Is a Culture of ‘Lawful but Awful’ Police Shootings the Best We Can Do?
Nine changes to police policy and practice to reduce the number of police shootings, don’t trust straw men holding red herrings, and yes, police are part of the criminal justice system
Photo credit: Julia Rendleman/Reuters (more of her work here). The photo of ballerinas Kennedy George and Ava Holloway was taken on June 5, 2020.
This was a brutal week and I set aside what I intended to write. The police shooting of Jacob Blake left me pretty despondent. Another young Black man, Jacob Blake, was killed by a white police officer. At some level, that is the only fact about this case that matters. That is the problem we have to solve.
But it is also true that judgment in this case is far more complex than it was in Breonna Taylor’s murder, or George Floyd’s murder or Ahmaud Arbrey’s murder. Jacob Blake’s death was a homicide. But was it murder? I know little that is not in the public domain. I don’t know what made officers initially draw their weapons—which is really the point of no return. I don’t know what is said, what Blake observed, and I don’t know what the police observed that we cannot. I just don’t have enough information to judge the totality of the case.
I am told that the place shooting of Jacob Blake was ‘Lawful, but Awful’. I am told this will happen many more times and we should get accept it as inevitable. Or something. We should not do that. Too many Black people are shot by police. We cannot simply throw up our hands, tell people to be more compliant with the police and hope shootings will magically decline. They will not. They will only stop if they are made to stop. Policy windows are short and policy reforms require substantial force. There is enormous pressure being applied, but more is needed.
There are a lot of proposals floating around to reform the police and reduce the number of officer involved shootings. I think many of these ideas are excellent. There should be independent investigative units to look into every police shooting. Qualified immunity should be outlawed. Dangerous restraints, including choke holds, should be prohibited (though see my note later on). Warrior training (and gear!) should be forbidden. No-knock warrants should be outlawed because of the chance for false positives with disastrous consequences. All of these are excellent ideas. Here is the text of the Minnesota police reform bill which has these and many other excellent reforms.
To these reforms, I would like to add nine other ideas that have received far less attention.
Culture Eats Strategy
Business in the 21st century is all about disruption. Facebook, Amazon, Netflix and Google (FANG) did not exist when I graduated from college but in three decades FANG has surely disrupted everything from newspapers to democracy. Although we are all exhausted by the cloying, teeth-grinding consultant speak that amplifies, leverages and leans into disruption, there is a real appeal in adapting ideas that have shown success. And one of the successful consulting maxims that accompanies disruption is that culture eats strategy or that culture eats policy. The idea is that if you change the nature of a place, you change the direction of the entire organization with far greater certainty than if you change a particular policy or set of policies. The culture of policing needs disrupting.
An overarching goal of police reform should be to reclaim civilian ownership over policing. Not just oversight, but ownership. Communities big and small have forgotten that the community determines the role of the police, not the other way around. Police are the personification of our shared values and they are not merely the shield and sword, but the measure of our benevolence. I don’t think most mayors are beholden to their local police chief in a Roadhouse kind of way, but there is a lot of genuflecting in most places. The chief of police should be just another leader within the executive branch not a separate branch of government. Minneapolis voted in June to move the job of Chief of Police down the administrative hierarchy. It is a critical step.
Along the same lines, one clever way to shift policing power to more democratic institutions is simply to make policing more democratic. One interesting proposal shifts the power within law enforcement agencies from a purely top-down hierarchy to one where the rank-and-file have more say in policy and more responsibility in implementing policy. The lack of rank-and-file input into police agency decision-making is well documented in the Department of Justice’s finding of a “pattern and practice of conduct that violates the Constitution and Federal Law.”
A more direct approach is to limit the power of police unions which have evolved to act against the public interest by not just by excusing the most egregious police behavior but by making limits on policing discipline their priority. Police unions are successful because they prey on the worst fears of every police officer—if you make a horrible mistake, without us, there is no one to protect you. This is obviously a little melodramatic given the myriad other protections sworn officers receive, but it nonetheless is well received. It is so convincing that it drowns out a much more compelling argument that this union behavior actually diminishes officer safety because it erodes the public’s trust which decreases the likelihood of compliance and increases the chance of a dangerous confrontation. Unions are politically powerful, but it is also a politically powerful moment.
One way to do this is to decouple police sanctions and police compensation in collective bargaining agreements. The Boston Globe editorial board lays out the logic of this approach clearly. Collective bargaining agreements give unions sway over discipline in all sorts of ways. According to a Reuters investigation, more than half of union contracts require departments to erase disciplinary records. They grant access to internal investigations before interviews, suppress public release of discipline decisions, forbid anonymous complaints, and restrict the window for complaints to be lodged to a few weeks and more. You can see the contracts here. These incentives to violate the public trust do not need to be included in collective bargaining and should not be. I think a more democratic police force will embrace this idea, because cities have long sought to trade less police accountability for lower pay.
When you get on a plane, you can be almost certain that the captain will make no mistakes. Your life is in her hands and she knows it. There are no ‘Lawful, but Awful’ plane crashes. Police have that same responsibility. Cops don’t get to make mistakes. The way to create a full proof policing system is to automate policing where practicable (such as traffic enforcement) and radically increase training. Through collective bargaining, most police agencies make continuing education voluntary or a small part of the job, often limited to facility with a gun. More training and constant training is the key to getting officers out of the tunnel vision that stressful encounters create. As Temple professor Jerry Ratcliffe has shown, police spend little of their time answers calls for service for violence-perhaps 4% of all calls. More and better training and more often is critical. That training should explicitly incorporate race and bias, and be ongoing all the time to create a culture of continuous improvement Training must also be tied to raises and promotions. Again, the culture must be that cops don’t make mistakes.
The idea of a mandatory review for all police use of force incidents has been around a long time but has had little traction. Called ‘Sentinel Event Review’ in public health, these are structured after action reviews of every use of force incident. The goal is two-fold: what went right and wrong in this encounter, and, what were the root causes for failure? These are mandatory for hospitals to maintain their accreditation and have led to very successful changes in practice, including the mundane but crucial pre-surgery checklist that keeps the wrong leg from being amputated. Participation in these civil proceedings must be mandatory for police officers and part of the collective bargaining agreement.
Mark Kleiman once said to me that the way to reduce shootings in general was to think much harder about the pre-cursor to every shooting—brandishing a gun. Whether a police shooting is justified or not, whether it is righteous or Lawful but Awful, it begins with a choice by a police officer to draw their weapon. We know very, very little about that decision. Frank Zimring’s foundational book When Police Kill only mentions brandishing three times and only in the context of the assailant’s behavior. The only time brandishing is seriously considered is in the murky world of defensive gun uses. We need to measure police brandishing and every time a weapon is brandished it should be subject to a Sentinel Event Review.
That leads to the obvious conclusion, again, that we need far more data about police behaviors. Shootings are the result of all sorts of precursor behaviors beyond brandishing a weapon, particularly the choice about who to arrest. Here are three ideas.
Police officers are rarely formally notified about the disposition of an arrest. They know little about how their arrest affected whether and how successfully a case was prosecuted. There are some interesting experiments in behavioral economics around judges and the effect on their decision-making of simply telling them what happened after they did or did not set bail (people released on bail generally return for court dates and simply knowing that people a judge has released tend to return increases the likelihood a judge sets bail in the future). In short, data can substitute knowledge for bias. Police officers should routinely be informed about the outcomes of their prior cases. They should certainly be informed what happened to people they shot, with regular updates. The idea is simple: if you know the ramifications of your actions, you can better evaluate future choices.
This leads to the complementary idea that police officers should be evaluated using different performance metrics. As James Q. Wilson points out in Bureaucracy, a police officer’s behavior as a largely autonomous agent can largely be predicted by their incentives. If rewards follow large numbers of arrests, then officers will seek to make a lot of arrests. Different performance metrics can change that behavior. Is the officer likely to arrest people who are ultimately convicted? Do they receive many complaints? Do they arrest chronic, serious offenders for serious crimes or just the poor and unlucky for disorder? Each arrest carries with it some chance of escalating, so being clear that every arrest was worthwhile is critical to reducing the number of shootings.
Paperwork is part of a police officer’s job. It is not process as punishment. It is not a waste of their time. A 21st century police culture would have data front and center. Policing is a customer-facing, client-centered job. Measuring the nature and characteristics of police-citizen contact is critical to measuring police effectiveness. Measuring police effectiveness is critical to avoiding mistakes. It’s worth repeating. Paperwork is part of a police officer’s job.
All of these ideas share the goal of changing policing culture and norms. A data-driven, client-centric, culture of continuous training and improvement is a critical foundation for change.
Watching for Unintended Consequences: Why banning choke holds might lead to more shootings
One danger of instituting many reforms within a narrow policy window is that those reforms are not sufficiently thought through and have unintended consequences. America’s leading modern metaphysician Donald Rumsfield called these the unknown unknowns, or things you don’t even realize you don’t know. Most of these problems can be avoided with some critical, objective thought. Among the commonly proposed reforms to reduce the number of officer-involved shootings is the outlawing of chokeholds and other dangerous restraints. The problem here is that by limiting the non-lethal options available to an arresting officer, they are more likely to choose a potentially lethal one. Police officers correctly note that the public wants them to ‘put their hands’ on people instead of shooting them, but rules like this create new reasons to reach for a weapon. Better training is a critical part of the solution. But that training has to reflect the realities of policing.
I did not include here the biggest reform I would propose, which is simply reducing the number of sworn law enforcement while dramatically increasing the number of highly trained emergency first responders with expertise in something other than order maintenance. You can read that here.
Reads
Taking on both police unions and Andrew Sullivan in one narrative is probably a poor idea. But I was struck by this passage from yesterday’s The Weekly Dish which seems to reflect the thinking of a lot of folks.
What I was struck by here is the casual conflation of disorder and violence. I was also struck by what appears to be an implicit assumption about crime and violence following a normal distribution. Bear with me a little here.
In criminology, disorder refers to either physical disorder or social disorder. Physical disorder is the presence of graffiti, a vandalized car or building, needles on the street. Social disorder refers to noncompliance with social norms such as a group of teenagers listen to loud music and yelling and maybe cursing. There is a long and frankly very tired debate about whether physical and social disorder leads to anything that is about as uninteresting as the debate over marijuana as a gateway drug. There is a more interesting inquiry into the role generally of incivilities as disruptors of public spaces.
But this conflation of violence and disorder is all a big digression, a strawman holding a red herring. There is a continuum of penalties in every criminal justice system in the world where the penalties for incivilities and social disorder are pretty trivial, often just a ticket if anything, and graduating in severity to low-level misdemeanors for small property losses and minor physical confrontations, to high misdemeanors for more serious property loss and physical attacks, to felonies for the most serious events.
It is also true that that distribution follows a Power Law rather than a normal distribution.
Both of the figures (from the amazing IDRE website at UCLA) above have a mean of 2 and a variance of 2, but obviously tell very different stories about the underlying phenomena. The incidence of disorder (and violence) in any sample almost certainly follows the Poisson distribution on the right rather than the normal distribution on the left. That means that if the average number of disorderly incidents by one person within a larger group is 2, it is almost certainly true that most people in the group will have less than one incident and a very few people will have many incidents.That tells a much different story about the behavior of the group as a whole than would be the case if the mean number of incidents followed a normal distribution and thus you were as likely to encounter a person with 3 incidents as a person with only 1 incident. And finally, if you are still with me, the Poisson distribution allows for really long tails, where one person could have dozens on incidents while most people have none. The normal distribution, in the special case of count data which must be positive, does not allow for any tail at all, and thus with a mean of two, and variance of two, the maximum number of incidents per person is 4.
Ok. So what? What does this have to do with violence and disorder? It turns out that crime data almost always follows a Poisson distribution rather than a normal distriubtion. If you measured the number (density) of criminal violations in a sample it would look like the data below, where more than half the sample has zero incidents and a few have a lot.But that’s just the tip of the iceberg. If you measured the number of violent icnidents you would likely see an even more exagerrated skew with even more zeros and even fewer non zeroes and rarer outliers. And finally, if instead of count on the x-axis you had seriousness of criminal incidents, with minor disorder on the left, misdemeanors in the middle and violent felonies on the far right, you would see the same Poisson distribution!
Obviously then, a big red herring in conflating disorder and violence is that they are implied to occur in equal frequency. That is nonsense.
One very final thought. There are a bunch of really interesting questions to be asked about Poisson distributions. In particular, it is worth asking how the zeroes come to be zeros. Is there something going on within the group of people who commit no disorder at all? Suppose we know how many people are at a protest from a drone picture. Some of the people in the image are part of the rally, but some are just bystanders. If we assume that bystanders never commit disorder, then the data are zero-inflated—the data generating process for some the zeros is just fundamentally different than it is for others.
More interestingly, it could instead be describing a ‘hurdle’ data generating process. That is, there is one process for converting a 0 to a 1 (from no disorder to some disorder), and a different process for converting 1 to a larger number. What happens that causes the hurdle is thus a very important bit of information if you are an organizer trying to keep your rally peaceful or the police trying to keep it peaceful. It is also important if you are a columnist writing about disorder and violence.
…
In looking around for more accessible descriptions of the Power Law ideas, I came across this amazing article describing the origins of Poisson Distributions in measuring and predicting war and peace. In a wild plot twist, the work by Lewis Fry Richardson, “became one of the ideas that inspired Mandelbrot’s theory of fractals.”
https://pdfs.semanticscholar.org/6e9c/de40cb861ac28c735748837650b9a40425d9.pdf
Graphic
I came across this poll result and it left me utterly flummoxed. These data are from after the Ferguson protests but before the George Floyd/BLM protests. According to Gallup, 53% of Americans have a great deal of confidence in the police but only 24% have a great deal of confidence in the criminal justice system. My question:
Why do people think the police are not part of the criminal justice system?
Musical Interlude
For the folks in Louisiana, hope you are well. Taj Mahal: