The Wire is often cited as the greatest TV show of all time, and for good reason. We’re drawn in by the charisma of characters such as Omar, Avon, Stringer, and McNulty, but its ultimate greatness lies in its realistic portrayal of the various forces surrounding Baltimore’s drug scene. By the end of the series, we have a pretty firm grasp on the entire drug ecosystem of Baltimore and how the police, city hall, the drug dealers and users, the media, and the education system all play their part in fueling that ecosystem, making it incredibly hard to solve. In this essay, though, we’re going to focus on how Goodhart’s law explains some of the specific perverse incentives in just one of those pieces, the Baltimore Police Department, by looking at the homicide unit in particular. Marilyn Strathern’s quote does a good job summing up Goodhart’s law and is as follows: “When a measure becomes a target, it ceases to be a good measure.” So, let’s go into how that ties to the police department.
Baltimore, like all other cities, needs to measure how well their police department is doing and does so by looking at crime. Yet “crime” as a concept isn’t quantified so in order to measure “crime” we quantify it by counting the number of murders, burglaries and other incidents that we deem bad. And so, we measure police departments by their impact on those numbers; we look at their murder clearance rate and the raw number of murders and other crimes on their watch.
Enter Jay Landsman, the sergeant of the homicide unit, and the quintessential company man. He understands what his unit and his bosses get measured on and does everything in his power to cover his ass and make them look great. How do we measure homicide? Easy, by counting the number of murders. So how do we measure the homicide unit? Also easy, by counting the number of murders they solve, a.k.a. their clearance rate. Though Jay is lazy, a total goof, and certainly comedic relief, he clearly understands that his unit is solely measured on its clearance rate, so he does everything in his power to ensure that the clearance rate looks as good as possible. This is why he’s the quintessential company man. Other characters, like Jimmy McNulty, also understand that the homicide unit is measured by its clearance rate, but he’s more concerned about solving actual crime than departmental optics, which is why he butts heads with so many bosses. So, you might ask, what’s so bad about focusing on the clearance rate? It seems like a pretty good measure of how well the homicide unit is doing to me. And certainly at first glance, you’d be right. After all, it makes intuitive sense to measure the homicide unit by how many homicides it solves, but as The Wire shows, the world is much more complicated and focusing solely on the clearance rate has other disastrous effects.
When McNulty and Lester Freamon finally discover that Marlo Stanfield’s gang has been hiding their murders in vacant houses and want to use departmental manpower to search other vacant houses for bodies, Jay immediately uses his authority to overrule them. The murders are in those vacants regardless of whether the police uncover them, but Jay understands that they won’t count as “murders” unless they’re uncovered by the police; if they just pretend that the murders aren’t there and never uncover them, then as far as the police department is concerned, the murders really aren’t there and the clearance rate remains unaffected. Though 22 murders are eventually uncovered when Freamon finally goes over Landsman’s head and straight to Daniels, we can imagine scenarios in which these murders never get uncovered. Freamon benefited from having someone like Daniels in charge, having Carcetti in the mayor’s office, and uncovering the murders when the statistics could still be attributed to the Royce administration. Without the confluence of all these rare events, the murders probably get buried and the investigation on Marlo never gets opened. Instead of uncovering and trying to solve the Marlo murders, the Baltimore Police Department was incentivized to try and bury the murders to juice their stats, an action that directly contradicts their ostensible raison d’être.
Another example of clearance rate maneuvering is shown in the beginning of season 2 when the thirteen girls show up dead in the harbor. First, the homicide unit tries to get the deaths written off as non-homicide related deaths so that they wouldn’t have to solve the case. When that fails, Rawls pawns the case off to Baltimore County by citing where the bodies were found and when McNulty foils him there, Rawls tries to meet with other jurisdiction heads to get the murders out of his jurisdiction. In an ideal world, unit leaders would devote all their time and resources to the cases they’re given, but because we have to measure their units, what ends up happening is that backroom politics becomes a large part of their job. After all, why spend all that effort trying to solve a futile case that if unsolved will affect your performance negatively when you can just hand those murders off to someone else?
I don’t mean to suggest that police are lazy, bad people who spend all their time on office politics and covering up crime so that they can go back to eating donuts. It’s just the unintended consequences that gets summed up by Goodhart’s law. When the clearance rate becomes the target instead of a measure, it’s no longer a good measure because the homicide unit starts to alter the clearance rate through methods other than solving murders. And that’s the hardest part of solving the issue. We have to measure the performance of police departments in some way so that we can identify whether it’s doing a good job. If we didn’t measure them in any way, we’d live in an even worse world. We can easily see that McNulty is one of the best detectives even if that’s not reflected in the clearance rate just like we can see that the system is quite messed up and that the incentives are often perverse. Yet, can we imagine a better system? The Wire certainly doesn’t offer any prescriptions. Unfortunately, there’s just no better way to measure the homicide unit than by looking at clearance rates. In an ideal world, we would measure the difficulty of cases, and then assign a difficulty-weighted clearance rate, but we don’t live in an ideal world. Assigning a rating for the difficulty of cases would just lead to an even worse system in which police game the weighting system. Ultimately, clearance rates are the worst measure of homicide units except all other measures that we have, and so we’re stuck with it for better or worse.