Eighteen years ago today, Dylan Klebold and Eric Harris walked into Columbine High School and murdered 12 students and a teacher, as well as injuring dozens more people. The mayhem ended when the two killers took their own lives as police closed in.

The massacre, perpetrated with guns and rudimentary explosives, created a political firestorm. Music, video games, and especially guns became lightning rods for outrage and demands for new legislation. The controversy re-energized gun control advocates and spawned Michael Moore’s award-winning anti-gun film Bowling for Columbine. Hundreds of new gun control bills were introduced, although few became law.

Subsequent school shootings, such as Virginia Tech in 2007 and Newtown, Connecticut in 2012, have generated similar cycles of gun control fervor followed by demands for new laws, but the fundamental debate remains the same: what can we do to effectively mitigate the risk of such tragedies?

In 2015 David Kopel attempted to answer this question by analyzing the efficacy of the types of gun control proposals that are so common after school shootings, including magazine bans, universal background checks, and assault weapons bans. He found little evidence that gun control legislation has been or could be effective at preventing spree shootings.

From the summary:

Although universal background checks may sound appealing, the private sale of guns between strangers is a small percentage of overall gun sales. Worse, the background check bills are written so broadly that they would turn most gun owners into criminals for innocent acts — such as letting one’s sister borrow a gun for an afternoon of target shooting.

Magazine bans are acts of futility because the extant supply is enormous. Today, magazines of up to 20 rounds for handguns, and 30 rounds for rifles, are factory standard, not high-capacity, for many of the most commonly owned firearms. These magazines are popular with law-abiding Americans for the same reason they are so popular with law enforcement: because they are often the best choice for lawful defense of one’s self and others.

Gun-control advocates have been pushing for a ban on assault weapons for more than 25 years. This proposal is essentially a political gimmick that confuses people. That is because the term is an arbitrarily defined epithet. A federal ban was in place between 1994 and 2004, but Congress declined to renew it after studies showed it had no crime-reducing impact.

What has occasionally proven effective at stopping spree shootings is the armed self-defense of would-be victims or bystanders. Kopel notes, for instance, that shootings at Pearl High School in Mississippi and at Appalachian School of Law in Virginia were halted by armed bystanders.

Highly motivated killers who plan their attacks weeks or months in advance (as Klebold and Harris did) have an inherent advantage over their unarmed victims, and are unlikely to ever be deterred by criminal penalties. The best examples of these crimes being stopped in their tracks are examples of armed defense, not legislative preemption.

Notably, the Columbine tragedy did produce one effective policy change, but it wasn’t about guns.

At the time of the shooting, standard police procedure for an active shooting situation was for the officer on the scene to cordon off the area and await the arrival of a SWAT team or other specialized unit to handle the crisis.

At Columbine, the armed sheriff’s deputy who served as a school resource officer followed protocol. He exchanged fire with the gunmen when they left the building, allowing several victims to escape. But the deputy did not enter the building himself, instead waiting for a better-equipped SWAT team to arrive. Klebold and Harris continued their rampage following the initial gunfire exchange, eventually killing themselves as SWAT officers closed in.

Following claims that the delay in police response allowed Harris and Klebold to kill more people, police departments around the country began to implement what would become known as Immediate Action Rapid Deployment (IARD), in which the first officers on the scene of an active shooting tries to confront and neutralize the threat even before the SWAT team arrives.

Gun crime remains a serious problem in America, not just spree shootings (which are exceedingly rare occurrences), but in general. Nearly two decades later, the lesson of Columbine is that an intense public desire to “do something” after a spree shooting is not enough to generate effective policy solutions. Our response to mass violence must be based on logic, available evidence, and compliance with the mandates of the constitution.