The Cases in Brief

Early in the pandemic, the Supreme Court exhibited a deferential attitude toward COVID-19–related restrictions on worship. In South Bay United Pentecostal Church v. Newsom, the Court voted 5–4 to reject a church's request to enjoin enforcement of COVID-19–related restrictions in California, which had imposed a stringent occupancy cap on worship that didn't apply to many secular businesses.Footnote ¹⁷

The Court didn't issue a written opinion in South Bay. But a concurrence by Chief Justice Roberts focused the constitutional analysis squarely on comparative risk. According to Chief Justice Roberts, California's restriction didn't violate the Free Exercise Clause because the state had applied “similar or more severe restrictions . . . to comparable secular gatherings,” where people “gather in close proximity for extended periods of time.”Footnote ¹⁸

Justice Kavanaugh dissented in South Bay along with Justices Thomas and Gorsuch.Footnote ¹⁹ These justices didn't contest the idea that the key question was whether California's restrictions treated worship worse than other “comparable” gatherings. But they took issue with Chief Justice Roberts's view that worship services were meaningfully different from places like shopping malls, factories, and cannabis dispensaries in terms of risk. “Assuming all the same precautions are taken,” the dissenters pressed, “why can someone safely walk down a grocery store aisle but not a pew?”Footnote ²⁰

As a one-justice concurrence, Chief Justice Roberts's opinion in South Bay wasn't technically binding on lower courts. But most courts took it as a signal and acted accordingly, concluding that more stringent restrictions on worship were permissible because—unlike other activities—a generic worship service typically involves people singing and chanting in a confined space for extended periods of time.Footnote ²¹ And with the Supreme Court's rejection of a challenge to Nevada's restrictions on worship a few weeks later, that approach seemed well cemented.Footnote ²²

Things changed in Roman Catholic Diocese of Brooklyn v. Cuomo.Footnote ²³ There, Catholic and Jewish claimants challenged restrictions on worship in certain parts of New York City. In a Red Zone, houses of worship were limited to a 25 percent occupancy cap or ten people, “whichever is fewer.”Footnote ²⁴ Restrictions in Orange Zones imposed a similar twenty-five-person cap.Footnote ²⁵ But in each of these zones, businesses and gatherings designated “essential” faced no such restrictions.Footnote ²⁶

Here unlike the prior church-closure cases, the Court issued an emergency injunction prohibiting enforcement of the restrictions.Footnote ²⁷ The contrast with South Bay was striking. Rather than citing generalized differences between worship and other gatherings, the Court noted that “[t]he Governor . . . stated that factories and schools have contributed to the spread of COVID-19,” yet treated those entities less harshly than the Catholic and Jewish claimants who had “admirable safety records.”Footnote ²⁸ Likewise, the Court expressed skepticism that a difference in risk could justify the differential treatment: “[A] large store in Brooklyn . . . could have literally hundreds of people shopping there on any given day,” yet a “nearby church . . . would be prohibited from allowing more than 10 or 25 people inside for a worship service.”Footnote ²⁹

The Court's opinion in Diocese of Brooklyn was per curiam, so it didn't identify an author or indicate which justices had supported it. But the accompanying concurrences and dissents revealed the vote had been 5–4, and that the new majority was composed of the prior dissenters in South Bay plus Justice Amy Coney Barrett, who had replaced Justice Ruth Bader Ginsburg just one month before.Footnote ³⁰ And after Diocese of Brooklyn, the Court enjoined officials in other jurisdictions from enforcing similar restrictions on worship, often exhibiting the same 5–4 split.Footnote ³¹

The debate in the church-closure cases was bitter. In Diocese of Brooklyn, Justice Gorsuch implied the dissenting justices were “shelter[ing] in place when the Constitution is under attack.”Footnote ³² In a later case, Justice Kagan retorted that the majority was engaged in “armchair epidemiology” rather than a reasonable reading of the Free Exercise Clause.Footnote ³³ But more fundamentally, these cases pointed to a bigger puzzle in the Court's jurisprudence.

Secular Comparators

In the early church-closure cases like South Bay, the Court didn't explain how its analysis tracked the requirements of neutrality and general applicability. But by focusing almost exclusively on whether worship services were “comparable” to less-regulated activities, it was clear the Court was invoking the general applicability requirement—and more specifically, the strand of general applicability focused on secular comparators. In the context of the church-closure cases, though, analyzing that issue proved extremely difficult.

Begin with the basic rule. As Douglas Laycock and Steven Collis have aptly said, general applicability is exactly what it sounds like—a requirement that a law or policy “applies to everybody, in all similar situations—or at least to nearly everybody and nearly all similar situations.”Footnote ³⁴ And as the Court has explained, one obvious way to enforce that requirement involves secular comparators. On this understanding, a law fails the general applicability requirement where it refuses an exemption for religious conduct while granting one to “secular conduct that undermines the government's asserted interests in a similar way.”Footnote ³⁵ The reason should be obvious: if religious and secular conduct are similarly situated in terms of the government's interests, granting a secular exemption but denying a religious one isn't traceable to a difference in circumstance. Instead, it reflects a negative value judgment about religion.

To illustrate the point, consider just one well-known example. Imagine a police department has a policy prohibiting facial hair, in order to give its officers a uniform appearance that's easy for the public to recognize. The policy contains no explicit exceptions or discretionary language. But as a matter of practice, the department allows some officers to grow beards for medical reasons but refuses to allow beards for religious reasons. Has the government acted without regard for religion? No, because the policy isn't generally applicable. Exemptions for medical beards and religious beards both undermine the government's asserted interest in officer uniformity. As a result, choosing to grant one exemption but not the other isn't traceable to a difference in circumstance. Instead, it reflects an official judgment that growing beards for medical reasons is more important than growing them for religious reasons. And by making a judgment like that, government has devalued religion in a way the Free Exercise Clause forbids.Footnote ³⁶

Here, though, it's worth noting an important complication. Some courts and scholars have insisted that, contrary to the description above, a law is still generally applicable even if government has denied a religious exemption while granting one for a similarly situated secular activity. On this view, the general applicability rule mandates heightened scrutiny only where a law or policy is so riddled with exceptions that it provides clear evidence of “something more sinister”—namely, raw religious animus.Footnote ³⁷

Prior to the church-closure cases, the Court hadn't spoken to this issue directly. But in Tandon v. Newsom, the Court offered two clarifications relevant to this debate.Footnote ³⁸ First, it said that when inquiring whether two activities are “comparable” for purposes of general applicability, what matters is “the asserted government interest that justifies the regulation at issue.”Footnote ³⁹ But second and more notably, the Court said that a law doesn't qualify as generally applicable just because it “treats some . . . activities as poorly as . . . religious exercise.”Footnote ⁴⁰ Instead, general applicability is lacking whenever government treats “any comparable secular activity more favorably than religious exercise”—that is, when it refuses to exempt religious conduct but exempts secular conduct that undermines its interest in a similar way.Footnote ⁴¹

In scholarly terms, Tandon signaled a setback for the animus interpretation of general applicability and a victory for what has been called the “most-favored nation” theory.Footnote ⁴² That theory argues that under the First Amendment, religious conduct must be treated at least as well as preferred secular conduct in terms of exemptions where the government's interest in restricting both is roughly parallel.Footnote ⁴³ Allowing police officers to grow beards for medical reasons but not religious ones may not reflect outright hostility toward Muslims or Sikhs. But judged against the purpose of the no-beard rule, it still signals a devaluing of religion when compared to needs the government considers more important.

Only time will reveal the full significance of Tandon. But even taking the Court at its word about the meaning of the secular comparators inquiry, the COVID-19 church-closure cases also highlighted a difficulty. They showed just how hard it can be to identify which activities are truly “comparable,” at least when the underlying controversy relates to highly technical questions like disease transmission.

Consider first the more aggressive stance. In choosing to invalidate many state restrictions on worship, the majority in cases like Diocese of Brooklyn made judgments about the comparative risk of virus transmission in churches versus retail stores. But as to a subject like that, judicial expertise is at its limit. Where the issue is likelihood of infection, deciding whether allowing churches to operate at 50 percent capacity endangers the government's interest “in a similar way” as a 50-percent rule for big box stores requires epidemiological expertise outside the core of judicial competence.Footnote ⁴⁴ To be sure, the difficulty of that question isn't a reason to avoid it. As one great justice observed long ago, judges in our system are obliged to defend the Constitution “not by authority of [their] competence, but by force of [their] commissions.”Footnote ⁴⁵ But in a context like the church-closure cases, where the questions arose as requests for emergency relief, the Court's decisions on those technical issues attracted understandable criticism.

But the problem runs the other way too. The justices who dissented in Diocese of Brooklyn and the later church-closure cases focused on the fact that, according to experts, the risk of virus transmission was especially strong in churches and other religious gatherings. But approaching the problem at that level of generality fails to grapple with what made the cases genuinely hard.

To understand the difficulty, think about just a few problems arising from Diocese of Brooklyn, the case about New York's ten-person cap on worship. Imagine the experts say that attending religious worship is one of the “riskiest activities” in terms of transmission, because “[l]arge groups of people enter an enclosed space, then sit or stand near each other for a significant amount of time, and typically talk and sing.”Footnote ⁴⁶ But what if, as the Catholic plaintiffs did, a church reduced its services to less than an hour, eliminated singing, required masks, and spaced attendees significant distances apart—even when entering and leaving the building?Footnote ⁴⁷ If an average trip to the grocery store takes roughly the same amount of time, involves brushing past other shoppers to grab the cereal or the apples, and may require standing in line behind someone talking loudly to their spouse, is it really so obvious that the government is acting evenhandedly by imposing a ten-person cap on one activity but not the other?

Likewise, imagine the government claims that some religious groups are ignoring safety requirements like masks or distancing—here, the city's Orthodox Jews—so it decides to place special across-the-board attendance caps on worship.Footnote ⁴⁸ If similar violations are discovered elsewhere—say, in restaurants serving alcoholFootnote ⁴⁹—but government opts for retail punishments rather than the same across-the-board restrictions it placed on worship, can that difference be fully justified by technical judgments about risk? Or at the very least, might it also depend on unconfirmed assumptions about what “usually happens” at a Shabbat service or worship services generally?Footnote ⁵⁰

The answers to these questions, I submit, are not obvious. And indeed, it is precisely because they are not obvious that Diocese of Brooklyn and the other church-closure cases were so divisive. Smith's general applicability prong requires courts to ask whether the government is actually treating religious conduct the same as nonreligious conduct posing a similar threat. But in the church-closure cases, that task involved an assessment of technical evidence susceptible to different readings depending on whether one focused on the claimants themselves or imagined generic conduct.

These problems probably can't be solved entirely, either in cases involving COVID-19 or in others presenting similar difficulties. Smith and its general applicability requirement are the law (at least for now),Footnote ⁵¹ so inquiries into comparative harm will often be a part of the framework. But even recognizing that, one might still wonder if there was another way to approach the problem.

The Other Side of General Applicability

As mentioned above, general applicability is more or less exactly what it sounds like: a requirement that laws apply across the board and treat similarly situated people and activities the same. In the free exercise context, that requirement has several purposes. One purpose is to provide vicarious protection for religious minorities, by tying their rights to those of the majority. It ensures that, as Justice Robert Jackson famously said, “the principles of law which officials would impose upon a minority must be imposed generally.”Footnote ⁵² But another purpose—and perhaps the more prominent one—is to identify instances of nonobvious bias or selective indifference toward religion, even if the law itself is facially neutral.Footnote ⁵³ Requiring that laws burdening religion be generally applicable is a way of ensuring that government is applying its rules even-handedly, and not in a way that treats religion with less solicitude than other similarly situated activities.

But how exactly does this requirement work? As discussed above, one aspect of general applicability—perhaps the most prominent one—focuses on comparison. Has the government refused religious exemptions while granting them to secular conduct that “undermines the government's asserted interests” in a similar way?Footnote ⁵⁴ But in fact, that's not the whole universe. General applicability also has another aspect. To see it, though, the first step is to return to the Court's decision in Smith.

In Smith, the Court considered free exercise claims brought by two workers who were denied unemployment benefits after they were fired for using peyote in a Native American religious ceremony. The Court said the Free Exercise Clause did not require an exemption because Oregon's ban on peyote was neutral and generally applicable. And according to Justice Scalia, that had always been the law—the Court “ha[d] never held that an individual's religious beliefs excuse him from compliance with an otherwise valid law.”Footnote ⁵⁵ In so insisting, however, the Court had to distinguish several past cases that suggested the opposite. In Sherbert v. Verner, the Court had held that the Free Exercise Clause forbade a state from requiring a claimant seeking unemployment benefits to work on Saturday where her religion observed that day as a Sabbath.Footnote ⁵⁶ The Court reasoned that forcing her to choose between receiving benefits and violating her faith was equivalent to imposing a “fine . . . for her Saturday worship.”Footnote ⁵⁷ And a series of later unemployment cases said the same thing.Footnote ⁵⁸

The Court in Smith dealt with this problem by arguing that—unlike Oregon's “across-the-board” ban on peyote—the laws in these unemployment cases were not generally applicable.Footnote ⁵⁹ Rather, they involved a “system of individual exemptions” that “invite[d] consideration of the particular circumstances” behind each claimant's conduct.Footnote ⁶⁰ So in Sherbert, for example, the problem wasn't that the law had forced Adell Sherbert to choose between observing her Sabbath or receiving benefits. Rather, the problem was that the law didn't actually impose a blanket requirement that claimants work on Saturday. Instead, it allowed officials to enforce or ignore the rule based on an “individualized . . . assessment of the reasons” for each applicant's refusal—and more specifically, based on an assessment of whether a claimant's refusal was justified by “good cause.”Footnote ⁶¹ And in a situation like that, the Court said, denying exemptions for religious claimants required heightened scrutiny.

Smith's discussion of the unemployment cases gave significant content to the general applicability requirement by focusing on laws that grant unbridled discretion to enforcement authorities. And, fittingly, commentators and courts have often viewed Sherbert in those terms, asking whether the government has created a system of entirely discretionary exemptions.Footnote ⁶² It was that aspect of Sherbert, for instance, that the Supreme Court relied on in Fulton v. City of Philadelphia.Footnote ⁶³

In Fulton, the Supreme Court held that the Free Exercise Clause forbade Philadelphia from terminating a foster-care contract with Catholic Social Services because the agency declined to place children with same-sex couples.Footnote ⁶⁴ To reach that conclusion, the Court held that Philadelphia's nondiscrimination policy in its foster care contracts was not generally applicable, since it allowed a city official to grant exemptions “in his/her sole discretion.”Footnote ⁶⁵ Citing Sherbert, the Court said that this “formal system of entirely discretionary exemptions” triggered heightened scrutiny, regardless of whether any exemptions had actually been granted.Footnote ⁶⁶

Fulton focused on the discretion that Philadelphia's law delegated to enforcement officials. But in fact, this isn't the only lesson to draw from Sherbert and the other unemployment cases. As the Court has repeatedly suggested, those cases also contain a second lesson—this one involving not unfettered discretion, but instead something else that's just as common. To see it, consider just a few other precedents.

In Church of the Lukumi Babalu Aye, Inc. v. City of Hialeah, the Court examined a series of city ordinances that, each in different ways, banned animal sacrifice—the primary ritual of a religion called Santeria.Footnote ⁶⁷ The Court enjoined enforcement of the ordinances, in part because they seemed designed to suppress Santeria sacrifice.Footnote ⁶⁸ Yet in its discussion, the Court also identified an additional problem. Besides the ordinances mentioned above, the city had also adopted the state's generic animal cruelty statute, which imposed punishment for anyone who “unnecessarily . . . kills any animal.”Footnote ⁶⁹ Unlike the other ordinances, this law didn't mention “sacrifice” or contain a labyrinth of explicit qualifications for favored activities (though at argument, the city said it wouldn't apply the law to things like hunting or pest control).Footnote ⁷⁰ Unlike the other ordinances, nothing on the face of this law mentioned religion or religious practices specifically. Fittingly, the city contended that it was “the epitome of a neutral prohibition.”Footnote ⁷¹ Yet according to the Court, that was not so.

The Court began by noting that, because the law explicitly prohibited only “unnecessary” killings, it “require[d] an evaluation of the particular justification for the killing,” and thus created a “system of individualized governmental assessment” just like the law in Sherbert.Footnote ⁷² Moreover, the Court continued, by interpreting that law as applying to religious sacrifice, the government's “application of the . . . test of necessity devalue[d] religious reasons for killing, by judging them to be of lesser import than nonreligious reasons.”Footnote ⁷³ The law banning unnecessary animal killings didn't explicitly target religious sacrifice. And, at least arguably, it didn't delegate unbridled discretion to enforcement authorities the way a “good cause” exception does. But according to the Court, that was beside the point. By its terms, this law still required officials to examine the reasons behind each party's act of killing to determine whether such killing was “necessary.” And when applied against religion, a law employing that kind of subjective or value-based differentiation wasn't neutral or generally applicable. Rather, it “devalue[d]” religion by explicitly requiring officials to declare sacrifice “unnecessary” as a precondition to regulating it.Footnote ⁷⁴ Thus, the Court concluded, the application of this law also “singled out religious practice for discriminatory treatment” and required heightened scrutiny.Footnote ⁷⁵

At this point, it's important to note the difference between the Court's treatment of the Sherbert-style cases in Lukumi as compared to its original treatment of the issue in Smith. In Smith, the Court described Sherbert and its progeny as cases involving laws that explicitly granted officials broad discretion to give or deny exemptions.Footnote ⁷⁶ But in Lukumi, the Court focused on a different problem. Besides granting wide discretion, the law in Sherbert also explicitly differentiated conduct based on an assessment of the reasons for engaging in it: refusing to work on Saturday was permissible, but only if done for “good cause,” just as killing animals in Florida was permissible, but only if it was “necessary.”Footnote ⁷⁷ And according to the Court, that textual fact presented a separate, additional concern.

By employing categories like “necessary” or “good cause,” these laws unambiguously required government to “asses[s] . . . the reasons for the . . . conduct” in question.Footnote ⁷⁸ But even more, when used to punish religious conduct, they revealed that government had “devalue[d] religious reasons,” because they allowed punishment only after determining that religious motivations didn't qualify as “good cause” or “necessary.”Footnote ⁷⁹ By regulating an activity contingent on a negative judgment about the religious reasons for undertaking it, these laws discriminated against religion in a way the Free Exercise Clause forbids.

The Court's discussion in Lukumi identified a new aspect of Sherbert and the unemployment cases—this one focused on governmental restrictions that employ subjective or value-based standards as a condition of their application. And in fact, those aren't the only cases where this rule appears. The Court applied the same rule in Masterpiece Cakeshop v. Colorado Civil Rights Commission, albeit without fully explaining it.Footnote ⁸⁰

In Masterpiece, the Court held that Colorado violated the Free Exercise Clause by failing to extend the “neutrality that the Constitution requires” to a religious baker who refused to bake a cake for a same-sex wedding.Footnote ⁸¹ Commentators have tended to focus on Justice Kennedy's analysis of the statements of a few commissioners as the basis for the decision.Footnote ⁸² But the more consequential point is what followed.

Prior to the religious baker's case in Masterpiece, the state had excused several other bakers from charges of religious discrimination after they refused to bake cakes with anti-LGBTQ messages for religious customers.Footnote ⁸³ Yet when the religious baker in Masterpiece raised that point, the state court said those refusals were allowed because the secular bakers were simply refusing an “offensive . . . requested message.”Footnote ⁸⁴ According to the Supreme Court, however, that explanation was impermissible—the Constitution doesn't allow differential treatment of religion based on “the government's own assessment of offensiveness.”Footnote ⁸⁵

Here, the Court wasn't saying that Colorado itself had called the religious anti-LGBTQ messages “offensive,” as some interpreters have claimed.Footnote ⁸⁶ Rather, it was reiterating the same rule about value-based distinctions it already laid out in Lukumi and Sherbert. Where a state creates an exemption from its public accommodations law based on a subjective standard—whether a provider finds a request “offensive”—it may not refuse the same exemption to providers who claim offense based on their religion. Why not? Because doing so devalues religious reasons by “elevat[ing] one view of what is offensive over another”—it suggests that secular reasons for taking offense are worthy of solicitude, but religious reasons are not.Footnote ⁸⁷

Appreciating this aspect of cases like Lukumi, Sherbert, and Masterpiece adds two additional features to the concept of general applicability beyond those already discussed. First, it provides an added textual requirement. As shown above, the Court has clearly indicated that a law isn't generally applicable where it provides officials wide discretion to give or withhold exemptions. But on Lukumi's reading, Sherbert and similar cases also established an additional rule—and one that the Court again reiterated in Masterpiece. A law is not generally applicable where it explicitly employs subjective or value-based categories as a basis for differential treatment while also punishing religion. Such laws do more than impose an “incidental” burden on religious practice.Footnote ⁸⁸ Rather, they explicitly require officials to consider the religious aspect of an activity and render a negative judgment about it as a precondition for regulation.Footnote ⁸⁹ Applied to religion, these laws are less than generally applicable, because their structure clearly reveals government has taken religion into account and “devalue[d]” it compared to activities it considers truly worthy of exemption.Footnote ⁹⁰

But there's also a second implication—this one just as important. As has already been mentioned, under the general applicability requirement, courts are often required to ask whether a proposed religious exemption would “endange[r] [the government's] interest in a similar or greater degree” when compared to other conduct eligible for exemption.Footnote ⁹¹ The point of that analysis is to smoke out instances in which government actually has devalued religion, even if the text of a law doesn't make it obvious. But notice that, under Lukumi's treatment of cases like Sherbert and also under Masterpiece, that inquiry into comparative harm isn't always necessary.

Where a law has been applied against religious conduct while explicitly indicating it only covers acts lacking “good cause,” killings that are “unnecessary,” and so on, there's no need to ask whether the government has implicitly considered religion. On the contrary, these laws explicitly require officials to render a negative judgment about the religious aspect of the conduct if the law is to apply at all. Under these laws, a religious claimant's benefits can be refused only after determining that Sabbath observance isn't “good cause,” and her animal slaughter can be regulated only after concluding religious sacrifice isn't “necessary.” And the same is true of a judge-made rule exempting secular bakers from fulfilling requests they deem “offensive” while refusing to exempt religious bakers who raise the same objection.Footnote ⁹² In these instances, there's no need to ask about comparative harm to make sure government isn't considering or devaluing religion. The law itself has already told us the opposite is true.

Commentators and courts have sometimes read Sherbert and related cases to be focused on laws explicitly delegating wide discretion. But in fact, those cases aren't so limited. Where a law or policy explicitly requires considering the reasons for conduct while also punishing religion, it fails the general applicability requirement. That kind of law isn't regulating without regard to religion. Instead, it has explicitly made religion part of the calculus and found it wanting. And understanding that point provides a new vantage point for thinking about Diocese of Brooklyn and several of the other COVID-19 church-closure cases.

General Applicability and “Essential” Activities

In Diocese of Brooklyn and the other church-closure cases, the Court focused almost exclusively on the aspect of general applicability involving comparisons—and more specifically, on comparisons between things like worship services and grocery stores. But in fact, that wasn't the only route available. Under a more complete understanding of general applicability, many of the church-closure cases could have pursued another avenue, and one that probably should have been relatively uncontroversial.

Begin with Diocese of Brooklyn, the case involving New York's color-coded regulations that marked the shift in the Court's approach.Footnote ⁹³ From the beginning of the pandemic, New York's COVID-19–related rules imposed varying capacity limits on a variety of entities, but provided special privileges to any “essential business or entity providing essential services.”Footnote ⁹⁴ State guidance defined an “essential business” as any that “provid[ed] products or services that are required to maintain the health, welfare, and safety of the citizens of New York State.”Footnote ⁹⁵ Under that definition, the state had predictably privileged entities like grocery stores and hospitals. But beginning early in the pandemic and continuing into the ensuing months, it also included many other entities—factories, airports, banks, bike shops, liquor stores, acupuncturists, cannabis dispensaries, and hardware stores to name just a few.Footnote ⁹⁶ Thus, as a matter of law, New York had determined that—in each of those instances—the entities in question provided services “required to maintain the health, welfare, and safety of [its] citizens.”Footnote ⁹⁷

The state's early executive orders did not indicate definitively whether religious institutions might qualify as “essential.”Footnote ⁹⁸ Eventually, though, the state declared that all worship services were “non-essential gatherings,” and that “individuals should not gather in houses of worship, homes, or other locations for religious services until the end of th[e] public health emergency.”Footnote ⁹⁹ And that basic distinction continued to operate behind the state's stringent numerical caps on worship in New York City's color-coded zones.Footnote ¹⁰⁰

Like the laws in Lukumi, Sherbert, and Masterpiece, New York's COVID-19 mitigation regime explicitly conditioned regulation on a subjective, value-based appraisal. Where a business or gathering was “essential,” it was subject to fewer if any restrictions; where it was “nonessential,” it was not. Likewise, as in these cases, there's no question that New York chose to regulate religious worship only after determining that gathering for religious reasons was not “required to maintain the health, welfare, and safety” of its citizens, regardless of the precautions houses of worship actually took.Footnote ¹⁰¹

Stating the matter in the Court's preferred terminology, then, New York's regime depended on an “individualized . . . assessment of the reasons” for various gatherings.Footnote ¹⁰² But even more than this, by designating worship as “nonessential,” New York's regulations “devalue[d] religious reasons for [gathering] by judging them to be of lesser import than nonreligious reasons” the state deemed truly “essential.”Footnote ¹⁰³ Parallel to Sherbert, Lukumi, and Masterpiece, the regulations explicitly required officials to consider the value of religious gatherings as a precondition for regulation—officials limited worship only after evaluating religious reasons for meeting and finding them wanting. And where religion is regulated under a regime like that, the First Amendment requires heightened scrutiny, irrespective of additional arguments about the similarities or differences between churches and hardware stores.

This aspect of general applicability—and specifically, the rule about regulations employing subjective or value-based categories—provides an explanation for the Court's conclusion in Diocese of Brooklyn that doesn't depend on more controversial claims about comparative harm. And, in fact, this rule also sheds light on other cases too.

Consider Calvary Chapel v. Sisolak, a challenge to Nevada's restrictions on worship that predated Diocese of Brooklyn.Footnote ¹⁰⁴ Early in the pandemic, Nevada's governor issued an executive order outlining a list of “Essential services and sectors,” which were allowed to remain open without occupancy caps, subject to social distancing requirements.Footnote ¹⁰⁵ As in New York, the list included a range of entities such as grocery stores, pharmacies, dry cleaners, and pet shops.Footnote ¹⁰⁶ This early order conspicuously declined to label houses of worship “essential.” Instead, it simply urged faith leaders to “find ways to tend to your congregation without bringing them together in person.”Footnote ¹⁰⁷ But in less than a month, the governor changed course, placing a hard ten-person cap on all religious gatherings—even “drive-in and pop-up services.”Footnote ¹⁰⁸

Eventually, Nevada adjusted its restrictions on worship to a hard fifty-person cap while choosing to treat retail stores and casinos more leniently.Footnote ¹⁰⁹ At the Supreme Court, the state defended that choice by relying on expert medical advice that worship services presented special risks.Footnote ¹¹⁰ But like New York's restrictions on worship, Nevada's restrictions were also arguably overbroad—they applied the same way regardless of the size of the church, a congregation's specific sanitation procedures, or adjustments to worship services that reduced risk.Footnote ¹¹¹

Here again, the Court divided on whether churches and other religious institutions were more likely to spread germs than casinos.Footnote ¹¹² But regardless, Nevada's rules ought to have been subject to heightened scrutiny. By their terms, these restrictions provided special carve-outs for “essential services and sectors.” As a result, they explicitly required officials to conclude that houses of worship were “nonessential” as a condition for regulating them more stringently. And under Lukumi, Sherbert, and Masterpiece, a regime like that isn't generally applicable.

To be sure, there are closer cases. For example, what if a law technically places religion on the favored side of a subjective designation like “essential,” but without the benefits it provides to other entities in that class? That was the situation in California, whose COVID-19 mitigation restrictions were the subject of the initial challenge in South Bay and numerous other emergency appeals at the Supreme Court.

Early in the pandemic, California issued a directive identifying sixteen “critical infrastructures” that were exempt from stay-at-home requirements and other restrictions.Footnote ¹¹³ Notably, these early regulations also classified worship services as “critical.”Footnote ¹¹⁴ But in fact, they didn't actually exempt worship from any restrictions, unlike all other “critical” activities. Instead, worship services could continue through “streaming or other technologies” only, while other beneficiaries in the “critical” category—film studios, cannabis dispensaries, and laundromats, to name a few—faced no such restrictions.Footnote ¹¹⁵ By the time the Court considered South Bay, the state had recategorized churches as “high-risk workplaces,” but it continued to provide special privileges to all other entities originally labeled “critical infrastructure.”Footnote ¹¹⁶

My view is that strict scrutiny should apply in this instance too, though I admit it presents a closer question. California technically placed houses of worship in the “critical infrastructure” category. But it didn't actually allow churches or mosques to meet the way movie studios could. And unless general applicability is just an empty formalism, it's hard to see why government should be able to avoid the requirement so easily. If the state in Sherbert had formally declared that Adell Sherbert's Sabbath observance qualified as “good cause” but still refused to pay her, it seems obvious that strict scrutiny still ought to apply. The same would seem to be true in these circumstances too.Footnote ¹¹⁷

At this point, it's important to make two final observations before turning to some larger issues. The first involves a basic question. In pointing to the rule against subjective or value-based distinctions as a way of analyzing the COVID-19 church-closure cases, much of what I've said turns on the fact that these regulations explicitly differentiated among gatherings using terms like “essential” or “critical.” But if that's true, couldn't New York or other states have dealt with the problem by just deleting the troublesome words while leaving the rest of their restrictions in place? And if so, isn't this theory just an empty solution?

I think it's fairly well settled that the answer to that question is no. The reason has to do with the relationship between neutrality and general applicability. Although these are separate requirements, they are mutually reinforcing. As a result, the Court has said that, when considering whether a law complies with the Free Exercise Clause, courts should consider the circumstances surrounding a law, including “the historical background of the decision” and the “specific series of events” leading up to it.Footnote ¹¹⁸ Under that rule, it seems clear that merely deleting words like “essential” or “critical” while leaving everything else in place wouldn't be enough to rebut the conclusion that a state's regulations still depended on a negative assessment of religion's value. Moreover, even assuming that a state chose to make larger changes, those changes must still satisfy the strand of general applicability focused on comparability—now informed by an earlier strict scrutiny analysis. Simply removing the troublesome words isn't enough to satisfy the Constitution, especially when the inference of a negative value judgment remains obvious.

A second observation involves the limits of my argument. Under my reading of the general applicability requirement, many of the COVID-19–related regulations considered at the Court (and many others around the country) presumptively violated the First Amendment by requiring officials to declare worship “nonessential” as a condition for regulating it. But that doesn't necessarily mean the Constitution requires treating churches or synagogues the same as grocery stores or hospitals. Rather, it means the government must produce evidence that it has a compelling interest in regulating a particular house of worship more stringently and that there really are no less restrictive means of protecting that interest. Saying that regulations may not discriminate against religion by drawing value-based distinctions doesn't prevent courts from ultimately concluding that some restrictions should be lifted while others can be maintained, as the Supreme Court sometimes did.Footnote ¹¹⁹ Nor does it mean the Court was necessarily wrong in early cases like South Bay, when little was known about the virus or the effectiveness of strategies to combat it. The point is that regimes regulating religion based on subjective or value-based categories trigger strict scrutiny, not that they trigger automatic exemptions.

In resolving the COVID-19 church-closure cases, the Court focused on comparability. But, in fact, the cases also might have been resolved with reference to the aspect of general applicability focused on subjective or value-based categories.

“Adequate Alternatives”

As I've argued above, a correct understanding of general applicability indicates that COVID-19–related restrictions that explicitly required officials to designate worship as “nonessential” likely violated the Free Exercise Clause. But one might respond—as some commentators and judges did—by simply defending the designation on the merits. On this view, acupuncture parlors and cannabis dispensaries actually are more essential than worship services, because “there may be no other way to procure an exempted essential service but to physically go somewhere in person, while alternatives exist for religious services.”Footnote ¹²⁰ Probably the most notable defender of this view, Judge Frank Easterbrook, put it memorably: “Feeding the body requires teams of people to work together in physical spaces, but churches can feed the spirit in other ways.”Footnote ¹²¹

Of all possible responses, this one is the least persuasive. From the point of view of religious claimants in these cases, online worship services were not an adequate alternative to in-person worship; that's the whole point. And very often, that's so for quite understandable reasons. For the Catholic claimants in Diocese of Brooklyn, for instance, the most important part of a worship service is receiving bread and wine consecrated by a priest—an act that can't be done remotely, and one that is impossible to accomplish through delivery where the number of parishioners runs in the hundreds or thousands. And the same is true of countless other religious practices, which depend upon rituals or communal activities that simply can't be replicated at a distance.Footnote ¹²²

But there's also a more fundamental point here—this one going to the heart of the Free Exercise Clause. It's true that some religious groups might consider remote worship an adequate substitute in some instances. But it's important to understand that the argument about adequate alternatives doesn't end there. Rather, the claim that “churches can feed the spirit in other ways” implies a power on the part of government to decide when “other ways” are good enough.

The logic of that position is odious to the First Amendment. As James Madison observed long ago, government is not “a competent Judge of Religious Truth.”Footnote ¹²³ As a result, it possesses no authority to tell religious believers which aspects of their worship are truly essential and which are not. If the government chose to ban religious circumcisions on the grounds that they were not “essential,” would anyone claim it could justify that law by simply asserting that, in fact, Jews have been wrong for two thousand years and less gruesome alternatives will suffice? In my view, the answer is self-evident, such that no more needs to be said. Government does not get to tell believers which parts of their worship are expendable and which are not. And that bedrock principle should be enough to put this argument to rest.

Individualized Assessments

A second possible response to my proposal would be to accept that many COVID-19 mitigation regulations likely violated the rule against value-based distinctions, but to argue that the rule itself is incoherent. The reason, the argument goes, is that all laws depend to some extent “individualized assessments” about the conduct in question, including value judgments about it. On this view, the individualized assessment doctrine—including the rule against value-based distinctions—simply collapses on itself.

Unlike the objection about adequate alternatives, this objection has some plausibility. After all, it seems clear that Justice Scalia recharacterized Sherbert and similar cases as involving “individualized assessments” to revamp the Court's free exercise doctrine in Smith without overruling those cases.Footnote ¹²⁴ Later, the Court reaffirmed the individualized assessment doctrine in Lukumi and Masterpiece Cakeshop, and extended it to include not just laws conferring unfettered discretion, but also laws employing subjective or value-based criteria.Footnote ¹²⁵ But given its dubious origins, one might question whether the doctrine itself actually makes sense.

In fact, though, Justice Scalia's instincts were better than he knew. Although likely invented as a device to distinguish Sherbert and similar cases, the rule against individualized assessments provides a sensible way to guard against the concerns over religious discrimination that animate the Court's post-Smith jurisprudence.

Begin with the original version of the rule as articulated in Smith, and recently reaffirmed in Fulton. In those cases, the Court said that laws containing a “formal system of entirely discretionary exemptions” weren't generally applicable, because they “invite” officials to grant or withhold exemptions based on impermissible reasons.Footnote ¹²⁶ In Fulton, the Court didn't explain the logic of that rule. But as others have observed, it mirrors rules governing licensing regimes in the free speech context, and is justified for many of the same reasons—where discretion is at a maximum, “the risk of bias is sufficiently high, and the effectiveness of after-the-fact review sufficiently low . . . that a prophylactic rule is necessary.”Footnote ¹²⁷ Where the law explicitly grants low-level officials broad discretion to discriminate, courts should closely scrutinize whether they are using that discretion in a way that actually conforms to the Constitution.

But what about the expansion of the rule in cases like Lukumi and Masterpiece? Even if the basic rule about individualized assessments is sound, does it really make sense to expand it to cover any instance where the application of a law is explicitly conditioned on a subjective standard? After all, aren't all laws that draw distinctions dependent on value judgments to a greater or lesser extent?

This objection fails because it's built on an exaggeration. To see why, an example is helpful. Imagine that a jurisdiction has a law requiring all vehicles to stop at red lights but exempting emergency vehicles. In drafting that law, legislators undoubtedly made a value judgment that saving citizens from serious injury or death was important enough to justify an exception to its traffic laws. But nothing on the face of that law suggests that religion was ever considered. Nor does applying that law to a religious driver—say, a worshiper late for church—require any such consideration. Unlike a law allowing people to ignore red lights for “good cause,” nothing in this law requires officials to consider a driver's specific reasons for failing to stop. Instead, it merely requires a binary judgment: is this car an emergency vehicle or not? A law like that might imply a value judgment in some general sense. But nothing in its text implies a value judgment about religion that signals a need for heightened scrutiny under the individualized-assessment framework that Lukumi and other cases set out.

To be sure, though, maybe this isn't the end of the matter. Critics of the individualized assessment doctrine have sometimes argued that the doctrine is incoherent because police officers and prosecutors frequently make case-by-case judgments when making arrests or charging crimes.Footnote ¹²⁸ Returning to our example, then, if applying the red-light law invariably involves discretion about whether to issue a ticket or bring a prosecution, doesn't this prove the rule against value-based assessments is fundamentally unworkable?

Here again the answer is no, but for a different reason. It's true that in our system, application of every law depends to some degree on discretionary choices by executive officials. But equating prosecutorial discretion with laws explicitly requiring individualized assessments elides a key distinction. As written, most civil and criminal laws don't authorize officials to discriminate on forbidden bases like race or religion. Nor could they, because doing so is almost always unlawful. Instead, these laws—like the red-light example above—employ objective, nondiscriminatory criteria, and the law presumes prosecutors will enforce them in a nondiscriminatory way unless a claimant proves otherwise.Footnote ¹²⁹ But where a law explicitly confers significant discretion or employs value-based distinctions, the situation is different. Unlike other laws, these laws on their face invite officials to consider the religious reasons for conduct when applying or enforcing the law against a religious actor. And as a result, the Court's free exercise jurisprudence demands that such laws be scrutinized more closely.

The rule against individualized assessments probably originated as a convenient way to distinguish cases the Court had to explain in Smith. But it is also a sensible way to police impermissible discrimination against religion, especially where laws explicitly depend on making a negative value judgment about religion as a condition of their enforcement.

Discrimination

This last point leads to a third objection, this one probably the most significant. So far, I've argued that a law isn't generally applicable where it both regulates religion and explicitly conditions its application on a subjective standard—whether conduct amounts to “good cause,” whether worship is “essential” and so on. But admittedly, this might differ from how the law conceptualizes discrimination in some other contexts—those involving free speech, for instance, and possibly discrimination on the basis of race. And if those other areas wouldn't recognize discrimination in these instances, why would the Free Exercise Clause be different?

To understand this objection, it helps to focus on a few hypotheticals. Imagine that a state designates a minority-owned movie theater as “nonessential” alongside others and thus subject to more severe restrictions under its COVID-19 mitigation regime than businesses in the “essential” category. Under current equal protection doctrine, the owner has a valid discrimination claim if she can show that the state shuttered her business “at least in part ‘because of,’ not merely ‘in spite of’” race.Footnote ¹³⁰ Yet it seems unlikely that a court would conclude that burden had been met solely based on the “nonessential” designation. And if that's true in this context, why shouldn't it also be true as to free exercise?

Or consider another example, this one involving free speech. Imagine an activist is prosecuted under a law banning the destruction of draft cards after burning his card during a protest. Under current free speech law, the fact that he burned his card for expressive reasons isn't a basis for applying strict scrutiny, because both the law and its application can be justified by “an interest . . . unconnected to expression” like protecting the integrity of the draft system.Footnote ¹³¹ And very likely, that same conclusion would follow even if the law contained an exemption for “essential” acts of destruction. The mere fact that applying the law requires designating protest burnings as “nonessential” isn't enough to prove government has targeted speech, because the same speech-neutral justifications still exist. And if that's true in the free speech context, why would religion be any different?

These are powerful arguments, at least on the surface. But they also suffer from a common flaw. They fail to appreciate why these areas of law analyze government motive the way they do, and consequently how the Free Exercise Clause differs. Although the Court's post-Smith jurisprudence has drawn on speech and equal protection doctrine in forming its requirements, the Free Exercise Clause is based in concerns separate from both of those areas. And in a context like the COVID-19 church-closure cases, recognizing those unique concerns is the key to understanding the kind of discrimination the Free Exercise Clause prohibits.

Begin with the case about the minority business owner whose movie theater is declared “nonessential” along with other theaters. Under current equal protection law, the owner has a discrimination claim if she can show the state regulated her business “at least in part ‘because of’” race.Footnote ¹³² Yet it seems unlikely that a court would draw that conclusion based on the “nonessential” designation alone. Why would religion be different?

Here, the answer has less to do with the Constitution than it does with a difference between the two cases. In drawing the parallel between the real church-closure cases and the hypothetical case about race, the argument assumes that labeling a minority-owned movie theater “nonessential” tells us the same thing about government purpose as labeling religious worship that way. But in fact, it does not. In the real church-closure cases, there's no question that states like New York took religion into account, because the structure of the regulations ensured that worship could be regulated only after officials determined that gathering for religious reasons was not “essential.” By contrast, classifying a movie theater as “nonessential” doesn't give rise to an inference of discrimination (at least without more), because owning movie theaters is not, in itself, a race-specific endeavor. Whereas “worship” is an activity fraught with religious significance, owning movie theaters is an activity that people of all races pursue. In these two cases, regulating the relevant activities with an “essential” designation just works differently.

There's also an additional point here. In the equal protection context, a plaintiff can make out a prima facie claim of discrimination by showing that a discriminatory purpose—that is, a purpose based on race or religion—was “a motivating factor” in the decision.Footnote ¹³³ Yet under the Equal Protection Clause, the Court has also mandated a burden-shifting framework, in which the government can prevail if it proves that “the same decision would have resulted even had the impermissible purpose not been considered.”Footnote ¹³⁴ In the church-closure cases, state governments frequently pointed to the special risks posed by churches and worship services as the reason for differential treatment. And if those kinds of arguments would be sufficient to rebut a charge of discrimination in the race context, why wouldn't they be sufficient as to religion?

Here again, the answer is more straightforward than it first appears. In the church-closure cases as in many racial discrimination cases, the government had a possible alternative explanation for differential treatment—worship services are riskier than other activities. Recall, though, that in the actual cases, those arguments depended almost entirely on broad generalizations about what “usually happens” at a worship service.Footnote ¹³⁵ Yet under the Court's equal protection doctrine, those kinds of broad generalizations aren't enough. The Constitution doesn't allow “even a true generalization about [a] class [to] justify class-based treatment.”Footnote ¹³⁶ Instead, government must produce evidence that the differential treatment was justified as to the specific party claiming discrimination. And in applying strict scrutiny in cases like Diocese of Brooklyn, that's exactly what the Court was doing—asking for religion-neutral reasons based on actual evidence about the parties rather than broad generalizations about “worship services” writ large. The Equal Protection Clause and the Free Exercise Clause are different. But as to the COVID-19 church-closure cases, at least, they don't actually point in different directions.

But what about free speech? As noted above, under the Court's current doctrine, a protestor can be prosecuted for burning his draft card under a law generally forbidding a card's destruction, and the same would likely be true even if the law required an official judgment that protest-motivated burnings are not “essential.” The reason, as I've already observed, is that there's also a sufficient non-speech–related justification available. And where that's so, the Court's expressive conduct doctrine says strict scrutiny doesn't apply. But that more deferential approach to government motive isn't just a given. Rather, it's justified by the specific concerns underlying the Free Speech Clause.

At its heart, the Constitution's protection of free speech aims to ensure that discourse on public issues remains “uninhibited, robust, and wide open” by limiting government's power to censor ideas or drive them from the marketplace.Footnote ¹³⁷ Yet in the context of laws regulating conduct, that danger is largely absent. In all but the rarest cases, regulating conduct that is also expressive doesn't actually prevent an idea from being expressed. Instead, it simply drives it to another medium: the spoken or written word. And to ensure that's true, even content-neutral regulations of speech must leave open “ample alternative channels.”Footnote ¹³⁸ To be sure, there may be instances where the message carried by some act is so evocative or powerful it can't be transmuted into speech. But those are precisely the cases in which government will be most tempted to regulate conduct based on its expressive character—and thus, the cases that the Court's more limited motive inquiry actually takes into account. The Court's expressive conduct jurisprudence employs a deferential motive analysis because it assumes that, in all but the rarest instances, regulations of conduct alone do not (and really cannot) significantly curtail speech.

The Free Exercise Clause is different, in at least two important respects. First, unlike the Free Speech Clause, it explicitly protects conduct—specifically, religious “exercise.”Footnote ¹³⁹ But just as importantly, it protects conduct that almost always does not have a readily identifiable substitute. If a police officer grows a beard to protest what he sees as the department's draconian grooming policy, enforcing the no-beard rule doesn't stop him from spreading his message in other ways. But if a police officer's religion demands he grow a beard and the department forbids it, there's no alternative means of fulfilling his duty to God—he can either obey the religious command and lose his job, or betray his faith.Footnote ¹⁴⁰

Of course, the mere fact that the policy puts the officer to that choice probably isn't enough to create a free exercise violation—that's the point of Smith. But recognizing that regulations of conduct can have a much more destructive effect on religion than they do on speech provides a reason to treat discrimination in free exercise cases differently, and with less deference to the government. And indeed, at least if the Court's discussion of the issue in Lukumi and elsewhere is our guide, there's a good argument that this more searching inquiry is the one the Court has already adopted.

Consider first Lukumi's treatment of Sherbert. When discussing the “individualized assessment” in Sherbert, the Court didn't ask whether some other reason was sufficient to explain the denial of benefits in that case, though it certainly could have based on its facts.Footnote ¹⁴¹ Rather, the Court was content to conclude that laws like the one in Sherbert triggered strict scrutiny because they explicitly instructed officials to “asses[s] the reasons for the relevant conduct” and allowed them to deny benefits after concluding religion didn't qualify as “good cause,” whatever other bases for denial might have also been present.Footnote ¹⁴²

The same thing is true of the other unemployment cases. In Hobbie v. Unemployment Appeals Commission—one of the cases in Sherbert's line—the Court considered a free exercise claim by a religious worker who had been fired from her job after converting to Seventh Day Adventism and informing her employer she could no longer work on Saturday.Footnote ¹⁴³ The state denied unemployment benefits, arguing that because Paula Hobbie changed her religion after taking the job, her refusal to work on Saturdays constituted “misconduct” that made her ineligible.Footnote ¹⁴⁴

There was no question that enforcing the rule against Hobbie could have been justified on a variety of religion-neutral grounds—protecting her employer's reliance interests, for instance, or protecting the unemployment system from the financial burden created by after-the-fact refusals.Footnote ¹⁴⁵ Nonetheless, Justice Brennan and seven other justices cast those concerns aside. Instead, they reasoned that a scheme which “labels and penalizes behavior dictated by religious belief as intentional misconduct . . . exhibits greater hostility toward religion than one deeming such resignations to be ‘without good cause.’”Footnote ¹⁴⁶ The same would seem to be true of a regime that explicitly labels worship “nonessential” and subjects it to harsher regulation, even if some other religion-neutral justification might also be available. Under these cases, the question is whether government took account of an activity's religious aspect when deciding whether to regulate it, not whether regulating religion is the government's only purpose. Free exercise and expressive conduct simply work differently, and with good reason.

In sum, the conclusion that heightened scrutiny ought to have applied to COVID-19–related regulations labeling worship “nonessential” doesn't require special pleading about discrimination. Instead, it just requires an appreciation of the differences between various constitutional provisions where questions of discrimination are relevant. The Free Exercise Clause is unique, even if it shares certain doctrinal tools with other areas. And once that's clear, it provides even more reason to think that the Court was correct in Diocese of Brooklyn and similar cases, albeit for different reasons than the ones the Court chose to focus on.