Small Arms Races

On November 19, 2021, Kyle Rittenhouse was acquitted of homicide charges stemming from his killing of two people—Anthony Huber and Joseph Rosenbaum—at a protest of police violence in Kenosha, Wisconsin. Rittenhouse had armed himself and traveled to the protest, purportedly to defend Kenoshans’ property against looting. The acquittal sparked substantial public outrage about the state of gun laws and about the legitimacy of the criminal justice system more generally. In a similar case, Travis McMichael, Gregory McMichael, and William Bryan were charged with murdering Ahmaud Arbery in Brunswick, Georgia.1 There, the defendants claimed to have suspected that Arbery was engaged in criminal activity and pursued him with a gun. When Arbery took action to protect himself, Travis McMichael shot and killed him. Here too, many were concerned that an acquittal would lead to greater vigilantism. And while the jury ultimately convicted, Georgia law would have also allowed acquittal in a similar or even identical case. The public discourse in these cases has tended to focus on the factual circumstances of these individual actors: defendants and victims. As a consequence, such cases have raised public concern that certain states’ gun-use and self-defense laws effectively invite malicious individuals—including vigilantes and white supremacists—to kill with impunity.

We argue that the situation is even worse than that. Certainly, premeditated intentional violence is of serious concern. However, under gun control regimes like Wisconsin’s and Georgia’s, even perfectly rational actors can easily find themselves committing deadly acts—no ill will required. Instead, when states combine generous open-carry policies, lax assaultive threat rules, and weak (or nonexistent) duties to retreat, the laws themselves create the danger. Together, such rules generate incentives for ordinary, rational individuals to issue progressively escalating threats of deadly violence for the sake of their own protection. The result is a scaled-down version of the brinksmanship that characterized midcentury nuclear strategy: a small arms race.

Here, we model the small arms race and suggest legal solutions. All practicable solutions involve breaking the cycle of escalation by imposing penalties on escalatory acts by one or more actors. Such penalties reduce actors’ access to self-defense in the short run while generating a less dangerous equilibrium for everyone in the long run. Because access to self-defense is at the core of the Second Amendment, we explore the potential constitutional consequences of our findings. We argue that the Second Amendment, properly understood, must sometimes allow governments leeway to make self-defense harder. At a minimum, such interventions must be constitutional when the resulting equilibrium makes society, on net, significantly safer than the old one.

I. Gun Laws, Self-Defense

We contend that the small arms race problem arises from a confluence of doctrines: (1) permissive firearm possession laws; (2) narrow definitions of assaultive threat; and (3) expansive self-defense and justification laws—including stand-your-ground laws and citizen arrest laws—that do not require retreat.

Both Wisconsin law and Georgia law exemplify this perilous confluence.

In Wisconsin, permits are not required to purchase firearms, including rifles, shotguns, and handguns. Nor, as in most states, are owners required to register their firearms. Permits are not required to openly carry firearms. Permits are required for concealed carry, but Wisconsin is a “shall issue” state, meaning that there are determinate criteria for obtaining such permit and that, if such criteria are met, the state authority has no discretion to deny the permit.

Wisconsin law has a high standard for imposing liability for assaultive threat, especially when firearms are involved. In the first instance, because openly carrying firearms is legal, one’s mere presence with a gun—which can in fact be threatening to others—is never actionable. Wisconsin’s criminal law regulates assaultive threats under the rubric of attempted battery.2 The attempt statute has been interpreted to require an intent to commit the crime—here, battering an individual—and acts done in furtherance of the commission. That is a high bar. As a consequence, many actions that, in actuality, would be threatening do not fall within the ambit of criminal liability. Indeed, even a literal threat to shoot someone may not be forbidden, so long as there is not actual intent to do so.3 This in turn incentivizes targets of such threatening conduct to engage in self-help.

Wisconsin has no duty to retreat with respect to self-defense.4 The statute reads “[a] person is privileged to threaten or intentionally use force against another for the purpose of preventing or terminating what the person reasonably believes to be an unlawful interference with his or her person by such other person.”

The statute further explains that “unlawful” is read to mean either criminal or tortious behavior. Thus, an actor may be justified in threatening deadly force even against others engaged in non-criminal conduct. And if those threatened with deadly force respond in kind, the use of deadly force by either party is, in turn, justified. Lastly, Wisconsin law appears to allow individuals to engage in investigation and citizen’s arrests. According to a 2008 letter from the Wisconsin Attorney General, “a citizen can make a felony arrest without a warrant based on probable cause but can make a warrantless arrest for a misdemeanor only if the misdemeanor is committed in the citizen’s presence and constitutes a breach of the peace.”

Consider the operation of these legal doctrines in the Kyle Rittenhouse prosecution. Protests erupted after police shot and paralyzed Jacob Blake in Kenosha, Wisconsin. Rittenhouse went to Kenosha with a semiautomatic rifle, he claimed, with the purposes of protecting businesses and offering medical aid if necessary. In the course of the protests, he was confronted by Joseph Rosenbaum. Rosenbaum, unarmed, threw a plastic bag at Rittenhouse, and a bystander Joseph Ziminski fired a shot in the air. Rittenhouse pointed his gun at Rosenbaum, who attempted to take it from him. Rittenhouse then shot and killed Rosenbaum. Rittenhouse then ran from the scene of the shooting and was confronted by two individuals, Anthony Huber and Gaige Grosskreutz.  Huber struck Rittenhouse with a skateboard, which knocked Rittenhouse to the ground. The evidence showed that Rittenhouse may have been struck by others as well. Rittenhouse then shot Huber and killed him. Grosskreutz believed Rittenhouse to be an active shooter and pointed a gun at Rittenhouse. Rittenhouse then shot Grosskreutz, injuring him. Rittenhouse was charged with several crimes, including intentional homicide, reckless homicide, attempted intentional homicide, and recklessly endangering safety (for his errant shots that did not strike other individuals). He was acquitted of all charges.

Whether Rittenhouse’s actions and verdict were morally and legally justified is the source of controversy. We do not intend to adjudge those questions here. Instead, we consider the events a tragedy and scrutinize how the law could produce such occurrences, even if no one involved had wished anyone else harm. To that end, we observe several important legal facts from the prosecution and verdict: Rittenhouse was not liable for assaultive threat for pointing the gun at Rosenbaum; Rittenhouse was not required to retreat when confronted by Rosenbaum, due to Wisconsin’s stand-your-ground law; Huber and Grosskreutz were likely not liable for confronting Rittenhouse themselves; and Rittenhouse was not required to retreat before rebutting the responsive actions of Huber, Grosskreutz, and others.

Like Wisconsin, Georgia has permissive gun laws. Georgia doesn’t require permits for possession of firearms or for open carry, and it is a “shall issue” state with respect to concealed carry. Furthermore, Georgia has a stand-your-ground self-defense law, imposing no duty to retreat, and it has a robust citizen’s arrest law.

With respect to threatening conduct, Georgia’s law is different from Wisconsin’s—perhaps importantly so. In Georgia, one is liable for simple assault if they “commit a violent injury to the person of another; or [they] [c]ommit[ ] an act which places another in reasonable apprehension of immediately receiving a violent injury.” Unlike in Wisconsin, this statute has not been interpreted to require a specific intent to injure, just a general intent to commit the acts in question. This rule might criminalize conduct that would begin perilous escalations. However, there are two reasons for possible skepticism. First, for liability to attach under Georgia’s law, the conduct must place another in “reasonable apprehension of immediately receiving a violent injury” (emphasis added). This may serve to exclude from criminal liability much factually threatening conduct that leads to escalation. Second, the theoretical operation of this statute may be overridden by Georgia’s broad stand-your-ground and citizen’s arrest laws. If the operation of Georgia’s law actually encompasses more conduct under simple assault than Wisconsin’s, that is welcome. Indeed, the mere fact that Georgia’s assaultive threat rule applies nebulously to gun threats might itself be sufficient to deter many unnecessary threats.

Consider now the murder of Ahmaud Marquez Arbery. On February 23, 2020, Arbery was running in a neighborhood near Brunswick, Georgia. Three men—Travis McMichael, Gregory McMichael, and William Roddie Bryan—pursued Arbery after seeing him run by McMichael’s home. They claimed to have suspected that Arbery committed a burglary or robbery in the neighborhood. They approached Arbery in two separate vehicles, and Travis McMichael exited his vehicle holding a shotgun. Travis McMichael and Arbery engaged in a physical confrontation, during which Travis shot and killed Arbery. After a significant delay, during which one prosecutor determined that charges were not warranted, all three were charged on various counts, including murder. Most relevantly, Travis McMichael was convicted of malice murder and felony murder; and Gregory McMichael and William Roddie Bryan were convicted of felony murder.

At first glance, these convictions may seem to suggest that Georgia law works to deter unwarranted violence. We don’t think so. Though jury verdicts are by design inscrutable, we think there were particular facts here that led to this verdict. Most importantly, Arbery was unarmed and, when pressed by the prosecution, the defendants acknowledged that they understood this in the moment. The prosecution also stressed that defendants had no specific knowledge that Arbery had committed any crime. Finally, the prosecution emphasized that the defendants engaged in continued aggressions in their pursuit of Arbery, who made initial attempts to flee the danger without violent engagement. All of this painted a picture that the defendants engaged in unwarranted violence (and perhaps unacceptable vigilantism), rather than self-defense. But we suspect a change in any of these facts—had the victim been armed, had there been reason to think the victim had committed a crime, or had the victim engaged in violent action himself (even if such action was justified by the circumstance)—might well have resulted in a different verdict. Indeed, Professor Alec Walen has presented a compelling case that, even on the case’s actual facts, the defendants could have plausibly been acquitted.

II. The Small Arms Race

Many commenters on the Rittenhouse and Arbery cases have argued that legal systems like Wisconsin’s and Georgia’s invite vigilantism—or even murder disguised as vigilantism. Their contention is that, under such legal rules, badly-motivated actors could seek out opportunities for violence and ultimately commit it with impunity. Perhaps a bad actor would travel to a disorderly protest, hoping to find someone arguably committing a crime. Then, the bad actor would confront that person, provoke them into a show of force, and kill them.

Without commenting on Rittenhouse’s or Arbery’s murderers’ actual motivations, we argue that the situation under laws like Wisconsin’s and Georgia’s is even more dire. Bad intentions are not necessary for violent results. Instead, such laws generate incentives for perfectly rational, well-intentioned individuals to end up in deadly exchanges.

This is because of arms races. “Arms race” is the name given by midcentury game theorists to a formal economic model designed to help them—and political actors—understand Cold War nuclear strategy. Why would the United States and Soviet Union amass enough destructive force to be able to end life on earth many times over? And why might they use it? The arms race model is instructive.

 In this Essay, we use nuclear strategy as a lens to help understand gun violence. But for early game theorists, the analogy ran in reverse. They used guns to shed light on warheads. Consider this passage from Thomas Schelling’s foundational book, The Strategy of Conflict:

[I]f I go downstairs to investigate a noise at night, with a gun in my hand, and find myself face to face with a burglar who has a gun in his hand, there is a danger of an outcome that neither of us desires. Even if he prefers to leave quietly, and I wish him to, there is a danger that he may think I want to shoot, and shoot first. Worse, there is danger that he may think that I think he wants to shoot. Or he may think that I think he thinks I want to shoot. And so on.

In Schelling’s model, my uncertainty about the burglar’s intent is not enough, on its own, to lead to a shootout. The question is what I then do about it. As Schelling explains, I would much prefer shooting the burglar first over either letting him shoot me first or shooting one another simultaneously. So I prepare myself to draw more quickly than the burglar, just in case he has murder on his mind. I place both hands on my pistol’s grip and tense my muscles. And why not? My movement from repose to readiness is cheap, and it does much to insure me against a surprise attack. However, my own readiness to shoot in an instant necessarily reduces the amount of care I can exercise when evaluating the burglar’s next move. Maximum speed, not maximum care, is the dominant strategy. This protects me, but it also increases the likelihood that I accidentally shoot the burglar, mistaking his innocent fidget for an attempt to draw his weapon.

The burglar notices my shift to high alert. He understands that my probability of shooting him—even because of a misunderstanding—has just gone up. What can he do but protect himself, just as I have? He puts his hand on his gun. I now notice his readiness and understand the increased risk it poses to me. I adjust even further. And now we are spiraling toward a bloody end.

It is worth noting here that the arms race model is a variation on another classic game-theoretical framework: the prisoner’s dilemma. As with the prisoner’s dilemma, arms races arise when actors’ individual payoffs are misaligned with net social payoffs. Arms races, however, are particularly pernicious because small, incremental steps that seem low-risk and high-reward can set off cascading series of escalation, triggering disaster. Perfectly-informed, perfectly-rational actors can anticipate these series and the inevitably-resulting attacks, but that alone does not solve the problem. Instead, it just means that such actors’ decision sets can be reduced to two choices: attack immediately or never. After all, why go through the motions of a slow escalation when one can see the inevitable result from the outset?

The emergence of an arms race depends on two factors. First, it depends on the baseline risk that one actor actually desires to do the other harm. Second, it depends on how much each escalatory move increases the risk of an accidental shooting. If either of these is sufficiently small, the applicable game-theoretical model may instead be the “stag hunt,” wherein rational actors may cooperate to avoid socially undesirable outcomes.

After Schelling, many scholars contributed to the arms race literature, applying the model to financial crises, modeling arms races under uncertain payoffs, and adding communication into the model. Yet Schelling’s original formalization still best captures the dynamics of many small arms encounters. In these encounters, as in Schelling’s model, agents do not know for certain one another’s intentions. Often, actors in armed encounters are either strangers or are acquaintances with a history of antagonistic, though non-lethal, encounters. This, along with the briefness and intensity of armed encounters, also undermines opportunities for effective communication. Furthermore, although agents may not be in a position to quantify exactly the (negative) payoff of being shot or the (less negative) payoff of shooting another person, the relative payoffs are utterly clear. It is much better to shoot than be shot. And it is very cheap to prepare oneself to shoot first, should the need arise. These factors again fit Schelling’s formalization, under which the optimal response to a mere probability of being shot is to increase readiness, decrease caution, and thereby make oneself more threatening. Thus, the deadly chain of escalation is triggered.

What does any of this have to do with law? Shelling imagined small arms races as arising in limited contexts—like late-night burglaries. But legal regimes like Wisconsin’s and Georgia’s broaden the fertile ground for small arms races substantially. They institute conditions promoting small arms races essentially everywhere: public streets, sports stadiums, bars, and more. Wisconsin and Georgia’s laws do this by refusing to apply the brakes. At every juncture where legal rules might be imposed to negate incentives to escalate–and thus break the arms-race cycle—Wisconsin and Georgia do nothing.

Consider the following vignette. Jesse needs to go out late at night in a high-crime area.5 There have been a number of acts of theft, vandalism, and assault in the neighborhood recently. The police cannot usually stop these crimes as they are happening. At best, they find and arrest the perpetrators later. Should Jesse rationally go out armed? It seems so. Jesse’s state—like Wisconsin and Georgia—makes owning and carrying a gun cheap, both monetarily and legally. And, assuming Jesse is a competent user of a gun, he will surely be better off armed than unarmed, if attacked by an armed assailant.

As Jesse is walking toward the bus stop, Lee begins to follow Jesse. Jesse is uncertain of Lee’s motives. Is Lee a robber, or just another public transit rider headed to the same station? Nervous, Jesse speeds up slightly. A minute later, Jesse sees that Lee has also sped up. Unbeknownst to Jesse, Lee has a transit tracking app and knows that haste is needed to catch the next bus.

Like Jesse—indeed, for the same reasons—Lee is armed. Lee’s gun is carried openly. In Lee and Jesse’s state, as in Georgia and Wisconsin, this is legal.6 Thus, as in those states, in Lee and Jesse’s state merely carrying a weapon openly never constitutes an assaultive threat. Glancing back at Lee, Jesse notices a hip holster. Now very nervous, Jesse looks around and realizes that there are no businesses open along this block where Jesse might seek the safety of company.

Jesse could run but worries that this will alert Lee to Jesse’s fear and leave Jesse defenseless while fleeing. Instead, Jesse turns and says, “Stop. I have a gun. I will shoot.” Under a law like Wisconsin’s and Georgia’s stand-your-ground statute, Jessie is entitled to do this, so long as it is reasonable to believe that Lee is a threat to Jesse’s person or property. That standard is quite plausibly satisfied here. Jesse is not certain that Lee is a threat. But it is not unreasonable to think that Lee might be.

Now it is Lee who is worried. Why has Jesse issued the threat? Is this a robbery? Lee likewise chooses not to flee, instead retorting, “Don’t try anything. I’m also armed.” Realizing that Jesse may already be prepared to fire, Lee draws the hip-holstered gun. Again, under the state’s stand-your-ground law, Lee may threaten force, rather than retreating, since it is not clearly unreasonable to believe that Jesse is a threat. Likewise, under a law like Wisconsin’s or Georgia’s, Lee is not liable for assaultive threat for brandishing a weapon, since Lee does not intend to shoot Jesse except in justifiable self-defense.

At this point, Jesse and Lee are engaged in a true arms race. Despite both parties’ benign intentions, it is now objectively true that Lee poses a nontrivial threat to Jesse. This is so just because Lee is poised to shoot, increasing the risk of “accidental” violence triggered by a misinterpretation of Jesse’s next move. Absent any external command to the contrary, Jesse’s only rational response is to prepare to shoot Lee even more quickly—and less cautiously. At some point, Schelling tells us, one must rationally shoot the other to head off what quickly becomes a very high probability of being shot themselves. We emphasize that this result is objectively rational. Thus, such a use of deadly force would be justifiable as self-defense under statutes like Wisconsin’s and Georgia’s. The arms race results in deadly violence, and law does nothing to stop it.

This is just one hypothetical example of legal rules touching off a needless arms race. But lest readers think this vignette idiosyncratic, consider how these dynamics could play out elsewhere. We think that laws like Wisconsin’s and Georgia’s are particularly likely to transform what would otherwise be low-level altercations into deadly shootouts. Imagine an argument in a bar between two people who happen to be armed. Neither needs to have any intention of using their gun to resolve a barfight for a quick-draw contest to ensue. Both know the other is armed and could easily shoot; neither knows for certain what the other intends; and each may be especially wary if the other is or may be drunk. Here, a runaway spiral of threats can easily ensue.7 Or consider instead a heated argument between domestic partners. Or an encounter between drivers who have just had a fender-bender. The list goes on.

Not all of these stories will necessarily turn deadly, even under legal regimes like Wisconsin’s and Georgia’s. Whether a given armed encounter will spark a small arms race depends on multiple factors. In part, it depends on the base rate of individuals who, in the relevant context, would actually desire violence. If violent intentions are sufficiently rare, and if both parties know this, then neither party needs to prepare for a first strike, even if they cannot know for certain what the other intends. In that case, the spiral of escalation will not ensue.

The cost of each escalatory act also matters greatly to the probability of an arms race. And as described above, laws like Wisconsin’s and Georgia’s make each escalatory move cheap. These laws impose zero (or near-zero) cost on acts like getting a gun, carrying it wherever one goes, intoning violence rather than fleeing from a moderately-probable threat, brandishing the weapon in response to an escalated threat, and ultimately shooting in self-defense.

III. Solutions

If certain combinations of laws trigger small arms races, then how can law help to prevent them? One option might be to make the world very safe. In a world where crime was extremely rare, even the modest cost of bearing arms—say a few hundred dollars to buy a handgun—would not be worth it for most people. This alone could collapse potential arms races, as encounters would rarely be armed, and escalations—even prophylactic ones—would become irrational. Unfortunately, reducing ambient crime rates is very, very hard. Violent crime, for example, dropped precipitously in the 1990s but has begun to rise again in the past few years. Neither the cause of the decline nor of the rise is well-understood.

What, then? If law cannot reliably reduce crime to near-zero, thereby reducing armed encounters, it must instead put the brakes on armed encounters that would otherwise spiral into arms races. Law can do this by making one or more escalatory steps more costly. It could make gun permits harder to obtain, guns more expensive to purchase, choices to stand one’s ground more legally perilous, brandishing of guns more likely to invite liability, or self-defense harder to prove.

It is difficult to say for certain which legal intervention would most efficiently short-circuit small arms races. The goal is to make at least one escalation for at least one party more costly (in terms of effort, money, or expected liability) than it is beneficial (in terms of ability to shoot first against a potential bad actor). Thus, which legal intervention is best depends on the baseline risk of shootings, the baseline cost of each escalation, and the baseline marginal increase in the risk of an “accidental” shooting from each escalation. None of these is easy to estimate a priori.

We suggest that the common law may contain some wisdom here. For centuries, English law recognized a duty to retreat from violent threats, rather than use deadly force—at least when retreat was feasible. And even in the United States, “stand your ground” has only been recently expanded outside the home—usually by statute. Perhaps this long tradition of requiring retreat, at least where reasonably available, reflects two intuitions relevant here. First, retreat might often be easy, meaning that escalation in the form of refusing to retreat has little marginal value in the arms race. Second, requiring parties to consider retreat before escalating might impose just enough cost on escalation to make retreat the rational choice in many important scenarios. If that is right, then repealing “stand your ground” rules (or otherwise reimposing a duty to retreat), might efficiently defuse many small arms races.

For similar reasons, another potential solution might be to expand assault liability to more readily cover brandishing of guns. In contrast to Wisconsin’s rule, many states—past and present—have held that pointing a gun at someone is an assault, even if one does not intend to shoot. Indeed, as discussed above, while Georgia’s suite of gun and self-defense laws is imperfect, this is one thing the state gets right. Expanding assault liability could again head off small arms races by raising the cost of marginal escalations. Expanding assault liability would not make self-defense impossible in cases where it is needed. First, states could continue to recognize true exercises of self-defense as an exception to assault liability. Second, even if assault liability were sometimes imposed on actors who truly thought they were in imminent danger, surely the threat of a minor penalty would not deter true acts of self-preservation.

We do not mean to suggest that these are the only potential solutions to small arms races. Indeed, we hope that, by identifying and framing this problem as a species of arms race, we might inspire others to invent other creative fixes. Nevertheless, we think our example solutions have the salutary feature of being well-rooted in the common law. This, we think, counts as some evidence that they might be efficient—having evolved over time to fit societies’ needs. Our solutions’ historical pedigree will also matter, as discussed below, to their constitutional analysis.

We pause here to concede that, though we have thus far focused on idealized rational actors, in the real world, people are not always totally rational. Sometimes they are, sometimes they aren’t, and sometimes they rapidly shift from rational to irrational. Especially in potentially-dangerous encounters, people may not act according to reasoned cost-benefit analysis. And in such situations, they may become less rational as the encounter escalates. Insofar as that is true, solutions to small arms races that operate later in a given armed encounter may be less effective than earlier interventions. On the other hand, we also think that law can shape culture, which can in turn shape non-rational instincts. This is especially true in communities where knowledge of substantive legal rules is unusually strong—as with gun owners. Thus, a gun owner who is not rationally weighing costs and benefits may nevertheless resort to legally-defined default actions. In a duty-to-retreat jurisdiction, retreat might become the first instinct, rather than standing and shooting. In this way, rationally designed gun laws might exert a rationalizing force on even non-rational actors.

Finally, we emphasize that it is crucial to evaluate any proposed solution to the small arms race problem in terms of its net effects on safety. One objection to imposing, for example, a duty to retreat is that there will be costly errors. Sometimes, one party in an encounter really will intend the other harm. And in some of those cases, forcing the potential victim to retreat—or even consider it—rather than defend themselves will be the difference between life and death. This is especially the case given that small arms races arise with entirely rational actors, who have assessed the risks of death or injury appropriately. This will be true for any of our proposed legal solutions to small arms races, since each is specifically designed to make self-defense marginally harder.

An increase in costly errors does not mean, however, that changing the law to prevent arms races will make people, on net, less safe. True, these changes would make self-defense slightly harder in some instances where one person has the genuine intent to harm another. But in exchange, they could greatly reduce “accidental” shootings—those that result from mutual fear and mistrust, not malice. This is an empirical question, but we think that the net effect likely favors legal reform—perhaps strongly. For one thing, as already discussed, legal reform need not produce many costly errors, since a true necessity for self-defense could always remain a defense to liability. Second, we think that serious, premeditated malicious intent calibrated to take advantage of the nadirs in the law is probably rare. Or, at least, it is rare compared with the situations that could produce unintentional small arms races, absent legal intervention.

IV. Constitutionality

As just discussed, the plausible legal solutions to the small arms race problem all increase the cost of self-defense. In District of Columbia v. Heller (2008), the Supreme Court held that the right to defend oneself—and to do it with a gun—forms the core of the Second Amendment’s protections. Thereafter, in McDonald v. City of Chicago (2010), a plurality of the Court determined that that the Second Amendment was incorporated against the states, as it related to the fundamental interest in self-defense. We thus briefly address whether interventions like re-imposing a duty to retreat or expanding liability for assaultive threats raise significant Second Amendment difficulties.

We think not. We contend that, under Heller and McDonald, self-defense is not an end in itself. The Second Amendment does not celebrate shootouts for shootouts’ sake. Rather, the right of self-defense is itself rooted in the fundamental liberty interests of safety and well-being. Thus, the Heller Court was clear that, at least sometimes, governments may legitimately curtail self-defense in ways that serve those interests. Heller said and McDonald reaffirmed that “longstanding prohibitions on the possession of firearms by felons and the mentally ill, . . . laws forbidding the carrying of firearms in sensitive places such as schools and government buildings, [and] laws imposing conditions and qualifications on the commercial sale of arms”—as well as prohibitions on “the carrying of dangerous and unusual weapons”—were permissible.

The Heller majority approved of other safety-promoting gun restrictions, too. The Court looked favorably on Founding-era laws that imposed liability for: taking a loaded weapon into any building, firing a weapon within city limits, and firing a weapon during certain public celebrations.

Each of these “longstanding prohibitions” makes self-defense with a gun more difficult or costly, at least in certain circumstances. A prohibition on carrying a gun into a bar makes it more difficult to defend oneself against a violent drunk. Nonetheless, a general rule forbidding intoxicated patrons of crowded bars from being armed almost certainly increases safety overall. And, as Heller says, this derogation of the self-defense right is acceptable.

In our view, Heller’s analysis must rest on some kind of balancing inquiry. Heller, we think, showed that, from the Founding to the present, the Second Amendment has been understood to allow well-tailored gun restrictions that produce significant net safety gains. Of course, the Founders understood, the state may constitutionally prohibit people from carrying guns in ways that threaten public safety. And it may do so precisely because the safety benefits to those same people outweigh the safety benefits of having the gun in the first place. Whether the Court is ready to explicitly acknowledge this principle is beyond our ability to predict.

Our cost-benefit reading of Heller may sound too pragmatic for certain originalist-leaning scholars. But let us reemphasize here that we understand our reading to be consistent with originalism. It is precisely history, as interpreted in Heller’s avowedly-originalist majority opinion, that reveals the Second Amendment’s amenability to net-benefits analysis.

For those who are still not convinced, let us additionally highlight the historical bona fides of our proposed interventions. As noted above, both a robust duty to retreat and broad assault liability for threats with guns are ancient, traditional rules. It is indeed states that lack a duty to retreat and lack liability for gun threats that are engaging in legal experimentation. Thus, if any legal rules count as “longstanding” and therefore constitutionally permissible regulations of gun use under Heller, our proposed interventions must be among them.

Other solutions to the small arms race problem might, as mentioned above, include further conditioning the commercial sale of firearms to make their acquisition more costly. Alternatively, legal rules could limit the use of firearms in sensitive places—which might include protest sites and areas that are expected to be crowded. These interventions might well pass muster under the same historically-grounded net-safety reasoning described above. Likewise, imposing certain versions of these rules might even represent a return to tradition, rather than a break with it. We are not certain. But at a minimum, re-imposing either a robust duty to retreat before using deadly force or assault liability for gun threats should pass both tests.

V. Conclusion

As we have shown, gun-regulation regimes like Wisconsin’s—and perhaps Georgia’s—are dangerous. But they are not dangerous primarily for the reason many people think. True, such regimes might make it easier for people who desire to commit violence to do so without consequence. But people like that are relatively rare. By contrast, ordinary people who wish to protect themselves against potential violence are extraordinarily common. And for people like that, regulatory environments like Wisconsin’s generate rational incentives to carry arms, brandish them during tense encounters, and ultimately commit violence—even when one does not wish to. This, however, is a legally-created problem amenable to legal solution. By imposing costs at some stage of escalation in small arms races, law can invert the incentive structure for ordinary citizens and thereby encourage peaceful cooperation. Potential solutions include the reimposition of certain traditional common law duties and liabilities. Because of these solutions’ historical pedigree—and because of their potentially-large net payoffs—the Second Amendment poses no obstacle to fixing small arms races.

• 1Race was a core component in Arbery’s murder, though we do not focus on it here. Instead, we show how pathologies in the law can lead to tragedy even between actors without bad motivations.
• 2See, e.g., State v. Steers, 495 N.W.2d 527 (Ct. App. 1992) (individual charged with “attempted battery” for holding a knife to another’s throat). Wis. Stat. § 941.20(c) criminalizes, as a misdemeanor, intentionally pointing a gun at another individual. Similar to battery, this requires intent—here the intention to point the firearm at or toward a person—which we think still serves as a high bar. Practically, as a matter of proof, it may be difficult to show that a person was intentionally pointing a gun at or toward another, rather than simply wielding the gun. See Joseph Blocher, Samuel W. Buell, Jacob D. Charles & Darrell A.H. Miller, Pointing Guns, 99 Tex. L. Rev. 1173, 1188-90 (2021) (discussing the shortcomings of laws criminalizing pointing guns at others). We thank Jake Charles for this note. Wis. Stat. § 943.30 also criminalizes inter alia communicative threats to injure, but there may be threatening behavior that isn’t communicated “verbally or by any written or printed communication,” as that section requires.
• 3Whether actual intent may be inferred from words is a factual question, but we can imagine scenarios in which verbal threats are found to be empty.
• 4Though Wisconsin has no duty to retreat, the availability of the opportunity to retreat may inform whether the defendant reasonably believed the force used was necessary to prevent an interference with his or her person. See State v. Wenger, 593 N.W.2d 467, 470–71 (Wis. Ct. App. 1999). We suspect that jury instructions correctly noting the state’s avowed lack of a duty make a functional duty to retreat unlikely.
• 5We use “high-crime area” here for two reasons. First, the base rate of crime is one of the key inputs into the perfect-information arms race model. Second, the term is legally salient in, e.g., the Supreme Court’s Fourth Amendment jurisprudence—though there are reasons to criticize its use there.
• 6Note that concealed carry might make the problem even worse by further obscuring the level of risk.
• 7Importantly here, some states have recognized the particular problems that intoxication may bring and have regulated to attempt to avoid the dangerous spiral. See, e.g., Wis. Stat. § 941.20(b); 21 Okl. St. § 1272.1. These are welcome regulations, but intoxication is not the only condition that might raise the risk of an arms race.