There’s a bit of a crisis in American murder policing in that no one can actually do it. Homicide clearance rates, or the rate at which murders result in an arrest or other solution (such as a suspect being found dead), have fallen from around 90 percent in the mid-sixties to approximately 65 percent today.

This is for a mix of reasons. One is that standards for arrests have risen, although usage of DNA-based evidence techniques, which make investigations easier to complete, have also been simultaneously improving. Additionally, murder rates have fallen dramatically overall, decreasing to half of the level of their peak during the ’80s. Theoretically, this should have allowed police to investigate each murder more thoroughly; however, this has not been the case.

Of course, one major factor has been a shift in police work from emphasizing the investigation of crime to emphasizing prevention, but this is not the full story. Police complain of a breakdown in the relationship between themselves and their communities, and the data can support that.

Between 2011 and 2013, clearance rates averaged 25 percent in Salt Lake City, Utah; 86 percent in New Orleans, La.; 25 percent in Austin, Texas; 93 percent in Detroit, Mich.; 35 percent in Oakland, Calif.; and 88 percent in El Paso, Texas.

The first and most startling pattern is the staggering amount of variation by city. Some maintain the near-100 percent rates of the 1960s, while in some cities, practically no murders are ever solved. One might notice that cities with large black populations have much lower clearance rates. (The Brennan Center’s recent and more sophisticated analysis “What Clearance Rates Say About Disparities in Crime and Prosecution” bears this out.) This fits perfectly with the narrative that the relationship between the police and the communities they operate in has broken down.

Let’s be clear: To have such a low clearance rate is not only a problem because a higher chance of being caught is one of the best deterrents for murder. This undermines law in its original purpose. Law is, at its base, not a system of communal protection. It is a commonly agreed-upon method for the redressal of grievance.

In the common-law tradition (the basis of legal systems in Anglophone countries), one of the few laws that predate writing was weregild, or “man-price.” This was the price that had to be paid by the family of a murderer to the family of a victim. This was developed by the community as an alternative to a “blood feud.” Blood feuds, in the case of “Romeo and Juliet,” appear in response to a murdered Montague. The Montagues could murder the Capulet they think is responsible. The trouble is, often, the Montagues are wrong, or the Capulets think they are. Then, the Capulets could murder the Montague they thought was culpable. In the end, Romeo and Juliet, as well as everyone else, are dead.

This is key to understanding the importance of low clearance rates: They don’t mean that murders go unsolved in the sense that no one is punished for murders. They mean murders go unsolved in the sense that blood feuds will continue forever. I believe that, ultimately, this means that the consequences for murders appear fewer and less severe.

The result is that in cities like New Orleans, Chicago, Detroit, or St. Louis, there are much higher murder rates. I hypothesize that this is not because these cities have more murder-prone people, but—due to underlying inequalities and disparities—there may be few other options for people to address long-standing grievances. In my opinion, this is what you might call a disaster.

Certain cities have bucked this trend. Washington, D.C averaged a clearance rate of 86 percent in 2011-2013, a startling rise since a low of 25 percent in 1996. At the same time, its murder rate has fallen by around two thirds. It is essential to ask how they did it.

Some policies have been simple. Reorganize the filing system so more eyes see each case. Devote more money to murder investigations. Still, the cleverest ones have been to encourage police to go to committee meetings in town and to participate in the communities in various ways. What this shows is that rebuilding relations between police and the community is key to both to solving murders and preventing them in the first place.

Here’s what I propose: In cities where there is an irreparable break between police departments and the community, create a separate department that only handles murder and other major investigations. These departments should not share information with the police.

Police are part of an era focused on crime prevention. Some argue that these policies have likely contributed to the dramatic decrease in crime since the ’80s. Hyper-targeted, “hot spot” policing means, almost by definition, that police go out of their way to harass innocent people. This will inevitably make police worse at investigations. Citizens will not act as witnesses for police who harass them. Similarly, acting as a witness for major crimes will often incriminate others for minor crimes, and many witnesses will not want to incriminate their friends and families for minor, non-violent crimes. With the current system, investigations are doomed. More murders continue to occur, and police, blamed for their failure to investigate, will continue to have even more splintered relations with their municipalities, and even more investigations will fail.

So why has the investigatory department been the same as the harassing department? These are different jobs, in the way that educators and plumbers are different jobs. Like educators and plumbers, they can be independent municipal departments. Granted, information sharing between these two spaces can be useful in many ways. They are closely related, so information gathered by one is often useful to the other. However, the links between these two entities can actually end up limiting their overall effectiveness. Since citizens aren’t going to trust the harassers, they’re not going to go to the investigators. As a result, in certain cities, citizens need to be absolutely sure that helping an investigation will not help the harassment department. This proposed separation would obviously be costly, in terms of transition as well as lost information to police, and should only be considered in certain cities. For cities looking for a change in the success of police investigations into murders, I urge them to start with a departmental separation.

Thomas Hanes is a member of the class of 2020 and can be reached at thanes@wesleyan.edu.