On April 3, 1987, William “Willie” Horton twice raped a woman after pistol-whipping and tying up her fiancé. He then stole the fiancé’s car and fled, before finally being apprehended by Prince George’s County, Maryland, police. Horton’s crime was particularly disturbing because, at the time of his offense, he was supposed to be 400 miles north, serving a life sentence for homicide in Massachusetts. He had been out on weekend furlough when, nine months earlier, he fled south, eventually resurfacing to commit new crimes.
Willie Horton’s story would have been forgotten, except that Massachusetts’s then-governor Michael Dukakis had defended the furlough program, including blocking legislative efforts to exclude murderers like Horton from participating. When Democrats nominated Dukakis for president in 1988, a political action committee plastered the airwaves with Horton’s face in his famous “weekend passes” ad, pinning the blame for Horton’s crimes on Dukakis. Dukakis went down in flames, losing to George H. W. Bush in November.
This story is the origin of what law professor and decarceration advocate John Pfaff, echoing other advocates of “criminal-justice reform,” has called the “Willie Horton effect.” The idea, per Pfaff, is that even if a reform is, on average, successful, one high-salience failure can still sink it. Horton was one such case, damning, in this view, both the furlough program and Dukakis’s potentially reformist presidency. Another was Darrell Dennis, paroled under an Arkansas criminal-justice reform bill that slashed the prison population by 10 percent. After Dennis committed a murder while on parole, Pfaff writes, “the parole board cut back releases so aggressively that by the end of 2013, Arkansas’s prison population had risen 17%, to a new high.”
Willie Horton effects show up in other areas of policymaking. George Mason University’s Alex Tabarrok, for example, has long lamented the Food and Drug Administration’s overcautiousness: a disastrously wrong approval will be caught, but deaths caused by delay or rejection of a drug that works won’t be. And it’s always easier to identify the homicide that a criminal-justice reform caused than the homicides that didn’t happen because of its diffuse benefits. We know, for example, that, on average, jail increases the marginal offender’s risk of reoffending. But we don’t directly observe the crimes people don’t commit because they aren’t jailed, while we do see the ones that people released—like Willie Horton—do commit.
We can’t, reformists insist, let a scary headline dictate the course of policy. They are correct. Well-designed policy evaluation tries to measure the things that do and don’t happen. While policy regimes can often be improved to minimize both kinds of events—as, for example, in the case of New York’s bail policies—advocates are right that the public often falls for Willie Horton–style traps. Serious analysts should not use rare, high-salience bad events that come out of a system to discount the potential bad events that that system helped us avoid.
Which brings us to the killing of Tyre Nichols.
Nichols died last month after being viciously beaten by five members of the Memphis police department. The officers have since been fired and arrested for second-degree murder. Almost everyone agrees that the assault, captured on body-worn cameras, was a heinous abuse of authority, for which criminal charges are obviously proportional.
The question is the inference that one can draw from this. One could observe—as my colleague Rafael Mangual did in City Journal—that police respond to millions of calls for service, and make hundreds of thousands of arrests, every year, with just a fraction resulting in even justified use of force. One could further note—as my colleague Heather Mac Donald did—that of the roughly 1,000 police killings every year, the majority involve an armed offender, from whom police were protecting themselves or others, and even in many of the unarmed cases, officers plausibly feared for safety or the safety of those around them.
Should we draw conclusions about the character and efficacy of a system from rare, salient, bad events? No, insist reformists in most contexts—unless doing so advances the goals of criminal-justice reform.
Thus did Nichols’s death become further proof that, as the New York Times’s Jamelle Bouie put it, the “institution of American policing lies outside any meaningful democratic control.” We know this is true, the argument goes, not only because of horrifying videos of police shootings but also because sometimes police unions fight accountability (though oddly, Bouie cites instances where they lose—never mind the evidence that unions have little impact on misconduct). Police also, Bouie notes, can stop and frisk on “reasonable suspicion,” and stop and search a vehicle on “probable cause” standards. Leave aside why, exactly, these practices are bad—the argument appears to be that Nichols was killed during a stop, and therefore stops are bad. The thousands of guns retrieved and warrants cleared in stops annually go unconsidered; only the horror story counts.
For another example, take the response of prominent police abolitionist Alex Vitale, who argued on MSNBC that Nichols’s death is proof that body-worn camera technology does not “deliver the goods,” i.e., stop misconduct. Vitale’s claim is wrong: the balance of the evidence supports the conclusion that cameras reduce complaints against police at the margins. But all the bad behavior they prevent matters little against a single, brutal video.
What unites these arguments, and many others made in the weeks since the Nichols incident, is the use of a horrific but rare consequence of policing to indict the institution as a whole. What goes unmentioned is all the horrors that policing stops: the murders, rapes, and other vicious crimes that police provably prevent before they happen. Much as when Willie Horton’s peers didn’t reoffend, these acts don’t go on the reformers’ ledger—only the horrible exception does.
In point of fact, reformists make Willie Horton arguments all the time. Following Nichols’s death, advocates have renewed their alarmingly successful calls to limit police traffic stops because, as one Bloomberg article put it, they are “a common catalyst for police violence.” This claim is not just misleading—as Temple criminologist Jerry Ratcliffe noted, 86 police stop deaths a year looks a lot smaller in the context of a 20-million-stop denominator—but also a canonical Willie Horton argument. We see the people killed in traffic stops, particularly the least justifiable ones. But over 40,000 people die every year in auto accidents, rates that have risen in the past three years; black drivers count disproportionately among those fatalities. High-quality evidence finds that traffic enforcement saves lives. But we don’t see the deaths that don’t happen.
A further example comes from another common target to which some returned after Nichols’s death: civil asset forfeiture. This involves the taking of property alleged to be involved in a criminal act through civil, rather than criminal, proceedings. Criminal-justice reform advocates call it “legalized theft,” pointing to high-profile cases of abuse to advocate dramatically curtailing or ending the practice. What goes unmeasured is the lives saved by the guns—the primary target of federal civil asset forfeiture—taken off the street through the practice.
Willie Horton arguments are persuasive. It’s hard to hear the story of Horton’s brutal violence and not question the wisdom of a furlough policy. And it is hard to watch the video of Nichols’s beating and not think something is broken in American policing. But critics of “tough on crime” policies ostensibly spent decades railing against Willie Horton arguments because they understood that such visceral examples could be used to enact systematic policy that they opposed.
If that logic was true then, it’s true now. When reformers use their own Willie Horton arguments, they deserve to meet the same skepticism they heaped on their opponents for doing the same.
Photo by John Lamparski/Getty Images