is suggestive. I'm convinced that publicized police murders + protests caused part of the spike. But I think the time series data were the weakest part of the argument.
The 'George Floyd's killing directly lead to the violence outbreak' claim is kind of shocking to me. If a bunch of black people went around killing more cops, white people, politicians etc. than normal, that would at least make sense (while being obviously morally indefensible to non-BLM-supporters). But what are they saying here? A black man got killed, and their response was to go out and kill a bunch of other black people? Putting aside the lack of empirical evidence for the claim, the implication of the claim strikes me as profoundly unflattering for black people, and if true should inform how we view black social pathology generally.
It's frustrating to see people devote so much time and energy theorizing about homicides without any understanding of the specific factors that distinguish homicide from other types of crime, even other violent crimes, leading to "explanations" for general increases in crimes and not specifically spikes in homicides.
The single most important thing to understand about homicides is that the majority of urban homicides are committed by relatively small groups of gang members, starting as drug/turf disputes and continuing as retaliations against previously unsettled homicides. Thus they are highly correlated in a way that has been compared to disease outbreaks - one homicide leads to another, which leads to another - which is not at all the case for other crimes likes robberies. (The excellent book "Don't Shoot!" by David Kennedy describes this dynamic as well as fascinating efforts to understand and break such cycles) Any explanation must take this into account. Trying to explain a percentage increase by looking for general increases in uncorrelated behaviors across large groups of the population is a waste of time.
It seems to me that *something* tends to happen after these high-profile incidents that leads to new retaliatory cycles, or more intense cycles, or some other changed dynamic from the previous status quo (which is already pretty bad in many urban areas but is apparently not the maximum possible rate). I wish everybody would spend their creative energies hypothesizing about what that something might be (changes in the success rate of detective work that can end cycles, changes in opportunities to start a new cycle or heat up a previously cooling one, etc) and not on all the other somethings out there that don't really have anything to do with the nature of urban homicide in the first place.
When I predicted this homicide spike in early June 2020 and provided the reasoning, I specifically mentioned that homicides would increase (not all crime) and that black people would be most affected. And now we still have criminologists try to come up with alternate explanations that don't even take these two things into account.
The question is how the Floyd incident triggered the spike in homicides, not if it triggered the spike (which Scott in my opinion convincingly enough substantiates). What are the causal paths between these two phenomena - the "intermediate variables"? Scott himself suggested several possibilities in his original post, and several additional hypotheses have surfaced during the debate. The causal path/s is important, because they are linked to different "what to do" suggestions.
Could it be as simple as: Americans just being awakened to their overall unpreparedness for emergencies like a pandemic (ie lines to get paper towels and toilet paper), decided that when they saw police stations and Target stores and CNN buildings getting set on fire....that they decided to be prepared for those activities and proactively bought a gun for protection? I have anecdotal evidence from friends that would have never owned a gun prior to May 2020, but have one in their home now. The increase in gun transfer is the effect, not the cause?
Totally hypothesising here and I don't have time atm to research this, but it seems weird to me that my first guess at the cause hasn't been mentioned anywhere I can see. Take a society with lots of guns and a strong suspicion of government and introduce a new scary phenomenon that cuases mass unemployment and house arrest. Take that increase in murderous intent and repress it with lockdowns and reduced mobility for the first few months (allowing for an increase in murderous intent but a reduced opportunity for murder) and then reduce movement restrictions and see what happens.
Hopefully will gget time to return to this and post some evidence.
If you're tired of us foreigners discussing a situation in US and want to engage in looking at problems in other countries for a change, this is for you:
In Germany, former UN expert Nils Melzer just accused the government of 'systemic failure' in the oversight of police. To my understanding the main accusation was, that goverment and judiciary system don't react adequately to cases of misuse of force by police. (https://www.zdf.de/nachrichten/politik/corona-demo-polizeigewalt-100.html).
In Germany, the rule of *Verhältnismäßigkeit* (proportionate reaction) is key. It rougly means, that police of course is allowed to use force, but only to the extend to which it is necessary in a given situation. Melzer found, that there were too many cases where force was used excessively (eg. taking sb. to the ground, even if the situation was already under control). The main problem here, according to him, is not only those cases, but that they are rarely prosecuted, and that the goverment doesn't really care much about them.
In case you care, the topic of excessive police force came up in the context of anti-Covid-measures - demonstrations. Not that I think this has any relevance.
Overall, I guess police is rather well trusted in Germany (except by the usual suspects like very leftists groups). Being part of the overall 'trust' vis-a-vis police, I think that most do a great job. A general tendency to protect police officers even if they overdid it, internally in the police, as well as in the judiciary system, + minimizing the problem by the goverment, is however plausible.
Here is a case that made it to the news: a girl that happened to be close to Anti-G20 protests wanted to go home, biked right towards a police barrier she thought was still open for passage, and ended up with a broken arm and severe PTSD. The article describes at lenght how the internal investigation was diverted and delayed for five years, until it was finally closed. A parallel lawsuit at an administrative court resulted in a statement that the force used by the police was illegitimate - this however has no concrete consequences. I don't think this one piece says all that much about the overall context; I would mostly trust that the main key points of the story itself have been well researched. (https://www.zeit.de/gesellschaft/2022-07/g20-gipfel-hamburg-polizeigewalt-gericht/seite-5)
Of course I wanted to know the number of killings by the police, if you're interested, here is some data: https://de.statista.com/statistik/daten/studie/706648/umfrage/durch-polizisten-getoetete-menschen-in-deutschland/ . Mostly between 5 and 15 persons killed / year.
Then of course the question, how many police officers were killed. I found the number to be > 400. Since 1945. (https://www.welt.de/politik/deutschland/article236666105/Mehr-als-400-Polizisten-im-Dienst-getoetet.html). If you scroll down, the article contains an overview of each single case in the last 20 years. No killings in 2018 and 2019, two police officers killed in 2020 and two killed in 2021. I remember the last case, it was in national news with lots of declarations of outrage and sadness.
The last two bits make me wonder, whether we live in a specifically peaceful country, or whether there is lots of crime and problems, but just below the level of killing each other (where police is involved). Probably a bit of both.
I want to make the argument that the media coverage of the murder itself is what led to the increased murder rates.
It is already well known that the wide-reporting of suicides leads to an increase in suicides (and, sadly, an increase in car and commercial aircraft crashes). It is similarly known that the reporting of murders leads to an increase in murders in the areas of reporting. This is most notably observed following, again sadly, school shootings, when numerous shootings occur in succession. So, I think the argument can be made that the massive media coverage of the murder of George Floyd led, in turn, to the murder of more black men.
This would track with the logic that in reporting suicides and murders you lead to copycat suicides and murders. Where ages are reported in articles discussing suicides, the increase in suicides occurs in those within that age group/range. It seems to follow, therefore, that massive reporting of the murder of a black man would lead to copycat events. This would seem to explain why deaths of other ethnicities do not see a staggering increase. Likewise, one imagines that coverage of the murder of George Floyd was more extreme in the United States hence why the increase is only observable there (albeit I would need to actually see if the coverage was greater in the US). I would add that this explanation would also cover why media outlets have discussed other reasons; to state this is the reason would make them, in some way, complicit.
Timings wise it would also make sense. The coverage and the protests would occur contemporaneously, each feeding off the other.
I appreciate that this thought is unsourced, and low-effort in that it may have been covered by other comments, or by ACT and I have just missed it. Regardless, any discussion based on this thought would be interesting.
Has there been any attempt to ask the murderers about this?
Would it be possible to get a hold of a sample of people who attempted or completed homicides in the summer of 2020, and ask them why they did it, what pushed them over the edge etc? Or perhaps just do a deep analysis of court proceedings from those cases.
I agree with most of Scott’s analysis but I think we could do a lot more to understand the causal pathways at play.
The "stock of guns" argument seems flaky to me. No data, but I expect that a "new" gun would be vastly more likely to be involved in an actual shooting. There's various reasons including: People buy/obtain guns for a reason, a lot of the stock is tied up in large collections which won't all be used, new gun owners may not have thought ownership through and modified their guiding narratives and reactions.
The CDC WONDER database this week now has pretty solid data thru December 2021 on homicide victimizations.
I come up with black deaths by homicide going up 37.8% in 2020 over 2019, Hispanics up 28.2%, non-Hispanic whites up 20.7%, American Indians up 20.4%, and Asians flat at 0.0%.
For 2021 vs. 2019, blacks are up 43.8% and Hispanics 42.6. Whites are up 19.2%, Asians up 7.2%, and Native Americans 5.5%.
These are not murders perpetrated but victims of homicide. The FBI collects data on deaths by murder, which are a smaller but closely related number. The FBI has demographics on perpetrators of cleared murders, but a large fraction aren't cleared. Also, the FBI numbers are all snarled up involving Hispanics and whites, whereas the CDC numbers follow the usual modern government method of giving priority to Hispanics or Non-Hispanics, then looking by race at Non-Hispanics, so the CDC numbers are pretty useful.
I have qualms about the “gun sales” argument. Unlike hot weather and a pandemic, increase in gun sales do not just happen for no reason. It can happen as a result of a change in legislation, it can happen because TV had a cool show about a gun nut, it can happen because people are preparing to overthrow the government. But it happens for a reason. It is not an engine of the change, only possibly a transmission belt. So you cannot explain something *by* an increase of gun sales, only *through* it to the root cause.
(I have the same qualms when Aella quotes fatherlessness as a cause for social disparity: fatherlessness is not an intrinsic trait of a certain population, it is a consequence of other past social factors.)
To me, the protests were a breaking point, perhaps a spark that set off a fire that was already burning: a resentment of the privileged, of the status quo, of rules and civil order and bourgeois niceties that we take for granted which has been growing more and more in the last years, but accelerated during the pandemic. I see it in the recklessness of drivers, in the dismissal of property crime as victimless and deserved, in the attitude that most jobs, and expectations within those jobs (dress codes, uniforms, punctuality, etc.) are de facto exploitation.... feelings that were already festering, but the protests freed it all up - hence more lawlessness, more traffic incidents, more violence, and more murders too.
Also, this is a tangential point but any treatment of the protests as spontaneous seems off-base to me. They appeared almost simultaneously in so many cities at once, large, medium, small ones, including communities where large protests and social unrest weren't ever a thing in my memory.... suggesting a well-organized operation that had been preparing or planning this sort of action. And around it of course lots of people came out of their own accord. I don't know how to test my hypothesis about this, or whether it changes the conclusion about WHY the murder spike.... But I think it is worth thinking about and taking into account.
HI FROM THE NEW GUY
I’m a newbie here, but I’ll get this right out: Ever since Ferguson in 2014 I’m not sure very many amateur writers have dived into this topic of crime, race, and policing in quite the same way I have. Mostly I’ve put out Medium pieces and social media posts, but below I’ll just rattle off a few resources that might be handy for folks.
I 100% endorse Scott Alexander’s conclusions. It’s obvious it's the anti-establishment, anti-policing protests that have caused the near 29% spike in crime in 2020, just as nationwide protests led to consecutive double-digit percentage increases in 2015 and 2016. I’m forever impressed how Scott addresses critics, and I’ll repeat his request: “Please find me any major country besides the US that had a homicide spike in 2020.” It galls me how many obfuscate and point to the COVID pandemic for explanations. Talk about motivated reasoning.
MY DATA
https://tinyurl.com/jails-police-save-lives
• This 50% increase over six years means that from 2015 to 2021 we’ve likely seen an excess of 26,500 murders if 2014 is the baseline (50% black victims). Compare this to the 40-50 unarmed killings a year (25-35% black). This 26,500 number I calculated very simply in this 2nd tab “Murder increase 2014-2021.”
• In 1st tab “The Real BLM Effect” I calculate and hypothesize that some significant part of 180,000 lives were saved by incarcerating and arresting violent offenders, as the USA dropped from an average of 24,000 annual murders (1990-1994) to a low of just above 14,000 murders in 2014. We’re now past 21,000 murders again in 2020 and 2021. I quote Stephen Leavitt of Freakanomics fame who claimed these four factors led to violent crime decreasing: "Increased incarceration, more police, the decline of crack, and legalized abortion." (Source: http://pricetheory.uchicago.edu/levitt/Papers/LevittUnderstandingWhyCrime2004.pdf)
• In the 4th tab I ran the numbers to show that a black person killed by police is 12 times more likely to get a news story than someone non-black. That’s what you get when blacks are 25% of the people killed by cops but 80% of the news coverage. Obviously, black lives DO matter. Yes, the media is brainwashing our collective brains on this topic, and it’s why so many people believe cops are routinely going around killing black people willy nilly. It’s all we see, from CNN to the New York Times to our local affiliates. (Sources: https://www.commentary.org/articles/wilfred-reilly/no-there-is-no-coming-race-war/ & https://www.skeptic.com/research-center/reports/Research-Report-CUPES-007.pdf)
THAT’S AMATEUR NUMBERS, GIVE ME ACADEMIC RESEARCH!
7 papers and analyses indicating a Ferguson Effect certainly occurred the past few years, defined as an uptick in violence and homicides as a result of increased police scrutiny and protests. This combination delegitimizes police in the eyes of citizens, causes police to pull back, and often leads to policies detrimental to public safety (like consent decrees).
1. https://scholarworks.uark.edu/edrepub/136/ (2022, Robert Maranto, University of Arkansas, Wilfred Reilly, Kentucky State University)
2. https://www.sciencedirect.com/science/article/abs/pii/S0047272721001936 (2022, Cheng Cheng, Department of Economics, The University of Mississippi, University and Wei Long, Department of Economics, Tulane University)
3. https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3715223 (2021, Deepak Premkumar, Public Policy Institute of California)
4. https://www.nber.org/papers/w27324.pdf (2020, Roland Fryer & Tanaya Devi, Harvard University)
5.
https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3145287 (2018, Paul G. Cassell, University of Utah College of Law)
6.
http://journals.sagepub.com/doi/full/10.1177/2378023117703122 (2017, Neil Gross, Colby College)
7th analysis on George Floyd Effect (2022, David Pyrooz, Justin Nix & Scott Wolfe)
https://jnix.netlify.app/files/pdfs/denpo_depolicing.pdf
AND THE PROTESTS WERE FOR NAUGHT
I’ve collected articles and research in Google Docs for the past few years, and some may find them useful.
• The system is not systematically racist: https://tinyurl.com/4-steps-crime-just-bias
• Also here: https://tinyurl.com/The-Real-Crime-Stat-Myths
• 13 studies indicating no proven racism in lethal use of force (started as nine studies, but keeps growing): https://tinyurl.com/9-studies-no-racist-police
• Dataset countering BLM and #SayHerName myths I put together on Fatal Force using 3 years of Washington Post data (and many other tabs with nerdy number crunching): https://tinyurl.com/http-blm-sayhername-myths
3 MEDIUM ARTICLES I’VE WRITTEN
I humbly serve these up for critique and/or for people to use as they see fit.
#1 Titled: “Kamala Harris: Biden’s VP Pick & Your Next Candidate to Libel Police & Criminal Justice”
https://agent-orange-chicago.medium.com/kamala-harris-bidens-vp-pick-your-next-candidate-to-libel-police-criminal-justice-1c797d705d05
EXCERPT:
The larger picture shows that African Americans are not being hunted down by police. But Americans are hunting down each other. In fact, from Chicago to New York City, black Americans are more than three times less likely to be killed by police than previous generations. Yet vast black-white disparities in murder haven’t changed at all.
For at least four decades African Americans have murdered others at a rate 8 times higher than their non-Hispanic white counterparts, and are killed at a rate 6 times greater. In 2016, CDC indicated there were more than 19,500 homicides and The Washington Post reported 234 black killings by police and 465 white killings by police. This means for every 40 blacks killed by fellow citizens, there’s only 1 by a cop. For whites, that ratio is 12:1. But even in August 2019 we get yet another questionable study omitting crime statistics, and incredibly arguing that police are a major threat to the lives of blacks, with this Harvard Kennedy School headline: “Black men 2.5 times more likely than white men to be killed by police, new research estimates.”
People simply need to do the basic math. Or does the math not add up to the social injustices activists, academics, and the media have been portraying?
[end excerpt]
-
#2 Titled: “MY OPEN LETTER TO CHICAGO: Stripping Context from Media & Government Reports on Police Abuse a Likely Cause of More American and Chicago Bloodshed (Subhead: Statistical-Based Evidence Undermines Consent Decree Logic that Throws Chicago Police Under the Bus)”
https://medium.com/@agent.orange.chicago/my-open-letter-to-chicago-stripping-context-from-reporting-on-police-abuse-a-likely-cause-of-more-225feacb9301
THE GIST: Chicago-centric hard analysis w/ stats and graphics I created to try to help prevent a consent decree from taking shape. Roland Fryer’s 2020 paper highlighted how these consent decrees have led to potentially thousands of lives being lost. Had more than 20,000 views back in 2018.
-
#3 - Written directly after the George Floyd protests as the writing was on the wall, but few knew: Only around 10 unarmed black people were killed by cops the year before. Skeptic Magazine’s polling shows how large numbers of liberals believe 1,000 or even 10,000 unarmed blacks are killed by police each year. Note: Those folks driving our “national conversation” are in media and academia, yet their beliefs undergirding their “reform” demands are wildly off.
https://agent-orange-chicago.medium.com/unarmed-killings-of-african-americans-numbered-under-10-last-year-a-400-reduction-since-2015-e54f3eeb67ae
EXCERPT:
This goes without saying: These unarmed deaths are real lives and real tragedies. Just as the 62 police killed by felony gunfire in 2021 are, a jump from 45 in 2020. 346 officers were shot in 2021.
These are the final “unarmed and killed by police gunfire” breakdowns in the context of “Black Lives Matter” protests for 2019, the year before George Floyd’s death:
• 26 white
• 12 black
• 11 Hispanic
• 5 in the category of “other”
But look at that Skeptic chart above again. It’s also crystal clear that more than 50% of “very liberal” Americans live in a fictional reality — a morose and racist universe where the number of unarmed Black Americans killed by police might be as high as “about 1,000,” “about 10,000,” or “more than 10,000.” And it’s those “very liberal” people making the biggest stink on Twitter, trust me.
[end excerpt]
Fin. And hope to have many fun conversations here.
David
"One reason might be if the people buying guns in 2020 were very different from the people buying guns in previous years. For example, if previous gun buyers were collectors who had 100 guns each, but 2020 gun owners were new buyers getting their first gun, then the share of people with at least one gun would go up by more than 2% over an average year."
Maybe the suicide rate didn't increase because the extra Americans who bought guns in 2020 were not people in suicide-prone demographics. Suicides are most common among older, white men living in rural areas, and I remember hearing about how many of the people buying guns in 2020 were women, black, or living in urban or suburban areas.
Late to the conversation, but this feels worth noting given Matty's point RE: the media's characterization of the homicide increase
Right on queue, on 7/8/2022, the Wapo published an article titled:
"The staggering scope of U.S. gun deaths goes far beyond mass shootings"
The print version, where this article was the header article on the front page, had the following subheader
"45,000 fatalaties in each of the past two years (line break) Increase conicides with record firearm purchases"
The article (and notably its actual interviewee quotes) goes on to hedge a bit but clearly wants to suggest that the gun purchase increase combined with covid stress is the likely culprit.
It cites lots of data on new gun purchases without ever comparing this to the existing gun stock or the existing gun ownership rate
It even correctly cites the race/skew of homicide and suicide victims (black male skew in homicide victims and white male skew in suicide victims) but it of course fails to cite the skew in homicide perpetrators or the race/sex characteristics of the perpetrator increase.
https://www.washingtonpost.com/nation/interactive/2022/gun-deaths-per-year-usa/
I think an under-ratedly weird aspect of the post-protest homicide spike is just how FAST it happened. Like, I wouldn't have been that skeptical of the protest/crime-spike link ex ante, but I would have guessed it would have taken, I dunno, months to set in gradually. When people try to deny the link, I think they're sometimes pattern-matching to other social phenomena in which there are meaningful lags between cause and effect, leaving all conclusions very fuzzy and uncertain. But here the whole thing played out in the blink of an eye. It seems almost too good to be true.
I have the same dizzy sensation regarding the CHAZ/CHOP debacle (https://en.wikipedia.org/wiki/Capitol_Hill_Occupied_Protest). I would have predicted ex ante that a miniature anarchist commune would run into trouble pretty quickly, but it was literally only a matter of days before these people reconstituted a police force that shot and killed an unarmed black teen! "Reductio ad absurdum" doesn't begin to cover it.
I guess the update is that...there ISN'T much ruin in a nation?
I'd like to signal boost the contributing factor of court closures to the perception of potential criminals that they would be unlikely to face (swift enough) consequences even if they did commit crimes.
Thread: https://astralcodexten.substack.com/p/what-caused-the-2020-homicide-spike/comment/7427848
Atlantic Article: https://www.theatlantic.com/ideas/archive/2022/07/covid-court-closings-violent-crime-wave/670559/
In my perception it seems like the argument about changed police behavior (in part due to BLM protests) is implicitly assuming a model where the police behavior causes perception-of-risk amongst criminals, which in turn is the actual proximate cause of (the increase in) crimes committed. I wanted to draw more explicit attention this part of the causal chain.
I'd be curious to know if there was a good way to estimate the effect size of changed court behavior (probably mostly caused by responses to the pandemic) versus changed policy behavior (probably mostly caused by responses to BLM protests). I do think that police, prosecutor, and court behavior is the main cause (feasibly manipulable contributing factor) of the magnitude of violent crime and are much more significant than small changes in gun prevalence or other pandemic-related factors, in partial agreement with Scott's thesis.
Regarding the Alexander/Yglesias disagreement about media coverage: I think both positions make a lot of sense if we model media coverage of a generic "what causes Y?" questions as a binary choice between the "A is THE ONLY cause of Y, if we control A we will completely control Y (we literally cannot conceive the concept of multiple contributing factors or causes)" template and the "wow Y is complicated, look we can sit here and make up contributing factors all day, it sure seems impossible to compare factors and draw conclusion" template. So from Yglesias's perspective the media chose the correct template for the situation, and from Alexander's perspective the templates suck.
I think it is better for our blood pressure to recognize that this is a structural problem with the way news media is written that affects every topic they cover, and that the media is not specifically trying to hide the truth on this issue.
(Note: the thesis here is reliant on new media using a small fixed number of framings / templates for everything where the choice of template is the entire analysis of the topic, not on there being exactly 2 templates.)