Disability Law and the Case for Evidence-Based Triage in a Pandemic
abstract. This Essay explains why model policies proposed or adopted in response to the COVID-19 pandemic that allocate scarce medical resources by using medical evidence to pursue two core goals—saving more lives and saving more years of life—are compatible and consonant with disability law. Disability law, properly understood, permits considering medical evidence about patients’ probability of surviving treatment and the quantity of scarce treatments they will likely use. It also permits prioritizing health workers, and considering patients’ post-treatment life expectancy. These factors, when based on medical evidence and not inaccurate stereotypes, are legal to consider even if they disadvantage some patients with specific disabilities.
It then discusses the ethical advantages of triage policies that use medical evidence to save more lives and years of life, which I call “evidence-based triage,” focusing on the benefits of these policies for patients with disabilities. In doing so, I explain why recent critiques err by treating people with disabilities as a monolith, overlooking the political disadvantages of less-visible victims, and treating the social origins of scarcity as a justification for saving fewer lives. Evidence-based triage parallels other policy responses to the COVID-19 pandemic, like physical distancing and postponement of medical procedures, which may burden patients with specific disabilities or medical conditions but are nevertheless justified because they save more patients.
The rapid spread of COVID-19 in the United States, fueled by a haphazard federal response, has led to a scarcity of potentially lifesaving treatments for COVID-19 patients. This scarcity has prompted states to develop or revise plans for fairly distributing access to these treatments. And it has prompted articles in high-profile medical journals delineating principles for fair allocation of ventilators, intensive-care beds, therapeutics, and other scarce treatments,1 and the development of model triage guidelines for hospitals.2 These proposals agree on the importance of saving more lives, saving more years of life, and not using quality-of-life judgments.3 They aim to save more lives by prioritizing frontline health workers who can then return to helping others, and by prioritizing patients who can benefit most from treatment—patients who are neither so healthy that they will likely recover without the scarce treatment, nor so ill that they are unlikely to recover even with it.4
The proposed adoption of these and other triage guidelines has prompted criticism.5 Some critics focus on state guidelines that use “quality of life” judgments, a criticism I endorse.6 But many also reject methods of triage that aim only to save more lives and years of life. Critics of triage instead urge alternative solutions such as:
2. Minimal triage: Considering only whether a patient can benefit, irrespective of likelihood or magnitude of benefit, or the likely quantity of resources required for benefit.9
This Essay argues that triage guidelines that use medical evidence about patients’ prospect of benefit to pursue the twin aims of saving more lives and more years of life (which I call “evidence-based triage”) are legally and ethically preferable to random selection or minimal triage.10 I argue that evidence-based triage is fully consistent with recognizing the legal and moral equality of each person and emphasizing our duties to the most vulnerable.
In Part I, I explain why the two core goals of evidence-based triage—saving more lives and saving more years of life—are compatible, and even consonant, with disability law. In Part II, I discuss normative arguments for evidence-based triage, highlighting the argument that evidence-based triage will benefit many patients with disabilities, and is even likely to save more patients with disabilities than random selection or minimal triage would. While random selection or minimal triage have the surface appearance of equity, they prevent individualized assessment of each patient’s medical situation, leading to worse outcomes for many patients with disabilities and ignoring relevant differences between disabilities. Although I present these Parts together, the doctrinal points in Part I do not require the correctness of the normative claims in Part II, nor do those normative claims depend on the persistence of the legal precedents Part I describes.
In this Part, I defend the legality of evidence-based triage, focusing first on saving more lives and then on saving more years of life. As an initial matter, recent Department of Health and Human Services (HHS) guidance does not preclude evidence-based triage. Rather, it distinguishes relevant and irrelevant considerations, stating that “persons with disabilities should not be denied medical care on the basis of stereotypes, assessments of quality of life, or judgments about a person’s relative ‘worth’ based on the presence or absence of disabilities,” and that decisions “concerning whether an individual is a candidate for treatment should be based on an individualized assessment of the patient based on the best available objective medical evidence.”11 Stereotypes and judgments of relative worth not only risk unfairness to patients with disabilities—they hinder efforts to save more lives and years of life by introducing irrelevant considerations. I likewise agree with HHS and with disability advocates that quality-of-life judgments are likely to incorporate unjust biases that preclude their use in pandemic triage.12
In contrast, the aim of saving more lives is widely championed in COVID-19 response both inside and outside triage contexts.13 Wide cross sections of both laypeople and theorists endorse it.14 Both this aim and that of saving more years of life can be pursued without reliance on unfair classifications; pursuing these goals only requires recognizing the equal value of all patients’ lives, and of every year within those lives.15 Notably, the HHS Office of Civil Rights closed a civil-rights complaint against the Pennsylvania Department of Health as satisfactorily resolved after Pennsylvania removed language that singled out specific disabilities and added language about individualized assessments; the revised guidelines continue to incorporate assessments of patients’ prospect of benefit with the aim of saving more lives and years of life.16
Although there is little precedent applying disability law to the allocation of scarce, lifesaving treatments, organ allocation is the closest analogy. Organ allocation policies align more closely with evidence-based triage than do random selection or minimal triage. Most notably, lung allocation policy considers candidates’ probability of post-transplant survival.17 Kidney allocation policy also considers survival, and allocation policies for all organs consider medical factors that affect survival, such as immunological matching.18 Legal precedent also recognizes that probability of survival is a legitimate basis for organ allocation.19 These precedents support applying a similar approach to allocating scarce COVID-19 treatments.
Cases applying the Americans with Disabilities Act (ADA) to medical decisions made in nonscarce contexts have often required individualized determinations,20 though distinguished scholars recognize that disability law “does not always require an individualized inquiry.”21 In a pandemic, the time needed to make individualized judgments may itself be scarce,22 although HHS asserts that pandemic triage decisions must nevertheless be individualized.23
Even if we assume, with HHS, that individualized judgments are required in pandemic triage, this counts for—not against—evidence-based triage, which employs individualized determinations.24 In contrast, individualization arguably prohibits random selection, which ignores medical differences between individual patients.25 In practice, random selection is also susceptible to biased or arbitrary decisions, because it requires initial decisions regarding who is eligible for selection but offers no guidance in making these judgments.26 While triage policies should recognize that medical professionals may be biased against patients with disabilities,27 the solution is not to encourage unguided eligibility decisions under time pressure, which are more likely to be biased or arbitrary.28 The better option is to employ triage policies that constrain and guide decisions by requiring that they be justified using consistently applied medical criteria.29 Evidence-based triage can reduce arbitrariness by ensuring that triage decisionmakers only have access to medically relevant information and are not privy to medically irrelevant and potentially biasing details, like a patient’s name, gender, race, nationality, or disabilities that are irrelevant to COVID-19 treatment.30
In this Section, I argue that disability law permits triage approaches that use medical evidence, including evidence about disabilities that affect a patient’s chances of survival or the quantity of resources they will require, to pursue the goal of saving more patients. This approach diverges from the view that a patient’s disability may only be considered if they “will die in the immediate term from that disability with or without that treatment,” and that triage guidelines may not deny “treatments to individuals because of their disabilities, when those individuals can benefit from them,” irrespective of potentially large differences in probability of benefit.31 Medicine is permitted to save the most lives, even when other available options would not be futile.
Some patients may have a lower probability of survival than others due to specific disabilities. Before defending minimal triage, Samuel Bagenstos concedes that “an individual’s disability may make it impossible or impracticable for them to satisfy certain eligibility criteria, even if the criteria are defined with no reference to disability,”32 and illustrates this via the example of a blind bus driver. Because blindness completely prevents safe driving, not hiring blind drivers is legal. But the bus driver analogy also justifies considering disabilities that make safe driving less probable, such as color blindness or epilepsy.33 The same should be true for disabilities that affect survival but do not make treatment futile.
Bagenstos argues that because the goal of rationing treatment is not universally agreed upon, appropriate eligibility criteria for scarce treatments are difficult to discern.34 But saving lives is a core goal of medicine and of COVID-19 response in particular, unlike maximizing the “prospective economic output of those we save.”35 It is endorsed in case law discussing the distribution of other scarce treatments, most notably transplants.36 It requires no judgments about the value of life with a disability.37
It is plausible—though contestable—that decisions about which patients will receive scarce medical treatments are governed by disability discrimination law.38 But the cases finding that providers have illegally discriminated involve groundless judgments or invidious stereotypes, not factually grounded prediction. The Supreme Court’s ruling against a provider who refused to treat a patient with HIV rested on the provider’s failure to provide “objective, scientific information” that would make refusal “reasonable in light of the available medical evidence.”39 Lower court opinions likewise differentiate groundless or biased decisions from those grounded in legitimate, even if disputable, medical evidence.40 And courts resolve uncertainty in favor of medical decisionmakers who are making good-faith efforts to ground their decisions in evidence.41
Patients with certain disabilities may require more of a scarce treatment than others do. The American Medical Association recognizes this may be a legitimate medical consideration in absolute scarcity.42 Yet some recent disability advocacy rejects it.43
Several disability law precedents are consistent with considering the quantity of resources a patient is likely to require. The first is Alexander v. Choate, which permits the provision of an equal quantum of resources to all patients, even if this produces unequal outcomes for patients with specific disabilities.44 The second is Olmstead v. L.C., which concludes that limited resources can justify refusing to maximally accommodate people with specific disabilities when doing so would deprive others—including people with other disabilities—whom the state is also bound to protect.45 The third is precedent defining reasonable accommodation: disability law does not require “an accommodation that would result in other employees having to work harder or longer.”46 The fourth is the permission provided in disability law to exclude a person from a program in order to protect others’ health or lives.47 In explaining this doctrine, courts state that disability law aims to counter “prejudice, stereotypes, or unfounded fear,”48 while giving appropriate weight to genuine threats to others’ health and safety, and that “the need to protect public health may at times outweigh the rights of disabled individuals.”49
Bagenstos argues that providers should be required to “take steps to ensure that those who are not qualified for life-saving treatments can become qualified.”50 Patients with disabilities should of course receive nonscarce resources that help them become qualified. But whether they, or any patient, should receive scarce resources to become qualified requires an evidence-based determination of whether doing so risks depriving other patients, including patients with other disabilities, of needed resources.51
Evidence-based triage also typically prioritizes healthcare workers.52 Doing so helps save more patients, especially those whose illnesses or disabilities make them likelier to require the assistance of skilled professionals. Although random selection and minimal triage would prohibit this prioritization, such a prohibition lacks legal support. Prioritizing healthcare workers, even if they are less likely to be disabled than others,53 does not constitute illegal discrimination, because prioritizing healthcare workers is “necessary for the provision” of scarce treatments.54
The American Medical Association’s discussion of fairly allocating scarce resources encompasses duration of benefit as well as likelihood of benefit.55 Importantly, evidence-based predictions about the number of years a patient can gain from treatment fundamentally differ from subjective judgments about quality of life. A patient’s unique perspective on their own quality of life merits deference.56 But patients lack similar expertise regarding their future lifespan. Life-expectancy ranges can be predicted for patients with certain illnesses or disabilities,57 based on verifiable outcomes—how many patients with a given condition survive for a specified length of time—rather than on subjective, unverifiable beliefs.58 Short- and long-term life-expectancy predictions are central to damages calculations in tort cases.59 Shorter-term predictions are employed to determine eligibility for scarce antiviral treatments60 and hospice care,61 as well as to prioritize patients for access to experimental treatments and consultations about end-of-life treatment preferences.62 Even those who question life-expectancy predictions are uncertain whether to categorically reject them, or instead to limit their use to short-term life expectancy.63
Bagenstos argues that even if some disabilities genuinely limit life expectancy, it would be unfair to consider this when setting priorities.64 As an ethical matter, I agree that using shorter-term life-expectancy predictions, as both lung allocation and model triage guidelines do,65 is desirable because it captures the direct effects of conditions that limit short-term survival regardless of social arrangements, while striving to screen out the effects of social injustice on overall lifespan.66 Short-term predictions also reduce concerns about uncertainty.
As a doctrinal matter, however, the law permits consideration of disabilities that limit lifespan, even when disability would not limit lifespan in a just world. As described in Section I.B.1, the law permits evidence-based medical judgments even when they disadvantage patients with certain disabilities. For instance, courts were legally justified to base their decisions on the risk HIV presented under prevailing social conditions,67 even when doing so exacerbated the disadvantage or inconvenience people with HIV experienced and even though HIV would have been less deadly had society acted justly and pursued a cure earlier. Similarly, serious limitations on long-term survival were judged to be an acceptable reason to deny access to a liver transplant.68 Notably, the law also permits triage policies to consider patients’ age.69
In this Part, I turn from doctrinal to ethical analysis of evidence-based triage. As a starting point, the aim of saving more lives is widely endorsed.70 Notably, many ethicists who reject utilitarianism endorse saving more lives, on the basis that saving more lives comes closest to meaningfully fulfilling our duties to each person,71 and recognizes the paramount and equal significance of each life.72 While saving more years of life engenders more debate, some nonutilitarian ethicists also recognize it as appropriate.73 Diverse participants in recent community engagement studies of triage policies similarly endorsed considering probability of survival and—though less strongly—years of life saved.74 These arguments provide compelling ethical support for evidence-based triage, particularly when it emphasizes saving more lives and regards saving life-years as a subordinate and constrained aim.75 Indeed, those willingly accepting social and economic burdens in order to save more lives might reasonably feel betrayed if triage policies abandoned that goal.
In this Part, I advance a more ambitious ethical argument: evidence-based triage not only saves more lives overall, but it likely saves more lives among patients with disabilities.76 While this argument is not necessary for evidence-based triage to be ethical, it demonstrates its consonance with the goals of cross-disability advocacy, and helps dispel the mistaken framing of evidence-based triage as primarily burdening people with disabilities while primarily benefiting people without disabilities.
How evidence-based triage affects a given patient or group depends on two factors: (1) whether they are likely to need scarce, potentially lifesaving treatments, and (2) whether they are likely to benefit from those treatments. These factors create three broad categories of patients:
1. Patients unlikely to need scarce, potentially lifesaving treatment, who will be largely unaffected by triage policies.
2. Patients who are more likely to need these treatments, but have comparatively good prospects of benefit. They will likely fare better under evidence-based triage, which makes more treatments available to patients who can benefit, than under random selection or minimal triage.
3. Patients who are more likely to need these treatments, yet have poor prospects of benefit even with them. They will likely fare worse under evidence-based triage than under random selection or minimal triage, although the difference may not be large given their limited prospects of benefit even with treatment.
Because evidence-based triage aims to save more lives, the gains for patients in the second group are expected to exceed the losses for patients in the third group. Imagine a stylized triage scenario where the second group comprises twenty patients with a fifty percent chance of survival and the third comprises twenty with a ten percent chance; only twenty treatments are available. Evidence-based triage prioritizes the second group, saving ten lives. In contrast, random selection, which treats the groups identically, saves five fewer lives in the second group and only one more in the third.
In this Section, I suggest that there is reason to believe that the second group of patients, those who need scarce treatments but can benefit from them, is likely to contain a greater proportion of people with disabilities than the first group. This, in combination with the fact that evidence-based triage aims at saving more lives, makes it plausible that the people with disabilities who would be saved only by evidence-based triage outnumber the people with disabilities in the third group who would fare better under random selection or minimal triage. The plausibility of this claim is not required for evidence-based triage to be legal or ethical: law and morality value all patients, not only those with disabilities. Nor is it sufficient for legality: a policy could illegally disadvantage patients with specific disabilities even if it is better for most patients with disabilities.77 But it helps to support the consonance of evidence-based triage with a disability advocacy approach that centers the interests of patients with disabilities.
The predicted benefits of evidence-based triage for many patients with disabilities undermine the generalization that evidence-based triage disadvantages patients with disabilities.78 They also challenge Bagenstos’s suggestion that triage policies that consider disabilities “place the burden of resource scarcity on disabled individuals” and his claim that “a process in which people with disabilities were equally represented vis-à-vis the nondisabled, and in which the interests of both groups were given equal concern” would exclude disability as a factor.79 And they similarly challenge the claim that considering probability of survival or quantity of resources required would “significantly disadvantage people with disabilities.”80 Disability advocacy has compelling reasons to endorse evidence-based triage policies that consider differences in patients’ prospect of benefit and the quantity of resources they need, because many of the additional lives saved by such policies will be the lives of patients with disabilities, since the overwhelming majority of disabilities do not affect these factors. Taking “equal account of the interests of disabled people”81 does not mean treating all people with disabilities the same—it means treating people differently when, and only when, this is justified by their differences.
Bagenstos questions whether we can “know that a policy explicitly denying treatment to some individuals based on their disabilities is going to benefit more people with disabilities overall.”82 There is uncertainty here, but we certainly cannot know that random selection or minimal triage will more effectively avoid “deny[ing] people with disabilities important benefits” or “forcing disabled individuals to face deadly consequences because of societal decisions not to invest in sufficient treatments.”83 In fact, there are good reasons to believe that random selection and minimal triage present greater risk of generating these undesirable results. First, forcing hospitals to ignore evidence about prospect of benefit may not be the best way of allowing “people with disabilities an equal chance to survive,”84 because disabilities likely also affect the need for scarce, lifesaving treatment, making the number of treatments available relevant. For instance, given the spread of COVID-19 in group housing,85 patients with disabilities (such as sensory and intellectual disabilities) that do not limit their prospect of benefit are likely to be overrepresented among COVID-19 patients.86 They therefore have much to lose from a policy that makes fewer treatments available to patients who can benefit, since such a policy not only leads to more deaths but concentrates those deaths among those likelier to contract COVID-19.87
Second, evidence-based triage could benefit patients with disabilities even under the implausible assumption that disabilities only limit prospect of benefit and do not increase the risk of contracting COVID-19. Assume, for instance, that only five of the twenty COVID-19 patients with a greater prospect of benefit in the example discussed above have disabilities, whereas all twenty patients with a lesser prospect of benefit do.88 Even under these assumptions, evidence-based triage would be expected to save more people with disabilities.89 A greater but unequal chance of survival seems ethically preferable to a smaller but more equal chance.90
Last, doubts that an evidence-based triage policy will in fact save more patients with disabilities would not end the analysis. As noted above, evidence that evidence-based triage will benefit many patients with disabilities importantly counters narratives that cast evidence-based triage as benefiting only patients without disabilities while burdening all patients with disabilities. Further, if desired, an evidence-based triage policy could be made more favorable to patients with disabilities by adding allocation criteria, rather than subtracting allocation criteria by retreating to random selection or minimal triage. For instance, patients with disabilities who can benefit, or patients who are disadvantaged more broadly, could be explicitly prioritized.91 Such an approach could still aim to save more lives, while trying intentionally to ensure that people with disabilities comprise a fair number of those saved. Meanwhile, though it is important to aim for accuracy in evidentiary determinations, randomness is not the solution to inaccuracy. Just as the problem of bias in medical care justifies anti-bias efforts rather than abandoning the potentially lifesaving enterprise of diagnosis and treatment in favor of minimal or random approaches, the same is true regarding concerns about bias in allocation.92
The more challenging question is when, if ever, we should accept saving fewer lives overall in order to benefit patients with specific disabilities. While disability policy sometimes allows for accommodations that present few trade-offs between potential beneficiaries,93 such options are unlikely to be available where scarce medical resources are concerned. In the face of scarcity, allocating scarce resources–even resources that are scarce because of social policy determinations–requires a normative framework.94 A triage policy that is worse for individuals with specific disabilities cannot be dismissed as unjust discrimination on that basis, but must be analyzed against a broader backdrop of normative analysis that considers the claims of different individuals and social groups.95 Saving more lives overall is a compelling starting point. By aiming to save more lives, evidence-based triage aligns with widely adopted measures like physical distancing and postponement of medical procedures; it aims to save more patients, including many patients with disabilities, even if this works to the disadvantage of patients with certain disabilities.
In this Section, I challenge the claim that random selection serves the interests of politically marginalized people, including marginalized people with disabilities.96 While Bagenstos is right to demand that “decisions that deny people with disabilities important benefits”97 be democratically legitimate, the tragedy of scarcity is that every decision denies some people with disabilities important benefits. Patients whose disabilities substantially reduce their prospect of benefit can organize to offer personal narratives in opposition to evidence-based triage,98 but allocation approaches that advantage those patients may poorly serve the interests of patients with other disabilities. In contrast, the patients with and without disabilities who fare worse under random selection are unaware of their fate until the decision has been made and have no unified narrative to offer policymakers. Advocacy should not mandate solidarity between people whose disabilities do not limit their prospect of benefit and those whose disabilities do limit their prospect of benefit.99 Rather, each patient, whatever their specific disability, should be considered as an individual.
Recognizing the multiple identities of patients with and without disabilities challenges the view that democratic legitimacy favors minimal triage or random selection,100 as well as the related view that minimal triage or random selection best recognizes the equality of people with disabilities. While the passage of the disability discrimination laws did not categorically exclude the nondisabled, it likely did exclude or limit the participation of disadvantaged people with and without disabilities. We should worry that elite decisionmakers might be attracted to formally equal approaches—like random or first-come, first-served allocation—that save fewer lives and thereby likely increase deaths among disadvantaged people with and without disabilities, but that impose few or no burdens on better-off people who are less likely to need scarce, lifesaving treatments and are likelier to reach the hospital first if they do. Significantly, the community engagement research discussed earlier reveals broad rejection of both lottery and first-come, first-served allocation, and underscores concerns that the latter approach favors the better-off.101
Ultimately, rather than assisting those “most likely to have been excluded”102 from policymaking, random selection and minimal triage would increase risk for many disadvantaged people and people with disabilities, while protecting the interests of a small group that is better positioned to organize.103 Cross-disability advocacy should acknowledge, not downplay, this problem. While Bagenstos recognizes potential tradeoffs when he suggests that “[m]aybe denying lifesaving treatment to individuals with cystic fibrosis could save more individuals who take daily medication for high blood pressure,”104 this example goes astray in two ways. First, evidence-based triage involves individualized determinations, not categorical denial of treatment to all cystic fibrosis patients. Second, Bagenstos compares cystic fibrosis to high blood pressure, a common condition many would regard as less significant than cystic fibrosis. Here is a better example: in scarcity, denying lifesaving treatment to those patients whose cystic fibrosis makes them comparatively unlikely to benefit could save more patients with cystic fibrosis who are nevertheless likely to benefit. This outcome is achievable under evidence-based triage, which considers individualized evidence about benefit, but is foreclosed both by minimal triage—which mandates ignoring the difference between the possibility of benefit and its likelihood—and by random selection.
Bagenstos and others rightly observe that scarcity stems not just from COVID-19, but from social and political decisions to underinvest in testing and equipment.105 But the social origins of scarcity do not support random selection or minimal triage.106 Some suggest that adopting these approaches would encourage decisionmakers to more vigorously prevent initial scarcity.107 This assumption lacks evidentiary support. And it becomes implausible when opposition to triage policies involves inviting the same federal officials who inadequately managed the initial COVID-19 response108 to now control how others respond to the scarcity they produced.109
Concentrating power in the hands of those responsible for scarcity will likely undermine, not increase, incentives to prevent scarcity. Disability advocates and others would therefore be wise to reject, for instance, recent proposals to allow the federal government to deny ventilators from the Strategic National Stockpile to any state that the Secretaries of HHS and of Homeland Security judge to be allocating ventilators in a discriminatory fashion.110 Such an approach would afford federal administrators largely unguided power to deny states ventilators, a particularly unwise decision given existing concerns about politicized and otherwise unjustified federal allocation of scarce COVID-19 treatments.111 It would also be inconsistent with HHS’s recognition that state decisionmakers “have the greatest insight into community-level needs in the COVID-19 response.”112 And, even if it accurately identified discrimination, this approach would impose double jeopardy on residents of states with discriminatory policies by denying them access to ventilators. A preferable approach, which would avoid undesirable “federal superintendence of treatment decisions traditionally entrusted to state governance,”113 would permit states and localities to adopt triage policies that mitigate harms resulting from federally exacerbated shortages, while allowing courts to use the established tools of disability law to address any concerns that these policies unfairly discriminate on the basis of disability.
I have argued that random selection and minimal triage are not required by law, and would be worse than evidence-based triage for patients with and without disabilities. Instead of advocating for federal intervention to impose these approaches, disability advocates would do better to support the adoption of evidence-based state triage guidelines that are responsive to public input, and to ensure that these guidelines are free of biased or unsupported assumptions about disabled patients.114 They could also effectively marshal disability law in support of efforts to reduce scarcity and limit the spread of illness to vulnerable patients.115 Developing evidence-based triage guidelines, while working to reduce scarcity, is the most effective way to save people with and without disabilities from COVID-19.
Govind Persad is Assistant Professor of Law, University of Denver Sturm College of Law and Greenwall Foundation Faculty Scholar in Bioethics. JD, PhD, Stanford University. I owe thanks to David Wasserman, Mark Kelman, Wendy Salkin, Joe Millum, Joseph Stramondo, Daniel Goldberg, Deborah Hellman, Kate Nicholson, and Theresa Loveless for conversations that shaped this project, to Glenn Cohen, Alan Chen, Sharona Hoffman, Rabia Belt, Doron Dorfman, Jessica Roberts, Valerie Blake, and Leslie Francis for written comments, and to Liza Sawyer and Ellen Blatt for research assistance. None of them is responsible for the content of this piece or any mistakes herein. Thanks also to the Yale Law Journal Forum’s editors, especially Joseph Daval, for their excellent work.